Optimization Of Adaptability Posture For Quadruped Robot In Cant Environment

This master's thesis mainly studies the quadruped robot which can adapt to the slope and other unstructured environment,respectively from the leg knee joints elbow configuration,leg kinematics analysis and simulation of the quadruped robot,analysis of the unstructured terrain environment and the selection of foot drop point.The main goal is to make the quadruped robot realized stable walking under the complicated topography,for example,walking straight,vertical movement and obstacle avoidance.The main content concludes the following three parts:First of all,analysis of the quadruped robot leg kinematics: through analyzing the characteristic of the slope environment to draw each leg joint configuration of quadruped robot,optimizing the structure model of quadruped robot.Motion equations were established according to the difference between the forelegs and hind legs,and the kinematics analysis was respectively carried out on the forelegs and hind legs of the robot.Using the theory of D-H coordinate system,analyzing the verse and inverse kinematics problem of the quadruped robot.Secondly,walk quadruped robot kinematics simulation: the quadruped robot to walk on a flat slope of gait is divided into time intervals for first second to sixteenth frames,measuring each frame under the angle of each joint,fitting the joint angle,fitting equation of each joint are obtained.Using the SimMechanics toolbox in the MATLAB/Simulink to build the structure model of quadruped robot on the inclined plane,the joint rotation fitting equation into the corresponding simulation joints input module,respectively use the simulation in the form of tracking the leg rotation of each joint,record its endpoint to the coordinates of the position.By solving movement of the forelegs and hind legs of the robot to analysis the movement characteristics of the four feet robot.Thirdly,a quadruped-legged robot terrain analysis and the obstacle-surmounting gait: the ground of the analysis of unstructured environment robot may meet kinds of obstacles,including flat slope and a variety of shapes,sizes,bumps and potholes ranging from hard and soft strength.Using binocular vision sensor sensing terrain environment,the extracting of swinging can be reached when the obstacles of road information.Analyzing the road information,looking for suitable for the laying of points,in order to realize the laying of action.At last,each posture under the joint angle is used to established the initial population,eliminate the unreasonable sample characteristic,the joint angle of decision support under the posture of leg in the horizontal projection size as fitness function,training program,and optimize the posture under the joint angle,realizing the adaptation of the unstructured environment.