Skip to content

论文arXiv 2611.253662 天前

M4M框架:利用操作策略隐空间实现无目标位姿移动操作 ​

本文由 AI 根据公开内容整理,可能有误,请以原文为准:Learning to Move from and for Manipulation Policies: Spatializing Latent Action Representations for Goal-Pose-Free Mobile Manipulation(arXiv 2611.25366)

核心内容 ​

针对移动操作中机器人位姿调整计算成本高且依赖预定义目标位姿的痛点,研究团队提出Learning to Move from and for Manipulation Policies(M4M)框架。该方法直接从预训练操作策略中学习具备操作感知的移动能力,无需联合训练即可协调导航与局部操作。在8个未见过的RoboCasa厨房铰接物体任务中,M4M取得47.0%的平均任务成功率,相比最强基线26.5%和直接操作执行19.2%实现大幅提升,且在冰箱关闭任务中对4种冻结操作策略均表现出一致性改进。

技术上,M4M利用操作策略动作生成网络的隐藏层表征作为密集监督信号,结合少量操作执行结果,学习捕获任务相关操作成功率的空间条件嵌入。该嵌入允许机器人仅凭当前立体视觉观测,即可估算未访问候选位姿的操作成功率,无需在每个位置实际展开操作策略推理。此表征被集成至导航策略训练中,并配合学习到的交接策略动态决策从导航到操作的切换时机,从而在统一框架下实现任务感知的移动与操作协同,解决了传统方法灵活性不足的问题。

关键看点 ​

  • 隐空间密集监督:提取预训练操作策略动作网络隐藏表征,结合少量执行反馈构建空间条件嵌入,避免全量联合训练的高昂算力与数据成本
  • 零样本位姿评估:基于立体视觉观测直接预测未访问位姿的操作成功率,无需逐点展开操作策略推理,大幅降低在线规划计算开销
  • 动态交接机制:通过学习型Handoff Policy自主决策导航至操作的切换时机,替代固定阈值或预定义目标位姿,提升非结构化环境适应性
  • 泛化性能验证:在RoboCasa 8类铰接物体任务中成功率47.0%,较次优基线提升20.5个百分点,且对4种不同冻结操作策略均有效

工程与产业观察 ​

M4M验证了复用预训练操作策略内部表征指导移动规划的可行性,为具身智能系统提供了低成本、高泛化的导航-操作协同范式。其核心在于将操作成功概率显式空间化,使移动决策真正服务于下游任务而非几何目标,这对缺乏精确位姿标注的真实场景部署具有重要工程参考价值。

英文摘要 (Abstract) ​

The success of manipulation policies is closely tied to the spatial relationship between the robot and the target object, as well as the task being performed, making the ability to adjust the robot's pose critical for successful execution. When the robot starts away from the task, such adjustments require coordinating navigation with local manipulation. Jointly training these capabilities incurs substantial computational and data costs, while navigating toward a predefined goal pose limits the flexibility needed to adapt to task-specific manipulation requirements. To address these challenges, we propose Learning to Move from and for Manipulation Policies (M4M), a framework that learns manipulation-aware mobility directly from pretrained manipulation policies. By leveraging the hidden representations of the policy's action-generating network as dense supervision, together with a limited number of manipulation execution outcomes, we learn a spatially conditioned embedding that captures task-dependent manipulation success across candidate robot poses. From the robot's current stereo observations, this representation estimates manipulation success at unvisited poses without rolling out the manipulation policy at each location. It is further integrated into navigation policy training, while a learned handoff policy decides when to transition from navigation to manipulation, enabling the robot to move and manipulate in a unified, task-aware manner. Across eight articulated-object tasks in unseen RoboCasa kitchens, M4M achieves 47.0% mean task success, compared with 26.5% for the strongest baseline and 19.2% for direct manipulation execution, and refrigerator-closing experiments show consistent improvements across four frozen manipulation policies.

作者:Siyuan Zheng, Zhong Cao, Henry X. Liu

基于本页技术方向关联的整机参数、核心零部件、实测报告与同赛道选型横评。

外部来源

本页事实对应的公开页面,以原文为准。(点击将离开本站)

  1. arXiv 2611.25366:Learning to Move from and for Manipulation Policies: Spatializing Latent Action Representations for Goal-Pose-Free Mobile Manipulation ↗访问 2026-11-20