Robotics · 7 November 2026 · 4 min read

E.ON will train a robot for the substations. It has not said how many

NEURA, E.ON and enviaM said on 5 November they will develop robot skills for grid inspections, including substations, at the Munich robot gym. MITNETZ STROM expects a first try on the real network during 2027. The note has no robot count, no price and no pass mark.

A man in black stands beside a white and black NEURA humanoid on a dark stage, in front of a large NEURA screen. — A photograph from Automatica 2025 of a NEURA humanoid beside a person, in front of a NEURA screen. It is a trade-show pose. It is not a substation, and it is not MITNETZ STROM.

NEURA Robotics said on 5 November, from Metzingen, that it is partnering with E.ON and enviaM to develop what it calls physical AI for critical energy infrastructure, and to put it into practice over time. The work starts in a gym, not on a live line. Skills are to be trained and tested at the TUM RoboGym, which NEURA and the Technical University of Munich opened in October at LabCampus by Munich Airport. E.ON is joining that gym. The release says it covers more than 2,300 square metres, and that companies from more than ten industries are already training workflows there. A place in the gym is not a robot in a yard.

The jobs named for the start are inspections, monitoring of critical assets including substations, and support for site security. Tasks in hazardous places will be assessed as possible uses. Assessed is not assigned. MITNETZ STROM, enviaM’s grid company, is the pilot. It is based in Kabelsketal and runs about 71,000 kilometres of distribution network in parts of Brandenburg, Saxony, Saxony-Anhalt and Thuringia. That length is the network it already operates. It is not a route a robot has walked. The partners expect initial practical experience on the real grid during 2027. Expects, during a year, is not a date and not a switch-on.

Who keeps the skill

The platform is Neuraverse, which NEURA describes as the link between training and later use: robots, models, developers and partners, with a skill trained in the gym meant to become reusable on compatible robots and further sites. Meant to is the aim. E.ON and enviaM, the note says, “always retain control of their own data and trained skills and decide for themselves whether, and with whom, to share them.” Control of a skill that has not been trained yet is a clause in a partnership, not a measured inspection. The release does not say how many robots, which model, what a pass looks like, or what anyone is paying.