Shawn Shen 团队推出 MemoriesUpload，为机器人采集第一人称工作视频数据
机器人圈常怪模型不行，这个团队反而去拍 20 多个行业里人类干活的视频，连手部动作都逐帧标注，已经有 3 家头部机器人实验室拿去训练了。做具身智能或关注数据瓶颈的都该看看这个思路。
Shawn Shen 团队发布 #MemoriesUpload 项目，核心观点是机器人行业瓶颈不在模型，而在于缺少高质量的人类真实工作第一人称视频数据。团队为此搭建了硬件、模型和基础设施协同的数据采集体系，开发周期超过一年。
该项目在 20 多个行业中实地拍摄人类执行真实工作的画面，并对每段视频中的手部动作进行逐帧标注，为机器人学习提供互联网规模的训练素材。目前已有 3 家全球排名前 5 的前沿机器人实验室在用这批数据训练模型。
机器人数据采集长期依赖遥操作或仿真，第一人称真实劳动视频逐帧标注的路线为数据供给提供了另一条路径，头部实验室的采用情况会是数据质量最直接的验证。
hot take: robots aren’t stuck because of models. they’re stuck because there’s barely any good video of humans doing real work this team is filming it across 20+ industries and labeling each hand frame by frame Shawn Shen @shawnshenjx Today we launch #MemoriesUpload . Robots need internet-scale first-person video of how people get things done. Collecting this at high quality takes hardware, model and infra working together. That's what we've spent over a year building. And already, 3 of the top 5 frontier robotics labs are training models with the data.
This report is categorized under AITOP's "Products" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include 机器人, MemoriesUpload, 数据标注, 具身智能.
Source attribution: curated from AI Breakfast (@AiBreakfast).