Robotic manipulators are entering unstructured environments, such as homes, offices, hospitals, and restaurants, where robots need to plan motions quickly while ensuring safety via obstacle avoidance. Motion planning in such settings is challenging in part because the robot must rely on real-world sensors such as laser scanners, RGBD sensors, or stereo reconstruction, which typically produce point clouds. In addition, enabling intuitive, interactive, and reactive user experiences requires that the robot generate plans of high quality as quickly as possible, without necessarily knowing in advance the maximum time allocatable to motion planning. Hence, motion planning in such settings should be implemented as an anytime algorithm, meaning the algorithm progressively improves its solution and can be interrupted at any time and return a valid solution.
展开▼