VEHICLE TECHNOLOGY ENGINEER
Monash Has An Autonomous Drive
Monash into a new era: M21 is the Monash answer for future Formula competition Humans in the Loop: From bilateral to autonomous control Self driving trucks: Trucks may benefit more than cars Critical scenarios: For co-operative and automated vehicles
June 2021 Issue 28 Representing mobility engineers since 1927 www.saea.com.au
VTE | Contents
Contents June 2021
SAE-A AGM the first since 2019
5
Auto Innovation Centre SA with BusTech
10
Autonomous buzz for self-driving trucks
14
Humans always in the loop
16
M21 the start of a new era for Monash University
18
Special Features 14
SAE AGM – Autonomous buzz for self-driving trucks
16
SAE AGM – Humans always in the loop
18
Monash – M21 the start of a new era for Monash University
VTE News 7
General News
8
Automotive News
10
Truck & Bus News
12
Defence & Aero News
13
Overseas News
Society News 4
Notes from the Chair - Welcome from Adrian Feeney
5
SAE-A News
Technical Feature 23
Technical – Simulation-Based Identification of Critical Scenarios for Cooperative and Automated Vehicles
about the cover M21 the new Monash autonomous car for future Formula competition
About the SAE-A SAE-A was founded in 1927 to address the need for further education for all facets surrounding Automotive Engineering, and now encompasses all mobility engineering industries in the Australasian region. The SAE-A is a non-profit organisation that works to serve the needs of its members and to promote the relevance of mobility related technologies to governments, industry and the community in general.
The editor, publisher, printer, the Society of Automotive Engineers – Australasia (SAE-A) and their employees, directors, servants, agents and associated or related entities (Publishing Entities) are not responsible for the accuracy or correctness of the text, pictures or other material comprising the contributions and advertisements contained in this publication or for the consequences of any use made of the products, services and other information referred to in this publication. The Publishing Entities expressly disclaim all liability of whatsoever nature for any consequences arising from the use or reliance on material contained in this publication whether caused to a reader of this publication or otherwise. The views expressed in this publication do not necessarily reflect the views of the Publishing Entities. The responsibility for the accuracy or correctness of information and other material is that of the individual contributors and the Publishing Entities do not accept responsibility for the accuracy or correctness of information or other material supplied by others. To the extent permissible by law, the Publishing Entities exclude all liability pursuant to the Competition and Consumer Act 2010 (Cth) or other applicable laws arising from statute or common law. Readers should make their own inquiries prior to the use of, or reliance on, any information or other material contained in this publication, and where necessary seek professional advice. All rights reserved. Reproduction in whole or part without the written permission of SAE-A is strictly prohibited.
www.saea.com.au
VTE | 3
Introduction | Secretary, CEO and Chairman Society of Automotive Engineers
VTE Published By: Society of Automotive Engineers - Australasia ABN:
95 004 248 604
Address: PO Box 103, Werribee Vic 3030 Phone: 0403 267 166 Email: info@sae-a.com.au Web: www.saea.com.au
Adrian Feeney
Membership & Subscriptions
Secretary, Chair and CEO Society of Automotive Engineers – Australasia
Rose De Amicis Email: rose@sae-a.com.au Events Melanie Webster Email: events@sae-a.com.au
Board of Directors: Chairman & CEO Adrian Feeney Board Greg Shoemark Michael Waghorne Noelle Parlier Bernard Rolfe David Young Luke Callaway Samsone Lagozzino (Sam)
Magazine Production: Editor Mandy Parry-Jones Trading Terms Media Email: mandypj@optusnet.com.au Mobile: 0409 806 986 Design Brigid Fraser Email: fraseram@optusnet.com.au Mobile: 0413 009 122 Advertising Jill Johnson Jill Johnson Media Email: jj@jilljohnsonmedia.com.au Mobile: 0409 217 624
VTE Industry Partner: Excellerate Australia
4 | June 2021
Australia, are we serious about Electric Vehicles? I strongly recommend you read an article in this edition of VTE which was provided by RACV and explains the impact of the new Victorian tax on electric vehicles for vehicle owners and operators. The estimated annual cost for a vehicle driven the average 13,000km is less than $300, and it will be offset by a $3,000 rebate on new electric cars plus a $100 annual reduction in registration fees. What’s more, the Victorian Government is committed to boosting its own electric fleet, so it will be subject to these taxes and concessions itself. So, does it really matter? We say yes, and it raises some significant questions around Federal and State leaders’ attitude and commitment to moving away from the traditional fossil fuels vehicle to electric zero emissions vehicle (ZEV) technology. It is estimated that less than one percent of new vehicles currently sold in Australia are plug-in electric, which is well behind many other nations (Norway current has over 75 percent), so what can our leaders do and why? It is SAE-A’s view that like all other developed countries, Australia needs a clear policy and a plan to move to EVs and it starts with a few simple steps: •
Meaningful cash incentives to buy electric ZEVs, whether battery electric or hydrogen fuel cell electric.
•
Generous tax concessions such as reduced registration fees, and tax write-offs that are brought forward for companies, and no Victorian type of new ZEV road tax and tax write-offs that are brought forward for companies.
•
All Governments – Federal, State and local mandate to purchase only ZEVs for their fleets.
•
Set achievable targets for new car sales to be EVs in line with Federal Government emission reduction targets, similar to what other countries have mandated (including China, Japan and the UK).
•
Charging stations infrastructure be created to meet the new demands of the travelling public.
Those of us who grew up on petrol engines loved that technology, but the world has changed and now electric has become the new challenge, already we see new companies popping up in Australia to develop this technology, all we need is a market to match that capability. It’s worth noting that many global manufacturers are phasing out IC vehicles completely, many within 10 years, which means that Australia, as a 100 percent passenger car importer, will be left with obsolete petrol vehicles with no further development within that timeframe. SAE-A’s very own Emergency Services Vehicle project has demonstrated the feasibility of bringing automotive manufacturing back to Australia and that program is based on moving away from petrol to Zero Emissions Electric Vehicle technology. Imagine if as a nation we embraced plug-in electric and hydrogen fuel cell electric technology. Not only would we not fall behind the rest of the world, but we could lead and again have a strong and viable automotive manufacturing industry. All it takes is the willingness of our leaders to accept that challenge.
SAE | News
New Members SAE-A AGM the first since 2019 On Thursday 20 May the SAE-A held its first Annual General Meeting since 2019, no AGM was held in 2020 due to COVID-19 restrictions. This was fortuitous as not long after the AGM Victoria was thrown back into a lockdown situation.
The SAE-A would like to welcome the following new members: Individual members Heath Borissow (Rejoin) Larry Collisson Hung Dang Mohammad Fard Corporate Members BTT Engineering Hendrickson Asia Pacific Motor Traders Association NSW Transport Accident Commission The SAE-A is where members enjoy many benefits and become a part of the advancement of the mobility and engineering profession across Australasia through the transfer of technical knowledge and skills, and an increased industry network. Individual and corporate memberships are available. More information at www.saea.com.au/membership Honorary Membership
This year’s SAE-A AGM was held at the Mulgrave Country Club with the formal part of the meeting held in a theatrette style environment followed by drinks at a private bar and then dinner and presentations in a private dining room.
Lawrence Geyer Milestone Members SAE-A would like to congratulate the following members on reaching a milestone year and thank them for their significant and ongoing contributions to the mobility industry.
CEO and Chair Adrian Feeney opened the floor to the 2021 AGM and presented a snapshot of the past two years’ activities and challenges, not the least of which was COVID19 which did not permit the usual Formula SAE event to be held. Lateral thinking by the SAE allowed a scaled down event to be organised which permitted teams to continue to develop their FSAE entries and experience.
50 years Harry Watson Ronald Kerwood 40 years Duncan Gilmore Peter Jaensch William Malkoutzis Trevor Parkes Geoff Senz Raymond Strong William Swinton
This year the event will return to greater normality for local teams, but due to travel restrictions it is unlikely overseas teams will be able to attend with the possible exception of teams from New Zealand. Despite the challenges of 2020, the SAE has prospered and continues to flourish; this was the sentiment expressed by SAE-A treasurer Michael Waghorne who presented the SAE’s financial report. This financial success should be further strengthened with the advent of the 21st APAC event, which will be held in Melbourne next year by the SAE-A on 3-5 October 2022. After a general networking opportunity over drinks the event moved to the dining room where the presenters; Saeid Nahavandi Pro Vice Chancellor and Director – Institute for Intelligent Systems Research & Innovation Deakin University, and Ross Cureton Director of Product Planning at PACCAR spoke. Both were extremely entertaining and informative. www.saea.com.au
Mr Nahavandi concentrated on the journey towards full autonomy and the shift from humans in-the-loop to humans-on-the-loop. Mr Cureton spoke on the challenges and opportunities introducing autonomy and increasing levels of ADAS (Advanced Driver Assistance Systems) to the heavy vehicle industry. He spoke on what is being offered and what the road ahead will look like. Both presentations are explored in greater detail in this issue of Vehicle Technology Engineer magazine.
30 years John Bamford John Baxter William Keramidas Tony Kotevski Mario Larocca Brian Milton Michael Paine Peter Tsavdaridis Jeffrey Watters 20 years Earl Gilchrist Eddie van den Berg John Varetimidis VTE | 5
News | SAE
Laurie Geyer recognised with Honorary Membership Well known industry figure Laurie Geyer has been recognised by the Board of Directors of the SAE-A with its highest accolade of Honorary Membership. The SAE-A awards Honorary Membership to only those persons it sees as having made a significant contribution, both to the engineering community and specifically to the SAE-A. Mr Geyer’s commitment to the industry and the SAE-A is unquestionable. He began his distinguished career with Toyota where he held a number of roles including Corporate Manager for Export where it was his task to understand the nature of quality issues that may be experienced with the Camry, which was being exported to five main markets in the Middle East. This was a highly technical and sensitive position which was very capably handled by Mr Geyer and resulted in a large number of improvements implemented in Toyota’s Altona plant. Mr Geyer led a team of around
30 quality, production, product and service engineers and technicians. He was later promoted to Manager – Export Operations & Warranty for all domestic and overseas products, and again he led a highly skilled team of technical experts – engineers and technicians alike. A further promotion saw him take on the role of Operations Manager – Quality Assurance at the Altona plant reporting directly to the Divisional General Manager of the quality control division. This new role saw him responsible for all aspects of the departmental activities which included technical, policy, team development and budget with around 100 staff. Following the retirement of the incumbent divisional general manager in 2007, Mr Geyer was appointed to the position of Divisional
General Manager Quality Assurance reporting to the Executive Director of Manufacturing. He held this position until his retirement in 2012. Mr Geyer was also heavily involved with the SAE-A and represented the society on various Standards Australia technical committees over the years. A formal presentation of Mr Geyer’s Honorary Membership will be held at a later stage as Covid-19 allows.
Registration open and call for papers for APAC conference in Melbourne The Society of Automotive Engineers Australasia (SAE-A) has been selected to host the 21st Asia Pacific Automotive Engineering Conference in Australia. The conference will focus on innovative applications, manufacturing, tools, platforms and solutions.
gone out and a full list of topics is available on the website with the key subject themes of: •
Automated and connected mobility
These will cover technology areas such as the application of signal processing, wireless communications, informatics and electronics, related to different types of vehicles such as cars, trucks, buses, off-road vehicles and trams.
•
Next-generation electric, hydrogen, fuel cell vehicle powertrain technology
•
Digital Transformation in the mobility industry
•
•
Human factor issues and challenges
•
ICT specialists
•
Crash and NVH performances
•
Design rule practitioners
•
Manufacturing, materials and lightweight solutions.
•
Traffic Accident Commission (TAC)
•
Human welfare experts
•
Local government planning - smart cities and road infrastructure
•
Representatives from Federal and State Government
•
Systems integrators/connectivity
•
Motoring bodies.
Registration for the conference is open and the cost of entry includes access to three days of conference sessions and the exhibition, morning tea, lunch and afternoon tea and catering breaks across all three conference days, tickets to the welcome reception and the gala dinner at Regent Theatre, access to the autonomous vehicle track and to the car launch display. The call for technical and industry papers has
The event will run from Monday 3 October until Wednesday 5 October 2022 at Hyatt Place, Essendon Fields, Melbourne. Delegates will come from all over the world representing: •
Vehicle manufacturers and importers
Component manufacturers
• Farming/agri-business
To start the event a welcome reception will be held on Monday 3 October 2022 from 6pm 8pm at the Hyatt Place, Essendon Atrium. And as a fitting end to the event a gala dinner will be held on 4 October from 7pm until 11pm at the Plaza Ballroom of the Regent Theatre. For more information visit: www.autonomous2022.com 6 | June 2021
General | News
Automotive Right to Repair passes the Senate and is now law The Senate has passed the Motor Vehicle Service and Repair Information Sharing Scheme Bill. Following nearly a decade of campaigning by the Australian Automotive Aftermarket Association (AAAA), the new law will make it illegal for car companies to withhold information from qualified independent mechanics allowing them access to vital information to service vehicles. The CEO of the AAAA, Stuart Charity, said the mandatory scheme will require all motor vehicle service and repair information to be made available for purchase by independent repairers at a fair market price. “It has been a long time coming but will be welcome news for the automotive industry. We started campaigning for this law a decade ago and have been through two government inquiries and even through a voluntary agreement in 2014 which was a complete failure,” he said. The new law is designed to provide a fairer playing field for the repair and service of the 74 automotive brands available in Australia in an industry worth $23 billion annually. In Australia the motor vehicle servicing and repair industry involves nearly 35,000
Polestar, the Swedish electric performance car has launched its 2021 vision of future mobility Design Contest under the theme of ‘progressive’. businesses employing more than 106,000 Australians. Mr Charity said around one in 10 motor vehicles taken to repair workshops are affected by a lack of access to service and repair information. “This can often lead to higher service costs for consumers,” he said. “What this law means is that the service and repair information that car manufacturers share with their dealership network must also be made available to independent repairers.” This new law is the result of unprecedented industry cooperation with over 75 workshops hosting visits from their local MPs to demonstrate what happens when vehicle manufacturers withhold software updates and technical service bulletins.
