Odyssey-3世界模型以20小时仿真数据实现真实驾驶
This is potentially a great news for robot learning:
物理AI领域的重要突破,展示了通用世界模型大幅降低机器人数据需求的可能性,值得关注其后续验证。
This is potentially a great news for robot learning:
这对机器人学习来说可能是一个重大利好消息:
Odyssey-3 (a foundation world model) just dropped and its attacking robotics' biggest bottleneck, task-specific data.
Odyssey-3(一个基础世界模型)刚刚发布,它攻克了机器人领域最大的瓶颈——特定任务的数据。
Odyssey's approach is to start with a pretrained world model, then attach an action decoder trained on tens of hours of robot demonstrations.
Odyssey 的方法是从一个预训练的世界模型开始,然后附加一个在数十小时机器人演示数据上训练的动作解码器。
Odyssey-3 drove real Indian streets after its driving policy trained on only 20 hours of simulation.
Odyssey-3 仅用 20 小时的模拟数据训练驾驶策略后,便能在真实的印度街道上行驶。
A lot of physical AI progress still comes from throwing more demonstrations at one narrowly defined system. Odyssey is arguing that broad world-model pretraining can absorb much of the knowledge that every downstream policy would otherwise need to relearn.
许多物理 AI 的进展仍然来自于向一个定义狭窄的系统投入更多的演示数据。而 Odyssey 的观点是,广泛的世界模型预训练可以吸收每个下游策略原本需要重新学习的绝大部分知识。
Their reported numbers are early but unusually small: tens of hours for robot arms, tens of hours of humanoid teleoperation, 20 hours of simulated driving, and tens of hours of simulated drone data.
他们报告的数据量处于早期阶段且异常之小:机械臂仅需数十小时,人形机器人的遥操作仅需数十小时,模拟驾驶仅需 20 小时,模拟无人机数据也仅为数十小时。
The adaptation layer then learns how to turn Odyssey-3's representations into the actions of that specific system.
随后,适配层学习如何将 Odyssey-3 的表征转化为该特定系统的动作。
This is basically a data-allocation bet. Spend the huge pretraining budget once on a general model of the world, then spend far less experiential data teaching each new body its controls.
这本质上是一种数据分配策略。将巨大的预训练预算一次性投入到一个通用的世界模型上,然后用更少的经验数据来教导每个新身体掌握其控制方式。
更进一步:量化金融体系
看懂新闻只是起点——沿量化金融路径,把它变成能交付的工程能力