# COMPASS: Cross-embodiment Mobility Policy via Residual RL and Skill Synthesis

arXiv: https://arxiv.org/abs/2502.16372 · HTML: https://mlsanity.com/paper/2502.16372/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 22 Feb 2025.

## Built on by (most cited first, 8 of 8)

- [From Seeing to Experiencing: Scaling Navigation Foundation Models with Reinforcement Learning](https://mlsanity.com/paper/2507.22028/) (2025)
- [TIC-VLA: A Think-in-Control Vision-Language-Action Model for Robot Navigation in Dynamic Environments](https://mlsanity.com/paper/2602.02459/) (2026)
- [Morphology-Conditioned World Model for Cross-Embodiment Quadrupedal Locomotion](https://mlsanity.com/paper/2604.08780/) (2026)
- [MM-Nav: Multi-View VLA Model for Robust Visual Navigation via Multi-Expert Learning](https://mlsanity.com/paper/2510.03142/) (2025)
- [Embodiment-Aware Generalist Specialist Distillation for Unified Humanoid Whole-Body Control](https://mlsanity.com/paper/2602.02960/) (2026)
- [Weave: Learning Whole-Body Dexterous Loco-Manipulation from Human-Object Interactions](https://mlsanity.com/paper/2609.16683/) (2026)
- [CrossTracer: Cross-Embodiment Navigation via VLA Model Reasoning and Trace Residuals Adapting](https://mlsanity.com/paper/2608.06688/) (2026)
- [Kinematic MeanFlow: One-Step Action Generation Policy for Robotic Foundation Models](https://mlsanity.com/paper/2610.00864/) (2026)

## Abstract (arXiv)

As robots are increasingly deployed in diverse application domains, enabling robust mobility across different embodiments has become a critical challenge. Classical mobility stacks, though effective on specific platforms, require extensive per-robot tuning and do not scale easily to new embodiments. Learning-based approaches, such as imitation learning (IL), offer alternatives, but face significant limitations on the need for high-quality demonstrations for each embodiment. To address these challenges, we introduce COMPASS, a unified framework that enables scalable cross-embodiment mobility using expert demonstrations from only a single embodiment. We first pre-train a mobility policy on a single robot using IL, combining a world model with a policy model. We then apply residual reinforcement learning (RL) to efficiently adapt this policy to diverse embodiments through corrective refinements. Finally, we distill specialist policies into a single generalist policy conditioned on an embodiment embedding vector. This design significantly reduces the burden of collecting data while enabling robust generalization across a wide range of robot designs. Our experiments demonstrate that COMPASS scales effectively across diverse robot platforms while maintaining adaptability to various environment configurations, achieving a generalist policy with a success rate approximately 5X higher than the pre-trained IL policy on unseen embodiments, and further demonstrates zero-shot sim-to-real transfer.