Path Planning Optimization for Teaching and Playback Welding Robot

Path planning for the industrial robot plays an important role in the intelligent control of robot. Tradition strategies, including model-based methods and human taught based methods, find it is difficult to control manipulator intelligently and optically. Thus, it is hard to ensure the better performance and lower energy consumption even if the same welding task was executed repeatedly. A path planning optimization method was proposed to add learning ability to teaching and playback welding robot. The optimization was divided into the welding points sequence improvement and trajectory improvement, which was done both on-line and off-line. Points sequence optimization was modeled as TSP and was continuously improved by genetic algorithm based strategy, while the trajectory between two welding points was on-line improved by an try-and-error strategy where the robot try different trajectory from time to time so as to search a better plan. Simulation results verified that this control strategy reduced the time and energy cost as compared with the man-made fix-order sequence. Our method prevents the robot from the computation-intensive model-based control, and offers a convenient way for self-improvement on the basis of human teaching.