Design And Algorithm Research Of Motion Control System For Quadruped Robots Targeting Complex Terrains

With the emergence and promotion of high-energy density torque motors,there has been a surge in design for electrically-driven quadruped robots.This has led to a trend towards miniaturization and multifunctionality of these robots,while also raising the bar for their dynamic performance.Due to the large number of joints in quadruped robots and their unique motion mode,the intermittent underdriven condition greatly increases the difficulty of controlling quadruped robots.This paper addresses the above issues and designs a high-dynamic quadruped robot motion control system based on an electric-driven quadruped robot.The specific content is as follows:1.Introduction of the quadruped robot platform and the hardware and software design of the control system: according to the design goals and bearing requirements of the quadruped robot,the experimental platform of the quadruped robot is built,and appropriate driving motors are selected for each joint.Based on the driving mode of the motor and the control frequency requirements,a hardware communication network is established,and suitable main control board and power supply are selected.According to the hardware communication method and the model of the main control board,an appropriate operating system and programming language are selected,and the software framework of the control system is built by writing software drivers for hardware communication.2.Control system algorithm design for quadruped robot: The role of robot control system algorithm is to solve the problems of where the robot is,where to go,and how to get there.In order to address the issue of where the robot is,a state estimation method is designed in this paper to real-time estimate the pose of the robot in the world coordinate system.To address the issue of where the robot should go,a centroid trajectory planner and a swing-phase trajectory planner are designed in this paper.To address the question of how the robot moves,a Convex Optimization-based Model Predictive Control(MPC)approach is devised in this work.By employing quadratic programming,this approach dynamically optimizes the optimal footground reaction forces for the upcoming gait cycle,effectively addressing intermittent underactuated conditions encountered by quadruped robots.In addition,to improve the robot’s motion stability on rugged terrain,an event-based gait planner is designed in this paper.It switches the gait phase according to the foot contact status estimated by the ground contact detection algorithm,improving the robot’s passability and stability on complex terrain.3.Simulation verification and physical prototype testing: Deploying the control system algorithm proposed in this paper on a four-legged robot experimental platform and simulation platform.Firstly,the effectiveness of the control algorithm was preliminarily verified by conducting slope walking simulations and non-structured terrain traversing experiments in the virtual simulation platform.Secondly,the stability and feasibility of the control algorithm were further validated through static pose adjustment experiments and diagonal gait walking experiments on the physical robot,while also testing the reliability and response speed of the four-legged robot platform and the control system hardware and software developed in this paper.