Pick-and-Place Motion Planning for Manipulators with Dual-Arms

Abstract

This paper proposes a method for planning pick-and-place motion by using dual-arm manipulators. In the offline phase of our algorithm, we model the surface of object and environment by using a set of planar regions. The contact between the object and the environment is established by selecting a cluster from both the object and the environment. In the online phase, our planner first generates multiple candidates of object pose stably placed on the environment. We then generate multiple motion strategies including the regrasping motion. The pick-and-place motion is generated by performing the grasp planning selecting one of the object pose and one of the motion strategies. We finally generate collision-free pick-and-place motion by using the PRM (Probabilistic Roadmap Method). The effectiveness of the proposed method is confirmed by simulation and experimental results using the dual-arm robot NX-HIRO.