New research on banning petrol and diesel cars If a ban was introduced on the sale of new petrol and diesel cars, and they were replaced by electric cars, the result would be a great reduction in carbon dioxide emissions. That is the finding of new research from Chalmers University of Technology, Sweden, looking at emissions from the entire life cycle from manufacture of electric cars and batteries to electricity used for operation. However, the total effect of a phasing out of fossil-fuelled cars will not be felt until the middle of the century, and how the batteries are manufactured will affect the extent of the benefit. A rapid and mandatory phasing in of electric cars could cause emissions from Swedish passenger cars’ exhausts to approach zero by 2045. The Swedish government has proposed an outright ban on the sale of new fossil fuel cars from the year 2030 but that alone will not be enough to achieve Sweden’s climate targets on schedule. “The lifespan of the cars currently on the roads and those which would be sold before the introduction of such a restriction mean that it would take some time – around 20 years – before the full effect becomes visible,” said Johannes Morfeldt, researcher www.saea.com.au
Polestar design contest for students and professionals
The contest features both student and professional categories, with initial designs submitted in the form of sketches or digital images. Designs should display new ways of thinking that could encourage positive change in society, including responses to the ongoing climate crisis. The progressive theme should be evident in the designs. Registrations start 7 June 2021 and close 30 June 2021. An exclusive feature of the global Polestar Design Contest is the coaching and support it offers. Shortlisted submissions will receive guidance and feedback with Head of Design Maximilian Missoni, Polestar designer and sustainability lead. The winning designs in each category will be brought to life as 1:5 scale models and an exhibition of the winning designs will then tour Polestar Spaces around the world. For more information visit: www.about.polestar. com/polestar-design-contest/2021
Electrify America Three years after Electrify America (EA) opened its first DC fast charging site in Massachusetts, the company currently sits at nearly 600 sites and 2,600 fast chargers throughout the United States.
in Physical Resource Theory at Chalmers University of Technology and lead author of the recently published scientific study. “The results from our study show that rapid electrification of the Swedish car fleet would reduce life cycle emissions, from 14 million tonnes of carbon dioxide in 2020 to between 3 and 5 million tonnes by the year 2045. The end result in 2045 will depend mainly on the extent to which possible emission reductions in the manufacturing industry are realised,” Mr Morfeldt said. A transition from petrol and diesel cars to electric cars will mean an increased demand for batteries. Batteries for electric cars are often criticised, not least for the fact that they result in high levels of greenhouse gas emissions during manufacture. To read the study visit: www.chalmers.se/en/ departments/see/news/Pages/Banning-the-sale-offossil-fuel-cars.aspx
A subsidiary of Volkswagen Group of America, EA is investing US$2 billion over 10 years in zero-emission vehicle (ZEV) infrastructure, education and access. The Reston, Virginia-based company expects to install or have under development about 800 total charging stations with about 3,500 DC fast chargers by December 2021, expanding to 29 metros and 45 states. Two cross-country routes already were completed in 2020, as were two coastal routes (north-to-south on each coast). VTE | 7
News | Auto
10 things you need to know about Victoria’s new EV road user charge This article was written by Victoria’s RACV. 1. What is road user charging? Road user charging simply means charging users for the use of roads they drive on. The Victorian example is a distance-based charge applied to zero and low emissions vehicles such as electric cars. They will be charged a set amount per kilometres driven. 2. Why should ZLEVs pay road user charges? All motorists that drive petrol and dieselpowered cars, also known as internal combustion engine (ICE) vehicles, pay the fuel excise, which is about 42 cents per litre. Much of the revenue the government collects from the fuel excise is used to pay for new roads, road maintenance and infrastructure. Given EVs are battery powered and don’t require fuel, EV owners don’t pay the fuel excise. The government says it introduced the road user charge as a way for EV owners to pay their fair share for road maintenance and infrastructure. 3. How much will it cost? Owners of battery electric (EV) and hydrogen fuel-cell electric vehicles (FCEV) will be charged 2.5 cents per kilometre driven, while plug-in hybrid electric vehicles (PHEV) will be charged 2.0 cents per kilometre. According to Vicroads, light vehicles in Victoria travel an average of 13,100km per year. That means EV owners are looking at an annual charge of about $330 and PHEV owners will pay about $260. 4. Will plug-in hybrids still pay fuel excise as well as the road user charge? Yes, they will. Plug-in hybrids use an internal combustion engine (usually petrol) combined with a battery or batteries that powers the
electric motor. Most PHEVs have an electric only driving range of about 50km. When the electric charge is depleted, the PHEV uses its engine for propulsion. As part of the road user charge, PHEVs are charged a lower per kilometre rate than electric vehicles because they incur both costs. 5. Will owners of regular hybrid vehicles have to pay the road user charge? No, they will not. Regular hybrid vehicles don’t require external electrical charge. The batteries are charged by the petrol engine and through regenerative braking and the vehicles still require fuel. 6. Can I still get a discount on vehicle registration for a ZLEV? Yes, you can. Owners of EVs, PHEVs and FCEVs are still eligible for a $100 discount on their annual registration fees through Vicroads. Regular hybrids, however, are no longer eligible for the discount. 7. What will the road user charge revenue be used for? The Victorian government says the money raised from the charge will be used to help fund a $100 million package of policies and programs designed to encourage the uptake of ZLEVs. The proposed package includes funds for subsidies for purchasing an electric vehicle, increasing electric vehicle charging infrastructure across the state, an electric public bus trial, a study looking at EV readiness of new buildings, and more. 8. How do you pay the road user charge? The charge will be payable through Vicroads. ZLEV owners will need to provide Vicroads
with their odometer readings to determine the charge, and it will be payable quarterly, half-yearly or annually. Vicroads says it will contact owners with more information on how to report odometer readings and the billing and payment process. 9. What happens if I don’t pay the charge? You may receive an invoice with additional charges, or your registration could be suspended or even cancelled. 10. Will I have to pay the road user charge if I drive in another state/territory? Yes. According to the Department of Transport, motorists driving a ZLEV that is registered in Victoria are required to pay the road user charge for all kilometres travelled within Victoria and interstate. There will be no road user charges for travel offroad (such as on private property farm tracks), but drivers must provide evidence of offroad use.
Premcar recruitment drive for engineers Automotive engineering firm Premcar has commenced an extensive recruitment drive to secure 35 skilled automotive specialists in Australia. The Melbourne-based company is expanding its team to accommodate the final development and production of a new Nissan vehicle model, which is a co-development between Premcar and the global automaker. Premcar is targeting specialists in engineering and manufacturing. Successful applicants will be dedicated to Premcar’s latest OEM (original equipment manufacturing) program with Nissan. Premcar’s new roles will be based at its production facility in the suburb of Epping 8 | June 2021
in Melbourne. The company established the state-of-the-art manufacturing and assembly
centre for its first project with Nissan the Navara N-Trek Warrior 4x4 pick-up.
Auto | News
Gross combination Mass Rating of Light Vehicles draft code Following the establishment of a Gross Combination Mass (GCM) Technical Working Group by the Australian Automotive Aftermarket Association (AAAA), a draft GCM Vehicle Modification code was released by Queensland’s Department of Transport and Main Roads (TMR). Customers requesting Gross Combination Mass upgrades has increased substantially, with larger vehicles and larger towing loads requiring modifications to the lead vehicle to ensure motorists are not overloading, damaging their vehicles or taking risks that affect road safety. While the industry has wanted to be able to respond to consumer requirements with confidence, this has been difficult as the pathway for GCM approval has been inconsistent between the Australian states, leading to uncertainty for all involved. “The AAAA has felt for some time that there was a range of issues and opinions regarding GCM and we took the step to hold
an industry forum to air these perspectives,” AAAA Director of Government Relations and Advocacy, Lesley Yates, explained. “Shortly after this forum, we formed the GCM Technical Working Group to design a suitable test protocol as a starting point for a dialogue with state regulators. “This process has been very rewarding, particularly as the AAAA has been able to bring together our well-respected industry engineers to collaborate on proposed industry standards which has been a first for our industry. “We certainly appreciate that this Queensland TMR draft is now open for consultation and see it as an avenue forward for consumers and for industry. “If we can have vehicle standards that are fit-
for-purpose and based on sound automotive engineering principles, that will be a good result for all. But if the Code requires testing that is either not available or not required, then this could discourage car owners from modifying their vehicles to safely carry more weight - clearly not a good outcome for road users or for the aftermarket Industry.” An initial analysis of the draft has highlighted some ambiguities and uncertainty, and it warns that in some areas, the draft strays into regulatory issues relating to Gross Vehicle Mass upgrades. This over-reach could have implications for the commercial viability of GVM upgrades in Queensland. The draft GCM Code is open for feedback. A copy of the draft code is available for review. Please contact the SAE-A for a copy.
FCAI not supportive of Victorian EV tax The Victorian Government’s proposed legislation to introduce a new road charging framework for zero and low emissions vehicles is premature and might hinder the uptake of these vehicles on Victorian roads according to the Federal Chamber of Automotive Industries. “An efficient road user charging framework associated with comprehensive tax reform has the potential to benefit governments and motorists and is an issue that must be considered for the future,” said FCAI Chief Executive Tony Weber. “However, it does not make sense to apply the charge to zero emission vehicles right now as these technologies are still in their infancy and account for a relatively small portion of vehicle sales across Australia. “Right now, Governments should be encouraging the uptake of these technologies with positive policy initiatives particularly around emissions targets, infrastructure development and appropriate incentives for www.saea.com.au
fleets and private consumers rather than introducing charges that potentially reduce the incentive for these customers to buy these vehicles.” Mr Weber added that a nationally consistent approach to future road user charging frameworks should be introduced to provide clarity and consistency across the country rather than the potential for different approaches across each State. “There is no doubt that Governments must consider future revenue streams to ensure continuing investment in road and transport infrastructure. The automotive sector is wanting to be a part of those discussions to support positive outcomes
driven by efficiency and effectiveness for all stakeholders. However, at current volumes, the funds raised through this proposed legislation will be minimal. “Until zero and low emission vehicles become more mature technologies, Governments should be avoiding the temptation to subject them to new taxes and charges that impact on their acceptance from consumers.” VTE | 9
News | Truck & Bus
Briefs Actros self-steering trucks Mercedes-Benz has launched an Australian validation program for an Actros truck that can help steer itself; it is the first truck with SAE Level 2 automated driving capability in Australia.
Trucking association wants zero emission truck buying incentive The Australian Government should implement a temporary zero emission truck purchase incentive if low and zero emission transport technologies are ever going to become a reality, CEO of the Australian Trucking Association, Andrew McKellar, said.
Twenty Actros models fitted with the Active Drive Assist feature will be validated by customers as part of the program that is designed to check how the system operates on Australian roads.
“New, low emission transport technologies will never become a reality if they are not viable commercial options for trucking operators,” Mr McKellar said.
The system is not fully automated, and the driver needs be at the wheel at all times. The MercedesBenz system helps to steer the truck and aims to prevent it getting out to the edge of the lane. The driver can switch the system off.
“There must be a strong focus on the roll out of these technologies, with targeted government investment and clear action on how to remove the barriers that are preventing industry from adopting them.”
Volta to venture into heavy duty trucks
To bring down these barriers, an ATA submission calls on the Government to implement a temporary Zero and Low Emission Vehicle (ZLEV) truck purchase incentive until these vehicles make up five per cent of Australia’s heavy vehicle fleet. “ZLEV trucks are almost non-existent on Australian roads. They won’t be commercially viable until they are deployed, tested and refined for Australian operations, and increase in scale to lower costs for businesses,” Mr McKellar said.
Volta Trucks has launched four fully electric commercial vehicles between 7.5t and 19t, multiple manufacturing facilities, and is targeting more than 27,000 vehicle sales per year across expanded markets. Building upon the Volta Zero, a purpose-built fullyelectric 16-tonne commercial vehicle designed for inner-city last mile deliveries, Volta Trucks plans to expand its product portfolio with three additional variants within the medium to lower end of the Heavy-Duty class. Volta Trucks will accelerate its market entry with a Europe-first strategy, followed by US and Asian cities. Volvo Australia new VP
Auto Innovation Centre SA with BusTech The Australian Automotive Aftermarket Association (AAAA) launched a new Auto Innovation Centre at the co-located BusTech Group and Brabham Automotive facility in Adelaide. The owners of BusTech Group and Brabham Automotive, Fusion Capital, is proud to partner with the AAAA in helping to advance the South Australian manufacturing industry by providing a space to test new automotive technology and services. The opening of the AIC saw almost 100 industry representatives, sponsors, dignitaries and key media gathered to watch Senator Rex Patrick cut the ribbon to officially declare the facility open for business.
