Research On Kinematic Control Of Redundant Manipulator With Multi-objective Constraints

The application of robot is more and more extensive with the rapid development of robot technology.Whether it is space walking or deep-sea exploration,whether it is disaster relief or daily life at home,are all the places where robots play their role.In order to make the robot more suitable for the complex work scene,but also to ensure its own safety,kinematic control theory was born accordingly.The main researcher of this paper include the kinematic control algorithm of redundant manipulator and the multi-obj ective evolutionary algorithm based on the 7-DOF redundant humanoid arm developed in the laboratory.The research contents of this paper are as follows:1)Aiming at the inverse kinematics control of redundant manipulator,the concept of"subtask precision" is proposed on the basis of General-Weighted Least-Norm method,and the transform matrix of the original algorithm is modified to keep its full rank all the time,which improves the numerical stability of the original algorithm.In this paper,the selection of the weight matrix is improved to reduce the calculation and avoid the vibration of joint velocity.The simulation results on the 7-DOF redundant manipulator also prove that the improved algorithm can avoid the vibration of joint velocity when tracking a given trajectory and has better effectiveness.2)For the problem that the traditional kinematical control algorithm can not guarantee the accuracy of primary and secondary tasks at the same time,the multi-objective evolutionary algorithm is to be adopted to solve the problem.A feedback phase that simulates the spare-time learning phenomenon based on the teaching-learning-based optimization algorithm is introduced.Students should compare their score with the average class score to select an appropriate means for further improvement.Poorly performing students can learn from the teacher directly for rapid improvement,whereas high-performing students prefer to motivate themselves for reinforcement learning.Non-dominated sorting is incorporated to permit this heuristic to solve problems with several objective functions.The crowding distance calculation is adopted to maintain the diversity of the obtained solution set in a single run.The performance of the MOFTLBO algorithm is compared with three well-known algorithms by using 11 unconstrained benchmark test problems and 4 performance metrics.The qualitative and quantitative results indicate that the MOFTLBO algorithm can provide considerably competitive results and outperforms the other algorithms on convergence,uniformity and spread.3)In this paper,the multi-objective evolutionary algorithm is introduced into the kinematic control of the manipulator in order to balance the completion accuracy of the main task and the secondary task Multi-subtasks correspond to multiple optimization objectives.Taking each joint variable of the manipulator as the decision variable,and taking the range of joint angle and joint velocity of the manipulator as the constraint conditions as well as taking the manipulability metric and joint-limit-avoidance metric as the optimized objective,the mathematical model of multi-obj ective kinematics control of redundant manipulator can be established.The single objective experiment with the joint-limit-avoidance metric proves the excellent exploratory ability of the feedback phase mentioned in this paper.The experiment with the manipulability metric and joint-limit-avoidance metric as double object also proves the validity of the established mathematical model and proves that it is feasible to introduce the multi-objective evolutionary algorithm into the kinematics control of redundant manipulator.