OmniBrain-0：面向具身智能的统一智能体模型
原标题：OmniBrain-0: A Unified Agentic Model for Embodied Intelligence
AI 摘要
研究团队提出 OmniBrain-0，一个轻量级统一具身智能体，通过共享高层决策模型协调操作与导航任务，整合任务理解、规划、进度跟踪与反馈驱动的失败恢复。其渐进式训练框架结合具身预训练、领域专家智能体蒸馏与多教师在线策略蒸馏（MOPD），并引入 Rubric-Guided Recursive Self-Improvement（RSI）数据飞轮。在 LIBERO-Pro、RoboCasa365、RoboTwin 2.0、R2R-CE 等基准及真实操作任务上，该模型以显著低于闭源教师的推理与执行开销实现较强跨域任务表现。
正文节选
OmniBrain-0: A Unified Agentic Model for Embodied Intelligence Abstract General-purpose embodied intelligence requires robots to coordinate diverse skills over long horizons, track task progress, and recover from failures. Specialist policies alone do not ensure task-level reliability, while powerful multimodal agents rely on costly reasoning and external experience that compact models struggle to reproduce. We introduce OmniBrain-0, a lightweight unified embodied agent that coordinates manipula