arXiv 2602.164442026-02-19cs.ROcs.AIcs.LG

RoboGene: 通过多样性驱动的代理框架提升VLA预训练以生成现实任务

RoboGene: Boosting VLA Pre-training via Diversity-Driven Agentic Framework for Real-World Task Generation

Beijing Innovation Center of Humanoid Robotics（北京人形机器人创新中心）；The School of Advanced Manufacturing and Robotics, Peking University（北京大学先进制造与机器人学院）；Beijing Institute of Technology（北京理工大学）；The School of Mechanical Engineering and Automation, Beihang University（北航机械工程与自动化学院）；State Key Laboratory of Multimedia Information Processing, School of Computer Science, Peking University（北京大学多媒体信息处理国家重点实验室）

Yixue Zhang, Kun Wu, Zhi Gao, Zhen Zhao, Pei Ren, Zhiyuan Xu, Fei Liao, Xinhua Wang, Shichao Fan, Di Wu, Qiuxuan Feng, Meng Li, Zhengping Che, Chang Liu, Jian Tang

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

RoboGene通过多样性驱动的代理框架生成多样化、物理合理的机器人操作任务，提升VLA预训练效果。

摘要：

通用目的机器人操作的追求受到多样且现实交互数据稀缺的阻碍。与从网络收集视觉或语言数据不同，机器人数据收集是一个主动过程，会产生高昂的物理成本。因此，自动化任务筛选以最大化数据价值仍然是一个关键但研究不足的挑战。现有的手动方法难以扩展且偏向常见任务，而现成的基础模型往往会产生物理上不可行的指令。为此，我们引入了RoboGene，一种代理框架，旨在自动化生成适用于单臂、双臂和移动机器人的多样化、物理上合理的操作任务。RoboGene集成了三个核心组件：多样性驱动的采样以实现广泛的任务覆盖、自我反思机制以强制执行物理约束，以及人机协作的优化以实现持续改进。我们进行了广泛的定量分析和大规模现实实验，收集了18k条轨迹的数据集，并引入了新的指标来评估任务质量、可行性和多样性。结果表明，RoboGene显著优于最先进的基础模型（例如GPT-4o、Gemini 2.5 Pro）。此外，现实实验表明，使用RoboGene预训练的VLA模型在成功率和泛化能力上表现更优，突显了高质量任务生成的重要性。我们的项目可在https://robogene-boost-vla.github.io上获得。

英文摘要：

The pursuit of general-purpose robotic manipulation is hindered by the scarcity of diverse, real-world interaction data. Unlike data collection from web in vision or language, robotic data collection is an active process incurring prohibitive physical costs. Consequently, automated task curation to maximize data value remains a critical yet under-explored challenge. Existing manual methods are unscalable and biased toward common tasks, while off-the-shelf foundation models often hallucinate physically infeasible instructions. To address this, we introduce RoboGene, an agentic framework designed to automate the generation of diverse, physically plausible manipulation tasks across single-arm, dual-arm, and mobile robots. RoboGene integrates three core components: diversity-driven sampling for broad task coverage, self-reflection mechanisms to enforce physical constraints, and human-in-the-loop refinement for continuous improvement. We conduct extensive quantitative analysis and large-scale real-world experiments, collecting datasets of 18k trajectories and introducing novel metrics to assess task quality, feasibility, and diversity. Results demonstrate that RoboGene significantly outperforms state-of-the-art foundation models (e.g., GPT-4o, Gemini 2.5 Pro). Furthermore, real-world experiments show that VLA models pre-trained with RoboGene achieve higher success rates and superior generalization, underscoring the importance of high-quality task generation. Our project is available at https://robogene-boost-vla.github.io.

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。