Mitsubishi Electric says it has built the first machining system in which humanoid robots and CNC machine tools work together through physical AI, a workplace step beyond a robot duck walking on its own. The Mitsubishi humanoid CNC system uses the company’s Maisart AI brand to link the two sides. Drives and Controls reported the announcement on November 3.
The system has three parts. An AI machining task planner takes natural language instructions and turns them into tasks. It assigns and coordinates the work, because one planner must see the whole job. A CNC machine tool with physical AI exchanges operational information in natural language. A general purpose humanoid carries out the physical jobs.
Those jobs cover set up, cleaning, removal and transport of finished products, inspection, and troubleshooting. The humanoid handles the work around the machine, while the machine tool handles the cutting. As a result, the Mitsubishi humanoid CNC system pair can run without a human operator standing by.
The company claims the system needs no costly retrofitting and no complex integration for each robot. Existing machine tools can join the setup without a rebuild, according to the announcement. That claim matters for smaller factories, because a full line rebuild would put the system out of reach.

Keeping existing machines matters, because a factory cannot halt production for a rebuild. The Mitsubishi humanoid CNC system lets the planner learn each machine’s rhythm, and the humanoid adapts to the machine. That order keeps costs down, so smaller factories can join.
Mitsubishi Electric demonstrated the system at the Japan International Machine Tool Fair in Tokyo in October. Visitors saw humanoids and machine tools coordinating tasks, at the same time, in a shared workspace. The demo was a proof of concept, not a production line sale, and the company frames it as a step toward autonomous unmanned factories.
For the Mitsubishi humanoid CNC system, unmanned operation does not mean an empty building. It means no operator stands by each machine through a shift. The humanoid covers set up, cleaning, transport, inspection, and troubleshooting. The planner watches the whole line and reassigns work when a task stalls. Night shifts are the clearest early use, with fewer people on the floor and the line still needing tending.
The commercial target is 2030. Mitsubishi Electric aims to sell the system by that year, after further testing and refinement. The timeline gives the company room to prove reliability, because a factory system must run for long shifts without errors.

The Mitsubishi humanoid CNC system fits a wider push to put humanoids into real workplaces. Humanoids have already entered public events, from a robot duck to a robot race up Mount Tai. The regional race in embodied AI is moving fast, and factory work is one of the first places where humanoids must prove they earn their keep.
CNC machines themselves are becoming more accessible, from a ToolDance X1 CNC mill for small shops to large factory lines. Mitsubishi Electric’s system sits at the factory end of that range, where uptime and precision decide value. A humanoid that tends a machine must match the machine’s pace, not slow it down.
The road to 2030 runs, therefore, through longer trials. The Mitsubishi humanoid CNC system must show it holds up across shifts, parts, and factory layouts. Buyers will watch uptime first, because a stopped line costs money every minute. Safety around human workers will also shape the verdict, since humanoids share space with people.
The system still has to prove itself outside a trade fair demo. Real factories bring dust, vibration, changing parts, and tight deadlines, a harsher test than a robot race up Mount Tai. A humanoid that handles those conditions day after day will have earned its place. Mitsubishi Electric’s 2030 target leaves time for that proof, and the next demos will show whether the pace holds. Each new demo should add harder conditions, so buyers see progress rather than a repeat show.
Mitsubishi humanoid CNC system claims like this one will test buyer trust in the years ahead. In short, the demo showed the concept working in a shared workspace. The factory floor will decide whether the concept becomes a product. Buyers will judge the system on uptime, safety, and cost, because demos end and shifts do not.
