Design And Gait Planning Of Biped Robot Based On Seven-link Model

The biped robot has a structure similar to that of human beings,and its unique bipedal walking mode can move well in complex or dangerous environments that wheeled robots are difficult to deal with,and complete tasks on behalf of humans.And rescue and other fields have a wide range of applications.However,due to the complex structure of the lower limbs of the human body,the degrees of freedom of most bipedal robots do not match the human body,the gait is too rigid,the human shape is poor,and the walking is unstable.Therefore,improving the anthropomorphism and stability of biped robots in mechanical structure and gait planning has become the core issue of biped robot research.This thesis focuses on the design and gait planning of a biped robot based on the seven-link model.The main work of the thesis is as follows:Firstly,design and implementation of biped robot based on seven-link model.Based on the analysis of humanoid robot requirements,the configuration of the degrees of freedom of the physical object model of the biped robot is carried out based on the seven-link model;The hardware system of the seven-link biped robot is designed and built,and the sensing module is equipped with a pressure sensor and a pose sensor,which improves the ability of the biped robot to interact with the environment;completed the component selection of each module of the biped robot;completed the design of its mechanical structure based on Solidworks;assemble the seven-link model and debug the module function;the simulation model is built in the V-REP simulation experiment environment,and the hardware and simulation research platform is established for the subsequent research on human-like gait planning.Secondly,the gait planning based on the CPG-ZMP hybrid walking control method is realized.The rhythm signal is generated by the highly bionic CPG(Central Pattern Generator)control method to drive the humanoid walking of the biped robot,a CPG network topology for walking control of a seven-link bipedal robot consisting of10 DOF oscillators and a central oscillator is built;aiming at the problem of poor anti-disturbance ability and walking stability of biped robot in CPG walking control method,the ZMP stability criterion is combined with CPG,when CPG generates the joint angle signal required for biped robot walking to drive the robot to walk,use the pressure sensor of the physical object model of the biped robot to obtain the position of the zero moment point in real time,and compare it with the expected ZMP point,and generate the corresponding joint angle compensation through the difference,forming a closed-loop control structure;using Genetic Algorithm to Optimise and Determine the Parameters in the Closed-loop Control Structure;the simulation results show that the CPG-ZMP hybrid walking control method can improve the instability of the robot’s walking in the CPG control method;the physical experiment results show that the CPG-ZMP hybrid walking control method can make the seven-link biped robot walk stably.Thirdly,CPG walking direction control is realized based on DDPG(Deep Deterministic Policy Gradient).Aiming at the deviation of the walking direction of the biped robot due to errors such as ground sliding,by analyzing the way the human body adjusts the walking direction during the walking process,added hip yaw angle oscillator needed to adjust orientation in the CPG network;introduce DDPG reinforcement learning network,design its Actor module and Critic module network structure,determine the reward function of DDPG network,the network uses the robot torso state and position information obtained by the pose sensor during the robot’s walking process as input,outputs variable weight gain,and controls the hip joint yaw angle servo of the biped robot by combining with the rhythm signal of the CPG.The angle of rotation changes the walking direction;complete the training of the reinforcement learning network;the simulation results show that the CPG walking direction control method based on DDPG can effectively improve the problem of walking direction deviation during the walking process of the seven-link biped robot.This thesis focuses on the design of the biped robot based on the seven-link model and its humanoid gait planning problem.The physical object model of the biped robot is designed,and the control method is designed based on CPG to ensure its humanoid and improve the walking stability.It has a certain reference value for the humanoid research of biped robot in mechanical structure and gait planning.