Machenta develops simulation-trained robot foundation models designed to learn new physical tasks from a single human demonstration, without task-specific retraining. The Imperial College London spinout focuses on instant robot learning across different robotic platforms, including humanoids.
Machenta is a Physical AI spinout from Imperial College London's Robot Learning Lab, founded by Professor Edward Johns and Vitalis Vosylius with colleagues from the lab. Its Instant Robot Learning approach uses robot foundation models pre-trained entirely in simulation to acquire new tasks from a single demonstration, without further task-specific model training. The team has demonstrated packing and tidying tasks across different robot platforms, including humanoids. In October 2026, Machenta announced a $40 million seed round co-led by Creandum and Union Square Ventures, which it plans to use to scale its model and expand its London team and laboratory. The technology remains at an early research and development stage.