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

Area of Science:

• Robotics
• Control Systems Engineering
• Mechatronics

Background:

• Wheel-legged robots offer adaptable mobility in unstructured environments but face balance control challenges due to underactuation.
• Hardware characteristics, like motor friction, negatively impact the dynamic convergence and stability of these robots.

Purpose of the Study:

• To develop an advanced balance control method for wheel-legged robots that effectively addresses motor friction.
• To enhance the stability, robustness, and convergence speed of wheel-legged robots during dynamic locomotion.

Main Methods:

• A Linear Quadratic Regulator (LQR) controller was designed based on the robot's dynamics model.
• A Stribeck friction model was identified using Particle Swarm Optimization (PSO) on data from a constant-speed excitation trajectory.
• The identified friction model was integrated as feedforward compensation into the LQR controller.

Main Results:

• Friction identification achieved a minimum standard deviation of approximately 0.30, with the model closely matching actual friction values.
• The friction feedforward LQR algorithm demonstrated superior convergence performance compared to the baseline LQR.
• Experimental results showed reduced oscillations, accelerated convergence, and improved stability and robustness across various terrains and disturbance scenarios.

Conclusions:

• The proposed friction feedforward LQR balance control method effectively compensates for motor friction in wheel-legged robots.
• This approach significantly enhances robot stability and dynamic performance, outperforming traditional LQR control.
• The method offers a robust solution for improving the adaptability and reliability of wheel-legged robots in complex environments.