基于扩张状态观测器和模型预测方法的四足机器人抗干扰复合控制

Anti-disturbance Composite Controller Design of Quadruped Robot Based on Extended State Observer and Model Predictive Control Technique

为提高存在模型不确定及外界扰动的情况下对四足机器人的控制效果,提出结合扩张状态观测器、二次型优化(Quadratic Programming,QP)及模型预测控制(Model Predictive Control,MPC)的复合控制算法。基于单刚体四足机器人模型,设计基于MPC的控制算法,实现对四足机器人的控制。考虑到模型及参数的不确定性,以及环境中存在的外界扰动,进一步利用变带宽非线性扩张状态观测器对机器人受到的总和扰动进行估计,并利用估计结果,设计基于QP的足端力重分配算法,实现对总和扰动的补偿。通过仿真实验验证了本文所提控制算法的有效性。仿真结果表明,在使用抗干扰复合控制方法之后,机器人在质量发生重大变化和受到外力作用下,依然可以保持良好的控制效果,与传统控制方法相比,机器人的控制精度和抗干扰能力得到了明显提升。

A novel composite control algorithm that combines the extended state observer,quadratic programming and model predictive control is proposed to improve the control performance of quadruped robot under model uncertainty and external disturbances.A model predictive controller is proposed for a quadruped robot based on the single rigid body model,and a variable bandwidth nonlinear extended state observer is developed to estimate the lumped disturbance,including the model mismatch and external forces.Based on the estimated result,a compensator is further constructed using the quadratic programming technique.The proposed control algorithm is validated through simulation.The simulated results demonstrate that the anti-disturbance composite controller is used to allow the robot to achieve the satisfactory control performance under the conditions of the changes in mass and the application of external forces.In comparison to existing studies,the proposed controller exhibits significant improvements in control accuracy and disturbance rejection capabilities.