Research On Locomotion Control On A Quadruped Robot With Bound Gait Based On Model Predictive Control

Quadruped robots have comprehensive advantages over biped or other multi-legged robots in terms of stability,loading capacity and movement speed.Kinematic performance has always been an important aspect of quadruped robot research.However,due to the characteristics of short touchdown time,strong nonlinearity of the model and underactuation,the high speed motion of quadruped robot becomes the key and difficult point to improve its motion performance.When the traditional control method is adopted to solve the motion control problem of the four-legged bionic robot,the theoretical derivation is complicated,and when there are uncertainties caused by model mismatch,distortion,interference and other factors,the improvement of control effect becomes very difficult.In recent years,integral control based on simple model has become a hot topic.Among them,model predictive control(MPC)is the most prominent.The research shows that the model prediction method has a low requirement for model accuracy,and the system is robust and stable,which provides a good prospect for solving the balance problem of dynamic robot.In this paper,a control framework based on model prediction is proposed to solve the high-speed motion control problem of quadruped robot.(1)it is not accurate to use a single information to estimate the ontology state because IMU will drift in application and the kinematic solution will be interfered by the shock.In order to improve the accuracy of estimation and the effect of control.This paper presents an estimation method of multi-sensor fusion.By using the rolling time domain estimation(MHE)method,inertial navigation and kinematic information are fused to make up their respective shortcomings.Finally,the simulation results show the effectiveness of the proposed method.(2)in this paper,based on the rigid body floating model of the quadruped robot,MPC controller is designed for the Bound gait,and the foot end position and contact force trajectory of the robot are automatically generated based on the planned centroid trajectory.The feasibility and rationality of this method were verified by one-leg experiment.(3)in order to improve the solving speed of MPC,iLQG algorithm with faster convergence speed is selected to optimize the solution.Compared with SQP,the solution speed is improved by about 10 times,and the lost accuracy is also within the acceptable range.