AVITA, Daicel to begin joint trial of quadruped robot for factory patrols and equipment inspections in December 2026

AVITA and Daicel announced on September 17, 2026, that they will begin a joint trial in December 2026 at Daicel’s Otake Plant in Otake, Hiroshima Prefecture, on using a quadruped robot for patrols and equipment inspections inside a manufacturing plant. The initial tests will use the “Unitree Go2,” and running tests have been underway since August 2026. The trial will assess the potential for use in an actual factory, the conditions needed for safe operation, inspection accuracy and operating procedures.

Daicel has pointed to the potential for quadruped robots and AI to support patrols and inspections that involve hazards or physical strain. AVITA describes its approach as Human-in-the-Loop physical AI that retains remote assistance, offering a way to bring automation into factory maintenance without assuming full autonomy.

What is new in this announcement is that a trial of a quadruped robot for patrols and equipment inspections at a chemical plant will begin in December 2026 at Daicel’s Otake Plant, an actual factory. The approach itself, however, builds on the Human-in-the-Loop physical AI AVITA announced on September 15. The model keeps people involved through remote assistance rather than assuming full autonomy, and the trial can be seen as extending that approach to factory maintenance. Our related article, AVITA launches Human-in-the-Loop physical AI, combining remote operation and autonomous control, described the combination of remote assistance and autonomous control as part of the company’s product and service strategy. The joint trial applies that approach to patrols and inspections at the Otake Plant, marking a shift from a conceptual announcement to on-site testing. However, what has been disclosed so far is the start of running tests and autonomous-navigation verification; the announcement does not go as far as confirming the adoption of production models or permanent operation. A broader implication for the industry is that the focus will be not only on robot performance but also on defining factory requirements, inspection accuracy, operating procedures and remote-assistance systems. Daicel will organize on-site conditions and evaluate the trial, while AVITA will develop AI and integrate it with robots. This could shift the main source of value from the robot alone to an operating package tailored to site requirements. The companies have also said they may select other robot types depending on the work and environment, not just quadruped robots. Which robots are suited to which patrol routes and inspection items will therefore be a key question. Many questions remain unanswered. For example, the companies have not specified how often autonomous navigation will be conducted after December 2026, what benchmarks will be used to assess inspection accuracy, or which robot and system configuration will be selected as better suited to the application. The announcement also does not detail the scope of patrols and equipment inspections or the conditions for moving to actual operation. A key next step will be determining whether the trial remains a one-off test or advances to the design of sustained operations.

Daicel will provide the factory environment for the trial, organize the work to be covered and on-site challenges, define requirements for equipment inspections, and evaluate the results and consider steps toward practical use. AVITA will develop AI tailored to the work, select and integrate robots, develop robots as needed, and provide technical support for the trial. AVITA says it has begun offering Human-in-the-Loop physical AI that extends to robots the expertise it gained in building systems for remote work using avatars, combining AI-based autonomous control with human remote assistance. The company also says it is pursuing a business partnership with KDDI in physical AI and developing its “AVITA OS” software platform.

Patrols and equipment inspections at chemical plants are areas in Japanese manufacturing that can depend heavily on labor and expert knowledge. Because this trial will need to address not only robot performance but also inspection requirements and operating procedures, whether robot operations tailored to on-site conditions can be standardized will be an important issue for factory maintenance in Japan.