Locomotion Control Strategy Of Optimal Feet Forces Allocation For Quadruped Robot Based On Virtual Model

In recent years,mobile robots have made great strides in delivery,rescue,exploration and military applications.Especially in the wake of the COVID-19 outbreak,mobile robots have gained prominence in areas such as disinfection,food delivery and express delivery.At present,wheeled,tracked and bionic foot mobile robots are primary types which are being widely used and studied.Compared with wheeled and tracked mobile robots,bionic foot robots have better adaptability to complex terrain because of their discrete contact with the environment.As a multi-input,multi-output and time-varying system,it is still difficult for a quadruped robot to achieve flexible and stable movement in complex terrain.This paper mainly focused on how to improve the environmental adaptability and locomotion stability of quadruped robot,and it mainly includes the following aspects:Firstly,a simplified structure model of the quadruped robot was established based on the actual physical prototype platform.Based on the simplified model,the kinematics model of the quadruped robot was established by D-H method.The forward and inverse kinematics solutions were solved,and at the same time the Jacobian matrix was calculated.The dynamic equation of the robot is established based on Lagrange equation method,which is the theoretical basis of dynamic feedforward part in below.Secondly,the characteristics of several common gaits of quadruped robot were analyzed,and the trot gait and walk gait are especially planned.Aiming at the different movement characteristics of the supporting phase and swinging phase of quadruped robot,the expected trajectory of the torso and swinging legs were planned respectively.The ground information was perceived according to the position information of the landing legs,and the terrain information was analyzed by the least square method.According to different motion environment and state requirements,trot and walk gait were switched to ensure the flexible and stable motion of the robot.Then,the overall motion control architecture of the quadruped robot was presented.In the support phase,the virtual spring and damping were constructed in the three-dimensional movement direction and rotation direction of the body centroid,and the required virtual forces were calculated according to the motion demand of the body centroid,and the virtual forces and resistance distribution problem was transformed into a quadratic programming problem(QP)to optimize the real-time distribution of the feet forces.In the swing phase,the virtual forces were calculated according to the planned swing trajectory of the feet,and the dynamic feedforward was considered to realize the forces control of the swing legs.Finally,the control algorithm of the robot is simulated and tested.Webots and VS2019 co-simulation platform are established to carry out omnidirectional movement test of the robot under diagonal gait,comparison before and after optimal allocation of feet forces,walking on flat ground and slope and other simulation experiments.The feasibility of the proposed motion control strategy was verified by observing and analyzing the body pose information,forward movement velocity,joint torques and feet forces.The omnidirectional movement test and walking experiment on flat ground and slope were carried out on the prototype experimental platform,which further verified the feasibility and stability of the proposed motion control method.