Research On Locomotion Stability Control And Parameter Optimization Of A Quadruped Robot

In recent years,the bionic legged robot has become a popular and difficult research topic in the field of mobile robot research,because bionic legged robots outperform other types of robots in many ways.For instance,the quadruped robot which mimics quadruped mammal can move at high speed with high flexibility and stability and can be deployed for either civil use in the field of material transportation,disaster relief and so on or military use such as battle reconnaissance and cooperative engagement with soldiers.Therefore,it gets the favor and attention of researchers.At the same time,the similarity between the structure of the quadruped robot and that of common tetrapods like cat,dog,and horse,makes it easier for people to get familiar with the quadruped robot.Thus,it could further lead to a broad application prospect of quadruped robots to human society.Hence,carrying out researches on the quadruped robot has both important theoretical and realistic significance.As an important direction in the study of quadruped robot,to enhance and improve the adaptability of quadruped robot in the complex environment and to boost the locomotion ability of quadruped robot in all kinds of environment has always been the research hotspot and difficulty.This thesis takes a hydraulic-driven medium-sized quadruped robot as the research object and focuses on the research and exploration in its locomotion stability control methods and key parameter optimization problems while the robot using different gaits in different environments and through different terrains.Considering carrying capacity is one of the design objectives for the quadruped robot,the key parameters of quadruped robot leg structure are designed and optimizatized,and the joint configuration form is set.The thesis details the design ideas,structure characteristics of the limb mechanical structure and the structural scheme,working mechanism of the hydraulic drive system and introduces the physical prototype after the completion of system integration.The robot kinematics model and the body motion planning model are established and served as the basis of motion control of the robot.Introducing comprehensive evaluation of the robot motion performance,two key parameters the stride length and the stride height of the Walk gait are analyzed and optimized.The results have guiding significance for the selection of key gait parameters under different locomotion conditions.Based on particle swarm optimization(PSO)algorithm,key parameters of trajectory planning of the center of gravity(COG)are optimized for a newly presented Walk gait with continuous horizontal adjustment of the COG.Applying this Walk gait,the quadruped robot could move with speed and efficiency,and the stability of the locomotion is guaranteed.To improve the locomotion stability of the quadruped robot on irregular terrains.The influence of foot-end trajectories with different local geometric features and velocity features on the performance of the quadruped robot walking on the surface with low coefficient of friction are studied and analyzed with amount of experiments.Thus,the stability control strategy is obtained to maintain the transverse locomotion stability of the robot moving on the ground with a low friction coefficient.In view of dynamic stability margin theory,the stability control parameters of the quadruped robot trotting on a slope is designed and numerical analyzed.Furthermore,by introducing the stability locomotion parameters of trotting on a slope,locomotion of continuously climbing stairs using Trot gait is realized.The stability control methods of the Trotting quadruped robot when encountering lateral impact and non-symmetric load distribution are studied respectively.Based on N-Step capturability theory,the lateral impact stability control framework merged with Yaw and Pitch cooperative control scheme is designed to achieve lateral locomotion stabilization after a lateral impact and meanwhile to restrain the offcourse movement.Based on inverse velocity kinematics,a position and posture feedback controller of Trot gait with four-legged support phase is established.By adjusting the position and posture during the four-legged support phase and introducing position and posture of the robot body into the swing leg control In this way the impact of the rolling moment created by asymmetric load distribution on the Trot locomotion is reduced effectively so that the trot locomotion stability is maintained.Inspired by the control method and control effect of the virtual model control theory which is applied to restrain rolling of the robot using Trot gait,an active lateral control method of the center of mass(COM)trajectory is introduced.According to the features of trotting locomotion on different directions,both bidirectional and unidirectional COM trajectory adjustment methods are adopted respectively.Combined with the variable height control strategy during stance phase,the active lateral control method will improve the stability of the trotting locomotion effectively.Reinforcement learning with the policy gradient method is further introduced to learn the approximately optimal lateral locomotion parameters.Preparing for the prototype experiments,a remote control software with realtime 3D motion monitoring function is developed.Then,experiments are conducted to validate the load ability of quadruped robot physical prototype,typical gaits locomotion ability,steering ability,slope locomotion ability and the ability to resist lateral impact.