Model-Free Control For Cable-driven Continuum Robots

With the development of modern intelligent technology,efficient intelligent robots are designed and applied.The cable-driven continuum robot inspired by elephant trunks and vines is a kind of bionic robot,which has superior flexibility and safety.Therefore,there are broad application prospects in medical services,post-disaster rescue and aerospace fields.However,the cable-driven continuum robots are difficult to accurately model.Meanwhile,they must not only overcome the influence of nonlinear uncertainties and disturbances,but also have high requirements for energy consumption and stability when completing tasks.So it is of great significance to design model-free control strategy combined with the optimal control principle.The main research contents and results of this paper are as follows:(1)A dynamic modeling method for cable-driven continuum robots is proposed.Under the assumption of piecewise constant curvature,the kinematics of an n-segment continuum robot is deduced from the geometric by matrix rotation and coordinate system transformation.Then combining the generalized force analysis and energy calculation,the dynamic model is completed by using the Euler-Lagrange equation.A dynamic model case and simulation verification analysis of a singlesegment cable-driven continuum robot are given.The work explores the shortcomings of the model and provides guidance for the design of modelfree controllers.(2)A continuous-time model-free optimal control strategy is proposed for the continuum robot system with linearization of the equilibrium point.The strategy is based on improved Q-learning which is combined with the Actor-Critic framework.Aiming at the problem that the system model is unknown,the Q function equivalent the optimal value function is introduced.The Critic network estimates the Q function and the Actor network approximates the system control strategy.Then the two networks are trained to update online by the gradient descent algorithm.Finally,the Lyapunov stability proof and simulation verification are carried out.(3)For nonlinear,time-varying cable-driven continuum robot systems,an optimal control strategy based on adaptive dynamic programming method is proposed for trajectory tracking.Firstly,an overall uncertainty term is defined to capture the unknown dynamics,and an auxiliary error function is introduced to transform the tracking problem into an optimal control problem.Secondly,based on the policy iteration method,an adaptive control algorithm including Actor-Critic network is designed to solve the Hamilton-Jacobi-Bellman(HJB)equation.The Critic network conducts policy evaluation,and the Actor network conducts policy improvement.And the synchronization optimization strategy of the two networks is given.Finally,the convergence to the optimal controller is proven,and the stability of the system is also guaranteed.The simulation experiment shows the feasibility of the algorithm.