CHINA'S WERIDE said it has launched WeRide WITT, a self-developed physical-AI
cognitive foundation model. Built on visual-language model capabilities, WITT
introduces the "minimum physical fact unit" concept to integrate video, image
and text multimodal data, decomposing continuous real-world scenes into
identifiable, verifiable fact units and forming a new AI understanding framework
centered on physical facts.