Physical AI Exposes Three 'Fracture Layers'; DeepRoute Launches First Physical AI Lab in China

量子位 · wechat · 2026-08-10

As physical AI technical pathways converge into VLA, world models, and foundation models, the industry is exposing three major 'fracture layers': between model and data, vision and action, and simulation and reality. Bridging these fractures requires integrating data, algorithms, simulation, and engineering into a single continuously iterating loop.

In this context, autonomous driving company DeepRoute has established SuperfluidLab, China's first AI lab dedicated to foundational physical world capabilities. Led by former DeepRoute Chief Scientist Ruan Chong, the lab focuses on foundation models, VLA, world models, and multimodal understanding. It has also restructured its AI Infra and R&D organization to eliminate 'friction' in technical collaboration and model iteration.

Original post →

More from Companies & People

Companies & People channel →