Online Motion Modification for Mobile Manipulators Based on Sequential Quadratic Programming

Le résumé fourni par la source

This paper describes a method of motion planning for mobile manipulators. The target task is to reach out and grasp an object on a table or a shelf. We especially focus on two things: (i) online motion modification, which is for avoiding obstacles that were newly detected during executing motion, and (ii) grasping motion generation, because the closer the reaching process is to the end, the more careful the robot's motion generation should be. We formulate these issues using Sequential Quadratic Programming (SQP) and connect them with our previous motion planner. The planner is a retrieval-based algorithm and utilizes Swept Volumes (SVs). Therefore, we view the motion planning problem as a combination of existing path search and generating transition motions. We also consider the allocation of processing time and present a method that allows online motion modification without stopping the robot motion for replanning. Experiments on simulation environments were conducted, and we confirmed that our unified motion planning is useful for not only global motion planning, but also online motion modification and grasping motion generation.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.