Bipedal Robot Fast Walking Based On Force Control

Due to the complex mechanical structure,high modeling difficulty and strong nonlinearity of biped robots,the existing motion control methods of biped robots cannot meet the requirements of rapidity and stability.This paper studies the rapid walking movement of the biped robot,and the two kinds of walking control methods are set up in the force control mode.Attitude controller,height controller,speed controller and steering strategy of the biped robot are designed by decoupling control,and the centroid state multivariable controller is designed based on the model predictive control and feedback control methods.Finally,the fast motion and anti-disturbance experiments are carried out on the WUKONG-? biped robot.The main researches and achievements of this paper are as follows:1)The system platform of biped robot is introduced and a simplified robot model is established.The mechanical design and electrical composition of the biped robot WUKONG-? are introduced,which can provide hardware support for the subsequent walking experiments.A simplified model of a biped robot with multiple joints and links is established.The kinematics and link dynamics of the biped robot are calculated and analyzed.2)Aiming at the fast and stable walking motion of biped robot,the state machine of periodic walking motion of biped robot is defined,and the attitude controller,height controller and velocity controller of biped robot are designed based on decoupling control.The attitude controller adopts the method of system identification to obtain the feedforward torque of attitude control.The height controller uses the vertical ground reaction force to control the height of the robot mass center.The velocity controller uses Kalman filter to estimate the centroid velocity by integrating the kinematics and accelerometer information,and uses the leg placement of the swing leg to control the centroid velocity.Finally,the WUKONG-? biped robot can realize the stable walking on the outdoor lawn with the maximum walking speed of 0.7m/s.3)In order to improve the walking speed and maintain the stability of the attitude of biped robot,a model predictive control method based on ground reaction force is proposed to realize the multivariable control of the center of mass state of the robot.Firstly,the prediction model of the robot centroid state is built,and the optimization function in the finite time domain is designed by introducing the constraint condition of ground anti-skidding.Secondly,the optimization function is transformed into a quadratic programming problem to solve the ground reaction force in real time.Finally,the ground reaction force is converted to the joint torque of the supporting leg.Finally,the experiment is carried out on the WUKONG-III biped robot,which achieved the maximum walking speed of 0.9m/s and effectively improved the effect of attitude control.