Aiming at the motion characteristics of welding robot and the obstacle avoidance problem of welding gun in the working process of welding robot, a path planning strategy for robot obstacle avoidance welding based on improved artificial bee colony algorithm is proposed. Firstly, aiming at the problems of traditional artificial bee colony algorithm, Lévy distribution was introduced into the process of detecting bees to find new honey source, instead of the random distribution process between 0 and 1. An improved artificial bee colony algorithm based on Lévy flight was developed and applied to the path planning problem of welding robot. Simulation experiments were carried out. The results show that the improved method can obtain the optimal obstacle avoidance path of welding robot, and the optimization speed is fast and the process is stable, which can be used to solve the problem of obstacle avoidance path planning of welding robot.