{"ID":2874446,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2509.03859","arxiv_id":"2509.03859","title":"Learning Multi-Stage Pick-and-Place with a Legged Mobile Manipulator","abstract":"Quadruped-based mobile manipulation presents significant challenges in robotics due to the diversity of required skills, the extended task horizon, and partial observability. After presenting a multi-stage pick-and-place task as a succinct yet sufficiently rich setup that captures key desiderata for quadruped-based mobile manipulation, we propose an approach that can train a visuo-motor policy entirely in simulation, and achieve nearly 80\\% success in the real world. The policy efficiently performs search, approach, grasp, transport, and drop into actions, with emerged behaviors such as re-grasping and task chaining. We conduct an extensive set of real-world experiments with ablation studies highlighting key techniques for efficient training and effective sim-to-real transfer. Additional experiments demonstrate deployment across a variety of indoor and outdoor environments. Demo videos and additional resources are available on the project page: https://horizonrobotics.github.io/gail/SLIM.","short_abstract":"Quadruped-based mobile manipulation presents significant challenges in robotics due to the diversity of required skills, the extended task horizon, and partial observability. After presenting a multi-stage pick-and-place task as a succinct yet sufficiently rich setup that captures key desiderata for quadruped-based mob...","url_abs":"https://arxiv.org/abs/2509.03859","url_pdf":"https://arxiv.org/pdf/2509.03859v3","authors":"[\"Haichao Zhang\",\"Haonan Yu\",\"Le Zhao\",\"Andrew Choi\",\"Qinxun Bai\",\"Yiqing Yang\",\"Wei Xu\"]","published":"2025-09-04T03:36:07Z","proceeding":"cs.RO","tasks":"[\"cs.RO\"]","methods":"[]","has_code":false}