A Motion Planning Algorithm for Redundant Manipulators Using Rapidly Exploring Randomized Trees and Artificial Potential Fields

Anoush Sepehri,Amirreza Mirbeygi Moghaddam

doi:10.1109/access.2021.3056397

Abstract

This paper presents a novel motion planner for redundant robotic manipulators by utilizing rapidly exploring randomized trees and artificial potential fields. Rapidly exploring randomized trees and artificial potential fields are two well-known navigational strategies in the robotics industry; however, their potential benefits and synergy when implemented together has been largely unexplored. In the proposed motion planner, rapidly exploring randomized trees is first used to determine a suitable path for the end-effector to follow that maneuvers around all obstacles in the robot's workspace. Once a path has been determined, attractive and repulsive potential fields are implemented at all points along the path and are used in a gradient optimization algorithm to determine joint trajectories to reach the desired location. To supplement the attractive and repulsive potential fields, an orientation field is proposed to minimize the error between the actual end-effector orientation and the desired orientation during joint trajectory planning. The motion planner is examined through both analytical and experimental evaluation using the 7 degrees of freedom Kinova JACO and Kinova Gen3 robotic arms. The conclusions drawn from this work substantiate the efficacy and superiority of the proposed two-stage motion planner for the control of redundant manipulators in obstacle-ridden environments.

Highlights

Robotic manipulators have proven to be extremely valuable at improving the efficiency and automation in many of today’s tasks
Artificial potential fields have been investigated for threedimensional motion planning of robotic manipulators; their applications have been limited to environments that only contain point obstacles, and when the orientation of the end-effector is not of interest [16]–[18]
SIMULATION RESULTS The proposed motion planner is tested in a kinematic simulation of a 7-DOF Kinova JACO robotic arm to evaluate its validity and performance

Summary

INTRODUCTION

Robotic manipulators have proven to be extremely valuable at improving the efficiency and automation in many of today’s tasks. Artificial potential fields have been investigated for threedimensional motion planning of robotic manipulators; their applications have been limited to environments that only contain point obstacles, and when the orientation of the end-effector is not of interest [16]–[18]. This paper proposes a novel motion planning algorithm that explores the strengths and benefits of combining rapidly exploring randomized trees for end-effector path planning, with artificial potential fields for joint configuration planning. Utilizing rapidly exploring randomized trees to supplement the artificial potential fields for motion planning eliminates the occurrence of local minima, demonstrating that a sampling-based algorithm and an optimization-based algorithm can be combined to exhibit synergistic benefits that would otherwise be difficult to achieve

DEVELOPMENT OF MOTION PLANNER

CASE STUDY III

EXPERIMENTAL RESULTS

CONCLUSION