arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2610.15425cs.RO
收藏

探索具身操作中组合泛化的极限

Exploring the Limits of Combinatorial Generalization in Embodied Manipulation

AutoLab, School of Artificial Intelligence, Shanghai Jiao Tong University(上海交通大学人工智能学院AutoLab)

Xupeng Zhang, Yantai Yang, Chang Guo, Zhaokai Yin, Zhipeng Zhang

AI总结:

该研究针对具身操作中VLA模型的技能重组泛化问题,提出HSE诊断框架与RoboCombine基准,发现起始状态差距后推出PRTI策略,提升了技能重组的有序执行成功率。

摘要:

在具身操作领域,视觉语言动作(Vision Language Action, VLA)模型可执行原子技能,但长程操作需在无每个技能组合演示的情况下对这些技能进行重组。全指令路线将技能链视为长程模仿。在一项受控研究中,对有限技能组合序列进行全指令训练后,模型在已覆盖组合上的有序成功率达40.0%,但在重组组合上仅为7.7%,这表明模型是对序列模板进行了记忆,而非实现稳健的技能重组。我们引入分层子任务执行(Hierarchical Subtask Execution, HSE),这是一个诊断框架,通过规划器、触发器和策略之上的共享模块化接口来揭示技能重组,将长程操作失败分解为可分离的模式。我们推出RoboCombine,这是一个无需组合技能训练轨迹的技能重组基准,包含子任务边界上的有序成功率和转换转换率等指标。RoboCombine表明,仅具备原子能力无法确保有序执行,失败集中在下游子任务从先前子任务的交接状态(而非规范原子起始状态)启动的转换环节,我们将这种不匹配称为起始状态差距。为解决该问题,我们推出姿态鲁棒任务初始化(Pose Robust Task Initialization, PRTI),这是一种轻量级原子任务后训练策略,可扩大起始状态覆盖范围,使其超出规范起始状态。在RoboCombine上采用神经常规计时HSE时,PRTI的有序成功率达18.16%,而随机扰动下为4.21%,机器人重置神经常规下为9.47%,且无需组合技能训练轨迹。这些结果表明,组合泛化需要具备交接稳健性的原子技能,能将第一阶段进展转化为可靠的下游执行,而非仅分解为已知子任务。代码:this https URL。

英文摘要:

In embodied manipulation, Vision Language Action (VLA) models can execute atomic skills, but long-horizon manipulation requires recombining these skills without demonstrations for every skill combination. The full instruction route treats skill chaining as long-horizon imitation. In a controlled study, full instruction training on limited skill combination sequences reaches 40.0% ordered success on covered combinations but only 7.7% on recombination, suggesting memorization of sequence templates rather than robust skill recombination. We introduce Hierarchical Subtask Execution (HSE), a diagnostic framework that exposes skill recombination through a shared modular interface over the planner, trigger, and policy, decomposing long-horizon failures into separable modes. We introduce RoboCombine, a benchmark for skill recombination without combined skill training trajectories, with metrics for ordered success and transition conversion across subtask boundaries. RoboCombine reveals that atomic competence alone does not ensure ordered execution. Failures concentrate at transitions where downstream subtasks launch from handoff states from preceding subtasks rather than canonical atomic starts. We identify this mismatch as the start state gap. To address it, we introduce Pose Robust Task Initialization (PRTI), a lightweight atomic task post-training strategy that broadens start state coverage beyond canonical starts. Under oracle timed HSE on RoboCombine, PRTI reaches 18.16% ordered success, compared with 4.21% under random perturbation and 9.47% under a robot-reset oracle, without combined skill training trajectories. These results show that combinatorial generalization requires handoff robust atomic skills that convert first stage progress into reliable downstream execution, not merely decomposition into known subtasks. Code: https://github.com/AutoLab-SAI-SJTU/RoboCombine.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