Decision And Behavior Planning For a Self-driving Vehicle At Unsignalized Intersections

When a self-driving vehicle has to pass through complicated unsignalized intersections, it is important to decide whether yield the right-of-way to a target vehicle or not and continuously plan a speed in order to sure everything goes safely and efficiently. This paper proposes a series of formulas to calculate suitable speed to cross an unsignalized intersection. It uses the map library built from the real roads and then creates the same scene in the simulation software. It tests the reaction of the self-driving vehicle in the scenario by driving the target vehicle with human in the simulation software. Decisions at unsignalized intersections need to estimate indicators of safety and efficiency, and how self-driving vehicles interact with human-driven vehicles is worth discussing.