Physical AI x IOWN x 60GHz Wireless LAN: Successful Demonstration of Advanced Industrial Plant Inspection

NTT Group (NTT East, NTT Docomo Business, NTT Docomo Solutions, NTT Data Group), 1Finity, and Mitsubishi Chemical are working to reduce the burden on workers inspecting outdoor factory equipment through the realization of smart maintenance using IOWN APN and AI. In February 2026, the three parties constructed a high-capacity, low-latency communication environment combining IOWN APN and 60GHz wireless LAN at the Mizushima Industrial Complex in Okayama Prefecture, confirming the ability to collect large amounts of data in real-time. This time, using this communication environment, they conducted a verification on the advancement of outdoor equipment inspection utilizing physical AI technologies such as autonomous robots and digital twins. Specifically, they verified remote control and autonomous driving of robots in outdoor environments, low-latency transmission of data acquired by cameras and microphones, abnormal vibration and sound detection by AI, and crack detection reflected in a digital twin environment, obtaining effective results for practical application. In the future, they will promote the reduction of the burden on field workers by integrating diverse data such as video, sound, odor, and temperature to realize advanced cognitive functions that can replace humans. In the verification, an IOWN APN environment was built between the Mitsubishi Chemical Okayama plant and an NTT Group building in Tokyo (approx. 700km), and a wide-area wireless communication environment was realized by placing WiGig access points on-site. The robots autonomously drove while avoiding obstacles, and the acquired video data was transmitted with a latency of less than 500ms. They also succeeded in detecting abnormal sounds from pumps and instantly visualizing concrete cracks on a digital twin. These results enable the centralization of inspection personnel and parallel inspection of multiple sites, contributing to the efficiency of factory equipment inspection at complexes. Moving forward, they plan to conduct further demonstrations toward the advancement of computing infrastructure compatible with multimodal AI processing.