AI Pulse
📄 论文解读

5520亿参数,机器人学得更快

机器人模型一直有个反常识的瓶颈:大家拼命堆机器人数据,却很少动「底座」。这篇把语言模型从 20 亿一路换到 5520 亿,训练方法完全不变,只换底座,看机器人能力怎么变。结果三件事同时变好:只给 10 次示范,就超过小底座喂 50 次的水平(87.8% 对 85.6%);换到没见过的场景,成功率 45.1%,小底座最多 31.3%;真机上插充电器,从 10 次成 4 次变成 9 次,提升集中在「对准接触」这种最难的环节。代价是每次动作要 216 毫秒,不算快但能用。它不是你明天能上手的东西,但它说明机器人不必从零学起,语言模型里那套「看懂世界」的能力可以直接借来用。

📄 原文摘要(英文)

Frontier Vision-Language Models (VLMs) such as GPT-6 Astra have demonstrated remarkable capabilities in robot control, suggesting that broad pretraining provides knowledge valuable for physical action. Yet most Vision-Language-Action (VLA) policies still rely on much smaller VLM backbones, leaving this potential largely untapped. We ask whether frontier-scale representations can provide a stronger foundation for robot learning. We introduce ScaleVLA, which connects a 552B DeepSeek-V4.1-Flash backbone to a trainable action DiT, and systematically compare VLMs from 2B to 552B under a shared training setup. We evaluate three fundamental capabilities of robot foundation models. In sample efficiency, just 10 demonstrations outperform a 2B backbone trained with 50 (87.8% vs. 85.6%). In generalization, LIBERO-PRO success reaches 45.1%, compared with at most 31.3% for smaller backbones. In precise manipulation, real-world charger insertion improves from 4/10 to 9/10 successes, with gains concentrated in contact-sensitive alignment. Despite its scale, ScaleVLA achieves 216 ms median latency per action chunk. Our findings suggest that frontier VLMs can serve not only as reasoning agents, but also as powerful representational foundations for efficient and generalizable robot policies.

arXiv 原文

订阅 AI Pulse

每天 08:00 · 12:30 · 18:30 · 23:50 更新