Probabilistic and Empirical Grounded Modeling of Agents in Partial Cooperative (Traffic) Scenarios

Abstract

Traffic scenarios can be regarded as problem situations with one or more (partial) cooperative problem solvers. Solving such a problem successfully requires (nonverbal) communication and distributed cognition. This is especially true when traffic is deregulated as in the shared space concept. Each traffic agent has a set of beliefs concerning his state, his goals, and the future behavior of the other agents. Risky but fortunately rare maneuvers can occur anytime. We call these risky maneuvers anomalies (eg. Verletzung der Vorfahrt, riskantes Auffahren, Einscheren, Überholen, ...). An experienced car driver is able to anticipate these anomalies. Novices and handicapped persons do not. It is expected that assistance systems could enhance the situation awareness and the communication competencies of unskilled or non-cooperative drivers. The design challenge for intelligent assistance systems is to model the behavior of traffic agents and diagnose these anomalies. We propose probabilistic models for these challenges.