Optimal Trajectory Planning for Robotic Arm Based on Improved Dynamic Multi-Population Particle Swarm Optimization Algorithm

Abstract

In response to the problem of easy falling into local optima and low execution efficiency of the basic particle swarm optimization algorithm for 6-degree-of-freedom robots under kinematic constraints, a trajectory planning method based on an improved dynamic multi-population particle swarm optimization algorithm is proposed. According to the average fitness value, the population is divided into three subpopulations. The subpopulation with fitness values higher than the average is classified as the inferior group, while the subpopulation with fitness values lower than the average is classified as the superior group. An equal number of populations are selected from both to form a mixed group. The inferior group is updated using Gaussian mutation and mixed particles, while the superior group is updated using Levy flight and greedy strategies. The mixed group is updated using improved learning factors and inertia weights. Simulation results demonstrate that the improved dynamic multi-population particle swarm optimization algorithm enhances work efficiency and convergence speed, validating the feasibility and effectiveness of the algorithm.