Research On Dynamic Modeling And Control Of Quadruped Robot Based On Spatial Vectors

Quadruped robot is more robust to tranvse rough and unstructured terrain than wheeled and crawler robot and it have showed amazing application potential.However,quadruped robot is a floating-base rigid-body system with high Do Fs which it is difficult to model and control.To solve these problems above,this thesis mainly studies the dynamic modeling and control theory of quadruped robot.The main research contents are as follows:Firstly,based on the common configurations of quadruped robot at home and abroad,a structure scheme of quadruped robot with linkage transmission is proposed.Based on the analysis of the Recursive Newton-Euler Algorithm,a more efficient Recursive Newton-Euler Algorithm based on spatial vectors is studied,and the inverse dynamics model of the quadruped robot is established by fusing the composite rigid body algorithm.Combining Model Predictive Control,Quadratic Programming and Whole-body Control,a control framework based on inverse dynamics is proposed.Among them,the Model Predictive Controller and Quadratic Programming provide the Whole-body Controller with estimated contact reaction forces;The Whole-body Controller enables a strict task hierarchy and the controller uses the full-body dynamics to calculate the joint torque.The dynamic model and control framework are implemented in the open source simulation environment to verify the accuracy of full-body dynamics and the robustness of the control framework.Based on the research above,an experimental prototype of quadruped robot was developed.Carried out multiple sets of foot position experiments and walking comparison experiments with or without a Whole-body Controller,verified the effectiveness and robustness of the control framework.The dynamic modeling method and control framework of quadruped robot based on spatial vectors proposed in this paper effectively realized the dynamic modeling and robust control of quadruped robot,which provides an important reference and experimental basis for the dynamic modeling and control method application of quadruped robot.