Research On Dynamic Walking And Control Method Of 3-D Bipedal Robot

Due to the human-like structure and walking style,bipedal robots are able to adapt to human's living and working environments,and they are expected to replace human beings to complete various operations.With the support of the National Natural Fund Projects,'Multi-source random uncertainty modeling,intelligent control and dynamic simulation experiment on biped robot(No.11372270)','Gait transition,multiobjective optimization and robust control on random uncertainty of 3-D biped robot(No.11772292)',and 'Dynamic modeling and whole-body control of humanoid robot with rigid-flexible-soft structure(No.91748126)',this dissertation was aimed at improving the gait planning efficiency,developing asymptotical stabilization and transition control strategies,and achieving robust control of 3-D dynamic walking.To achieve this goal,this dissertation performed research on the dynamic walking and control methods of 3-D bipedal robot.This dissertation is organized as follows.In Chapter 1,the background and significance of the study of bipedal robots was introduced and the development of the research on the gait planning,control methods,gait transition control methods,and robust control strategies of 3-D bipedal robots was summarized.By analyzing the important problems that need to be further studied,the content of this study was proposed at last.The second chapter introduced the preliminary theories and methods that would be used in the study of 3-D dynamic walking,including dynamic modeling method,the virtual constraint approach,the periodic gait planning procedure,the hybrid zero dynamics and Poincare map method,and the event-based control method.Chapter 3 studied the gait planning of 3-D dynamic walking.First,the quasi-periodic gait was planned based on the virtual constraint approach and nonlinear constrained optimization.Second,the virtual constraints of the quasi-periodic gait were parameterized and the mathematical relationship between the periodicity constraint and the parameters was established,and then the gait modification method was proposed based on the Newton-Raphson algorithm.Finally,by analyzing the sufficient conditions for the iterative equations of the gait modification,a novel modification strategy for the quasi-periodic gait was proposed.In Chapter 4,the asymptotical stabilization of the dynamic walking of a 3-D bipedal robot with asymmetric structure was studied.First,by using heuristic state variables as control variables,a heuristic motion controller was proposed based on the virtual constraint approach and the hybrid zero dynamics.This motion controller was designed to asymptotically drive the state of the robot to the zero dynamics manifold.Second,to render the hybrid zero dynamics locally asymptotically stable,an event-based controller was designed in each continuous phase of the asymmetric 3-D biped.By establishing an explicit relationship between the control goal and the controller parameters in each continuous phase,an analytical method for the design of the controller parameters was proposed.Chapter 5 studied the gait transition control of 3-D dynamic walking.First,based on the Poincare map method,an event-based controller with adaptive feedback gains was proposed to achieve asymptotical stabilization of the target gait.To design the even-based controller,the mathematical relationships between the control goal and the control parameters was first established.Second,a transition controller was designed to guide the robot from the current gait to a neighboring point of the target gait.To guarantee the physical constraints for the gait transition,the control parameters of the transition controller was obtained by solving the following optimization.In Chapter 6,the robust control of 3-D dynamic walking subject to uncertain disturbances was studied.First,the dynamic model for the 3-D biped under uncertain disturbances was established,and the stability analysis of the robot system was studied.Second,based on the sliding mode control method,an adaptive sliding mode controller was proposed.Finally,the adaptive sliding mode controller was further extended into the robust control 3-D dynamic walking on uneven terrains.For this goal,an invariance condition for impact velocity was presented and the joint trajectories were planned online to control the landing velocity.In Chapter 7,the simulations and experiments were performed based on the biped GTX-?,including the experiments of bipedal walking subject to external disturbances and bipedal walking on uneven terrain.First,the mechnical structure and control system of GTX-? was introduced.Second,the virtual prototype of GTX-III was established based on ADAMS.Then,the joint trajectories were planned based on the landing-velocity control,and the torso angles were chosen to be the control variables.Conclusions of this work were summarized in Chapter 8,and in which some advanced topics for future work could also be found.