ARTIGO TÉCNICO Ruben Benkmann, Uwe-Philipp Käppeler, Oliver Zweigle, Reinhard Lafrenz, Paul Levi Institute of Parallel and Distributed Systems (IPVS) University of Stuttgart
RESOLVING INCONSISTENCIES USING MULTI-AGENT SENSOR SYSTEMS ABSTRACT Agents acting in physical space use perception in combination with their own world models and shared context models to orient. The shared context models have to be adapted permanently to the conditions of the real world. If a measurement of an agent’s sensor does not fit to the corresponding data in the shared context model the system contains an inconsistency. In this case it is necessary to decide whether the reason for the discrepancy is a change in the real world or a measurement error. If there is a change in the real world the shared context model has to be corrected. A single agent can hardly answer this question using its local world model. This work describes the scenario of a context model that is shared with and updated by many agents that possess one or more sensors. In occurrence of an inconsistency it is possible to call other agents to validate a measurement. The functions to call the other agents are provided by the Nexus platform, a federation of systems that manages users and objects in shared dynamic context models. The study reported in this paper evaluates procedures that combine a multitude of measurements to a single result that can be integrated in the shared context model. The statistically optimized procedure based on ratings of the participating agents is enhanced using scaled weighted arithmetic means which prevents the system from running into singularities caused by the feedback from the ratings. The method is combined with an additional preprocessing based on fuzzy clustering that detects aberrant measurements which can be excluded from further processing.
I. MOTIVATION
II. NEXUS PLATFORM AND AGENT NEGOTIATION
The main objective of the Nexus Center of Excellence is the definition and realization of dynamic shared context models for context-aware applications. In this scope, issues concerning communication, information management, methods for model representation and sensor data integration are covered. Based on these digital world models, new innovative applications become possible, which can access information of the real world originating from sensors and additional, aggregated information. We currently witness the rapid proliferation of different kinds of sensor systems. These systems allow the acquisition of context information and make the integration of the sensor data an important research aspect. Open questions are which sensors are suitable for providing context information to the world model with as little redundancy as possible. The problem in updating the world models by sensor measurements is to reduce uncertainty and inconsistency.
This section gives an overview of the architecture of the nexus platform and the process of the agent negotiation to exchange sensor data. More details are described in [1]. The Nexus platform provides methods to detect inconsistencies between models, to call agents and start the negotiation about measurements.
If a local world model of an application or agent conflicts with data in the shared context model the system contains an inconsistency. It is hard to decide, whether the data of one single local world model based on sensor data is erroneous or if the shared world model is out of date and needs a correction. If the application needs to be sure about this specific information to work properly, the inconsistency needs to be resolved. One solution to this problem is to repeat the measurement with different sensors and to reduce uncertainty by redundancy. Methods to address other agents with access to sensors and to communicate the sensor data are provided by the Nexus Platform. The study described in this paper examines methods to statistically optimize the reduction of uncertainty where only a few measurements of corresponding physical values are available. This is necessary to make relevant contributions to the shared world model, maintaining a high degree of reliability while keeping costs and expenditure of time within an acceptable range.
[22]
robótica
The goal of the Nexus Platform is to support all kinds of context-aware applications by providing a shared global world-context model. To achieve this goal, the platform federates local context models from so-called context servers. The local models contain different types of context information: representations of real world objects like streets, rooms or persons and virtual objects that link to digital information spaces. Sensors keep the data of local context models up to date (e.g. the position of a person).
Figure 1 . Architecture of the Nexus Platform.