Robust and topology-aware motion planning with moving obstacles

A novel robust solution to motion planning in the presence of moving obstacles is presented in this paper. The proposed method addresses the uncertainty in the robot state estimate, as well as in the estimate of the a-priori unknown obstacles motion. Our scheme employs the globally optimal state-feedback velocity field for static workspaces and exploits its topological properties to enable moving obstacles avoidance. In contrast to previous relevant contributions, our method tackles polygonal obstacles and accounts for both the obstacles geometry and the approximate motion profile to extract a reactive solution. Furthermore, the proposed kinematic planner is extended to a class of second-order systems. Our multi-layered approach integrates robust safety filtering and exhibits superior performance w.r.t. relevant methods. In summary, this work consists an important step towards bridging the gap between open-loop planning and state-feedback control.