Volvo Group Australia announced that Tom Chapman, has been appointed to the role of Vice President, Mack Trucks Australia. Mr Chapman has worked within Volvo Group Australia since 2015 in a variety of roles, ranging from marketing and communications to more recently supporting the VGA retail network working in branded commercial support roles. 10 | June 2021
The ATA submission also recommended the ZLEV strategy address vehicle design rules to implement additional mass and width for zero emission and cleaner trucks, as well as extending investment in hydrogen refuelling stations.
The South Australian AIC branch launch follows the opening of the Victorian head office in December 2019 and is targeted at supporting the vast range of South Australian Automotive product manufacturers in their research and development.
Truck & Bus | News
Volvo Bus adapting the BZL for Australia Due for release in 2022, Volvo Bus Australia’s (VBA) electromobility team, led by Dean Moule, Electromobility Product Manager, Asia Pacific Region, is working to ensure that the BZL chassis, a Volvo Sweden product, is adapted to the Australian market.
Daimler on the long road to hydrogen trucking Daimler Trucks is focusing on hydrogen-powered fuel-cells for the electrification of its vehicles for flexible and demanding long-haul transport. It aims to achieve ranges of up to 1000 kilometres and more, without any stops for refuelling.
“As we look to our next 50 years in Australia, we are excited to further embark on our journey towards electromobility,” said Mitch Peden, General Manager Volvo Bus Australia.
rigorous undertaking, but we are confident that with nearly a decade of electromobility experience, that this process is one that is worth waiting for and seeing through properly.
“Our S-Charge product has already demonstrated great success and we are eager to see our awaited BZL revolutionise our product offering and electromobility efforts.
“Volvo has established itself as the electromobility expert across Europe and with our consistent and steady BZL development efforts locally, the sentiment is the same in Australia.”
“Our BZL project has been a thorough and
SEA showcases full range of electric trucks At the Brisbane Truck Show automotive technology company SEA Electric showcased the first-ever public appearance of a full range of operational-ready electric trucks utilizing proprietary SEA-Drive power systems, new SEA Electric branding, and the announcement of senior global leadership taking the company’s helm throughout the Asia Pacific region.
The Brisbane Truck Show highlighted five new SEA-Electric-branded truck models, including the launch of the SEA 300-45 EV and the SEA 300-85 EV. Both models are fully ADR compliant and assembled in Melbourne for Australian distribution. On the heels of its recent $42 million equity financing announcement, SEA Electric also www.saea.com.au
closed its latest purchase of 1000 electric vehicle batteries from long-time technology partner Soundon New Energy Technology. Most of the initial units are for the United States,with the balance to go to SEA Electric inventories in Australia, New Zealand, and Southeast Asia, as well as the company’s first entry into the European market.
The truck manufacturer has begun its rigorous tests of the first new enhanced prototype of its Mercedes-Benz GenH2 Truck, which was unveiled in 2020. This marks an important milestone on the path to series production. According to Daimler Trucks’ development plan, the vehicle will also be tested on public roads before the end of the year. Customer trials are scheduled to begin in 2023. The first series produced GenH2 Trucks are expected to be handed over to customers starting in 2027.
The development engineers at Daimler Trucks are designing the GenH2 Truck so that the vehicle and its components meet the same durability requirements as a comparable conventional Mercedes-Benz Actros. This means 1.2 million kilometres on the road over a period of 10 years and a total of 25,000 hours of operation. VTE | 11
News | Defence & Aero
Briefs BAE opportunities for STEM Returners BAE Systems is the first Australian company to join with STEM Returners to provide new opportunities for skilled professionals in its national defence and security business.
Boxers delivered to Australian Army Rheinmetall has delivered the first 25 Boxer 8x8 Combat Reconnaissance Vehicles (CRV) to the Australian Army under the $5.2 billion LAND 400 Phase 2 Mounted Combat Reconnaissance Capability project.
STEM Returners is a program that targets people with STEM (Science Technology Engineering Maths) skills who have had extended career breaks or who are keen to move from another sector into the Defence industry and matches them to new career opportunities. The available positions will be in Melbourne and Adelaide working on some of the most critically important programs that BAE Systems is delivering for the Australian Defence Force. The recruitment and interview phase will occur mid-year ahead of 12-week internships and formal job offers being made at the end of the year. Space positing systems to benefit Australia The safe operation of autonomous vehicles may be enabled through enhancements to space-based positioning systems being investigated in Europe, with benefits flowing to Australia. Thales Alenia Space, a joint venture between Thales (67%) and Leonardo (33%), was selected for a new strategic contract to assess the extension of the Safety of Life system for aviation into the road, rail and maritime sectors. It will focus on the development of a new approach to combine several sensors (sensors fusion) including and complementing evolutions of EGNOS the European SBAS (Space Based Augmentation System) in order to provide the necessary Safety of Life integrity level to serve the high reliability and high accuracy positioning needs of new demanding applications such as road autonomous vehicles but also autonomous transport in maritime and rail sectors. The extension of Safety of Life integrity services beyond the traditional aviation certified capability, could deliver significant benefits to Australia and New Zealand. Teaming agreement for guided weapons manufacturing Lockheed Martin and Thales Australia have finalized a teaming agreement advancing the delivery of an Australian guided weapons manufacturing capability in support of a sovereign national guided weapons enterprise. The agreement will see them cooperate in the design, development and production of Lockheed Martin’s Long Range Anti-Ship Missile – Surface Launch (LRASM SL) variant, with a specific focus on booster and rocket motor technologies. Joe North, Chief Executive, Lockheed Martin Australia, said that the teaming agreement represents not only a significant commitment to the delivery of sovereign guided weapons manufacturing capabilities in Australia but recognises that local industry is also investing in opportunities for local manufacturing and production. 12 | June 2021
Rheinmetall will deliver a total of 211 Boxer 8x8 Vehicles in different versions, 131 will be the CRV variant. They will replace the Army’s Australian Light Armoured Vehicles (ASLAV) under LAND 400 Phase 2. Delivery of the first 25 vehicles enables Army to continue towards Initial Operating Capability on schedule as Rheinmetall moves into the next phase of the LAND 400 Phase 2 program.
manufacturing techniques for highly complex military vehicles.
Rheinmetall Defence Australia Managing Director Gary Stewart said delivery of these initial vehicles was only possible by taking advantage of the current production lines in Germany, and using this approach as part of technology transfer activities to ensure Australian workers and suppliers become familiar with
“Australian engineers, project managers, welders, technicians, trainers and more are living and working with their colleagues in Germany to build a deep understanding of Rheinmetall products and, crucially, acquire the skills and certifications to transfer this expertise and intellectual property to Australia,” Mr Stewart said.
Rheinmetall Defence Australia has more than 30 Australians currently living and working in Germany, working at Rheinmetall sites and learning from German colleagues. This is fostering close co-operation and a genuine partnership to realise the capability for the Australian Army.
Defence Autonomy Centre of Excellence at Fishermans Bend SYPAQ Systems will establish an SYPAQ Defence Autonomy Centre of Excellence at its new global headquarters located in Fishermans Bend, Victoria. The SYPAQ Defence Autonomy Centre of Excellence will continue to develop innovative technologies and intellectual property related to autonomous systems, sensor systems, military systems integration, artificial intelligence and cyber security. The SYPAQ Corvo family of autonomous systems builds upon SYPAQ’s aerospace engineering pedigree, harnessing its worldclass guidance, navigation & control (GNC) and software development capabilities. These systems deliver innovative, complete solutions to the most challenging tasks. Corvo autonomous systems build upon an entirely sovereign autonomous control system
that delivers a range of air, land and maritime platform solutions to meet specific mission requirements. “We are excited about our prospects in Victoria and our ability to grow valuable jobs for Victorians and we’re extremely grateful for the support of the Victorian Government,” SYPAQ’s Managing Director David Vicino said. “The opportunities available in Fishermans Bend will allow us to foster collaboration with other innovative companies and academia.”
Overseas | News
$68M hydrogen industry mission A new Hydrogen Industry Mission launched by CSIRO, Australia’s national science agency, will help support the world’s transition to clean energy, create new jobs and boost the economy.
Coregas build hydrogen refuelling station Coregas, the largest Australian-owned gases company, is building a hydrogen refuelling station for heavy vehicles at its existing Port Kembla hydrogen production facility. The project was given the go-ahead after Coregas received a $500,000 grant from the NSW government’s Port Kembla Community Investment Fund.
The research mission will help drive down the cost of hydrogen production to under $2 per kilogram, making the fuel more affordable and helping to position Australia to lead the world in exporting hydrogen by 2030. Over the next five years, more than 100 projects worth $68M have been planned by partners including: Department of Industry, Science, Energy and Resources (DISER), Australian Renewable Energy Agency (ARENA), Fortescue Metals Group, Swinburne University, the Victorian Government, the Future Fuels CRC, National Energy Resources Australia (NERA), and the Australian Hydrogen Council, along with collaborators Toyota and Hyundai. CSIRO and Boeing – research partners for
Coregas says the facility, located at the BlueScope site, will enable the deployment of Australia’s first prime mover fleet of hydrogen powered vehicles. Coregas will be making data from the project available and expects converting trucks from diesel to hydrogen will halve its vehicle emissions. Driving simulator market set to grow
more than 30 years – will also continue to explore hydrogen’s future use in the aviation industry.
$30M commercialisation fund The Morrison Government is supporting Australia’s manufacturers to turn their good ideas into world-beating realities that create more local jobs, with a new $30 million fund. According to the Morrison Government the Commercialisation Fund will foster projects that bring industry and researchers together to commercialise new manufacturing products and processes. Consistent with all programs under the Modern Manufacturing Strategy, this fund will support projects within the Government’s six National Manufacturing Priorities.
Briefs
These priorities are: Medical Products, Food and Beverage, Resources Technology and Critical Minerals Processing, Recycling and Clean Energy, Defence, and Space. Commercialisation Fund grants will be between $100,000 and $1 million and must be matched by industry. These smaller-sized grants will complement the larger projects that will be supported through the $1.3 billion Modern Manufacturing Initiative.
The automotive driving simulator market is set to grow by US$194.26 million during 20212025, Technavio’s latest market research report estimates the automotive driving simulator market to register a CAGR of about 3%. Asia Pacific is expected to be the fastest growing segment. This growth can be attributed to a demand of skilled drivers due to high road accident rates, growth in air traffic, high speed train projects and R&D investments in autonomous vehicles. New Blue petrol Following on from R33 Blue Diesel, Bosch, Shell and Volkswagen have come up with a low-carbon gasoline. This new fuel, called Blue Gasoline, contains up to 33 percent renewables, ensuring a well-to-wheel reduction in carbon emissions of at least 20 percent per kilometer driven. The initial plan is to make the fuel available at regular filling stations over the course of the year, starting in Germany. The aim is for the price at the pump to be in the range of premium fuels such as Shell V-Power.
Inkjet vehicle spray painting Robotics specialist ABB has developed painting robots that use an inkjet-type printer head with over 1000 individually controllable nozzles, enabling higher accuracy and transfer efficiency in the automotive paint shop, the company claims. The new robots have been designed to meet end-customer demand for vehicle individuality, particularly in the SUV segment. Typically, special OEM finishes demand extensive masking and repeated paint applications, adding significant time and cost during manufacture. Transfer efficiency refers to the ratio of sprayed coating that adheres to the substrate versus the overspray that ends up as wasted material. www.saea.com.au
Called PixelPaint, the system avoids paint flow, but instead uses the multiple nozzles to apply individual droplets of paint to a surface, much like digital pixels. Variable droplet control, combined with a novel inkjet design incorporating 1,000 nozzles within a 100-mm (4-in.) spread, facilitates faster and more accurate high-resolu!on printing of two-tone
and customized designs directly onto vehicle bodies. The result is a high level of detail. ABB has also introduced a compact interior paint station, which combines 12 robots to offer a space-saving alternative to traditional robotic interior systems. It is claimed to reduce the footprint of current paint booths by up to 33 percent. VTE | 13
Feature | SAE AGM
Autonomous buzz for self-driving trucks Autonomous vehicles are the engineering buzz words of the decade, but it may be trucks rather than cars that could benefit more from autonomy. Ross Cureton, Director of Product Planning at PACCAR, was one of the two speakers at the recent SAE-A AGM, and after spending 27 years in the truck industry he certainly has the credentials to speak about the way truck engineering is heading. As he said at the start of his talk “if there’s one thing I do know a little bit about, it’s trucks”. “So, I’m going to spend a bit of time talking about how autonomy might play out with heavy vehicles, there’s a lot around about passenger cars but trucks are different,” he said. “Now the industry is finding out that autonomy in trucks is a) more do-able b) more important and c) offers a better return,” he said. Mr Cureton said it was worth asking what the purpose of a vehicle with high levels of autonomy is, as it sometimes gets a little lost in the excitement of the prospect of achieving it or even designing something that clever. Currently there are various levels of autonomous driving with Level 0 where the driver does everything. Those older than 50 years of age will no doubt remember what that really means; no cruise control, no power steering, and intermittent wipers meant they had an electrical fault. Level 0 nowadays is a bit different but remains basic, it’s up to the driver to control all functions of the vehicle. Level 1 is where the vehicle is doing something for you like cruise control or automatic headlights – simple, functional but not intrusive.
