arXiv cs.AI· Martin Schuck, Maks Sorokin, Simone Manni, Duy Ta, Angela P. Schoellig, Marco Hutter, Simon Le Cleac'H, Jan Br\"udigam·· 14 小时前AI 评分34
从 SMPC 演示中学习移动操作：稀疏奖励离线到在线 RL 方法
Learning Loco-Manipulation From SMPC Demonstrations With Sparse Offline-to-Online RL
AI 导读
研究团队提出用仿真中的 SMPC 作为自动专家生成大规模离线数据集，配合纯稀疏任务奖励训练离线到在线 RL 智能体，省去密集奖励塑形的人工调参。该方法将高层智能体与低层动态稳定控制器结合，学到的策略可超越原始最优控制教师。团队在配备机械臂的 Spot 四足机器人和 G1 人形机器人上成功部署了复杂移动操作技能。
正文 · 原文
Abstract:Integrating locomotion and manipulation is essential for robot autonomy, but scaling standard Reinforcement Learning (RL) to complex tasks is severely bottlenecked by the slow, manual process of dense reward shaping. To bypass this limitation, we leverage Sample-based Model Predictive Control (SMPC) entirely in simulation as an automated, rapidly tunable expert to generate massive offline datasets. Because this data solves the fundamental exploration problem, we can train an off-policy RL agent using purely sparse task rewards, drastically reducing the time required to learn new skills and eliminating the need for manual tuning. Integrating this high-level agent with a low-level dynamic stability controller yields more optimal behaviors that strictly align with true task objectives, ultimately allowing the learned policies to surpass the original optimal control teacher. We validate the robustness of this sim-to-real framework by successfully deploying complex loco-manipulation skills across different morphologies, including an arm-equipped Spot quadruped and a G1 humanoid.
| Comments: | CoRL 2026 Camera-ready version |
| Subjects: | Robotics (cs.RO); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2608.12063 [cs.RO] |
| (or arXiv:2608.12063v2 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2608.12063 arXiv-issued DOI via DataCite |
Submission history
From: Martin Schuck [view email]
[v1]
Wed, 12 Aug 2026 13:48:56 UTC (40,902 KB)
[v2]
Mon, 26 Oct 2026 11:46:26 UTC (40,903 KB)
来源：arXiv cs.AI · arxiv.org