Intelligent trajectory optimization in industrial manipulators by PSO algorithm

Abstract

This work presents an initial approach toward an intelligent methodology for trajectory optimization in industrial manipulators, based on the metaheuristic algorithm Particle Swarm Optimization (PSO). The evaluation is carried out in a MATLAB simulation environment that incorporates a dynamic model of the robot, including mass, volume, and inertia of the links, as well as obstacle representation and collision detection. Three cost functions are proposed and compared, each considering different physical performance metrics: (i) total trajectory time, (ii) joint effort, estimated through the average and maximum torques applied, and (iii) a weighted combination of both criteria. The optimization follows a staged strategy: it first ensures geometric feasibility by eliminating trajectories with collisions, and then refines the solution by optimizing the physical metrics. The results show that the proposed optimization technique leads to more efficient and realistic trajectories, with higher potential for deployment in demanding industrial environments.