Data-Driven Physics Embedded Dynamics with Predictive Control and Reinforcement Learning for Quadrupeds

TL;DR

This paper presents a novel approach combining Lagrangian Neural Networks with reinforcement learning and MPC to improve quadruped robot locomotion by enhancing physical consistency, efficiency, and real-world applicability.

Contribution

It introduces a physically consistent neural dynamics model integrated into an RL MPC framework, significantly improving efficiency and long-horizon accuracy for quadruped control.

Findings

Up to 4x faster computation with minimal performance loss

Improved sample efficiency and reduced long-horizon errors

Successful real-world deployment on Unitree Go1 robot

Abstract

State of the art quadrupedal locomotion approaches integrate Model Predictive Control (MPC) with Reinforcement Learning (RL), enabling complex motion capabilities with planning and terrain adaptive behaviors. However, they often face compounding errors over long horizons and have limited interpretability due to the absence of physical inductive biases. We address these issues by integrating Lagrangian Neural Networks (LNNs) into an RL MPC framework, enabling physically consistent dynamics learning. At deployment, our inverse dynamics infinite horizon MPC scheme avoids costly matrix inversions, improving computational efficiency by up to 4x with minimal loss of task performance. We validate our framework through multiple ablations of the proposed LNN and its variants. We show improved sample efficiency, reduced long-horizon error, and faster real time planning compared to unstructured neural dynamics. Lastly, we also test our framework on the Unitree Go1 robot to show real world viability.