Level 2 and the vehicle is doing two things simultaneously like longitudinal speed control and steering. Level 3 starts to deal with situations and scenarios – this is the level at which the driver no longer needs to monitor the vehicle’s behaviour. Mr Cureton said this is a level nobody loves because when a vehicle runs out of ideas it must hand control back to the driver who may be distracted and not prepared for an emergency. Level 4 is where OEMS are currently working, and Level 5 is where a driver can be asleep for the entire trip. However, people in the industry have said that this last level is probably not going to happen, it no longer seems achievable. Currently we are sitting at L2 with cars, and with some but not all trucks. Other than having wheels and tyres in common – generally more and in a different configuration – the difference between trucks and cars is huge. Both in terms of objectives and lifespans. Apart from uses, trucks are kept in service much longer than cars. The lifespan of a heavy rigid truck in Australia is an average of 15.6 years and articulated trucks an average of 11.5 years. Therefore, it takes longer for new technology to be generally available to most truck drivers or companies; for it to filter through.
explored such as closed course autonomy, on-highway autonomy and urban autonomy.
Emerging levels of truck autonomy are being
“People would have you think that is quite close to the next level,” Ross said. “But they are worlds apart.”
Closed course is by far the easiest to realise as it may be a truck running around a closed environment like a port or a distribution facility (think of the autonomous picking machines in a distribution centre). The next level though is a leap not a step.
The next level in this instance is highway autonomy, and while there is a lot of merit in highway autonomous trucks, they are not on the radar just yet. Speaking of radar, and Lidar, these are just some of the technologies that are needed to implement autonomy but in an urban setting it’s just too hard. PACCAR is currently in the process of developing trucks with ADAS (Advanced Driver Assistance Systems) incorporating a mix of cameras, lidars and radars but not for urban consumption. “For instance, in a lidar view you can see fixtures, you can see the other road users, but those humans – that’s tough because you can train a computer to know what a human 14 | June 2021
SAE AGM | Feature
“Everything you know about the world is provided to you by an organ that itself has never seen that world. That exists in silence and darkness like a prisoner. To your brain the world is just a stream of electrical impulses like taps of morse code and out of this bare and neutral information it creates for you, quite literally creates, a vibrant three-dimensional sensory engaging universe. Your brain is you; the rest is just plumbing and scaffolding.” When considered in this light it is not surprising that a totally autonomous Level 5 car or truck is impossible or at the very least very, very far into the future. You are trying to build a brain for vehicle. But even if Level 5 is not likely there are lower levels that for the truck industry could prove to be far more advantageous than for the car driver. Certainly, not having to have a driver in a truck could revolutionize the commercial reality of running a heavy vehicle or a fleet of them. PACCAR is working on Level 4 and Level 2 proof of concept, that means using L4 to drive this demand for technology and software then trickling it down to Level 2. And according to Mr Cureton it seems to be a pretty effective way of doing it.
looks like, but which way is that human facing, are they about to step off the kerb, are they looking my way – all those things. If a human is riding a bicycle or carrying a box, does it still recognise it as a human,” Mr Cureton asked. “We tend to focus on the sensing when that’s not the important area. This is an overview of the process, as a human we sense the environment, we perceive it, we interpret it, then we decide what to do. There about a billion lines of coding to do that.”
Basically, it’s turned out to be harder than anyone in the industry thought. Maybe it was people over promising with this technology. Mr Cureton went on to describe the vast differences between what a human does every minute and what can possibly be taught to Ai. In the book The Body: A Guide for Occupants by Bill Bryce he says that in 30 seconds of doing nothing your brain churns through more information than a Hubble space telescope can process in 30 years.
Highway autonomy might be a viable next step for trucks because some of the infrastructure is already in place. Many freight terminals are already on city fringes due to our urban sprawl so it may be possible to have autonomous trucks only employ drivers for the last 30 kms of a trip much like a ship’s pilot to lead the truck through the urban fringe to the terminal. Platooning is another option that is already being trialled in real life situations overseas particularly in industries like forestry where logging trucks are used on closed roads. With platooning, only the lead truck has a driver with those trucks following taking their instructions from the lead. Truck platooning is the linking of two or more trucks in convoy using CAV technology. The trucks automatically maintain a set close distance, the truck at the head of the leads with the vehicles behind mimicking its movements. To watch a video of platooning visit https://cohdawireless.com/platooning/ which was a vehicle demonstration shot in Australia at The Bend motor racing circuit near Adelaide. The video demonstrates two autonomous vehicles following a lead vehicle around the track at a gap of 20m. Subsequent tests confirmed Cohda’s platooning solution can deliver gap management of 16ft (5m) + 0.4 seconds at 59 mph (95 km/hr). Which way we go with truck and car autonomous driving will be interesting to see but it won’t be this year or the next.
www.saea.com.au
VTE | 15
Feature | SAE AGM
Humans always in the loop Professor Saeid Nahavandi, director of the Institute for Intelligent Systems Research and Innovation at Deakin University, presented at the SAE Annual General Meeting on the topic of Human-on-theLoop; from bilateral to autonomous control. The talk focused on the transition from “humanin-the-loop” to “human-on-the-loop” in the field of robotics and haptics. The role of humans in the current robotics ecosystem will shift from being “in the loop” to “on the loop” where robots are deployed. In this talk, Professor Nahavandi argued that most existing robotic systems have human involvement in some way, shape or form. To achieve a robot assisted mission, the human is often in the loop, performing a critical task to the mission that comes with its own weaknesses and disadvantages. He predicted that the role of humans in this ecosystem will
shift from being in the loop to on the loop. In this talk, he covered aspects of robotics and haptics research and their applications and also highlighted major challenges in the journey towards full autonomy, and the shift from humans in the loop to humans on the loop but said it was never intended to have humans out of the loop. What is a Human-in-the-Loop? The definition is when a machine performs a function for some time then waits for human input before continuing. Human-on-the-Loop is when a machine can
perform a function entirely autonomously but has a human in a supervisory role with the ability to intervene if and when necessary. Professor Nahavandi then took the audience on a journey with robot developments since 1983, those days robots were new but quite primitive and have advanced greatly since. In 1986/87 he said he had an idea of creating a ‘humanless factory’ now called industry 4.0. “We set a series of machines like CNC machines, PLC machines and robots to work autonomously,” he said. “With direct numerical controls we were using drip feed – all old technology now. I managed to demonstrate full autonomy.” However, by 1991 Professor Nahavandi had come to conclusion that a human out of the loop was not a good idea and a ‘humanless factory’ was the totally wrong idea as a human is the most important part of any system. In early 2000 while at Deakin University Professor Nahavandi set up a research lab in haptics; haptics is the science and technology of transmitting and understanding information through touch. The most well-known examples of haptics are probably the vibration in a mobile phone or the rumble in a game controller, but there are a huge variety of applications. In 2006 haptics was trialled in the automotive industry and used to assemble a virtual car cockpit where the cockpit did not exist in reality but could be assembled and disassembled. Where this is important is that not only can you see how to assemble a part you can feel it, feel the space, the materials and really be able to judge how something could be put together. “We arrived at the conclusion that haptic devices are all single point and then looked
16 | June 2021
SAE AGM | Feature
offer remote diffusing of bombs safely. This was around 2007 or 2008.” In recent times this technology has been trialled with hospitals. Before this trial surgical operations with robots were used but a surgeon had no feel when using a robotic arm. These surgeries were generally facilitated with a Da Vinci robot. With Professor Nahavandi’s new haptics technology incorporated, it was possible to not only perform remote surgery but feel it as well as see what was happening. As he said there wouldn’t be anyone who would prefer the former if they were under the knife. “Then I had crazy idea in 2006,” he said. That idea was to obtain a RoboCoaster, which was ostensibly designed as an amusement ride using a Kuka robot. Professor Nahavandi wanted one to work with at Deakin University an eventually obtained ARC funding to purchase one.
at the technology and came up with what is called multi-point haptics and patented this technology,” Professor Nahavandi said. “Then we built robots, and these have been in service in Victoria with the police and special operations groups. We call them Ausbot – the Aussie robot. “What we did is we actually combined the two technologies: a robot arm, a mobile platform and a haptic device. By combining these technologies, we were the first in the world to
www.saea.com.au
Now Deakin University’s Universal Motion Simulator is employed to overcome the limitations of current motion simulator platforms by employing an anthropomorphic robot arm that provides the motion fidelity necessary to exploit the potential of modern simulation environments. Full motion simulators frequently utilize Stewart platforms to mimic the movement of vehicles during simulation. However, due to the limited motion range and dexterity of such systems, and their inability to convey realistic acceleration, they are unable to represent accurate motion characteristics. Professor Nahavandi’s Universal Motion Simulator aims to close the gap between the limitations of the current motion technology and the real world, by introducing a flexible,
modular, high-fidelity motion system that can be used for a variety of immersive training applications. The modular nature of the design allows interchangeable and configurable simulation pods to be attached to the end effector. Results of this research show that a simulation-control layer allows the novel motion simulator to accurately reproduce the motion of a simulated vehicle while introducing minimal latency within the control loop and greatly increase the overall functionality of the simulator. This motion platform will open new research opportunities in the study of human-machine interaction, human psychophysics, and the evaluation of human performance in virtual training situations. After conducting tests with the Australian army, the army asked to purchase one of these for its own use. Deakin has since commercialised the project and the army has paid $54M to buy six of these robots for its own use. Since then, Professor Nahavandi has developed a robotic simulator where you can create acrobatic manoeuvres for JSF aircraft, and as Professor Nahavandi said not even Lockheed Martin has this type of simulator – but in a village called Geelong we have one. Saeid Nahavandi has a BSc (Hons), MSc and PhD in Control Engineering from Durham University, UK. He is an Alfred Deakin Professor, Pro ViceChancellor and the founding director for the Institute for Intelligent Systems Research and Innovation at Deakin University. More information is available at https://www.deakin.edu.au/iisri
VTE | 17
Feature | Monash
M21 the start of a new era for Monash University Monash students have been leading the way for years with new vehicle innovations for Formula SAE from combustion engines to electric vehicles and now autonomous vehicles
After witnessing the many attempts, and wins, that Monash has had at Formula SAE it comes as little surprise after attending the Monash Motorsport 2021 Design launch night that they are spearheading the charge to autonomous vehicles. This is a very professional team run very much as a company complete with a Chief Operating Officer in Malhar Palkar leading an Upper Management Team with Jack Church as CEO and Ben Robinson and Chief Technical Officer. But all that can’t happen without the support of the university and the Monash Motorsport Team has that in spades with the Dean of the Faculty of Engineering, Professor Elizabeth Croft opening the night and expressing her delight at the work done by her students. Professor Croft and Mr Palkar reflected on the work that has gone before by students now gainfully employed in industry and more recently with the 2019 driverless car project which Monash had developed alongside its other Formula cars under the banner of One Team Three Cars. “At the Australasian competition in 2019, both M19-C and M19-E battled teams across the world in static and dynamic events. After a long weekend of intense competition, we were crowned the victors in both combustion and electric classes,” Mr Palkar said. “Our victory was marked with an update of the world rankings, which are updated to reflect the results of the latest competition. After our win at the Australasian competition, we have retained the world rankings of number one for the Combustion Class, and number three in the Electric Class.” Unfortunately, 2020 threw up an obstacle that could not be overcome – COVID-19, and so any competition as it had been held before was off the table regardless of whether it was in Australia or overseas. Rather than see this as a setback the team regrouped and reset to preparing for the future format of student competitions: driverless cars. With Monash’s current driverless platform, the M19-D performing at a high level, the team was 18 | June 2021
Monash | Feature
ready to take a leap up to the next level of the competition. This involved a restructuring to encourage the development of an integrated electric/driverless technology, and to foster an integrated approach to the design process.
designated heads of departments with Jack Bell as Head of Dynamics, Maksis Darzins Head of Structures, Jessica Lee Head of Electrical Systems, Jordan Esh Head of Autonomous Systems and Ben Wang as Head of Business.
“In order to take advantage of our position and the extended time available to us when working from home during lockdown, the team made the decision to cease development of combustion vehicles and commence the design of our most ambitious entry yet, the integrated electric/driverless vehicle, M21,” Mr Church explained.
The M21 all-wheel drive platform
“In 2021, the easing of COVID-19 restrictions allowed us to resume our in-person activities, this included us moving to our new home in Monash Makerspace. Our new workshop’s facilities include dedicated specialised bays for machining, composites, electrical, fabrication and sanding, as well as more area for us to accommodate a growing team.” As with industry automotive design departments Monash has
www.saea.com.au
M21 will be Monash Motorsport’s first car designed from the ground up with driverless capabilities in mind. “After extensive concept development, we are proud to announce our new aerodynamics focused concept with an all-wheel drive electric powertrain – M21. However, the greatest change in this vehicle is the integration of our autonomous pipeline. “For the first time ever, we have designed all of the systems on M21 to be compatible with our autonomous systems, future proofing our concept for future competitions and enabling us to push the frontier of our Autonomous technology development,” said Ben Robinson Chief Technical Officer.
