Path tracking of snake-like robot based on neural network identifier

Abstract

Due to its under-actuating characteristics, the snake-like robot has become a research hotspot and a difficult point. Based on the model established by Pal Lilebak, a linear path tracking algorithm is proposed, which is based on neural network identifier and can automatically search PI parameters. At the same time, the line-of-sight guidance law with variable forward distance is used to speed up the side shift and track the target straight line. More over the algorithm is extended to curve path tracking. The simulation results show that this path tracking algorithm can stabilize the control system and speed up the tracking of the direction angle by dynamically adjusting PI parameters. The proposed line-of-sight guidance law can significantly accelerate the side shift speed. The result also show that the neural network identification control can avoid the problem that PI control will be unstable due to the change of the forward distance. And the path tracking algorithm also performs well in curve path tracking.