Integrating Combinatorial Reasoning and Continuous Methods for Optimal Motion Planning of Autonomous Vehicles

Motion planning involves deciding among combinatorial options. Local, continuous methods can yield global optimal solutions only in absence of such options. In order to attain global optimal solutions in real world traffic, integration of combinatorial reasoning into continuous methods is required. This work introduces the necessary constraints on the planner to restrain the motion into distinct combinations and compares the results of complete planning and path-velocity decomposed planning.