基于动力学模型预测控制的Stewart型轮足腿控制方法

为了提高轮足机器人Stewart型轮足腿在不同位姿条件下的精度和动态特性，提出将模型预测控制（model predictive control，MPC）应用于轮足腿的位姿跟踪.首先对Stewart平台进行受力分析，利用牛顿-欧拉方程建立动力学模型的状态空间表达式，建立MPC预测模型和优化目标函数；然后通过Adams和Matlab联合仿真对MPC控制器进行参数整定，检验其对未建模摩擦力的鲁棒性；最后通过对比实验验证MPC控制器在负载位姿变化条件下可以获得优于PI控制器的姿态跟踪精度和动态特性. 

Stewart Type Wheel-Foot Leg Control Method Based on Dynamics and Model Predictive Control

To improve the motion precision and dynamics of Stewart type wheel-foot leg under variable posture,a dynamics model predictive control (MPC) was proposed.Firstly,the force analysis was carried out for Stewart platform and a dynamics model was derived with the Newton-Euler for state-space representation.Then,the MPC predictive model and cost function were established for optimization.Next,Adams and Matlab joint simulation was performed to regulate the MPC controller and evaluate its robustness to the un-modeled friction.Finally,the comparative experiments were conducted to verify the superiority of MPC controller to PI controller in terms of motion dynamics and precision under variable posture.