Research On Walking Control Strategy For Static And Dynamic Gait Of A Quadruped Robot

Robotics is a multi-discipline technique,including mechanics,electronics,information,control and computer etc.It is currently the most prevalent area in both academic research and engineering applications.With decades of development,the robot moves towards commercialization from a science fiction concept and starts to serve humans.Among various robots,quadruped robots have many advantages,including high speed movement,flexibility,environment adaptability and stability,which result in a great application prospect of quadruped robots in the fields of material transportation,engineering exploration,emergency rescue and military inspection etc.Therefore,the research of the quadruped robot has important theoretical significance and practical value.Due to the advantages of hydraulic drive,including high response speed,power density and load capacity,the quadruped robots with high dynamic walking performance and load capacity have become an important research branch.This thesis takes a hydraulic quadruped robot as research subject,and focuses on four aspects: drive control,active compliance control,static/dynamic balance control and optimal learning decision control.(1)Electro-hydraulic servo valve control cylinder system is a basic drive unit of hydraulic quadruped robots,and its control performance directly affect whether the quadruped robot could track the desired trajectory or even affect the walking stability.Its system model is characterized by high nonlinearities,parameter uncertainties and model errors,and in the actual robot walking,there are also problems of sensor noises,external disturbances and load changes.To solve the above problems,a new design method of adaptive robust controller is proposed and the control performance of electro-hydraulic servo valve control cylinder system can be improved effectively.The method combines with a state observer,and the proposed controller is obtained by designing a control state-space expression with PID-like auxiliary control components.The auxiliary control components in each order system contains an adaptive term(proportional term)and a robust term(integral term),which are employed to address the parameter uncertainties and unknown terms of the order system,respectively.Meanwhile,the state observer is introduced to observe the state that is not easy to measure.Then,the Lyapunov method is utilized to prove the convergence and stability of system.Finally,comparative experiments are implemented to verify the effectiveness and advantages of proposed method.The method considers most system uncertainties and sensor noises,and it has certain universality and can be applied to other similar systems.(2)The stable walking is the premise for other tasks of quadruped robots.However,when quadruped robots walk on uneven terrains,the contact between foot end-effector of robot and ground will produce a great impact leading to the body shakes or even falls over.To solve the problem,an active compliance control in Descartes space based on force control in outer loop and position control in inner loop is investigated.The conditions for the robot to achieve the target impedance are deeply analyzed.Also,the stable impedance control for a single leg of hydraulic legged robots and environmental interaction stability are studied by virtual decomposition control technique.Meanwhile,to improve the transient transition performance of foot end-effector position and contact force when robot touches ground and reduce the tracking error of foot end-effector gradually,a novel fractional order impedance-like control model is proposed based on traditional impedance control model and it can improve active compliance control performance effectively.(3)Although the static walking speed of quadruped robots is slow,its stability is high and its ability to pass complex terrains is strong.Thus,three typical static gaits are studied and their advantages and disadvantages are analyzed,which put forward the exploratory gait with ability to perceive the ground.Under the condition that the robot is only with hydraulic cylinder displacement sensors and a body attitude sensor,the terrain estimation method and position/posture adjustment strategy when the robot walks on unknown ground environments are studied.The research can improve the environment adaptability and ability to pass unknown complex terrains effectively.Trot gait is the most commonly used gait in quadruped animals,and it has a related high speed and coordination.Therefore,trot gait is an important gait in quadruped robot gait planning.To improve the dynamic walking stability,disturbance rejection ability and complex environment adaptability of quadruped robot,a dynamic gait balance control algorithm combining virtual model control and three separate parts control method is proposed.On this basis,further researches about disturbance rejection adjustment control strategy,speed planning,trajectory tracking control and omni-directional walking planning are done,which lay the theoretical basis for high dynamic motion performance of quadruped robots.(4)Robot intelligence is an important trend of robot development in the future.In order to make the robot more intelligent,the relationships among terrain complexity index,stability,speed and cost of transport are deeply analyzed and an optimal learning decision control strategy is proposed.The optimal learning decision control strategy contains: 1)a gait parameters optimization method based on genetic algorithm and a position/posture adjustment optimization method based on particle swarm optimization algorithm are proposed to improve the robot static walking ability;2)a dynamic gait balance control strategy based on supervised learning is proposed to improve the robot dynamic walking ability;3)a path planning method based on convex optimization combined with a-star algorithm is proposed to find the best passable path in known environments;4)a gait autonomous selection method based on C4.5 decision tree algorithm is studied to make the robot choose the right gait and do the movement autonomously in unknown or untrained environments.The above researches have guiding significance and provide an insight for autonomous control and intelligence of quadruped robots.