Priority-Based Multitask Coordinated Control Of Quadruped Robot

Bionic quadruped robot has attracted more and more researchers’ attention in recent years because of its strong flexible movement ability and complex terrain adaptability.The motion process of quadruped robot is a process of multiple tasks executing at the same time.From the perspective of motion and force,it can be divided into constraint task,speed task,attitude task,torque task and so on;From the perspective of body,it can be divided into trunk task,supporting leg task,swinging leg task and so on;From the perspective of coordinate system,it can be divided into tasks under body coordinate system and tasks under world coordinate system.Due to the highly redundant characteristics of quadruped robot,it is possible to coordinate and complete multiple motion tasks.Taking the 12 DOF bionic motor-driven quadruped robot as the research object,aiming at the problem of motion planning and multitask coordinated control of quadruped robot running in a straight line at a fixed speed on the flat ground with a diagonal trot gait,this thesis studied the motion planning algorithm and coordinated control algorithm of quadruped robot based on position mode.The main work is as follows:1.Aiming at the problem that the traditional kinematics model is difficult to solve the problem of multitask coordination,this thesis established a differential kinematics model which can uniformly express all motion tasks by using the form of differential equations based on Jacobian matrix.The model consists of two parts: leg differential kinematics,which describes the relationship between leg motion state and joint driving,is used to express the planned motion task in the body coordinate system;The wholebody differential kinematics,which describes the relationship between the whole-body motion state(legs,body)and joint driving,is used to express the planned motion tasks in the world coordinate system2.Aiming at the control goal,the motion planning of quadruped robot is divided into three parts: gait,foot trajectory and attitude adjustment.Based on the movement of biological dogs,various motion parameters of diagonal trot gait were planned;The compound cycloid was optimized from three aspects of position,velocity and acceleration to obtain a continuous,smooth and mutation free trajectory of the foot;Based on the standing standard posture of the robot,the posture description system and the planning posture adjustment algorithm were established.Finally,the independent effectiveness of the three algorithms was verified by simulation experiments.3.When the three independently planned motions of gait,foot trajectory and posture adjustment were combined to solve the joint drive,there were task conflict and had no solution.Taking minimizing the 2-norm error of the overall task as the optimization goal,this paper uniformly expressed each motion task into the differential kinematics model,and used the least square approximate solution as the joint drive.The simulation results showed that this method enabled the quadruped robot to perform speed tasks and attitude tasks at the same time.4.For the control aiming at minimizing the 2-norm error of the overall task,there were some problems,such as compromise of all task accuracy and potential safety hazards.The multitask coordination strategy based on priority was adopted to classify the motion task into three tasks with decreasing priority: safety task,speed task and attitude task.Then,using the method of dimension reduction,all tasks were expressed in the differential kinematics equation of the unified dimension,and the concept of null space in the least square is used to project the low priority tasks into the high priority null space for execution,so as to complete the low priority tasks as much as possible without affecting the accuracy of the high priority tasks.Through simulation and experiment,the safe and effective control of quadruped robot was realized.