VTE | 19
Feature | Monash
Electric charge even more critical The goal for this design period was to facilitate the AWD concept and optimise previous designs but longer term the team is targeting higher performance taking into account brake regeneration. “This car is built from the ground up as a driverless vehicle. We’re also continuing development of parts for future MMS cars, including development of a custom Silicon Carbide inverter, and HV to LV converter and distribution system,” Ms Lee explained. “For M21 we have designed an accumulator with a single p config of low impedance Melasta cells. This gives us 138s1p 580V pack that we can push to supply our twin Lenze inverters for 80kW discharge. “Currently we are looking at an estimated car mass of 237kg, simulated CLA of 5.02 and CDA of 1.81. We’ve chosen a track width of 1300mm and wheelbase of 1550mm. “Based on our current lap time simulations, we are also looking at an estimated power limit of 80kW in both charge and discharge over the endurance event.” Mr Robinson said that the CFRP monocoque design supported the vehicle’s high downforce concept, which gave the team the geometric freedom to utilise direct acting suspension at the front of the car, and pushrod suspension at the rear due to the constraints imposed by the aerodynamics. “For our powertrain, we have gone with four Fischers, each driving a wheel directly through a custom gearbox at each corner of the car. These are powered by two Lenze inverters, each driving two motors at both ends of the car. Finally, our battery pack is centred around lithium-Ion pouch cells, provided by Melasta,” he said.
The goal for autonomous systems For the autonomous systems department the goals were to increase the robustness of the autonomous software pipeline and provide a solid foundation for future developments. The team has a new suite of features and upgrades that it is keen to integrate into the vehicle.
Jordan Esh it is a ‘staggering improvement’ that paves the way for further advancement in software including a newly developed nueral network which is able to detect objects from pointcloud images at up to 50 frames per second. “This computational load will be handled on an nVidia Jetson AGX Xavier, neatly packaged in our new, more deeply integrated computing enclosure, shaving almost two kilograms off our design for our current driverless vehicle, M19-D,” Mr Esh explained. “With functionality and accessibility in mind, the computing units on M21 will be split into two separate housings, with the rear main enclosure containing the more commonly accessed units, such as the Xavier, and our autonomous state machine circuitry. The front housing, neatly hidden away inside the front access hatch of the monocoque, contains additional circuitry, such as our GPS boards, which we don’t access as frequently. “We’re excited to also announce that we are starting development this year into a custom Model Predictive Controller implementation, using the FORCESPRO software. This will allow us to better understand, and optimise the control problem, which is how the vehicle should react, for our specific application.”
“In addition, we have designed a fully selfdeveloped distributed battery management system. Our BMS design is the culmination of work that began in 2019 and grants us massive advantages in mass and packaging.” Currently Ms Lee’s department is completing the bulk of manufacturing in the lead up to testing on the MA+E department mechanical dyno. This is a big milestone and will form the major component of integrated system testing before driving. “We will be mounting our systems on the dyno for motor control tuning. We’re also aiming to validate our low voltage, safety systems, and tractive interactions. Concurrently, our LV system is being developed with wiring harness assembly scheduled to begin shortly,” she said.
Providing structure Maksis Darzins’ area faced working closely with all other areas to accommodate their requirements and so has integrated and developed new components such as a new accumulator, planetary gearboxes, a monocoque chassis and outboard motors. The was achieved after countless hours of CAD and simulation techniques. Mr Darzins said that his area now faced the busiest part of the year with a raft of machining and composite layup tasks ahead.
Dynamic performance
Outside the square
Many of the new features and upgrades come courtesy of the companies that have chosen to assist the team with products and technical know-how, knowing that that these are the engineers they will be employing in the future.
Extracting as much performance as possible aerodynamically and vehicle dynamically was the task for Jack Bell’s team and so far, the success has translated to a 60 percent increase in aero performance over the M19 model.
One of these is Baraja who has provided their Spectrum-Flex LiDAR system which is able to see up to 40 metres ahead and according to
LEAP provides Mr Bell’s team access to the Ansys Fluent program in order to simulate the aerodynamic performance of the package.
Engineering by itself is only part of the plan, without the support of other departments it stalls and that is also where Monash is ahead of the game with the business area headed by engineer Benjamin Wang commingled with students from other faculties and areas of expertise such as communication, design, marketing, finance, and law faculties.
20 | June 2021
Not only do these add to the fullness of the
Monash | Feature
engineering process but those facilities and students then have the opportunity to work in a real-life environment which can only add to their preparedness for industry work. And hopefully, working with the engineering team will inspire some to work in this area which is often not perceived as glamourous as others.
The end is in sight “Moving forward, we will continue to work towards our rolling, driving and flying deadlines, which will culminate in our first planned autonomous drive in August,” Mr Robinson said. “To achieve this ambitious concept, we have set a rather aggressive timeline: with the team returning in early January this year, we set our design freeze deadline at the start of February to give our part designers time to polish their designs and finalise any integration matters with the team now that we could be back on campus. “From there we have been in full manufacturing and outsourcing mode to begin the construction of M21.
of the 19th of July, followed shortly by the completion of our aerodynamics package at our flying deadline and then culminating in our first autonomous drive with M21 in early August.
“To ensure that we have sufficient time to test and refine all of our systems on the car, we set a relatively early driving deadline
With all this in mind, we also set a date for our rolling chassis (which includes all suspension components) to be completed
in only a few weeks from now at the start of May. This gives us approximately five months to manufacture our most ambitious vehicle concept yet, and approximately 3-4 months to complete our rigorous testing and validation period. All in the lead up to the main event for the year: the 2021 Formula SAE-Australasian competition in Winton at the start of December.”
Presenting a deserving organisation CEO and Chairman of SAE Australia Adrian Feeney and Monash Director of Student Teams, Scott Wordley were invited to the stage for a special presentation. As Mr Feeney explained, from the inception of Formula SAE until 2016 there was just one class for internal combustion engines and hence just the one major trophy to be won. Since then, there was the addition of an Electric Vehicle team trophy and so the original trophy was outdated. What to do with such a large piece of important silverware became the question for the SAE-A. Keeping it at the SAE-A offices was an option but a more fitting destination was Monash. Since 2009 until 2015 the Monash University team won every year; a huge achievement to gain the top spot six years running. So, it was thought that to reward such an outstanding result it was time to pass the trophy to Monash University permanently. Mr Wordley accepted the trophy on behalf of the Monash alumni past and present who made such a massive contribution to these wins. www.saea.com.au
VTE | 21
Technical | Feature
Sven Hallerbach, Yiqun Xia, and Ulrich Eberle, Opel Automobile GmbH, Germany Frank Koester, German Aerospace Centre, Germany
Simulation-Based Identification of Critical Scenarios for Cooperative and Automated Vehicles Article courtesy SAE– International Introduction The release of cooperative and automated vehicles is impeded due to a small market penetration in the beginning. The task to include these vehicles into an existing infrastructure with human drivers is challenging. It would be simpler to introduce these vehicles without having to consider the driving behavior of other traffic participants. Even though the automation functions are working properly and result in a normative driving behavior, other traffic participants can cause a critical scenario for the automated vehicle. Figure 1 shows such a critical scenario caused by the complex interactions of human drivers. The initial situation on a German “Autobahn” includes three involved vehicles where vehicle 1 is a cooperative and automated vehicle and the other participants are controlled by humans. The driver in vehicle 2 shows an aggressive behavior, while the driver of vehicle 3 drives defensively. In the beginning, the automation functions of vehicle 1 spot an ideal situation to perform a lane change. While this lane change is performed, vehicle 2 disregards the safety gap and tailgates vehicle 1. At the same time, vehicle 3 is driving extremely careful, which can result in a very serious situation for the automated vehicle shown in Figure 1. Regardless of who has caused this critical scenario, the vehicle automation has to be capable to handle this scenario properly. Thus, these scenarios are critical and have to be tested. Keeping this in mind, the identification of every possible critical scenario is quite difficult. Even with the use of existing methods, for example, developing a scenario cata-logue, gathering expert opinions, or the investigation of existing data, critical scenarios may be missed [1]. Additionally, malfunctions in the automated vehicle can cause critical scenarios. For example, errors can occur in the sensor percep-tion, actuator faults, wrong data in precision maps, etc. The specific
FIGURE 1: Critical scenario for automated vehicle V1, caused by traffic participants due to violation of the required minimum gap. Initial state shown at the top segment, final state shown in the bottom segment. www.saea.com.au
influences of these errors have to be investigated, evaluated, and tested. The main contribution of this article is the composition of a generic simulation-based toolchain for the identification of critical scenarios. This toolchain provides the possibility to use exchangeable automated driving functions, evaluation metrics, and parameter spaces suitable for the intended iden-tification process. To show the capability of the toolchain, a cooperative and automated vehicle, called digital prototype, is embedded in a coupled simulation environment. The performed simulation runs, including the digital prototype, are evaluated by suitable metrics. This approach enables the test and improvement of the automated driving functions depending on requirements. After having reached a mature development stage, such a digital prototype is called a “digital twin” of the cooperative and automated vehicle. In order to provide an overview of the toolchain’s capabilities, an exem-plary automated driving function on a highway entrance ramp is tested. To identify critical scenarios based on an exemplary metrics set, the requirements presented here are twofold: they are either safety or traffic related. Note, the simulation-based toolchain is generic, meaning that the driving functions, metrics, etc. are exchangeable. Currently, only a few partial solutions for this formidable challenge exist, mainly provided by academia as well as industrial activities. Insightful work on testing and validating cooperative and automated vehicles in general can be found at [2, 3, 4]. To get a better understanding of scenario-based approaches, we recommend [5, 6]. Recently, new players like Apollo [7] and Nvidia [8] are entering the field with the objec-tive to provide comprehensive simulation platforms in the near future. The methodology presented here shows a generic meth-odology to address this overall toolchain approach in a systematic way. Additionally, the proposed toolchain does not depend on a specific simulation tool, vehicle automation function, metrics, spatial application, test method, or develop-ment status. All of these aspects can be organized by a modular approach to the scenario identification process. Simulation-Based Toolchain The major challenge of releasing cooperative and automated vehicles is the vast driving distance that has to be tested in real traffic as mentioned in [9]. To estimate the effort of a validation process, assessing accident statistics seems reason-able. Nevertheless, the derivation of a general distance based on these assumptions is questionable. In this consideration, neither location nor variation of the driven route is
ABSTRACT One of the major challenges for the automotive industry will be the release and validation of cooperative and automated vehicles. The immense driving distance that needs to be covered for a conventional validation process requires the development of new testing procedures. Further, due to limited market penetration in the beginning, the driving behavior of other human traffic participants, regarding a mixed traffic environment, will have a significant impact on the functionality of these vehicles. In this article, a generic simulationbased toolchain for the model-inthe-loop identification of critical scenarios will be introduced. The proposed methodology allows the identification of critical scenarios with respect to the vehicle development process. The current development status of the cooperative and automated vehicle determines the availability of testable simulation models, software, and components. The identification process is realized by a coupled simulation framework. A combination of a vehicle dynamics simulation that includes a digital prototype of the cooperative and automated vehicle, a traffic simulation that provides the surrounding environment, and a cooperation simulation including cooperative features is used to establish a suitable comprehensive simulation environment. The behavior of other traffic participants is considered in the traffic simulation environment. The criticality of the scenarios is determined by appropriate metrics. Within the context of this article, both standard safety metrics and newly developed traffic quality metrics are used for evalu-ation. Furthermore, we will show how the use of these new metrics allows for investigating the impact of cooperative and automated vehicles on traffic. The identified critical scenarios are used as an input for X-in-the-Loop methods, test benches, and proving ground tests to achieve an even more precise comparison to real-world situations. As soon as the vehicle development process is in a mature state, the digital prototype becomes a “digital twin” of the cooperative and automated vehicle. VTE | 23
Feature | Technical
FIGURE 2: Simulation-based toolchain for the verification and identification of critical scenarios for cooperative and automated vehicles. Automated segments of the toolchain are surrounded by straight lines. Toolchain segments where human intervention is needed are surrounded by dash-dotted lines.
included. If a test is conducted always on the same route, it can easily be argued that the distance of the tests is not a sufficient indi-cator for the validation of cooperative and automated vehicles. Further, the identification of critical scenarios is a key factor in the validation of these vehicles. Critical scenarios are defined as scenarios that need to be tested, regardless of whether the requirements are functional or nonfunctional. Aspects like traffic efficiency, driver comfort, etc. are not considered in the estimation of the validation effort so far [10]. Thus, important questions for the release of these vehicles have to be raised: • Which scenarios have to be tested with respect to the vehicle development process? • What are the specific functional and nonfunctional requirements for the evaluation? • Which test should be carried out in what test environment? • What are the general advantages and constraints of a specific test environment? Therefore, a generic simulation-based toolchain to address these questions is shown in Figure 2. This toolchain is a significant advancement of our rough concept presented at [1] and allows us to start with a logical scenario [11], which is a scenario description based on parameter spaces, defining a domain that confines possible scenarios. Table 1 shows an example of a logical scenario and possible parameter spaces. Within this parameter space, there are many concrete scenarios [11] that are determined by Attribute Entrance ramp length Number of lanes Speed limit highway Traffic flow Driver behavior Curve radius Coefficient of friction
Parameter space Unit lmin - lmax m Nmin - Nmax n.a. vmin - vmax m/s Qmin - Qmax veh/s Defensive - Aggressive n.a. rmin - rmax m μmin - μmax n.a.
TABLE 1: Logical scenario with parameter spaces (eg entering the highway).
