Minerva Emerges From Stealth With Roger, a Humanoid Built to Defuse Bombs

Most humanoid robots want a job in a warehouse. Roger wants the job nobody else will take. Minerva Humanoids, a San Francisco startup that emerged from stealth on October 6, has unveiled a semi-autonomous humanoid built for bomb disposal, hazardous-materials response, and work on offshore energy rigs. The launch came with roughly $10 million in pre-seed funding led by General Catalyst, with Long Journey Ventures and Credo Ventures as co-leads, and backers including Hugging Face and Codecademy.

The premise is blunt. The most dangerous jobs on the planet still put human bodies inside them, and the humanoid form may be the best way to take the person out of the blast radius while keeping human judgment in the loop.

Why a humanoid, and why now

Bomb-disposal robots already exist. Most are tracked vehicles with a single manipulator arm, capable machines with clear limits. According to Minerva, tasks that require two-handed dexterity can still force a technician in a bomb suit to approach a device in person, which defeats the purpose of sending a robot at all.

The humanoid form closes that gap. Quadrupeds and tracked robots are excellent at everything up to roughly a foot off the ground, but poor at anything above it, and a device can be placed exactly where they cannot reach. Because the person who placed the device is human, reaching it with a human-like body is intuitive, and it lets teams follow the procedures they already have instead of inventing new ones. The work stays the same; the person in the bomb suit goes home.

The work stays the same; the person in the bomb suit goes home. Roger's job is to be the body nobody has to risk.

Shared autonomy: human judgment, robot reflexes

Roger is not autonomous in the way most humanoid hype implies. Minerva calls its approach "shared autonomy," and the division of labor is deliberate. A trained specialist wears a VR headset, sees through Roger's cameras, and performs fine manipulation with haptic gloves, while the onboard Minerva Intelligence system handles balance, navigation, and recovery after a fall. As CEO Sandor Felber puts it, the high-level intelligence is the human operator; the robot handles the reflexes.

The sensing suite is built for that partnership. Roger grants the operator multiple points of view and sensing modes on demand, including thermal imaging, giving the specialist superhuman sight rather than merely human-equivalent vision. The kinematics are tuned to minimize what Felber calls the human-to-robot embodiment gap, down to wrist axes that coincide the way a human arm's do, so the operator's movements translate naturally into the machine's.

Crucially, Minerva Intelligence runs on the robot itself, with no live cloud connection required. If Roger loses communications or power, it freezes and executes a controlled fall away from identified critical assets, rather than toppling into the thing everyone is trying to protect.

The sacrifice option

The starkest detail from Felber's launch interviews concerns a scenario no bomb squad wants to contemplate. If an explosive is not safely approachable or cannot be safely detonated, Roger can sacrifice itself. It can take on the render-safe process to the best of the operator's judgment, absorbing risk that no human should ever absorb, with no fear clouding the decision.

That framing matters. Roger is engineered, in Felber's words, to be the best possible avatar for the operator. The robot is expendable in a way the specialist is not, and the design accepts that honestly rather than treating it as an edge case.

Who is building it

The founding team is notable. Felber worked on humanoid AI research at MIT CSAIL and on Tesla's Optimus program. Co-founder and CTO Maurice Rahme spent more than five years at Boston Dynamics, where he helped take the Stretch warehouse robot from an early prototype to a commercially deployed product. Minerva says that experience let the company take Roger from a clean-sheet design to a walking humanoid in just over five months, in part by developing its software stack on third-party robots while its own hardware was being designed.

The supply chain tells a second story. Most software development takes place in the United States, the first prototypes were built in Germany, and the company says it plans to manufacture in both countries while avoiding dependence on single-source Chinese components. For a startup selling to public-safety and energy customers, that is as much a sales feature as an engineering choice.

From demos to paid pilots

Roger has already been shown where it will have to work. The company demonstrated it at the NATO EOD Centre of Excellence in Slovakia and completed on-site testing at one of the world's 10 busiest airports, where a visitor with no training and a language barrier reportedly put on the headset and began operating a humanoid immediately. Minerva is now launching its first paid pilots in energy and explosive ordnance disposal this fall, is working with the FBI on testing and upcoming pilots, and plans field testing across the US, UK, and Germany.

Felber also disclosed that Minerva is working with several governing bodies in the EU on force-protection efforts, and that roughly 70 to 80 percent of the hardware requirements overlap between energy and public safety applications. That overlap lets one core platform serve offshore rigs and bomb squads alike, configured per mission.

The next iteration of the hardware is already on the drawing board, with better sensor integration and higher payload capacity, since the central engineering tension is balancing dexterity against the ability to handle heavy loads. For now, though, Minerva's strategy stands apart in a crowded field: while everyone else builds general-purpose humanoids for factories, Roger is a specialist built for the rooms no human should have to enter.