Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2512.09101 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912848015261696 |
|---|---|
| author | Zhuang, Lipeng Fan, Shiyu Audonnet, Florent P. Ru, Yingdong Ho, Edmond S. L. Camarasa, Gerardo Aragon Henderson, Paul |
| author_facet | Zhuang, Lipeng Fan, Shiyu Audonnet, Florent P. Ru, Yingdong Ho, Edmond S. L. Camarasa, Gerardo Aragon Henderson, Paul |
| contents | We present Masked Generative Policy (MGP), a novel framework for visuomotor imitation learning. We represent actions as discrete tokens, and train a conditional masked transformer that generates tokens in parallel and then rapidly refines only low-confidence tokens. We further propose two new sampling paradigms: MGP-Short, which performs parallel masked generation with score-based refinement for Markovian tasks, and MGP-Long, which predicts full trajectories in a single pass and dynamically refines low-confidence action tokens based on new observations. With globally coherent prediction and robust adaptive execution capabilities, MGP-Long enables reliable control on complex and non-Markovian tasks that prior methods struggle with. Extensive evaluations on 150 robotic manipulation tasks spanning the Meta-World and LIBERO benchmarks show that MGP achieves both rapid inference and superior success rates compared to state-of-the-art diffusion and autoregressive policies. Specifically, MGP increases the average success rate by 9% across 150 tasks while cutting per-sequence inference time by up to 35x. It further improves the average success rate by 60% in dynamic and missing-observation environments, and solves two non-Markovian scenarios where other state-of-the-art methods fail. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_09101 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Masked Generative Policy for Robotic Control Zhuang, Lipeng Fan, Shiyu Audonnet, Florent P. Ru, Yingdong Ho, Edmond S. L. Camarasa, Gerardo Aragon Henderson, Paul Robotics Artificial Intelligence We present Masked Generative Policy (MGP), a novel framework for visuomotor imitation learning. We represent actions as discrete tokens, and train a conditional masked transformer that generates tokens in parallel and then rapidly refines only low-confidence tokens. We further propose two new sampling paradigms: MGP-Short, which performs parallel masked generation with score-based refinement for Markovian tasks, and MGP-Long, which predicts full trajectories in a single pass and dynamically refines low-confidence action tokens based on new observations. With globally coherent prediction and robust adaptive execution capabilities, MGP-Long enables reliable control on complex and non-Markovian tasks that prior methods struggle with. Extensive evaluations on 150 robotic manipulation tasks spanning the Meta-World and LIBERO benchmarks show that MGP achieves both rapid inference and superior success rates compared to state-of-the-art diffusion and autoregressive policies. Specifically, MGP increases the average success rate by 9% across 150 tasks while cutting per-sequence inference time by up to 35x. It further improves the average success rate by 60% in dynamic and missing-observation environments, and solves two non-Markovian scenarios where other state-of-the-art methods fail. |
| title | Masked Generative Policy for Robotic Control |
| topic | Robotics Artificial Intelligence |
| url | https://arxiv.org/abs/2512.09101 |