OSCAR: 面向机器人的全具身骨架条件世界动作模型
OSCAR: Omni-Embodiment Action-Conditioned World Model for Robotics
AI总结：
提出OSCAR，一种基于动作条件的视频世界模型，通过大规模数据管道和2D骨架渲染统一表示，实现跨机器人具身的泛化，并用于策略评估。
摘要：
我们提出OSCAR，一种精确的动作条件视频世界模型，能够泛化到不同的机器人具身并支持机器人策略评估。现有的视频世界模型在真实机器人评估中面临三个主要挑战：当前机器人训练数据集的场景多样性有限、动作跟随不精确、以及跨具身泛化能力差以支持广泛采用。我们从两个角度应对这些挑战。其核心是一个大规模标准化数据管道，用于整理、过滤和去重广泛的机器人和以自我为中心的人类数据集，产生一个涵盖多样化任务、场景、动作和机器人具身的干净联合训练数据集。为了给视频模型提供条件，我们采用2D运动学骨架渲染作为统一的条件表示，能够泛化到不同的机器人手臂甚至人类手部。我们在单个GH200 GPU上微调Cosmos-Predict2.5-2B模型。与现有基线相比，我们的模型在动作跟随、外观质量和运动一致性方面取得了显著改进，而基线要么模型规模大得多，要么需要更多GPU。我们进一步将OSCAR部署到RoboArena中评估机器人策略。大量实验表明，OSCAR中的虚拟策略评估与真实世界评估之间存在显著相关性，为未来机器人策略可以纯粹在虚拟生成的世界中评估铺平了道路。
英文摘要：
We present OSCAR, a precise action-conditioned video world model that generalizes across different robot embodiments and enables robot policy evaluation. Existing video world models face three main challenges for real-world robot evaluation: limited scenario diversity in current robot training datasets, imprecise action following, and poor generalization across embodiments for broad adoption. We tackle these challenges from two perspectives. At its core is a large-scale standardized data pipeline that curates, filters, and deduplicates broad robotics and egocentric human datasets, yielding a clean joint-training dataset that spans diverse tasks, scenarios, actions, and robot embodiments. To condition the video model, we adopt 2D kinematic skeleton rendering as a unified conditioning representation that generalizes across different robot arms or even human hands. We finetune the Cosmos-Predict2.5-2B model on a single GH200 GPU. Our model achieves significant improvement on action following, appearance quality, and motion consistency, compared to existing baselines, which either have a much larger model size or require more GPUs. We further deploy OSCAR to evaluate robot policies from RoboArena. Extensive experiments demonstrate the significant correlation between our virtual policy evaluation in OSCAR and real-world evaluation, paving the way for the future where robot policies can be purely evaluated in virtual generated worlds.
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。
