Task Planning of Manipulator Based on Dynamic Space Constraint and Torque Sensor

Abstract

It is a main direction of industrial application that manipulator performs grasping task in task space and combines obstacle avoidance planning. This paper presents an improved RRTConnect path planning algorithm IRRTC based on dynamic constrained space and admittance control. The sliding constraints on the Cartesian position space is used to search for the reachable path of the traction robot by the algorithm. For orientation control, a priori path is formed by artificially given constraint interval, and then the dynamic admittance control is modified by using the end grasping load moment estimated by the joint torque sensor of the manipulator as the constraint. At the same time, the algorithm designs a local escape algorithm based on the current state and target state of the robotic arm, which overcomes the problem that the traditional RRTConnect and RRT cannot escape the local minimum. The simulation and experimental results of the Xmate3 redundant manipulator show that a feasible path can be quickly searched based on constraints and local escape strategies, and a relatively optimal trajectory can be planned based on the original path by the optimized path and also meet the constraints. Within 3S, the average planning success rate is increased to over 90%, and without time constraint, the success rate is 100%, the efficiency is higher than other algorithms. Because the admittance control is used in conjunction with the joint torque feedback to maintain the dynamic correction of the end orientation, which improves the stability of the manipulator. The algorithm can be artificially designed to solve the obstacle avoidance and task constraints in disorderly grasping.