arXiv：cs.LG· Seohong Park, Sergey Levine·· 4 小时前AI 评分33
OCBench：机器人操作基准如何复现行为克隆之谜
Behavioral Cloning Mystery
AI 导读
研究者提出 OCBench，一个机器人操作基准，其脚本化策略模仿人类示范的关键特性，可在可控环境下复现行为克隆的多个反常现象，例如模型越过拟合性能反而越强、无 action chunking 的闭环策略完全失效。OCBench 配备 GPU 加速环境和脚本化策略，支持对已有行为克隆现象进行科学分析并检验、反驳各类假设。
正文
Abstract:Behavioral cloning (BC), despite its simplicity, exhibits many counterintuitive phenomena in the real world. For example, the performance of BC often keeps increasing as the model overfits more to the dataset, and fully closed-loop policies often completely fail without action chunking. Unfortunately, properly studying these anecdotal phenomena ("behavioral cloning mysteries") is challenging: in the real world, datasets and experiments are costly and not fully controllable; in simulation with synthetic data, these phenomena are often not easily observed partly due to the discrepancy between scripted policies and human demonstrations. In this work, we propose OCBench, a robotic manipulation benchmark with controllable scripted policies that have similar properties to human demonstrations. We show that, by mimicking key properties of human demonstrations, OCBench reproduces many anecdotal BC-related phenomena in controlled settings. With its GPU-accelerated environments and scripted policies, we demonstrate how OCBench enables scientific studies of previously reported BC-related phenomena by analyzing and refuting various hypotheses. Project page: this https URL
| Subjects: | Robotics (cs.RO); Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.07056 [cs.RO] |
| (or arXiv:2610.07056v1 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2610.07056 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Seohong Park [view email]
[v1]
Mon, 5 Oct 2026 04:33:28 UTC (6,996 KB)
来源：arXiv：cs.LG · arxiv.org