Research On The Movement Control Of The Quadruped Robot Based On Model Predictive Control

While wheeled robots are common,legged robots,with their flexibility and adaptability to the environment,are widely used in post-disaster rescue,field exploration,and military fields.Moreover,it has become one of the most popular research fields in the robotics industry.However,breakthroughs in motion control for quadruped robots have only been made in recent years,so in this paper we will look at the following aspects of quadruped robots.Leg kinematics modeling and foot-end trajectory planning for quadruped robots.The positive kinematic model of the quadruped robot's leg between the joint variables and the Cartesian coordinates of the foot end was established by the D-H method,and the inverse kinematic solution of the leg was obtained by the closed-form solution;Then refer to the Bezier curve to plan the swing phase trajectory of the foot end according to the characteristics of the robot,so that the quadruped has better motion characteristics in the swing phase.Research on leg impedance control.Based on the generalized impedance control expression,two different specific control methods can be deduced: admittance control and impedance control.The admittance control is that the amount of position control is obtained based on the deviation between the desired moment and the actual moment,which is then entered into the leg motor.Whereas,impedance control is where the amount of torque control is obtained based on the deviation between the desired position and the actual position,which is then fed to the leg motor.Both are achieved through dynamic control of the relationship between force and position for the purpose of compliance control.In addition,for the desired forces in impedance control,which may vary on different ground surfaces,a system identification method is used to estimate ground stiffness,which is then applied to impedance control.attitude stabilization control in the standing state and movement control based on model predictive control.The quadruped robot is first treated as a parallel mechanism in the standing state,the IMU data is read and the leg vector is solved by vector,then the attitude control is realized by inverse kinematics;the movement control based on model predictive control is divided into three control layers: the foot position planning controller,the trajectory generator and the impedance controller.The input of the foot position planning controller is the desired speed,the output is the next foot position which is also the input of the trajectory generator,the output of the trajectory generator layer is the foot end Cartesian coordinates based on the hip coordinate system at each moment,this position is input to the impedance control layer and subsequently the output of the impedance control layer controls the motion of the quadruped robot's legs.Simulation and experimental research of quadruped robot.In response to the above,the correctness of forward and inverse kinematics,that narrowly defined impedance control is faster,more flexible,and more stable than admittance control,and that ground system identification can achieve the purpose of distinguishing between different materials was experimentally verified,respectively.Attitude stabilization control was then achieved with the better PID parameters determined by simulation.Through simulation,movement control of the quadruped robot was accomplished using model predictive control algorithm to achieve trajectory following with good stability for different desired speeds.