Racing a Wheeled Quadruped: Active Load Transfer Mitigation via Model Predictive Control

2026-06-24

AI Summary

This work addresses the degradation in stability and performance of wheeled quadrupedal robots during high-speed autonomous racing caused by lateral load transfer. To mitigate this issue, the authors propose a hierarchical control framework that synergistically integrates model predictive control (MPC) with end-to-end reinforcement learning (RL). For the first time, MPC and RL are jointly employed for active roll control: an upper-level MPC layer, based on a bicycle model, online-optimizes trajectories to minimize the lateral load transfer ratio, while a lower-level 16-degree-of-freedom whole-body RL controller actuates the legs to generate anti-roll moments and coordinates active knee suspensions for dynamic stabilization. Experimental results demonstrate a 44% reduction in average lateral load transfer ratio, an 8.7% improvement in fastest lap time, and a peak lateral acceleration of 1.98 m/s²—representing a 21.3% increase—significantly outperforming existing approaches.

Abstract

This paper presents a hierarchical control framework using model predictive control (MPC) and reinforcement learning (RL) for active roll control to manage lateral load transfer during autonomous racing of a wheeled quadruped. The framework integrates offline time-optimal raceline generation, an online MPC planner that actively minimizes the lateral Load Transfer Ratio (LTR), and a low-level, whole-body RL policy deployed directly onto the robot's 16 actuators. The MPC is based on a vehicle dynamics bicycle model of the Unitree Go2-W platform. The robot's leg actuators act as active suspension where knee joints generate anti-roll torque to bank into turns. Physical track experiments demonstrate that active roll control reduces mean LTR by up to 44%, improves the fastest lap time by 8.7%, and boosts peak lateral acceleration capability by 21.3% to 1.98 $m/s^2$, maintaining robust high-speed stability beyond the range of a non-tilting baseline controller. Supplementary code and video can be found at https://github.com/meisman-ucb/go2w-roll-control-mpc

Problem

• lateral load transfer
• wheeled quadruped
• autonomous racing
• roll control
• high-speed stability

Innovation

• Model Predictive Control
• Reinforcement Learning
• Active Roll Control
• Load Transfer Ratio
• Wheeled Quadruped