NTT Docomo Develops Robot Coordination Platform

NTT Docomo announced on October 27, 2026, that it had developed a “Physical AI Coordination Platform” connecting personal AI agents with multiple robots. The platform aims to control robot behavior according to users’ attributes and surrounding conditions, supporting safe and comfortable use in people’s living spaces.

The platform combines three functions: coordination between AI agents and robots, control to determine whether an action is allowed, and consideration to generate behavior suited to the surroundings. It manages the process from a user’s instruction to an AI agent through the robot’s action and tracks execution. Connections use A2A, a standard communication protocol for AI agents; Docomo developed the permission controls, action-history tracking and context-aware behavior generation itself.

For permission controls, the platform checks user attributes such as age against rules registered for each situation. It stops disallowed actions before they occur and sends only approved instructions to robots. The platform manages user attributes internally and sends robots only action instructions. It also records whose instruction was given, when it was given and what action was carried out.

As a possible future restaurant use, Docomo envisions selecting a serving robot for orders placed through an AI agent, while preventing alcohol from being served to minors and foods containing allergens from being delivered to users with allergies. The robot could also move around a person who enters its path while it is delivering food, Docomo says. The company plans to assess the platform’s practicality through internal deployment, joint research with robot makers and field trials with businesses.

NTT Docomo announced on October 27, 2026, that it had developed a “Physical AI Coordination Platform” connecting personal AI agents with multiple robots. The platform aims to control robot behavior according to users’ attributes and surrounding conditions, supporting safe and comfortable use in people’s living spaces.

The platform combines three functions: coordination between AI agents and robots, control to determine whether an action is allowed, and consideration to generate behavior suited to the surroundings. It manages the process from a user’s instruction to an AI agent through the robot’s action and tracks execution. Connections use A2A, a standard communication protocol for AI agents; Docomo developed the permission controls, action-history tracking and context-aware behavior generation itself.

For permission controls, the platform checks user attributes such as age against rules registered for each situation. It stops disallowed actions before they occur and sends only approved instructions to robots. The platform manages user attributes internally and sends robots only action instructions. It also records whose instruction was given, when it was given and what action was carried out.

As a possible future restaurant use, Docomo envisions selecting a serving robot for orders placed through an AI agent, while preventing alcohol from being served to minors and foods containing allergens from being delivered to users with allergies. The robot could also move around a person who enters its path while it is delivering food, Docomo says. The company plans to assess the platform’s practicality through internal deployment, joint research with robot makers and field trials with businesses.