Reinforcement Learning Based Motion Planning for Robotic Arm of Welding Robot

With the development of intelligent manufacturing and the expansion of industrial automation, welding robots have become an important production unit, and how to accurately plan the movement of the robotic arm of the welding robot is important for realizing efficient welding work, reduce energy consumption and improving production efficiency. This paper presents a novel approach using Reinforcement Learning to address these challenges by autonomously optimizing the motion planning of welding robots. We developed an improved Proximal Policy Optimization (PPO) algorithm, tailored specifically for the intricacies of robotic welding. Our method incorporates a custom-designed reward function that directly assesses weld quality and operational efficiency, promoting precise and efficient welding paths. We conducted simulation experiments to validate our method and compare it with other methods. The results show that our RL-based method not only achieves higher weld quality, but also improves the adaptability and efficiency of the robotic welding process.