Toward Reliable Sim-to-Real Predictability for MoE-based Robust Quadrupedal Locomotion

TL;DR

This paper presents a MoE-based locomotion policy and RoboGauge assessment suite to improve sim-to-real transferability and robustness of quadrupedal robots across diverse terrains using proprioception.

Contribution

It introduces a unified MoE policy framework combined with RoboGauge metrics for reliable transfer from simulation to real-world terrains.

Findings

Robust locomotion demonstrated on unseen terrains like snow, sand, and stairs.

Achieved high-speed movement of 4 m/s with emergent gait stability.

Enhanced generalization and transferability without extensive physical trials.

Abstract

Reinforcement learning has shown strong promise for quadrupedal agile locomotion, even with proprioception-only sensing. In practice, however, sim-to-real gap and reward overfitting in complex terrains can produce policies that fail to transfer, while physical validation remains risky and inefficient. To address these challenges, we introduce a unified framework encompassing a Mixture-of-Experts (MoE) locomotion policy for robust multi-terrain representation with RoboGauge, a predictive assessment suite that quantifies sim-to-real transferability. The MoE policy employs a gated set of specialist experts to decompose latent terrain and command modeling, achieving superior deployment robustness and generalization via proprioception alone. RoboGauge further provides multi-dimensional proprioception-based metrics via sim-to-sim tests over terrains, difficulty levels, and domain randomizations, enabling reliable MoE policy selection without extensive physical trials. Experiments on a Unitree Go2 demonstrate robust locomotion on unseen challenging terrains, including snow, sand, stairs, slopes, and 30 cm obstacles. In dedicated high-speed tests, the robot reaches 4 m/s and exhibits an emergent narrow-width gait associated with improved stability at high velocity.