Using natural voice commands, a human operator interacts effortlessly with a 5G-connected robot dog powered by edge-hosted Vision-Language Models (VLMs). The demo system follows AI-RAN architecture design, which integrates AI capability directly into cellular network infrastructure to enable real-time perception, reasoning, and action execution required for physical AI use cases.

To deliver low-latency and reliable Physical AI for autonomous navigation, inspection, and task execution, we use an intelligent Service Management and Orchestration (SMO) solution behind the scenes to orchestrate radio, compute, and AI resources. The SMO solution is currently commercialised by NeuroRAN, a spinoff company from SUTD.

This demonstration was developed by the 𝗙𝘂𝘁𝘂𝗿𝗲 𝗖𝗼𝗺𝗺𝘂𝗻𝗶𝗰𝗮𝘁𝗶𝗼𝗻𝘀 𝗖𝗼𝗻𝗻𝗲𝗰𝘁𝗶𝘃𝗶𝘁𝘆 𝗟𝗮𝗯 (𝗙𝗖𝗖𝗟𝗮𝗯) at the Singapore University of Technology and Design (SUTD) and NeuroRAN, leveraging the Open RAN and AI-RAN testbed supported by the Future Communications R&D Programme (FCP). The demo showcases how AI-native 5G networks brings edge intelligence to next-generation robots, enabling real-time human-AI interaction and new physical AI applications.

YouTube Link: https://www.youtube.com/watch?si=QDFSS6nMskoB627U&v=LrLoQgVKZNI&feature=youtu.be