Comparison of motion planners in an environment with removable obstacles

This work deals with the problem of motion planning for a robotic system with two arms, considering the possibility of using one arm to remove potential obstacles in order to get a collision-free path to reach a desired object with the other arm. The paper compares different motion planning algorithms based on random sampling methods. In the used framework the planners do not discard the samples that imply collision with removable objects, instead these samples are classified according to the obstacles that produce a collision and then processed to decided whether it is necessary to remove these obstacles to get a proper free path. The efficiency of the motion planners are compared by solving planning problems in three different scenarios.