OmniGibson
PulseAugur coverage of OmniGibson — every cluster mentioning OmniGibson across labs, papers, and developer communities, ranked by signal.
3 天有情绪数据
-
李飞飞团队发布ESI-Bench，用于评估具身空间智能
李飞飞团队发布了一个名为ESI-Bench的新基准，用于评估AI的具身空间智能。与以往假设最优观察的基准不同，ESI-Bench要求AI代理主动采取行动来收集信息，从而闭合了感知-行动循环。使用GPT-5和Gemini等领先模型进行的初步测试表明，当前的AI在主动探索和决策方面存在困难，表现出“行动盲区”和元认知缺陷，这表明主要挑战在于战略行动而非纯粹的感知。
-
New ESI-Bench benchmark tests AI agents' active spatial reasoning
Researchers have introduced ESI-Bench, a new benchmark designed to evaluate embodied spatial intelligence in AI agents. This benchmark focuses on the perception-action loop, where agents actively explore their environme…
-
StereoPolicy enhances robot manipulation with stereo vision
Researchers have developed StereoPolicy, a new framework designed to enhance robotic manipulation by utilizing synchronized stereo image pairs. This approach strengthens geometric reasoning for robots, overcoming the de…