让机器人操作模型自动适应旋转平移，提升泛化能力
Eq.Bot: Enhance Robotic Manipulation Learning via Group Equivariant Canonicalization
- 通过空间归一化将视觉输入转到标准坐标系
- 在多种模型上实现最高50%的性能提升
- 无需修改网络结构，通用性强适合实际部署
机器人操作系统正广泛应用于多个领域。然而现有多模态学习框架缺乏几何一致性保障，难以处理旋转、平移等空间变换。尽管近期工作尝试通过定制化架构引入等变性，但存在实现复杂、计算开销大、可移植性差等问题。受人类空间推理认知启发，我们提出 Eq.Bot，一种基于 SE(2) 群等变理论的通用归一化框架，用于机器人操作学习。该框架将观测数据映射到规范空间，应用已有策略，并将动作反向映射回原空间。作为模型无关方案，Eq.Bot 不需修改网络结构即可赋予模型空间等变性。大量实验表明，在基于 CNN（如 CLIPort）和 Transformer（如 OpenVLA-OFT）的架构上，Eq.Bot 在多种机器人操作任务中均优于现有方法，性能提升最高达 50.0%。
原文摘要 · Abstract (English)
Robotic manipulation systems are increasingly deployed across diverse domains. Yet existing multi-modal learning frameworks lack inherent guarantees of geometric consistency, struggling to handle spatial transformations such as rotations and translations. While recent works attempt to introduce equivariance through bespoke architectural modifications, these methods suffer from high implementation complexity, computational cost, and poor portability. Inspired by human cognitive processes in spatial reasoning, we propose Eq.Bot, a universal canonicalization framework grounded in SE(2) group equivariant theory for robotic manipulation learning. Our framework transforms observations into a canonical space, applies an existing policy, and maps the resulting actions back to the original space. As a model-agnostic solution, Eq.Bot aims to endow models with spatial equivariance without requiring architectural modifications. Extensive experiments demonstrate the superiority of Eq.Bot under both CNN-based (e.g., CLIPort) and Transformer-based (e.g., OpenVLA-OFT) architectures over existing methods on various robotic manipulation tasks, where the most significant improvement can reach 50.0%.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。