Hitachi Construction Machinery Starts R&D on Physical AI to Learn Hydraulic Excavator Operations

The company will develop technology that learns lever operations from camera footage and work instructions, then determines and performs appropriate operations based on the situation.

According to Hitachi Construction Machinery's announcement, the R&D aims to implement autonomous operation technology by teaching the system to operate hydraulic excavators. In construction equipment automation, operational data and learning design can be competitive factors alongside the machines themselves, making it important to determine which field tasks to include in training and how far to take that learning.

What is new in this announcement is that Hitachi Construction Machinery is starting industry-academia collaborative R&D under the GENIAC framework on physical AI for hydraulic excavators, teaching the system lever operations from camera footage and work instructions. The effort extends the established push toward autonomous operation and remote control of construction equipment. Hitachi Construction Machinery lists debris removal at disaster sites and excavation and loading at construction and mining sites as potential future applications, suggesting that it is building a foundation with practical use in mind. In our related article, [Yanmar Holdings and Hitachi Construction Machinery Reach Basic Agreement to Explore Collaboration on Compact Construction Equipment](/en/news/yanmar-holdings-and-hitachi-construction-machinery-agree-to-explore-co-bdcabd) (August 10, 2026), Hitachi Construction Machinery was exploring potential collaboration on compact construction equipment. This time, the collaboration shifts from finished vehicles and machine development to the intelligence layer of learning operator actions. It appears to show that, apart from competition over machine design and sales, how operational data is collected and used for training is becoming another competitive axis. Jizai's Advance Sales Begin for Palmimo DevKit concerns tabletop robots, but Hitachi Construction Machinery's initiative differs in its focus on a data foundation for learning field operations rather than on robot form. In terms of industry structure, the focus is on three layers: input data, learning methods and field deployment. Autonomous operation of hydraulic excavators depends not only on automating the machinery, but also on how much data can be accumulated to imitate and abstract operator actions. Selection for GENIAC indicates that the effort has moved beyond an experiment by a single company to a stage of testing feasibility through industry-academia collaboration. However, it is not yet clear how many machines will be used to gather training data or how much of the operation will be made autonomous. Please refer to the original source for details.

Japanese construction machinery manufacturers have strong positions, but individual companies often handle the collection of field data and development of learning infrastructure on their own. By highlighting GENIAC and industry-academia collaboration, Hitachi Construction Machinery has brought into focus how Japan's construction machinery sector can advance the accumulation and training of operational data, as well as safety validation, alongside machine development.