Hitachi, Agile Robots Partner to Bring Physical AI Into Factories

TOKYO — Hitachi and Agile Robots have formed a strategic partnership to develop physical-AI systems for factory work that is too variable for conventional automation, with plans to commercialize the resulting technology through their industrial businesses.

The companies announced the partnership Sept. 29. Hitachi will integrate its edge‑AI learning software with Agile ONE  humanoids and mobile systems to address variable factory tasks.
the firms will merge edge‑AI learning with humanoid and mobile robots to handle variable tasks such as parts handling, machine loading, assembly and line changeovers. The companies explained that these jobs often demand workers adjust to various parts, positions or operating conditions instead of following a fixed sequence.

Agile Robots said the partners will collaborate to commercialize the systems: Hitachi will market them via its HMAX Industry portfolio, and Agile Robots will sell to its own customers. No financial terms, named customer deployments or timetable for the first commercial system were disclosed.

Hitachi has been building the technology separately. In March, it described a physical-AI system that learns from operational data and can issue robot-control commands at up to 100 Hz. The company cited wire-harness handling as an example of work requiring both vision and tactile sensing. Those performance claims came from Hitachi rather than an independent evaluation.

A month later, Hitachi and Hitachi High-Tech announced an edge-AI semiconductor intended for industrial robots and other equipment. Hitachi said the chip processes sensor information locally and uses less power than conventional computing approaches.

Launched in November 2025, Agile ONE targets industrial tasks and features hands with 21 joints and force and tactile sensing. Agile Robots says its robot foundation models use real industrial, simulated and teleoperated data.

The Hitachi agreement does not specify which factory will host the technology, leaving timing and scale unclear.