Real-Time Conflict Resolution of Task-Constrained Manipulator Motion in Unforeseen Dynamic Environments

摘要

This paper introduces conflict resolution in task-constrained real-time adaptive motion planning (RAMP) to enable a robot manipulator performing tasks in an environment with dynamically unknown obstacles. The method continuously improves and maintains diverse task constrained as well as unconstrained robot trajectories to allow the manipulator switching to a better trajectory at any time and seamlessly resolving conflicts between satisfying task constraints and avoiding dynamically unknown obstacles. If dynamic obstacles block all available task-constrained trajectories, the algorithm allows the manipulator to change goals on the fly to be free of task constraints and resume the task whenever there is a collision-free, task-constrained trajectory. The method is validated in different dynamic environments with different task constraints in both simulation and real-world experiments.