Previous state-of-the-art motion planning methods employ a ‘hub and spoke’ approach, using precomputed graphs of a finite number of fixed configurations, which are known to be safe. Motion planning using Graph-of-Convex-Sets (GCS) enables robots to easily adapt to different configurations within precomputed convex regions — allowing the robot to ‘round the corner’ as it makes its motion plans. By blending graph search and convex optimization, GCS can find paths through intricate environments and simultaneously optimize the robot trajectory. Moving precisely across each vertex in large graphs and high-dimensional spaces, GCS had clear potential in robotic motion planning. The team is also exploring applications of GCS trajectory optimization to robot task and motion planning.
Feb 20 2023