{"ID":2863726,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2509.24956","arxiv_id":"2509.24956","title":"MSG: Multi-Stream Generative Policies for Sample-Efficient Robotic Manipulation","abstract":"Generative robot policies such as Flow Matching offer flexible, multi-modal policy learning but are sample-inefficient. Although object-centric policies improve sample efficiency, it does not resolve this limitation. In this work, we propose Multi-Stream Generative Policy (MSG), an inference-time composition framework that trains multiple object-centric policies and combines them at inference to improve generalization and sample efficiency. MSG is model-agnostic and inference-only, hence widely applicable to various generative policies and training paradigms. We perform extensive experiments both in simulation and on a real robot, demonstrating that our approach learns high-quality generative policies from as few as five demonstrations, resulting in a 95% reduction in demonstrations, and improves policy performance by 89 percent compared to single-stream approaches. Furthermore, we present comprehensive ablation studies on various composition strategies and provide practical recommendations for deployment. Finally, MSG enables zero-shot object instance transfer. We make our code publicly available at https://msg.cs.uni-freiburg.de.","short_abstract":"Generative robot policies such as Flow Matching offer flexible, multi-modal policy learning but are sample-inefficient. Although object-centric policies improve sample efficiency, it does not resolve this limitation. In this work, we propose Multi-Stream Generative Policy (MSG), an inference-time composition framework...","url_abs":"https://arxiv.org/abs/2509.24956","url_pdf":"https://arxiv.org/pdf/2509.24956v2","authors":"[\"Jan Ole von Hartz\",\"Lukas Schweizer\",\"Joschka Boedecker\",\"Abhinav Valada\"]","published":"2025-09-29T15:50:51Z","proceeding":"cs.RO","tasks":"[\"cs.RO\",\"cs.AI\",\"cs.LG\"]","methods":"[]","project_urls":"[\"https://msg.cs.uni-freiburg.de\"]","has_code":false,"code_links":[{"ID":609048,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_id":2863726,"paper_url":"https://arxiv.org/abs/2509.24956","paper_title":"MSG: Multi-Stream Generative Policies for Sample-Efficient Robotic Manipulation","repo_url":"https://github.com/robot-learning-freiburg/CURB-SG","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0},{"ID":609049,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_id":2863726,"paper_url":"https://arxiv.org/abs/2509.24956","paper_title":"MSG: Multi-Stream Generative Policies for Sample-Efficient Robotic Manipulation","repo_url":"https://github.com/robot-learning-freiburg/MSG","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}