Trajectory Optimization of Industrial Welding Robots based on Artificial Hummingbird Algorithm

Abstract

In industrial welding scenarios, the smoothness of welding trajectories, uniformity of energy distribution, and control of thermal deformation directly determine weld quality and production efficiency. Traditional trajectory planning algorithms struggle to balance robustness, real-time performance, and executability under multi-objective constraints, exhibiting problems such as unstable convergence and a tendency to fall into local optima. This paper proposes the application of the Artificial Hummingbird Algorithm (AHA) and its improved variants to the trajectory optimization of industrial welding robots, constructing a multi-objective optimization framework to achieve closed-loop optimization of weld feature perception - trajectory generation - execution. By introducing chaos mapping in the population initialization phase, the global search capability and convergence stability of the algorithm are enhanced. A simulation model is built based on the Matlab platform, combining visual perception and robot kinematic models to model and optimize high-dimensional multi-objective constraints (smoothness, heat input, energy distribution, etc.). The research results show that the improved Artificial Hummingbird Algorithm can effectively enhance the smoothness of welding trajectories and the uniformity of energy distribution, reduce thermal deformation and residual stress, improve the consistency of weld quality and production efficiency, and provide an efficient and concise trajectory optimization scheme for industrial welding in complex environments.