Evolutionary Minimum Cost Trajectory Planning for Industrial Robots

Abstract

A general method for computing minimum cost trajectory planning for industrial robot manipulators is presented. The aim is minimization of a cost function with constraints namely joint positions, velocities, jerks and torques by considering dynamic equations of motion. A clamped cubic spline curve is used to represent the trajectory. This is a non-linear constrained optimization problem with five objective functions, 30 constraints and 144 variables. The cost function is a weighted balance of transfer time, mean average of actuators efforts and power, singularity avoidance, joint jerks and joint accelerations. The problem is solved by two evolutionary techniques such as Elitist Non-dominated Sorting Genetic Algorithm (NSGA-II) and Differential Evolution (DE). Numerical applications for a six link robotic manipulator – STANFORD robot (pick and place operation) and a two link planar manipulator (motion in the presence of obstacles) are illustrated. The results obtained from the Proposed techniques (NSGA-II and DE) are compared for different values of weighting coefficients. The influences of the algorithm parameters and weight factors on algorithm performance are analyzed. The DE algorithm converges quickly than NSGA-II. Also DE algorithm gives better results than NSGA-II in majority of cases. A comprehensive user-friendly general-purpose software package has been developed using VC++ to obtain the optimal solutions of any complex problem using DE algorithm.