24 | June 2021
the parameter space of the logical scenario. The selection of these parameters can be done as shown in the left part of the toolchain by deter-mining critical scenarios using expert opinions and peer review, recorded data, scenario catalogues, etc. The major drawback of these methods is that they neglect critical scenarios. Hence, the toolchain possesses another path with a parameter variation module. In this article, we will mainly focus on the path on the right-hand side of the toolchain. The parameter variation module creates concrete scenarios by changing the parameters of the logical scenario. Parameter variation is achieved by automatically changing the param-eters of the logical scenario with a certain bandwidth. This is a straightforward approach to identify critical scenarios over the entire parameter range. A systematic way to achieve more efficiency while performing this task is part of our current research. The major challenge using these approaches is that the critical scenarios and their specific nature are unknown before applying the toolchain, and therefore, it is a difficult task to determine which parameter combination has to be tested more carefully and comprehensively than others. The resulting concrete scenarios are used as an input for the simulation environment. This environment consists of a coupled traffic simulation, a vehicle dynamics simulation, and a cooperation simulation. Each simulation environment has particular advantages and a specific purpose for this method. The traffic simulation provides the surrounding environment for the automated vehicle. The vehicle dynamics simulation contains a detailed model of the vehicle and includes the automation functions that have to be tested. In order to capture the cooperative aspects of these vehicles, the environment provides a cooperation simulation in which cooperative aspects and communication models can be included. An example overview of some state-of-the-art simulation tools compatible with the presented toolchain comprises among others Apollo Simulation, DLR-Simulation of Urban Mobility, IPG CarMaker, Nvidia Drive Constellation, TASS International PreScan, and Vires VTD. The automated classification of these concrete scenarios into critical or not critical is done with the help of tailored metrics. Metrics are used to evaluate the quality of certain aspects. The toolchain can be used with various types of metrics depending on the domain of interest. In this toolchain, two possibilities of metrics are shown. It is possible to exchange metrics or extend this approach with other metrics depending on the functional and nonfunctional requirements. The evaluation of the simulation using metrics results in a verification or identification of scenarios that are critical, depending on the previous chosen path of the toolchain, and have to be investigated further. The sole use of a Model-in-the-Loop approach is only sufficient for the development part of the V-Model [12]. The use of a test method is always derived from requirements and the quality of the results. Validation engineers on the right side of the V-Model need far more
FIGURE 3: Test methods with respect to the V-Model, see also [12]. MiL = Model-in-the-Loop, SiL = Software-in-the-Loop, HiL = Hardware-in-the-Loop, ViL = Vehicle-in-the-Loop.
resilient results than devel-opment engineers on the left side designing a basic control concept. Fortunately, the level of detail for simulation models, benches, and driving tests increases along the development process. It is not possible to use a vehicle model with an exact parameter set in early stages of the development process, because the vehicle has not been produced in this phase, components are not yet built, and even the design of the development vehicle is not finished. Nevertheless, development engineers are still able to use vehicle dynamics to develop, for example, a control concept. These models will differ from the actual vehicle model in a later process phase. But the basic concepts developed in an earlier phase can be adapted to changes in the model occurring in subsequent development. Figure 3 shows a V-Model and the resulting test methods. On the left side of the V-Model, basically only Model- and Software-in-the-Loop tests are possible. As already mentioned, the vehicle components are not yet built. After the component development, it is possible to use more complex models with detailed knowledge about model parameters and component setups. On the right side of the V-Model, benches and driving tests can be applied. With growing knowledge about the vehicle and the use of real components as well as test benches, the level of detail increases, meaning that the validity of the tests rises. The drawback of this aspect is that the test effort increases simultaneously. The term “test effort” is mainly determined by costs for simulation, operating of benches, and performing driving tests. There are various constraints for possible test methods. In addition to the test effort, which is mostly an economic constraint, there are a lot of different technical properties to consider. The limits of each test environment have to be inves-tigated separately. The performance of a simulation strongly depends on the complexity of the used mathematical models. The more detailed the model quality, the more processing power is needed. In addition, the use of mathematical models always has the drawback that these models contain a certain inaccuracy. Further, it is more difficult to perform parameter identification and assess the robustness against parameter changes. Bearing that in mind, driving tests are always valid, and due to the absence of mathematical models, there are no inaccuracies in the test results. Driving tests have drawbacks,
Technical | Feature
besides their economical infeasibility, too. It is not possible to run driving tests faster than real time and some scenarios are too dangerous to be performed safely, when these borderline scenarios are particularly important for validation. Benches serve as a compromise between simulation and driving tests. Hardware-in-theLoop methods require the dismantling of vehicle parts. The part of interest is separated from the vehicle and mounted on a bench, which can be controlled by a simulation interface. For that matter, it is possible to simulate a scenario and simultaneously investigate the effects on the hardware. An extension of this is the Car-in-theLoop method. The benefit of this method is that the actual test vehicle can be mounted on a bench. Dismantling is, there-fore, not necessary. This is the closest state before having to perform driving tests. Driving tests considered in the tool-chain are not constrained to a certain domain. These tests can be performed on proving grounds and in real traffic. As some tests cannot be performed in real traffic, because this would endanger other traffic participants, proving grounds are there-fore a valuable alternative. The benefit of testing in real traffic is the possibility to sample data that shows the realistic behavior of the surrounding world of the vehicle, including other traffic participants, infrastructure, sensor characteristics, and the impact of disturbances due to, for example, map errors or sensor shortcomings. Recorded data can be obtained from various sources. Depending on the domain of interest, data can be gathered from driving tests, accident studies, traffic observation systems, etc. The use of this data is a good way to compare simulation and driving tests with the existing data. The major drawback can be the incompleteness of the data sets. For example, traffic observation systems are mostly based on camera systems. Therefore, the states of the vehicles are estimated during the process. If recorded data is used, it will be necessary to take a closer look at how the data was acquired. All of these aspects have an influence on the last part of the toolchain. The analysis is the feedback to the function development and consists of a specific description of the performed tests. Depending on the domain of interest, the analysis looks different. The function development department can choose the quality criteria. The toolchain is a generic setup, and therefore, it is possible to change scenarios, parameters, simulation models, metrics, test methods, and analyses. With that in mind, the toolchain supports a generic way to identify scenarios for cooperative and automated vehicles independent from used vehicles, sensor setups, implemented functions, locations, criticality criteria, test methods, required analyses, etc. In this article, we will demonstrate the toolchain using the example of automated highway chauffeur [13]. However, it is important to mention that the toolchain can be extended to other functionalities and domains, such as rural roads and, most importantly, urban areas. www.saea.com.au
Simulation Environment The basic requirement for the application of a coupled simulation is that the virtual environments are comparable. These environments do not necessarily need to be identical, while the properties depend on the level of abstraction and the objective of the particular simulation. For example, the vehicle dynamics simulation requires more detailed information about the environment than the traffic simulation. Nevertheless, the basic geometrical information has to be similar enough so that the dynamic coupling works in both simulation tools properly. The static coupling is based on a measured data set, provided in the OpenDrive format [14]. For the functionality of the toolchain’s simulation environ-ment, the map data has to be converted for each simulation tool. The map formats of the simulation tools differ quite strongly. A scheme of the static coupling and the map data conversion is shown in Figure 4.
FIGURE 4: Static coupling and map data conversion for the simulation environment.
The map data is acquired through high-accuracy measurements of the selected road. The Opel proving ground near Frankfurt (Germany) is chosen in this article, and the data is stored in an XML document following the OpenDrive standard. The map data converter shown in Figure 4 is a data conversion tool parsing the data from OpenDrive format to the specific formats needed in the particular simulation envi-ronment. As already indicated, the static environment of each simulation tool has different requirements with respect to accuracy of the converted data. The geometrically consistent virtual environment, with respect to the required accuracy, is the foundation for the coupled simulations used in the toolchain. Further, the consistency of the virtual environment is vital for the implementation of the dynamic coupling. Figure 5 shows the proving ground in both traffic and vehicle dynamics simulation environments. The dynamic coupling in this article focuses
FIGURE 6: Dynamic coupling with an adjustable region of interest.
mainly on the traffic and vehicle dynamics simulation. The objective of the simulation framework is to include a cooperative and automated vehicle into a traffic simulation environment. Therefore, traffic participants, for example, passenger cars, busses, pedestrians, etc., are controlled by the traffic simulation, and the virtual automated vehicle comprises driving functions of an existing automated vehicle and can be considered as a digital twin of the real vehicle. For example, vehicle model, sensor setup, and parameters are virtual copies of a real test vehicle. In addition, the use of the same driving functions ensures a similar behavior of the digital twin and the existing vehicle. The perception capabilities of the vehicle depend on the sensor setup. Due to the fact that this setup can change, we decided to introduce an adjustable region of interest, as shown in Figure 6. The traffic simulation provides the behavior of traffic participants, which is dynamically included in the vehicle dynamics simulation. Simultaneously, the vehicle dynamics simulation provides the behavior of the cooperative and automated vehicle including driving functions, sensor setup, etc. The state of the automated vehicle (ego-vehicle) is used by the traffic simulation to determine the behavior of the surrounding traffic participants. The dynamic coupling ensures that the cooperative and automated vehicle is able to drive in a surrounding traffic environment, which provides a dynamic traffic environment that is able to react according to the driving behavior of the vehicle’s driving functions. Each traffic participant possesses its own adjustable driver model. One of the major issues for the dynamic coupling between traffic and vehicle dynamics simulation is the discrepancy of sample time. The simulation of the vehicle dynamics requires a high sample rate (TS, Veh−Dyn = 0.001 s) for an accurate calculation of the vehicle behavior,
FIGURE 5: Proving ground representation in different simulation environments. 1: Traffic simulation. 2: Vehicle dynamics simulation. 3: Proving ground overview. 4: Ego-vehicle surrounded by traffic. VTE | 25
Feature | Technical
while the traffic simulation works sufficiently with a lower sample rate (TS,Traffic = 0.1 s) for traffic evaluation. Therefore, the absent motion of the traffic partici-pants provided by the traffic simulation is predicted with a constant-velocity model [15]. The motion prediction is shown in Figure 7.
FIGURE 7: Motion prediction with a constant-velocity model, required due to the discrepancy of sample times between the simulation environments.
This prediction is necessary to avoid large step changes in the sensor perception of the cooperative and automated vehicle. Simulation results state that the gap of the simulation environment’s sample times is sufficiently small so that the results with enhanced motion prediction do not influence the sensor perception significantly. Besides, this behavior is not unusual for environmental modeling. The use of motion models for the time frame of sensor fusion itself, which gener-ates a new perception for the surrounding environment, is conventional. The interaction between sensor models and virtual environment of the automated vehicle is performed in the vehicle dynamics simulation. The enhancement of a cooperation simulation is not elaborated in this article. It is worth mentioning that the tool-chain is expandable with respect to cooperation simulations including V2V, V2X, etc. Currently, the simulation environ-ment is being enhanced in our group to simulate more than one cooperative and automated vehicle simultaneously. This provides the possibility to evaluate features like cooperative merging, traffic distribution, overtaking, etc. To complete the simulation environment, functions of the digital twin have to be embedded. It is important that the vehicle model represents the test vehicle sufficiently. If this is the case, the driving functions, such as driving strategy, trajectory planning, vehicle control, maneuver intention prediction, and localization, can be implemented. All of the aspects mentioned above including driving functions complete the simulation environment of the toolchain. The simulation framework builds the foundation of the following investigations and evaluation results presented in this article. Metrics In order to identify critical scenarios with a simulation-based toolchain, the term criticality needs to be specified. It is obvious that the understanding of criticality can vary significantly, depending on the specific requirements. The generic design of the toolchain allows the use of different terms of criticality. Usually, criticality is determined by the application of metrics. The best-known criticality metric is called “time to collision” [16]. In the following, a short introduction to some standard safety-related metrics is given, before an approach to developing new metrics 26 | June 2021
will be shown. These new metrics are designed to identify critical scenarios for the coop-erative and automated vehicle (ego-vehicle) regarding traffic quality. As already mentioned, the most commonly used safety-related metric is called “time to collision” [16]:
where Dp is the difference of the vehicle positions, vego the ego-vehicle velocity, vobj the object velocity, and vrel the relative velocity between both vehicles. TTC is defined as the time until a collision between the ego-vehicle and an object would occur, if the velocity of both does not change with respect to the point of time when the TTC is calculated [17]. Another standard metric is called “time to brake,” which can be defined as [16]
where aego,max denotes the maximum deceleration the ego-vehicle is able to execute. This safety-related metric is defined as the time span until a collision with 0 m/s is unavoid-able, depending on the maximum deceleration ability of the ego-vehicle [18]. The last safety-related metric introduced in this article is called required deceleration and describes the deceleration of the ego-vehicle needed to generate a collision with 0 m/s [19]. This metric can be stated as [16]
where aobj is the acceleration of the object. The introduced toolchain allows for using different metrics depending on the specific requirements. This possibility will be demonstrated in this article. In order to improve the performance and robustness of the identified scenarios, it is convenient to use different metrics simultaneously. Furthermore, the implementation of different metrics allows for collecting more information about an investigated scenario. The introduced standard metrics have one major drawback: they are only defined for the lane the ego-vehicle is driving in. That is not sufficient for automated vehicles, because the relevant traffic participants can enter the ego-vehicle’s lane. Therefore, a maneuver intention prediction is used to identify relevant traffic participants, which are performing a lane change toward the egovehicle’s lane. The concept of using machine learning algorithms and training data to extend the usability of criticality metrics, which are only defined for a single lane, is part of our current research, but will not be elaborated in detail in this article. For further information about maneuver intention prediction and interaction modeling, we recommend [20, 21]. For the investigation of cooperative and automated vehicles and their impact on traffic quality, the use of a metric combination is proposed. This approach aims at collecting more information for the criticality evaluation of a scenario. The benefit of this methodology
is that diverse aspects can be analyzed simultaneously, which increases the robustness and validity of the results. The methodology is based on already known traffic quality metrics that are adapted for this application. General requirements for the metrics can be stated as follows: • Every critical scenario should be identified. • The “false-positive rate” (FPR) should be low. • A grading system should be used for the assessment. • There should be a threshold allowing a binary classification for the combined metrics. Usually, the investigated time interval for traffic quality varies from several minutes to hours. For our purpose, this time interval is too large and is adjusted to 15 s. The reason for this is that a longer interval would not capture short-term impacts of the automated vehicle, and the assessment whether the ego-vehicle is responsible for the critical scenario or not is difficult to make. The spatial “domain of interest” (DOI) is chosen to be 450 m following the suggestions in the highway capacity manual [22]. To achieve an additional indication about surrounding influences of the vehicle, a moving DOI is introduced. This DOI follows the egovehicle and takes the direct surroundings into account. A circle with an adjustable radius moves with the ego-vehicle and considers all traffic participants included in this area. Lastly, there will be an addi-tional moving DOI, which just takes the ego-vehicle into account. Figure 8 shows the overall concept.
