Physical AI News (Japan)

Physical AI, robotics and compute — reported from Japan.

日本語
· Partnerships & M&A· 日本語で読む

ACCESS supports NTT DOCOMO's VLA model research and validation

ACCESS announced that it supported NTT DOCOMO's research and validation of VLA models.

What happened

ACCESS announced that it supported research and validation related to VLA models being pursued by NTT DOCOMO. The support included research and development of VLA models that handle force information, testing on physical robots, and a demonstration experiment on remote robot control under conditions with communication delays.

Why it matters

The announcement confirms that ACCESS added an extension to a model linking vision, language and action to handle force information, and tested it on physical robots. It also tested the effects of communication delays on remote robot control. For researchers and developers working on remote robot operation, the announcement identifies areas for testing when considering control designs for tasks that are difficult to handle using vision or language alone, or for situations with unstable communications.

Sources

Translated from the Japanese original: /news/access-ntt-dokomo-no-vla-moderu-kenshou-kaihatsu-wo-shien-fi-d826e8. The Japanese version is authoritative if they differ.