Path Tracking Control Of Snake-like Robot Based On Neural Network

In recent years,snake-like robots have attracted more and more attention because of their easy camouflage,high concealment,compact structure and high integration.Because of their under-actuated,strong coupling and strong non-linearity,snake-like robots have attracted more and more attention in the field of control.Joint angle tracking and path tracking of snake-like robot are the premise and foundation of its autonomous motion,obstacle avoidance motion and target detection in complex environment.Therefore,the research on joint angle tracking and path tracking of snake-like robot has important theoretical and practical value.Firstly,according to the actual mechanical structure design of snake-like robot,the dynamic equation of snake-like robot with separated driven wheels and eccentricity of single link mass is deduced.The physical model of the snake-like robot is obtained by abstracting the mechanics of the actual model.Considering the separation friction and eccentricity effect,the dynamic equation is established based on the viscous friction model,and the driving force term in the dynamic equation are decoupled.Secondly,aiming at snake-like robot's winding gait,the research of snake-like robot's joint angle tracking and straight-line path tracking is performed.In this paper,the dynamic equations are separated from the driving ones and the non-driving ones.The snake-like robot joint angle tracking is based on local state feedback-PD controller,the snake-like robot linear path tracking is based on RBF neural network identifier-PI controller and the line-of-sight guidance law are improved for the snake-like robot lateral speed and linear path tracking speed.At the same time,the performance of control algorithm under external disturbance is tested.Thirdly,aiming at eel-like gait of snake-like robot,the joint angle tracking and linear path tracking of snake-like robot are studied.In order to solve the problem of joint angle tracking and path tracking instability of snake-like robot in eel-like gait,we use synovial controller to track joint angle,and neural network identifier-PI controller with non-linear term to track target path.Then Check the performance of the controller under disturbance.Fourthly,the research on snake-like robot curve path tracking is studied.Based on the piecewise guidance law,the snake-like robot curve path tracking is carried out to speed up the tracking speed of snake-like robot in the phase of approaching the target curve.At the same time,considering the guidance law switching surface,and the switching surface function is designed based on the idea of filtering,so that the snake-like robot can transit smoothly when the guidance law is switched.Finally,considering the influence of uncertain friction moment between joints on the system,an adaptive neural sliding mode controller is used to track the snake-like robot's linear path.Based on Lyapunov stability condition,the adaptive law is deduced.The RBF neural network is used to approximate the uncertainties of the friction moment between joints in the dynamic equation.Linear path tracking of snake-like robot is carried out based on inner-loop sliding mode controller and outer-loop neural network identifier-PI control.Then Check the performance of the controller under disturbance.The adaptive algorithm is extended to curve path tracking,and then we combine the piecewise guidance law to complete the curve path tracking...