FIGURE 8: Domains of interest for different evaluation metrics.
The first traffic quality sub-metric is a macroscopic description. The concept uses a fixed DOI and is based on the highway capacity manual [22] and can be calculated as
where D is the traffic density, vP the traffic flow rate, and S the average travel velocity. The traffic density can be compared to an evaluation table grading the traffic quality from A to F, meaning that A is the best grade and F the worst. This metric includes some empirical correction factors used such as peak-per-hour factor, driver-population factor, and a heavy-vehicleadjustment factor [22]. The second sub-metric used is a microscopic metric introduced by Zhu Weiha et al. [23], considering the velocity deviation and the average velocity in a fixed DOI. The concept aims to interpret the velocity deviation divided by the velocity mean value of the ego-vehicle as an indication of the microscopic traffic quality. The focus on the overall traffic quality requires the enhancement from one vehicle to
Technical | Feature
the consideration of every vehicle, referenced by index j, in the specified DOI and can be stated as [23]
where σvj is the standard velocity deviation and vj the mean velocity of every vehicle, respectively. The resulting mean coefficient of variation CV is calculated by the mean values of Equation 5. The microscopic metric gives further information about the traffic conditions, but still more information is required for the traffic quality evaluation. Therefore, a circular DOI is attached to the ego-vehicle and travels with the ego-vehicle’s position. The first moving DOI is a circle surrounding the ego-vehicle that allows for investigating close-range interactions. This metric is called nanoscopic and the calculation is based on velocity deviation and mean value with respect to the DOI and can be written as
where σvCircle, j is the velocity standard deviation and vCircle,j is the mean velocity with respect to the DOI. By using the mean value DV, including every vehicle driving inside the DOI, the overall traffic quality can be evaluated again. The last used sub-metric is called individual metric [24] consid-ering just the ego-vehicle’s quantities. The DOI surrounds the ego-vehicle only, considering solely the vehicle’s behavior to gather further information for the evaluation process. The decision whether a scenario is critical or not and if it should be investigated further is binary. It is desirable to have one specific threshold for the differentiation of critical and uncritical scenarios, even though many indicators influence the decision. The individual grades can be used later on to achieve a better understanding of why the scenario is classified as critical. To achieve an overall grading system, it is obvious that every individual metric should have the same specified range. Therefore, every grade will be normalized so that the mathematical set ranges from zero to one, where zero is defined as the best grade and one the worst. The normalized grading system can be stated as follows:
two time intervals from grade A to F according to the highway capacity manual [22]. The other grades are equipped with reference values that represent good traffic quality, which ensures that the metrics are adjustable to the specific situation. The determination of the reference values is based on the observation of representative traffic scenarios. Each individual indicator is observed over the entire range for representative scenarios and adjusted to a subjective evaluation by experts. Attributes with the subscript “ref” are adjustable to different DOIs, for example, rural roads and urban areas. The set of equations in Equation 7 can be interpreted as a grading system to evaluate different aspects of traffic quality. The macroscopic grade represents the change of traffic density between two time intervals. Therefore, negative changes in traffic density caused by the ego-vehicle result in a critical classification of the scenario. Figure 9 shows the domain of interest and the traffic quality indicator in detail.
www.saea.com.au
acceleration changes and low average velocity result in critical classification regarding traffic quality. The concept of the individual metrics is shown in Figure 12.
FIGURE 12: Individual metric with traffic quality indicators.
The simplest form of an overall grading would be to weigh every single metric equally as shown in Equation 8:
FIGURE 9: Macroscopic metric including the DOI and traffic quality indicators.
The microscopic grade is a trade-off between the coefficient of variation and the mean velocity of a time interval. This trade-off is used to achieve a better understanding of the overall traffic quality. Strong velocity fluctuations and a low average velocity result in a critical scenario, while small fluctuations and high average velocity can be classified as an optimal traffic condition. Figure 10 shows the microscopic metric.
FIGURE 10: Microscopic metric including the DOI and traffic quality indicators.
The macroscopic grade for our purpose is determined by the change of the traffic quality. This ensures that only negative changes weigh into the overall grade. The denominator of the macroscopic grade is determined by the maximum change of traffic quality between
FIGURE 11: Nanoscopic metric including the DOI and traffic quality indicators.
Similarly, the nanoscopic metric and individual grade use the average velocity as an additional indicator. The coefficient of variation and average velocity for the nanoscopic metric are calculated only for the vehicles inside the moving circle, to consider the traffic participants with close-range interactions. Figure 11 represents the concept of the moving DOI to consider close-range interactions between traffic participants inside the moving circle. The individual metric is calculated by the standard deviation of acceleration and the average velocity of the ego-vehicle itself. Similar to the other metrics, in this case, strong
It is obvious that the metrics differ in their sensitivity. Therefore, an optimization of the weighting coefficients is done. The goal is to increase the robustness of the overall evaluation based on training data. The final grade can be rewritten in parameter form
where x1…x4 are the individual grades for every metric and β1…β4 are the corresponding weighting factors. Figure 13 shows the optimization scheme.
FIGURE 13: Weighting factors optimization and metrics performance evaluation.
The training data is produced with the coupled simulation environment and evaluated and graded by expert opinions. The quadratic error of the simulation data and the evaluation grade is minimized. The optimization problem can be stated as follows:
where X is a matrix containing the individual grades, β a parameter vector including the weighting coefficients, and y a vector with the VTE | 27
Feature | Technical
grades based on expert opinions and constraint vectors b1 = [0 ... 0]T, bu = [1 ... 1]T. The optimization is based on data from 836 scenarios in a training data set. This training data set consists of representative critical and uncritical traffic scenarios created by a traffic simulation tool. The traffic density in the training data varies over a band-width from low to high. For example, scenarios with traffic congestions, the need for full stops on the highway, strong decelerations, and velocity fluctuations were generated. This data represents a set of possible traffic conditions and is used to evaluate the introduced traffic quality metrics. The binary threshold is determined by the socalled “receiver operating characteristic” (ROC) graphs and the related confusion matrix shown in Figure 14 [25].
FIGURE 14: Confusion matrix for the binary classification of critical scenarios.
It is desirable to have a “true-positive rate” (TPR) of 100% while keeping the false-positive rate low. Figure 15 shows the results of the optimization in an ROC graph.
FIGURE 15: Metrics performance results (training data).
It can be seen that the combined and optimized metric (Figure 15, black solid line) shows the best results. The FPR is around 10%, while the requirement of a 100% TPR is fulfilled. Considering a scenario where the ego-vehicle follows a leading vehicle on the same lane and the leading vehicle performs a braking maneuver causing the ego-vehicle to also brake, the root for the traffic quality decreasing behavior is triggered in this case by the leading vehicle, and therefore, the ego-vehicle is not responsible for the decrease in traffic quality. The metrics evaluation would still classify this scenario as critical, which results in a false-positive case. To prevent similar false-positive flagging, the FPR can be decreased even further by designing a 28 | June 2021
filter, which minimizes the false-positive cases by considering the velocity behavior of the leading vehicle in the immediate past and the gap changes between the ego and the leading vehicle. In some special cases, the use of a filter can result in an undesired omission of some true-positive cases. The decision if a filter should be used depends on the desired results. For the development phase, it is convenient to neglect some special cases in order to achieve a smaller FPR. In the validation process, every true-positive case has to be captured. The metrics can now be used to identify critical scenarios with the simulation-based toolchain and provide additional information on how the ego-vehicle automation functions influence traffic quality. Identification of Critical Scenarios The investigated automated driving function is an SAE Level 3 highway chauffeur [13, 26]. The highway chauffeur is able to perform standard driving tasks on highways such as entering/leaving the highway, overtaking, etc. The cooperative and automated functions are implemented on an Opel Insignia equipped with a sensor and actuator setup suitable for auto-mated driving tasks. The vehicle under test (VUT) possesses features like robust localization, path planning, maneuver prediction, driving strategy, etc. This VUT is embedded as a digital prototype in the simulation environment. The scenario of entering the highway with different exemplary disturbances is chosen as an example use case. It is worth mentioning that this methodology can also be used for other automated driving functions, use cases, and different spatial domains, for example, rural roads and urban areas. The first step in the toolchain is the definition of a logical scenario and its corresponding parameter spaces. It is obvious that a lot of different parameter spaces can be listed here. In order to keep a clear view only a part of possible, for this use case necessary, parameter spaces are shown in Table 2. Attribute
Parameter Determined space by example
Entrance ramp length Number of lanes Speed limit highway entrance Traffic flow
lmin - lmax Nmin - Nmax
410 m 4
vmin - vmax
36.1 m/s
Qmin - Qmax
1 veh/s
TABLE 2: Logical scenario: entering the highway.
Most parameters are determined by the static attributes of the highway entrance, which in this case is located at the Opel proving ground. Figure 16 shows the initial condition for the simulation runs investigated in this section. The traffic flow is set to a value, where a lot of
FIGURE 16: Initial conditions for the performed simulation runs.
traffic participants take part without causing a congestion before the ego-vehicle enters the highway. The behavior of other traffic participants is set to a normative way in the first three scenarios. In the fourth scenario, this behavior is varied. The cooperation simulation is neglected in this example, and therefore, the traffic and vehicle dynamics simulations are used for evaluation. The disturbances can be included in the logical scenario. Depending on the representation, it is difficult to represent every disturbance in Table 2. For example, it would be possible to vary the driving behavior of other traffic participants stepwise from defensive to aggressive. For a full definition of the scenario nomenclature used in this article, the reader is kindly referred to [11]. The evaluation process will be done by using the already introduced metrics for safety and traffic quality. In this case, the focus will be on the identification and evaluation of critical scenarios, where the corresponding thresholds for the criticality classification are Gfinal = 0.279, TTC = 3.9 s, TTB = 3.8 s, and areq = −2 m/s2 [16]. The verification of critical scenarios is not carried out here, and the example is solely based on a Model-in-the-Loop approach. The enhancement of further X-in-the-Loop, recorded data, and driving tests to the methodology is part of our current research. Based on the previous steps, it is possible to analyze the scenarios and prepare test results for the function development. The first scenario is carried out with no disturbance at all. The ego-vehicle enters the highway in a normative way. This scenario is chosen to illustrate that the metrics classify a so-called optimal behavior as not critical. The results are shown in Table 3. Scenario characteristics Results Disturbance Safety metrics Traffic metrics Criticality
None TTCcrit = ø, TTBcrit = ø, areq,crit = ø Gfinal = 0.15 Not critical
TABLE 3: Concrete scenario: entering the highway without disturbances.
