New AI Lets Factory Robots Work Together Without a Boss

In factories, robots often need to work together to move materials and get things done. Traditionally, a central computer tells each robot what to do. But what if robots could coordinate on their own, like a team of workers who figure out tasks without a manager? That's what researchers have been exploring with a new AI method called FMAPPO. It combines simple sensors (like the ones in robot vacuums) with task information to help robots decide where to go and what to pick up.

The team first trained the robots in a realistic computer simulation, then tested the same AI on actual robots in a factory-like setting. They found that the robots could successfully collect parts and bring them to storage without a central boss. The AI even handled changes in how often it received commands, which is important because real-world communication isn't always perfect.

This research is a step toward more flexible factories where robots can adapt to changes and work together more naturally. Instead of programming every move, you could just tell the robots the goal and let them figure out how to achieve it. That could make factories more efficient and easier to reconfigure for new products. While the code isn't public yet, the team plans to share it after their paper is published, so others can build on this work.