Physical AI 前沿雷达 · 2026-09-21

语言:中文 · EN · JA

每日自动更新的具身智能前沿雷达:八条主线、三语精炼、证据可查。

lane distribution

今日重点

1. Physical AI Takes the Wheel: How the World’s Robotaxi Leaders Are Building With NVIDIA Technologies

The global robotaxi market — physical AI’s first commercial breakthrough — is projected to reach $400 billion by 2035, with over 6 million commercial vehicles in operation as driverless fleets are already moving people through some of the…

2. Fine-tuning a 350M Model for Better Structured Outputs in 100 GRPO Steps

3. NVIDIA Brings Real-Time AI to Broadcast, Sports and Global Streaming at IBC

At the IBC conference, running Sept. 11-14 in Amsterdam, the creative, technology and business communities are coming together to turn ideas into action and discuss innovations across the media and entertainment industries.

4. Quantization-Aware Healing: a compressed, 4-bit model that outperforms its full-precision original

5. The Best Way to Explore Lunar Craters Is a Giant Robot Ball

On a good day, the rock quarry in central Texas is about 370,000 kilometers (230,000 miles) from the moon. But last February, when Rishi Jangale watched his 1.8-meter-wide, 150-kilogram inflatable robot roll effortlessly over rocks,…

6. Piloting the world’s first double-blind AI evaluations

Piloting the world’s first double-blind AI evaluations

7. How to Train a Cross-Embodiment Robot Navigation Policy with AI Agents

Navigation enables a robot to turn perception and motion into purposeful autonomy. Unlike locomotion, which produces stable movement, navigation must be used to…

8. Introducing agentic video understanding with Gemini

前沿基线(人工策展)

1. Z-1: efficient RL post-training for flow-based VLA models

2. Gemini Robotics 2 brings whole body intelligence to robots

3. LeRobotDataset v3.0: a standardised, streamable format for robot learning data

4. RD-VLA: implicit test-time compute scaling via latent iterative reasoning

5. Newton: open-source GPU physics engine for robotics (NVIDIA, Google DeepMind, Disney Research)

6. NVIDIA Jetson Thor: up to 2070 FP4 TFLOPS and 128 GB at 40-130 W

主线分布

主线 #
基座模型(VLA / WAM) 3
数据引擎(采集 / 清洗 / scaling) 6
训练与自我改进(RL / 后训练) 5
仿真与评测(sim2real / 闭环) 5
边缘与实时性(端侧推理 / 控制频率) 7
安全、权限与合规 5
系统与编排(长程 / 多策略) 3
本体与供应链(硬件 / 成本) 3

近期节奏

cadence

evidence mix

合规与标准观察


本仓库为技术观察,不构成投资建议、产品承诺或安全认证结论。论文结论以作者自报为准。

方法论 · 历史归档