Every explosive-ordnance-disposal (EOD) operator knows the long walk. That is the phrase Dominic Asquith uses for the moment a bomb-disposal specialist has to put on the suit and approach a suspected device alone, because the tracked, single-arm robots the field has relied on for decades need a second hand they do not have.
Asquith, a British Special Forces veteran, now runs hazardous operations at Minerva Humanoids. The San Francisco company emerged from stealth on 6 October with about $10 million in pre-seed funding, led by General Catalyst, with Long Journey Ventures and Credo Ventures as co-leads and other backers including Hugging Face, Codecademy and RunwayVC.
Its robot, Roger, is a humanoid aimed squarely at the jobs that kill people.
A humanoid steered by a person
Roger is not an autonomy demo. A trained specialist wears a VR headset, sees through the robot's eyes, and performs fine manipulation from a safe distance. A control layer the company calls Minerva Intelligence handles balance, fall recovery and navigation, and translates the operator's input into movement. Minerva says operators learn full-body control in minutes.
Minerva is targeting onshore and offshore oil and gas, high-voltage plants, explosive ordnance disposal, hazardous-material response and public safety. It is exploring deployments at two of the world's ten busiest airports.
That division of labour is the bet. Roger does not try to solve unsupervised manipulation, a problem that has consumed years at far better-funded rivals. It moves a human out of the blast radius and leaves their judgement in the loop.
Five months, idea to walking
Roger went from concept to a working walking humanoid in five months, including bespoke design, manufacturing and assembly. Investors keep returning to that speed. "Building a humanoid from scratch is often a years-long project, but Sandor and Maurice did it in less than six months," said Robin Dechant, a partner at General Catalyst.
Sandor Felber, the chief executive, helped launch MIT CSAIL's research line on humanoids and worked on Tesla's Optimus program. Maurice Rahme, the chief technology officer, spent six years at Boston Dynamics as a technical lead on the Stretch robot, taking it from prototype to commercial product.
The tracked, single-arm systems we've relied on for decades send a lot of people on that walk, because the task needs two hands and real dexterity.— Dominic Asquith, Head of Hazardous Operations, Minerva Humanoids
The crowded middle
Roger sits between two crowds. General-purpose humanoid makers — Figure AI, Apptronik and Agility Robotics — chase warehouse, factory and home work, where dangerous-materials disposal is not a core market. On the other side sit the legacy single-arm EOD robots that have done the job for decades. The closest functional analogue may be the UK's Shadow Robot Company, which supplies dexterous teleoperated arms for nuclear decommissioning but does not build a full humanoid.
As we wrote in September, the humanoid race is splitting between machines that try to learn any task and machines built to learn one. Minerva has picked its corner. Capital has followed the whole category: London's Humanoid raised a $152 million Series A in July.
By the funding-data firm Dealroom's count, Minerva's pre-seed sits above the 95th percentile of US AI seed rounds. The signal is that investors will write large early cheques for a narrow, hazardous use case instead of a general-purpose promise.
A supply chain made of two countries
Minerva is explicit about where it builds. Software and AI are developed mainly in the United States; the first Roger prototypes are engineered and assembled in Germany, and the company plans to produce in both countries without relying on single-source Chinese components. That is a pointed line in a category where China builds almost every humanoid shipped worldwide. The company has also set up research partnerships, including one with the Mohamed bin Zayed University of Artificial Intelligence, to push humanoid capability into the energy sector.
What Minerva is not claiming is autonomy. Roger still needs a person at the other end, and the first paid pilots start this autumn. That may be the honest near-term humanoid: a remote body for a specialist who would rather not take the long walk. "We didn't set out to replace the people doing them," Felber said. "We set out to build a robot that lets them do the job efficiently from the comfort of their office or home."