Hitachi Creates GenAI Tech to Interpret Industrial Blueprints

Hitachi, Ltd. has announced the development of a specialized learning technology that enables generative AI to accurately read and interpret industrial drawings, including electronic circuit diagrams, piping diagrams, and power distribution schematics commonly used in infrastructure and manufacturing.

The technology works by creating pairs of correct textual data that map the diagram image to its corresponding device connections. Training generative AI with these pairs allows it to flexibly handle a wide variety of industrial blueprints and accurately recognize complex element connections. In tests on electronic circuit diagrams, this approach improved the AI’s accuracy in identifying lines and arrow directions, previously prone to errors, with connection recognition increasing by approximately 220% compared to conventional methods.

Also Read: CTCSP Corporation unveils “Digital Workplace for AI” solution

By enabling precise digitization of industrial drawings, Hitachi’s technology facilitates rapid access to on-site knowledge, such as past design cases and maintenance records. This can support knowledge transfer from experienced engineers, improve maintenance and design efficiency, and enhance decision-making on-site. The innovation represents a significant step in applying generative AI to industrial knowledge management, making previously complex and error-prone diagram interpretation faster, more reliable, and more accessible.