Online replanning for reactive robot motion: Practical aspects

摘要

We address practical issues to develop reactive motion planning method capable of replanning the path online when the environment changes during the execution. By introducing planning and execution threads running in parallel, the robot can keep moving even during the replanning process as long as the motion is safe. The proposed method can be applied to discontinuous input of environmental changes that may occur due to incomplete perception. We then address a roadmap reuse method for efficient replanning to make use of the increasing knowledge about the environment, by introducing working and learning roadmaps. A general interface with robot motion controller is also defined so that the method can be applied to various types of robots. The proposed method is validated through planning simulations with moving obstacles.