MIT researchers use AI to convert hand gestures into training data for robots

Researchers at MIT have developed a method that uses artificial intelligence to interpret hand gestures. The system translates these gestures into data that can train robotic systems. By capturing natural motions, the approach simplifies the creation of robot learning datasets. The technique could reduce the time and cost of programming robot behaviors. Early tests demonstrated successful training of robots to mimic demonstrated actions. The work highlights the potential of gesture‑based interfaces in robotics. MIT plans to refine the technology for broader industrial applications. Future research will explore scaling the method to more complex tasks.