BiRoAD：用于双臂操作的学习共享与角色自适应表示
BiRoAD: Learning Shared and Role-Adaptive Representations for Bimanual Manipulation
浏览论文内容
中文总结 AI 辅助
BiRoAD提出双臂角色自适应分解框架，通过交换对称/反对称分量学习共享与角色自适应表示，在不改变策略输入下提升双臂操作在角色配置上的鲁棒性。
中文摘要 AI 辅助
双臂操作需要策略在协调两只手臂的同时，根据场景几何、物体配置和任务上下文调整其功能角色。学习这种场景条件下的角色自适应仍然具有挑战性，因为演示数据可能包含不均匀的角色分布，从而限制了对代表性不足的手臂-角色配置的泛化能力。此外，许多双臂策略在固定的左臂和右臂动作空间中预测动作。虽然这为机器人控制提供了自然的参数化方式，但它并未明确指定当功能角色在手臂之间交换时行为应如何转换。在不同的场景初始化下，两只手臂可能遵循相似的协调模式，但分配给每只手臂的角色特定行为应随场景而变化。因此，我们提出了BiRoAD，一个双臂角色自适应分解框架，用于在双臂策略中学习共享和角色自适应的表示。给定双臂轨迹或动作令牌特征，BiRoAD将这些特征分解为交换对称和交换反对称分量：前者捕获对手臂交换不变的协调结构，后者捕获随功能角色分配一致变化的角色特定差异。然后，这两个分量作为对原始配对手臂表示的残差更新进行重组，使BiRoAD能够作为模块化特征变换，而不改变策略输入、模仿学习目标，也不需要手动定义的角色标签。在多个具有平衡和不平衡角色分布的双臂操作任务中，BiRoAD相比相应的基础策略，在角色配置上的鲁棒性有所提高，尤其是在代表性不足的角色配置上取得了显著改进。
英文摘要
Bimanual manipulation requires policies that coordinate two arms while adapting their functional roles to scene geometry, object configuration, and task context. Learning such scene-conditioned role adaptation remains challenging, as demonstrations may contain uneven role distributions that limit generalization to underrepresented arm--role configurations. In addition, many bimanual policies predict actions in fixed left- and right-arm action spaces. While this provides a natural parameterization for robot control, it does not explicitly specify how behaviors should transform when functional roles are exchanged across arms. Across different scene initializations, the two arms may follow a similar coordination pattern, but the role-specific behavior assigned to each arm should change with the scene. Therefore, we propose BiRoAD, a Bimanual Role-Adaptive Decomposition framework for learning shared and role-adaptive representations in bimanual policies. Given bimanual trajectory or action-token features, BiRoAD decomposes these features into swap--symmetric and swap--antisymmetric components: the former captures coordination structure invariant to arm exchange, and the latter captures role-specific distinctions that vary consistently with functional role assignment. The two components are then recomposed as residual updates to the original paired arm representations, allowing BiRoAD to serve as a modular feature transformation without changing the policy inputs, imitation-learning objective, or requiring manually defined role labels. Across multiple bimanual manipulation tasks with balanced and imbalanced role distributions, BiRoAD improves robustness across role configurations over corresponding base policies, with notable gains on underrepresented role configurations.
发表机构
- Peking University（北京大学）
- PrimeBot Research Institute（PrimeBot研究院）
- Hong Kong University of Science and Technology (Guangzhou)（香港科技大学（广州））
- UC Berkeley（加州大学伯克利分校）
机构由 AI 辅助整理，请以论文原文为准。
