Research On Gait Planning And Motion Control For Humanoid Robots Based On Optimization Methods

With the vigorous development of the field of legged bionic robots,humanoid robots have attracted much attention in related research fields as robots that can best adapt to human production and living environments and replace humans for work services.The humanoid robot integrates motion control,visual processing,artificial intelligence interaction and other technologies,which is a robot research direction with great development potential.Compared with general legged robots,the biggest feature of humanoid robots is that they have at least six degrees of freedom of legs and at least three degrees of freedom of arms,which makes humanoid robots more flexible in motion and larger in workspace.At the same time,the planar structure of the foot end can also ensure that the contact area between the robot and the ground is larger,so that it can maintain a more stable posture relative to the point foot.However,due to the high complexity of humanoid robots,how to ensure its stable motion control algorithm has become the focus of current research.Based on the overall motion characteristics of humanoid robots,this paper analyzes and models the structure of humanoid robots.Based on the optimization method,the gait planning and motion control algorithm of humanoid robots are studied mainly from the main body of motion control-leg.The main research contents are as follows:1.The overall structure analysis and whole body kinematics dynamics modeling of humanoid robot are completed,which lays a model foundation for joint control,gait planning and motion control algorithm of humanoid robot.Firstly,the joint topology of typical humanoid robot is analyzed,and its coordinate system and mathematical description framework are established.After determining the overall model framework of the robot,the kinematics analysis of each part is carried out,and the kinematics model of the torso to the head,arm and leg is established.Finally,the method of robot kinematics and dynamics parameter configuration based on KDL library is studied,and the dynamic model of humanoid robot is generated.2.The gait planning method based on model predictive control is studied,and the stable centroid and foot trajectory are optimized and generated for gait planning.Based on Intrinsically Stable Model Predictive Control(IS-MPC)algorithm,a humanoid robot motion planning framework is built to generate humanoid robot gait trajectory.The framework uses the dynamic extended linear inverted pendulum as the prediction model,the zero moment point(ZMP)speed as the control input,and the stability constraint is added to ensure that the generated centroid trajectory is bounded relative to the ZMP trajectory.Finally,the optimal gait planning obtained by solving the IS-MPC optimization problem sequence is processed by the kinematics controller to generate the desired humanoid robot joint angle in the walking state.3.The motion control method based on quadratic programming to optimize the plantar force is studied.Based on the gait planning of the humanoid robot,the plantar force optimization is added to improve the terrain adaptability and motion stability of the robot.The swing phase and support phase controllers of the robot are designed respectively.In the support phase,firstly,the virtual mapping model from the center of mass to the support leg is established,and the quadratic programming under the constraint of the full-dimensional friction cone is used to optimize the plantar force of the support leg.Finally,the support phase is obtained by servo control of joint torque through inverse dynamics.The swing phase is achieved by the servo joint torque following the stable swing trajectory in the workspace.In order to improve the response speed and accuracy of the robot joint,a dynamic feedforward compensation is added to achieve a more stable torque control.4.The algorithm is implemented on the robot platform,and its effectiveness and reliability are verified by experiments.Firstly,the software framework of ROBOTIS OP3 humanoid robot platform is sorted out,its control framework and specific implementation process are analyzed,and the underlying software framework is modified to apply to the algorithm in this paper.Then the gait planning and motion control algorithm proposed in this paper is transplanted to the robot platform,and simulation and prototype experiments are carried out.The stable walking under the gait planning method is completed in the simulation environment and the prototype,which verifies the real-time and reliability of the gait trajectory generation.At the same time,the anti-disturbance experiment and stable walking under the quadratic programming optimization plantar force method were completed in the simulation environment,which verified the robustness of the method.