Research And Realization On Locomotion Control Technology Of Quadruped Robot

Quadrupedal mammals, as a species with high evolution and rich characteristics, possess various advantages in locomotion performance, such as high dynamic property, strong terrain-adaptivity, impressive stability, outstanding heavy-load capacity, etc. They almost can reach all kinds of terrains on earth. Based on such facts, increasing numbers of researchers begin to study on bionics of quadrupedal mammals so as to design and develop a mimic quadrupedal robot which can imitate quadrupedal mammals’ outstanding athlectic abilities and be applied to applications, such as military detection, resource exploratioin, goods transpotation, disaster rescue, etc. Essentially, the quadrupedal robot is a blending product bridging multi-disciplines, multi-majors and multi-domains, whose research contents include mechanism design, gait planning, power driven, locomotive transmittion, sensor measuring, signal processing, decision making, system control, etc. Most of these contents are really popular and difficult. This thesis discusses those technology challenges we will face during quadrupedal robots are operating, which include multi-joint coordination control, dynamic stability control and energy consumption control of power system. Specifically, several key technologies of a quadrupedal robot is systematically and deeply discussed including bionic mechanism design, locomotion control theories and methods, construction of motion control system and hydraulic servo system design, which will lay the foundation of robot whole system optimization and further related researches.According to previous researches in related fields, we carried out bionic studies on several kinds of quadrupedal mammals, concluded valuable bionic laws in designing quadrupedal mechanism, and then analysed and compared several robot control methods so as to obtain control laws with simplicity and validity. Aiming at designing a supported quadrupedal robot, we designed mechanism parameters such as number of femora, tilt angles of joints, positions to deploy hydraulic cylinders and operating strokes; established kinematic model and simplified dynamic model(based on dynamic balance) of this hydraulic quadrupedal robot with all-inward limbs of 2 femora each; proposed a mimic locomotion control system scheme for the robot, which integrated hierarchical control scheme and distributed control scheme; explicitly stated the system framework and control goal of the power and electrohydraulic-servo system; constructed the simulation model platform and coordinated the simulation model of such quadrupedal kinematic, dynamic and control systems.Supported by above researches, a quadrupedal trot-gait dynamic model was built and laid the theorectical foundation for the existence of the optimal stride with lowest energy consumption; multi-cycloid and Bezier curve distal trajectory planning were planned to obtain zero-impact performance during distal commutation and flexibility in parameter setting respectively; analyzed and compared(based on lots of simulation experiments) both distal trajectories in aspects of velocity, stability and ernegy consumption, and their impacts of variation in gaits, strides, periods and leg-lifting heights on the robot; studied the parameter setting of optimal energy consumption in the condition of ensuring walking stability of the robot. Additionally, a velocity-based gait transition control strategy is proposed in accordance to the conclusion that the optimal stride of a quadrupedal robot varies along with different gaits.On account of drawbacks of model-based quadrupedal gait control methods, such as modeling complexity, single period planning and bad performance in adaptability, we here focused on distal trajectory planning and analyzed the positional relation between the center of gravity(COG) and each distal; designed conversion between stationary and motion status and gait transition in consideration of both bionic observation and CPG gait transition theory; developed a control method of which stride and period keeped changing continuously. Meanwhile, we discussed the impact on the robot brought by the variation of adhesive force and gradient and surface elevation and also the related strategy to solve these problems; calculated the relation between variations of joint equilibrium position according to slope motion law of quadrupedal mammals. In order to improve walking efficiency, we here used properties of Bezier curve and took slipping into account to broaden its workspace for walking, designed the control strategies for crossing obstacles. Besides, we amended distal trajectory in order to ensure that it walks smoothly.Additionally, we carefully chose and constructed electro-hydraulic servo system, used simulation experiments to analyze its traits comparing to the field of traditional servo control and designed a fuzzy adaptive PID control method with forward compensation and the related controller which guaranteed similar bang-bang control effect when big error occurred. We applied a trajectory planning interpolation method using quantic spline curves to guarantee the continuous smoothness of velocity and acceleration of hydraulic valves. Finally, we tested the trot gait using multi-cycloid trajectory planning on the quadrupedal prototype and results met our design objectives. Key problems solved in the thesis may show some hints and be valuable to related researches on quadrupedal robotics.