A strategy on situation evaluation for driver assistance systems in commercial vehicles considering pedestrians in urban traffic

Current developments of advanced driver assistance systems (ADAS) are concerned increasingly about the protection of vulnerable road users. This paper suggests an approach for the required, complex situation evaluation that decides if a detected pedestrian is relevant for the application. In order to account for the uncertainty in the behavior of the road users and the measurement uncertainties, their positions are derived probabilistically using a combination of Monte Carlo simulation, stochastic reachable sets and Markov chain abstraction. Prototype paths are assigned to the ego vehicle and the pedestrian movement with probabilities defined from previous measurement cycles. An efficient online algorithm computes the partial, conditioned collision probability of the ego vehicle with a pedestrian by intersecting their stochastic reachable sets and considering the path probabilities.