As expected, the evaluation results show that the metrics do not deflect (indicated by the empty set ∅) and the corresponding conclusion is a negative criticality classification. In the toolchain, the scenario is not identified as critical and will be neglected. The first run-through the whole toolchain is therefore concluded. The second concrete scenario remains a highway-entry scenario affected by a sensor error disturbance causing the ego-vehicle’s trajectory-following controller to result in fluctuating behavior. To keep the examples sufficiently brief, the same logical scenario is chosen with the same specified attributes. Keeping that in mind, it is possible to carry out
Technical | Feature
the examples directly with the toolchain step “concrete scenario,” which is the input for the corresponding simulation environment. Table 4 shows the results of the simulated concrete Scenario Results characteristics Disturbance Safety metrics Traffic metrics Criticality
Sensor errors TTCcrit = 2.9 s, TTBcrit = 1.2 s, areq,crit = −6 m/s2 Gfinal = 0.29 Critical
TABLE 4: Concrete scenario: entering the highway with sensor errors.
scenario. Correctly, the classification result of the scenario is stated as critical. The traffic quality metrics respond to strong fluctuations of the ego-vehicle, and especially the individual metric takes this aspect into account. In this particular simulation run, the safety-related metrics responded as well. In general, this disturbance can lead to safety-critical behavior. Due to usage of the Model-in-the-Loop approach, it is possible to skip the next step and go directly to the analysis step. This critical scenario and the complete simulation data are saved in a database together with a test result description. The function development department can use this simulation data and test result documentation to improve their functionalities. Cooperative and automated vehicles use precision maps to increase their foresight compared to the operation of their sensor setup only. Precision maps can be used to develop driving strategies. Especially on highway entries, these maps provide helpful information, such as the distance to the end of the lane and the start of the dashed line allowing the vehicle to enter the highway. The third scenario is a consideration of what can happen when the information of the precision map is incorrect. The map thus provokes an error by not allowing our ego-vehicle to enter the highway directly when the dashed line begins, but 40 m later, shortening the possible length of the lane change from 140 m to 100 m. Table 5 contains Scenario characteristics Results Disturbance Safety metrics Traffic metrics Criticality
Map errors TTCcrit = ø, TTBcrit = ø, areq,crit = ø Gfinal = 0.46 Critical
TABLE 5: Concrete scenario: entering the highway with map errors.
the outcomes of this scenario. On account of the map error, the ego-vehicle does not enter the highway and performs a full stop on the entry ramp. The level-3 function demands that the driver takes control while the vehicle blocks the entry ramp. The traffic metrics also correctly classify this scenario as critical. Therefore, the results are saved and passed on to the function development department. The ego-vehicle caused no safety-related problem in this scenario. Therefore, the safety metrics did not deflect, which is also accurate. It is worth mentioning that this behavior has a very negative effect on customer acceptance of these systems. Thus, the reliability of the egowww.saea.com.au
vehicle functions can be considered as a key requirement for the release. The fourth scenario highlights the behavior of other traffic participants and their influence on automated driving functions. As already indicated in the introductory example, the driving behavior of other participants can maneuver the ego-vehicle into critical scenarios. Therefore, this driving behavior is changed to aggressive in the traffic simulation. The coupling allows for changing this parameter very easily by adjusting Scenario Results characteristics Disturbance Safety metrics Traffic metrics Criticality
Aggressive traffic participants TTCcrit = 1.9 s, TTBcrit = 1.2 s, areq,crit = −5.7 m/s2 Gfinal = 0.19 Critical
TABLE 6: Concrete scenario: entering the highway with aggressive traffic participants.
the driver models inside the coupled environment. The results can be seen in Table 6. It is obvious that the aggressive driving behavior influences the safety-related metrics very strongly. The values of each single metric fall below the critical threshold at a certain time in the simulation. These scenarios are dangerous and have to be tested further. In this simulation run, the traffic quality did not decrease significantly, which correctly led to an uncritical classification result regarding the traffic metrics. The use of the entire toolchain for different scenarios and the correct classification of the relevance of the scenarios prove that this methodology is able to identify critical scenarios. Normative driving behavior is correctly classified as not critical and, therefore, not saved and passed on to the function development department. Conclusions In this article, a simulation-based toolchain to identify and verify critical scenarios for cooperative and automated vehicles is introduced. The methodology is illustrated and demonstrated to work with different example scenarios containing sensor errors, map errors, and aggressive driving behavior of other traffic participants. Two different metrics are used for the classification, considering standard safety metrics and newly developed metrics to evaluate traffic quality. Critical scenarios are flagged and the corresponding data is passed on to the function development department as test results in order to improve the implemented auto-mated driving functions. The results presented in this article show an example for the realization of the generic toolchain and its capabilities. The combination of different metrics to evaluate traffic quality allows us to make a binary classification for the criticality of a concrete scenario. Instead of using every single traffic quality indicator separately, this approach focuses on finding one final grade with a criticality threshold for the overall identification process. Due to that fact, by calculating the final grade and comparing this value with the threshold determined by the optimization scheme using training data, it is
possible to decide whether a concrete scenario is critical or not. This approach can be adopted for other metrics as well, for example, driver comfort, fuel efficiency, and driving foresight, just to name a few. Even though the results presented here are examples, the capabilities of the toolchain to identify critical scenarios independent of the implemented driving functions, different simulation tools, the application-specific environ-ment, and the use of various metrics derived from the chosen requirements are demonstrated. The proposed methodology is limited due to the imper-fections of current simulation tools as well as of the simulation models. Especially parameters used for the vehicle dynamics including the sensors are highly sensitive to various real-world conditions, for example, weather, road surface proper-ties, etc. To counter these limitations, we are currently devel-oping a so-called Prototype-in-the-Loop method, where the vehicle dynamics simulation is replaced by a real-world coop-erative and automated vehicle. This vehicle is interacting continuously, via both a static and a dynamic coupling, with the traffic simulation while driving on a real-world proving ground. Regarding future activities, the methodology will be extended with X-in-the-Loop methods, benches, and driving tests to prove the plausibility with a real test vehicle. Such an approach enables a high degree of awareness whether the simulation results are sufficiently robust and provides insight into the interferences and uncertainties of the used mathematical models. The influences of these uncertainties will be investigated further. A proposition can be made about which model precision is required in what part of the V-Model. References 1. Hallerbach, S., Eberle, U., and Köster, F., “Absicherungs- und Bewertungsmethoden für kooperative und hochautomatisierte Fahrzeuge,” AAET - Automatisiertes und vernetztes Fahren, Brunswick, Germany, Feb. 8-9, 2017, 368-384, ISBN:978-3-93765541-3. 2. Koopman, P. and Wagner, M., “Autonomous Vehicle Safety: An Interdisciplinary Challenge,” IEEE Intelligent Transportation Systems Magazine 9(1):90-96, 2017, doi:10.1109/MITS.2016.2583491. 3. Pütz, A., Zlocki, A., Küfen, J., Bock, J. et al., “Database Approach for the Sign-Off Process of Highly Automated Vehicles,” 25th Enhanced Safety of Vehicles Conference (ESV), Detroit, 2017. 4. General Motors, “2018 Self-Driving Safety Report,” 2018, https://www.gm.com/ content/dam/gm/en_us/english/selfdriving/ gmsafetyreport.pdf. 5. Ulbrich, S., Menzel, T., Reschka, A., Schuldt, F. et al., “Defining and Substantiating the Terms Scene, Situation, and Scenario for Automated Driving,” IEEE 18th International Conference on Intelligent Transportation Systems, Canary Islands, Spain, 2015, doi:10.1109/ITSC.2015.164. VTE | 29
Feature | Technical
6. Menzel, T., Bagschik, G., and Maurer M., “Scenarios for Development, Test and Validation of Automated Vehicles,” 2018 IEEE Intelligent Vehicles Symposium, Changshu, China, 2018, https://arxiv.org/ abs/1801.08598. 7. Baidu Artificial Intelligence, “Apollo Simulation,” accessed July 27, 2018, http:// apollo.auto/platform/simulation.html. 8. Nvidia Cooperation, “Nvidia Drive Constellation,” accessed July 27, 2018, https://www.nvidia.com/en-us/self-drivingcars/drive-constellation/. 9. Maurer, M., Gerdes, J.C., Lenz, B., and Winner, H., Autonomes Fahren - Technische, rechtliche und gesellschaftliche Aspekte Teil 4 Sicherheit (Berlin, Germany: Springer Verlag GmbH, 2015), 454-458, doi:10.1007/978-3-662-45854-9. 10. Hallerbach, S., Eberle, U., and Köster, F., “The Challenges of Releasing Cooperative and Highly Automated Vehicles - A Look beyond Functional Requirements,” AmE - Automotive meets Electronics, GMM-Fachbericht, VDE, Dortmund Germany, Mar. 07-08, 2017, 102106, ISBN:978-3-8007-4369-8. 11. Bagschik, G., Menzel, T., Reschka, A., and Maurer M, “Szenarien für Entwicklung, Absicherung und Test von automatisierten Fahrzeugen,” 11. Workshop Fahrerassistenzsysteme und automatisiertes Fahren, FAS 2017, UniDAS e.V., Walting, Germany, Mar. 29-31, 2017, 125-135, ISBN:978-3-00-055656-2. A related English-language article “Scenarios for Development, Test and Validation of Automated Vehicles” by the Braunschweig Group is to be found at: https://arxiv. org/ abs/1801.08598. 12. Winner, H., Hakuli, S., Lotz, F., and Singer, C., Handbuch der Fahrerassistenzsysteme-
Grundlagen, Komponenten und Systeme für aktive Sicherheit und Komfort Third Edition (Wiesbaden: Springer Fachmedien, 2015), 128-132, doi:10.1007/978-3-658-05734-3. 13. Bartels, A., Eberle, U., and Knapp, A., “Deliverable D2.1. System Classification and Glossary,” Adaptive Consortium, Wolfsburg, Germany, Feb. 6, 2015, 63. 14. Dupuis, M. et al., “OpenDrive Format Specification, Rev. 1.4,” VIRESSimulationstechnologie GmbH, Nov. 4, 2015. 15. Schubert, R., Richter, E., and Gerd, W., “Comparison and Evaluation of Advanced Motion Models for Vehicle Tracking,” 11th International Conference on Information, June 30-July 3, 2008, doi:10.1109/ ICIF.2008.4632283. 16. Junietz, P., Schneider, J., and Winner, H., “Metrik zur Bewertung der Kritikalität von Verkehrssituationen und - szenarien,” 11. Workshop Fahrerassistenzsysteme und automatisiertes Fahren, FAS 2017, Uni-DAS e.V., Walting, Germany, Mar. 29-31, 2017, 149-160, ISBN:978-3-00-055656-2. 17. Hayward, J.C., “Near Miss Determination through Use of a Scale of Danger,” 51st Annual Meeting of the Highway Research Board, Washington, DC, Jan. 17-21, 1972. 18. Hillenbrand, J., Kroschel, K., and Schmid, V., “Situation Assessment Algorithm for a Collision Prevention Assistant,” in Proceedings of Intelligent Vehicles Symposium, 2005, IEEE, Las Vegas, June 6-8, 2005, 459-465, doi:10.1109/ IVS.2005.1505146. 19. Karlsson, R., Jansson, J., and Gustafsson F., “Model-Based Statistical Tracking and Decision Making for Collision Avoidance Application,” Proceeding of the 2004 American Control Conference, Boston, June 30-July 2, 2004, 3435-3440, ISBN:0-7803-8335-4.
20. Augustin, D. and Hallerbach, S., “InteractionAware Motion Prediction for HighlyAutomated Driving Function on Highways,” AmE - Automotive meets Electronics, GMM-Fachbericht, VDE, Dortmund Germany, 127-132, Mar. 7-8, 2017, ISBN:978-3-80074369-8. 21. Lefèvre, S., Vasquez, D., and Laugier, C., “A Survey on Motion Prediction and Risk Assessment for Intelligent Vehicles,” ROBOMECH Journal 1(1), 2014, doi:10.1186/ s40648-014-0001-z. 22. Transport Research Board- National Research Council, Highway Capacity Manual (Washington, DC, 2000). ISBN:0- 30906681-6. 23. Zhu, W., Boriboonsomsin, K., and Barth, M., “Microscopic Traffic Flow Quality of Service from the Drivers’ Point of View,” Proceedings of the 2007 IEEE, Intelligent Transportation Systems Conference, Seattle, WA, Sept. 30-Oct. 3, 2007, 47-52, doi:10.1109/ ITSC.2007.4357790. 24. Ko, J., Guensler, R., and Hunter, M., “Variability in Traffic Flow Quality Experienced by Drivers: Evidence from Instrumented Vehicles,” Transportation Research Record: Journal of the Transportation Research Board, No. 1988, Transportation Research Board of the National Academies, Washington, DC, 1-9, 2006, doi:10.3141/1988-02. 25. Fawcett, T., “An Introduction to ROC Analysis,” Pattern Recognition Letters 27:861-874, Dec. 19, 2005, doi:10.1016/j. patrec.2005.10.010. 26. SAE-International, “Taxonomy and Definitions for Terms Related to On-Road Motor Vehicle Automated Driving Systems,” SAE Standard J3016.201401, Jan. 16, 2014.
Nomenclature Symbol/acronym Description
Unit
Symbol/acronym Description
Unit
TTC
Time to collision
s
σvCircle,j
Nanoscopic velocity standard deviation
m/s
TTB
Time to brake
s
v vCircle,j
Nanoscopic mean velocity of every vehicle m/s
a req
Required deceleration
m/s
LOSt
Macroscopic grade
Δp
Difference of vehicle positions
m
v ego
Ego-vehicle velocity
m/s
CVref , v ref , DVref Reference values for normalization v ref , σa,ref
n.a., m/s, n.a., m/s, m/s2
v obj
Object velocity
m/s
Relative velocity between vehicles
m/s
Mean velocity of every vehicle with respect to time interval
m/s
v rel
v
a ego,max
Maximum deceleration of the ego-vehicle
m/s2
v ego
Average velocity of the ego-vehicle with respect to time interval
m/s
a obj
Acceleration of the object
m/s2
σa
Individual acceleration standard deviation m/s2
D
Traffic density
veh/m/lane
n.a.
vp
Traffic flow rate
veh/s/lane
Gfinal , Gmac , Gmic Normalized traffic quality grades Gnan , Gind
S
Average travel velocity
m/s
X
Matrix of grades evaluated by metrics
n.a.
CVj
Microscopic coefficient of variation
n.a.
β
Parameter vector
n.a.
σvj
Microscopic velocity standard deviation
m/s
y
Vector of grades evaluated by experts
n.a.
vj
Microscopic mean velocity of every vehicle m/s
bl
Vector of lower bound constraints
n.a.
DVj
Nanoscopic coefficient of variation
bu
Vector of upper bound constraints
n.a.
30 | June 2021
2
n.a.
n.a.