AI model masters new terrain at NASA facility one scoop at a time

Researchers at the University of Illinois Urbana-Champaign developed an AI model that enables robots to autonomously assess and scoop materials on unfamiliar extraterrestrial terrains. The model was trained using data from various materials and then tested on a robotic arm at NASA's Jet Propulsion Laboratory.

Remarkably, the AI successfully adapted to the new terrains without prior training, identifying and scooping suitable materials after an initial failed attempt on a rock-like surface. The robotic arm then moved to a finer grain material and successfully scooped it. The model's ability to quickly learn and adapt demonstrates its potential for autonomous exploration and resource utilization on other planets.

The team plans to expand the research for more complex autonomous excavation and construction tasks.