Daicel to Begin Physical AI Equipment Inspection Trial at Hiroshima’s Otake Plant in December

Daicel announced on the 17th that it will begin a trial in December at its Otake Plant in Otake, Hiroshima Prefecture, using physical AI for chemical plant patrols and equipment inspections. AVITA will lead AI development, using a quadruped robot to test its potential applications at the plant and establish operating conditions. The company said it aims to reduce employees’ workload and improve the working environment by automating inspection tasks.

Daicel is targeting inspection work at a chemical plant, which involves risks around the clock. If quadruped robots and AI can take over some patrols and anomaly checks, the trial could offer a basis for rethinking how workloads are allocated at the site. For AVITA, it is also an opportunity to demonstrate the conditions for applying its technology in a real-world setting—factory inspections—in addition to avatars and collaborative software.

What is new in this announcement is the application of physical AI to factory equipment inspections, with a trial at a high-risk site: a chemical plant. Using a quadruped robot to assess operating conditions, meanwhile, appears to be an extension of the established approach for the trial stage. The key question is not simply whether automation is possible, but how much of inspection work that spans day and night can be shifted to machines. In related stories, this newspaper reported that AVITA had focused on dividing roles between people and machines and enabling safe collaboration in AVITA launches human-in-the-loop physical AI combining remote operation and autonomous control and AVITA announces AVITA OS software platform for safe collaboration between avatars and robots. The Otake Plant trial can be seen as an effort to bring discussions of remote operation and collaboration into on-site inspection processes. However, the announcement does not yet make clear which inspection steps will be automated, under what conditions people will intervene or how far autonomy will extend. See the original announcement for details. From an industry perspective, the focus is less on the robot itself than on implementation that connects patrols, detection, anomaly assessment and verification of operating conditions. Even with quadruped robots, safety requirements, sensor integration and decisions on when to stop in response to anomalies can become bottlenecks at chemical plants. The significance of this trial will therefore depend not only on whether it reduces the labor required for equipment inspections, but also on whether it can establish on site which tasks AI will handle and which people will perform. The next points to watch are the equipment covered by the trial, the operating structure, the robot’s operating conditions and which parts of inspection work it is designed to replace.

Chemical plant inspections are an area of Japanese manufacturing where safety requirements are stringent and reliance on human labor tends to persist. If trials combining quadruped robots and AI like this one expand, they could create implementation-related demand for domestic companies involved in inspection equipment, sensors, control systems and safety design.