2026/02/09 by Zichen Jeff Cui, Omar Rayyan, Haritheja Etukuru +19 · 1 voice
Computer Science · Engineering · #Conditioning #Language model #Modular design #Multimodal Machine Learning Applications #Reinforcement Learning in Robotics #Robot #Robot Manipulation and Learning #Robotic arm #Robotics #cs.LG #cs.RO
paper · pdf · doi:10.48550/arxiv.2602.09017
openalex publication_date 2026/02/09 · arxiv published 2026/02/09 · arxiv updated 2026/02/09 · openalex created_date 2026/02/11 · openalex updated_date 2026/07/28
The prevalent paradigm in robot learning attempts to generalize across environments, embodiments, and tasks with language prompts at runtime. A fundamental tension limits this approach: language is often too abstract to guide the concrete physical understanding required for robust manipulation. In this work, we introduce Contact-Anchored Policies (CAP), which replace language conditioning with points of physical contact in space. Simultaneously, we structure CAP as a library of modular utility models rather than a monolithic generalist policy. This factorization allows us to implement a real-to-sim iteration cycle: we build EgoGym, a lightweight simulation benchmark, to rapidly identify failure modes and refine our models and datasets prior to real-world deployment. We show that by conditioning on contact and iterating via simulation, CAP generalizes to novel environments and embodiments out of the box on three fundamental manipulation skills while using only 23 hours of demonstration data, and outperforms large, state-of-the-art VLAs in zero-shot evaluations by 56%. All model checkpoints, codebase, hardware, simulation, and datasets will be open-sourced. Project page: https://cap-policy.github.io/