{"ID":5937678,"CreatedAt":"2026-07-07T03:14:33.014478982Z","UpdatedAt":"2026-07-08T12:22:32.149807025Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2607.04265","arxiv_id":"2607.04265","title":"HALO-WA: Hybrid-Attention Latent-Guided Online Reinforcement Learning for World-Action Models","abstract":"World-action (WA) models can generate long-horizon action chunks for general-purpose robotic manipulation, but they remain vulnerable to calibration, perception, and contact-dynamics errors in real-world precision tasks, often failing in the final few millimeters of alignment or insertion. We propose HALO-WA, a hybrid-attention latent-guided online reinforcement learning (RL) framework for WA models, which leverages latent features and action priors from the WA generation process through a lightweight actor-critic adapter to enable fast online adaptation to real deployment errors. HALO-WA introduces a hybrid-attention structure that preserves the temporal consistency of action chunks while reading task-relevant information from WA latents conditioned on visual context and end-stage correction requirements, thereby producing refined action chunks. We validate HALO-WA on four real-world precision manipulation tasks, where it improves the average success rate from 26.4\\% for WA-base to 87.1\\%, outperforming the strongest baseline by 19.2 percentage points while requiring only 45--75 minutes of online training per task. To facilitate reproducibility, we further conduct supplementary simulation experiments in RoboTwin and release the code at https://github.com/YeanRoot/HALO-WA.","short_abstract":"World-action (WA) models can generate long-horizon action chunks for general-purpose robotic manipulation, but they remain vulnerable to calibration, perception, and contact-dynamics errors in real-world precision tasks, often failing in the final few millimeters of alignment or insertion. We propose HALO-WA, a hybrid-...","url_abs":"https://arxiv.org/abs/2607.04265","url_pdf":"https://arxiv.org/pdf/2607.04265v1","authors":"[\"Angen Ye\",\"Weijie Ke\",\"Xiaofeng Wang\",\"Xinze Chen\",\"Chaojun Ni\",\"Guosheng Zhao\",\"Boyuan Wang\",\"Zheng Zhu\",\"Junjie Xie\",\"Dapeng Zhang\"]","published":"2026-07-05T12:24:14Z","proceeding":"cs.RO","tasks":"[\"cs.RO\",\"cs.AI\"]","methods":"[\"Reinforcement Learning\"]","has_code":false,"code_links":[{"ID":613978,"CreatedAt":"2026-07-07T03:14:33.014478982Z","UpdatedAt":"2026-07-07T03:14:33.014478982Z","DeletedAt":null,"paper_id":5937678,"paper_url":"https://arxiv.org/abs/2607.04265","paper_title":"HALO-WA: Hybrid-Attention Latent-Guided Online Reinforcement Learning for World-Action Models","repo_url":"https://github.com/YeanRoot/HALO-WA","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}