Online Verification of Cognitive Car Decisions

Abstract

Verifying a safe locomotion of cognitive cars is indispensable for their participation in road traffic. This paper suggests an approach for verifying safety by computing reachable sets for the position of relevant traffic participants, i.e. a cognitive car as well as moving objects in its environment. In order to account for the uncertainty in the behavior of traffic participants, a stochastic setting is chosen, in which Markov chains represent the positions probabilistically. An efficient online algorithm is presented that leads to the result whether the reachable sets of different traffic participants can intersect, meaning that the control strategy of the cognitive car is possibly unsafe.