基于改进落足点的四足机器人爬楼梯MPC控制算法研究

Research on MPC Control Algorithm for Quadruped Robot Stair-Climbing with Improved Foothold Placement

摘要: 本文针对四足机器人爬楼梯这一特定场景，在传统MPC控制的基础上，提出一种改进落足点的腿部规划方法，并将其应用到MPC控制器中。通过在Matlab中建立四足机器人与楼梯的仿真模型，实现了四足机器人爬楼梯的控制过程。并通过引入姿态角的标准差这一概念，将本文所提出的优化后MPC控制方法与传统MPC控制算法下四足机器人爬楼梯过程中的姿态角标准差作对比，验证了所提出方法的优越性。

Abstract: This paper addresses the specific scenario of a quadruped robot climbing stairs. Building upon traditional Model Predictive Control (MPC), an improved leg planning method for foothold placement is proposed and integrated into the MPC controller. By establishing a simulation model of the quadruped robot and stairs in MATLAB, the control process for the robot’s stair-climbing is realized. Furthermore, by introducing the concept of the standard deviation of posture angles, a comparison is made between the standard deviation of posture angles during stair-climbing for quadruped robots controlled by the optimized MPC method proposed in this paper and those controlled by traditional MPC algorithms. This comparison validates the superiority of the proposed method.