Machenta Raises $40 Million for Robotics Model Pretrained in Simulation

London robotics startup Machenta has raised $40 million in a seed round to expand a simulation-pretrained model designed to help robots learn physical tasks from one demonstration after deployment.

Creandum and Union Square Ventures jointly led the round. Its foundation model is pretrained entirely in simulation, without real-world demonstration data. After deployment, one demonstration can teach a robot a new task, such as packing or organizing.

Professor Edward Johns of Imperial College London’s robotics learning lab and doctoral student Vitalis Vosylius co-founded Machenta. The company emerged from stealth this week with more than 10 team members and no product on the market.

Machenta plans to use the funding to expand its models and grow its team in London’s King’s Cross district.