Multi-objective particle swarm optimization algorithm for trajectory planning of obstacle avoidance in dual-arm robots

ObjectiveTo address the problems of a single-objective optimization easily falling into local optima and the insufficient coordination between the collision detection and the planning in the obstacle avoidance trajectory planning for dual-arm robots, an adaptive multi-objective particle swarm optimization algorithm was proposed.MethodsThe kinematic model was established by using the D-H parameter method. Trajectory planning in the joint space was carried out by means of the quintic polynomial interpolation. Accurate collision detection was achieved by combining the oriented bounding box method, the separating axis theorem, and continuous collision detection. The path length, collision risk, and motion smoothness were taken as optimization objectives. The weights and algorithm parameters were dynamically adjusted, and the final strategy was selected according to the environmental risk classification.ResultsThe results show that the adaptive multi-objective particle swarm optimization algorithm exhibits the fast convergence speed and the high precision. After optimization, the numbers of changes in the joint angular velocity and the angular acceleration are significantly reduced, and joint impact is obviously decreased. Under both no-load and load conditions, the end-effector position repeatability satisfies the accuracy requirements, and all joint velocities remain within the safe constraint ranges. The proposed algorithm achieves an upgrade from “post-collision replanning” to “active prevention”, effectively reducing joint energy consumption and damage risk, and improving motion smoothness and service life.