Research On Gait Optimization Of Electrically Driven Quadruped Robot Based On Virtual Model Control And Its Realization In A Single-leg Experimental Device

In recent years,with the rapid development of quadruped robot technology around the world,physical prototypes with better dynamic motion ability and terrain adaptability continue to emerge.And the continuous progress of electrically driven technology provides an important engine for the development of legged robot technology.This paper takes a heavy-load electrically driven quadruped robot which is being developed by the research group as the research object and carries out research and experimental work on gait optimization and single-leg drive control of the robot,in order to provide theoretical and technical support for the development of the prototype.In this paper,firstly,a gait generation controller is designed based on the virtual model control method(VMC).Specifically,the stance phase is constructed for achieving the targets of body translation control,body rotation control and gravity compensation,and the foot trajectory of the swing phase is planned.A gait control system based on intuitive virtual force is constructed in MATLAB/Simulink environment,which realizes the direct mapping from the whole machine motion control targets to the joint level control targets.Then,aiming at the main working conditions such as diagonal trot,in-situ take-off and free fall,the whole model of quadruped robot is simulated,and the simulated joint power,joint angular velocity,joint torque and other motion parameters are analyzed.And it is used as the basis of gait optimization and driving control.In this paper,a gait optimization method based on motion energy consumption and trajectory error is proposed,in which an objective function is constructed by the targets weighting method,and a genetic algorithm is used to optimize the stiffness coefficient and damping coefficient in the VMC model.This method is used to optimize the combination of several groups of objective functions and weighting coefficients in the Simulink-Adams simulation environment.Based on the optimization results,the single-leg motion energy consumption and trajectory error are simulated and tested.The test results show that this method can effectively reduce the motion energy consumption,reduce the trajectory error of the swing phase,ensure the obstacle-surpassing ability and endurance performance of the robot.The drive hardware system for single leg motion is built,and the setting debugging and gain scheduling are studied.It includes the selection of a single-leg drive unit according to the data of the condition simulation,the parameter tuning of single-joint Elmo driver,the establishment of Ether CAT ring communication network based on Elmo controller,the realization of multi-joint synchronous drive control,and the development of single-leg drive control software.The gain scheduling algorithm based on activation function is studied.Exponential function scheduling and hyperbolic tangent function scheduling are used for proportional gain and integral gain,respectively.The simulation results in MATLAB/Simulink environment show that the gain scheduling algorithm can effectively improve the high-frequency alternating output performance of the motor and significantly reduce the speed error of the foot linear trajectory.A single-leg test platform for a large electrically driven quadruped robot is developed,and the verification experiments of single-leg straight-line trajectory swing,standing squatting,pedal and gait optimization are carried out.The experimental results show that the theoretical analysis and simulation results are accurate and effective,and the related technologies and data can be used for the development of the target system of the electrically driven quadruped robot.