Hitachi initiates Physical AI project for excavators

Hitachi Construction Machinery is starting a Physical AI research project in Japan that will train an excavator using recorded human lever operations in combination with camara data and task instructions.

The OEM will use operator work footage and operation logs to train the AI with the goal of achieving “highly accurate and autonomous decision-making and construction even on actual construction sites”.

The application focus will be excavation and loading in construction and mining as well as debris removal at disaster sites.

The project has been selected for the country’s GENIAC (Generative AI Accelerator Challenge) programme, implemented by Japan’s Ministry of Economy, Trade and Industry (METI) and the New Energy and Industrial Technology Development Organization (NEDO).

Image: Hitachi Construction Machinery

Hitachi said Physical AI – interacting with the world of machines and sensors - was a technology that could autonomously execute processes including “understanding the surrounding environment (perception), work plan optimization (decisions), and operation execution (actions).”

‘Not just mechanical automation’

Physical AI is not simply mechanical automation, said the OEM, because it “understands the machinery and surrounding environment to autonomously carry out construction tasks.”

The manufacturer said the technology would help construction and mining industries facing labour shortages and improve safety and productivity in difficult work environments.

“Moreover, the importance of rapidly responding to increasingly severe natural disasters and ensuring a stable supply of mineral resources has also risen in recent years”, said Hitachi.

Working on the project alonside Hitachi will be Jizai Inc, a Tokyo-based generative AI and robotics specialist, and the Nara Institute of Science and Technology (NAIST), located near Osaka.