부산대학교 도서관

A highly restricted workspace of the robotic arm may hinder to perform safely any autonomous mobile manipulation task with planetary exploration rovers. To ensure mission safety as well as high efficiency, a coupled path and motion planner for mobile manipulation is presented in this work. First, a Fast Marching Method based path planner generates a safe trajectory to reach the goal vicinity, avoiding obstacles and non-traversable areas in the scenario. The path planner is able to control the final rover orientation to ensure that the goal is finally reachable by the arm. Second, a 3D Fast Marching Method based motion planner generates the arm joints motion profile, by creating a 3D tunnel-like cost volume surrounding the already computed rover base trajectory. This tunnel makes use of an offline-computed safe workspace of the manipulator, thus ensuring that no self-collision will occur during the planned motion. The presented algorithm has been tested with multiple simulation experiments, benchmarked with an off-the-shelf motion planning library, and validated in a field test campaign with the rover SherpaTT of DFKI Robotics Innovation Center. The tests consisted in SherpaTT approaching an interesting area on the scenario and performing a mobile manipulation sample scanning operation. These experiments have demonstrated that the proposed motion planner increases efficiency as well as ensures mission safety. This is thanks to, on the one hand, a coordinated base-arm movement that results in maximized efficiency in time terms, and, on the other hand, considering the manipulator workspace offline in the mobile manipulation motion planner to guarantee self-collision avoidance.