Ex-DeepMind team launches startup that lets customers teach robots on the job
Former DeepMind engineers have founded Reimagine Robotics, a venture that enables users to train robot arms by demonstrating tasks directly.
Jonathan Scholz, former head of DeepMind's applied robotics team, launched Reimagine Robotics together with ex-Google engineers Oleg Sushkov, Akhil Raju and Misha Denil. The startup, emerging from stealth this month and funded by Fly Ventures and Firstminute Capital, aims to close the "100,000-year data gap" in robot learning by turning end-users into teachers. Its approach lets workers show a robot a task, watch its attempt, and correct it by moving the arm, a process the founders call post-training.
In a pilot with a company that recovers valuable materials from used hard drives, the method reduced the time needed to teach a new task from one day to ten minutes. Reimagine expects a new market of robot trainers to develop, allowing manufacturers to adapt robots without calling specialist teams. The company believes this hands-on model will prevent robots from becoming expensive paperweights and help bring them into everyday production lines.
Why it matters
It shows a practical way to make industrial robots adaptable without massive data, potentially speeding automation adoption.
In this story
