| The traditional rigid humanoid robot has a stiff gait and low energy utilization,and the collision with the ground will cause a large energy loss.There are many problems of the humanoid robot,such as too large impact force between the robot and the ground,unclear adjustment mechanism of joint stiffness,and low precision of flexible joint control.Therefore,it is of great significance to study the compliance of humanoid robots with variable stiffness joints.The main research contents of this paper include:(1)A variable stiffness ankle joint was designed for a humanoid robot,and the stiffness mathematical model of the variable stiffness joint was established.The stiffness characteristics were analyzed by ADAMS simulation.The humanoid robot has carried on the gait planning,the simulation,and the joint force analysis,has completed the position motor and the stiffness motor selection,finally has carried on the check to the key parts strength.(2)Based on the designed variable stiffness ankle joint,a mathematical model of variable stiffness ankle joint including parameter error term was established and verified by ADAMS.Abstract:Aiming at the problem that the position control accuracy of a humanoid robot with variable stiffness joint is reduced when the load and disturbance are constantly changing,the deficiencies of PID controller and BP-NN neural network PID controller are analyzed.A dynamic surface robust controller based on RBF is designed by using the dynamic surface method.RBF neural network with better approximation ability is used to approximate the uncertain and unknown nonlinear functions in the system online.In order to compensate for the unknown disturbance,the damping term was added to the control system to improve the accuracy of variable stiffness ankle joint position control.Finally,the effectiveness of the controller was verified by Simulink simulation.(3)Aiming at the vertical balance control problem of a humanoid robot with variable stiffness joints,an inverted pendulum model was established and linearized by using a simplified model method.The LQR controller is designed,and the validity of the proposed control method for the balance control of a flexible humanoid robot is verified by the software ADAMS.At the same time,the influence of the variation of joint stiffness on the upright balance recovery process of the humanoid robot is analyzed.According to the simulation results,a variable stiffness LQR balance control strategy is proposed,and the results show that the reasonable stiffness variation strategy can make the flexible humanoid robot quickly restore stability under the condition of ensuring stability.Finally,it is verified by MATLAB/ADAMS co-simulation.(4)Based on the principle of energy conservation,the mathematical models of the stiffness of the knee,ankle,and the whole leg of the humanoid robot with variable stiffness were established.The effect of knee and ankle joint stiffness on the equivalent stiffness of the leg was analyzed,and the validity of the equivalent stiffness model of the leg was verified by theoretical calculation and ADAMS simulation.Based on the mathematical model of equivalent stiffness,the foot contact force model of the robot is established,and the influence of different changes of knee joint stiffness and ankle joint stiffness on the foot contact force is analyzed,which provides a useful reference for further study of the application of stiffness in the humanoid robot. |
