Decision making for automated driving at unsignalized intersection

As automated vehicles begin operating on complex urban roads, precise decision making for automated driving has been increasingly important for safe automated driving. In particular, decision making at unsignalized intersections is one of the most challenging problems of automated urban driving. This paper presents intention-aware automated driving at unsignalized intersections. The intention of the traffic participant is modeled as a Dynamic Bayesian Network (DBN). Given the inference result, an intention-aware decision-making problem is modeled as a Partially Observable Markov Decision Process (POMDP), which is regarded as one of the most widely used models for sequential decision-making problems under uncertain environments. We implemented the proposed system in a passenger car, and the effectiveness of the proposed algorithm is evaluated through experiments at unsignalized intersections on our university campus road.