Balance Control Method for Bipedal Wheel-Legged Robots Based on Friction Feedforward Linear Quadratic Regulator

被引:0
|
作者
Zhang, Aimin [1 ]
Zhou, Renyi [2 ]
Zhang, Tie [3 ]
Zheng, Jingfu [3 ]
Chen, Shouyan [4 ]
机构
[1] GAC R&D Ctr, Guangzhou 511434, Peoples R China
[2] Guangdong Univ Technol, Sch Electromech Engn, Guangzhou 510006, Peoples R China
[3] South China Univ Technol, Sch Mech & Automot Engn, Guangzhou 511442, Peoples R China
[4] Guangzhou Univ, Sch Mech & Elect Engn, Guangzhou 510006, Peoples R China
关键词
bipedal wheel-legged robots; balance control; LQR controller; Stribeck friction model; PSO algorithm; DYNAMIC-MODEL;
D O I
10.3390/s25041056
中图分类号
O65 [分析化学];
学科分类号
070302 ; 081704 ;
摘要
With advancements in mobile robot technology, wheel-legged robots have emerged as promising next-generation mobile solutions, reducing design costs and enhancing adaptability in unstructured environments. As underactuated systems, their balance control has become a prominent research focus. Despite there being numerous control approaches, challenges remain. Balance control methods for wheel-legged robots are influenced by hardware characteristics, such as motor friction, which can induce oscillations and hinder dynamic convergence. This paper presents a friction feedforward Linear Quadratic Regulator (LQR) balance control method. Specifically, a basic LQR controller is developed based on the dynamics model of the wheel-legged robot, and a Stribeck friction model is established to characterize motor friction. A constant-speed excitation trajectory is designed to gather data for friction identification, and the Particle Swarm Optimization (PSO) algorithm is applied to determine the optimal friction parameters. The identified friction model is subsequently incorporated as feedforward compensation for the LQR controller's torque output, resulting in the proposed friction feedforward LQR balance control algorithm. The minimum standard deviation for friction identification is approximately 0.30, and the computed friction model values closely match the actual values, indicating effective and accurate identification results. Balance experiments demonstrate that under diverse conditions-such as flat ground, single-sided bridges, and disturbance scenarios-the convergence performance of the friction feedforward LQR algorithm markedly surpasses that of the baseline LQR, effectively reducing oscillations, accelerating convergence, and improving the robot's stability and robustness.
引用
收藏
页数:23
相关论文
empty
未找到相关数据