Researchers at the University of Tehran have introduced a distributed artificial intelligence framework designed to let multi-agent autonomous systems self-coordinate without relying on a central controller. By combining a distributed optimization framework with an actor-critic reinforcement learning architecture, the method enables individual machines to make real-time decisions using only local data and peer-to-peer communication with neighboring units, all while adhering to shared operational constraints.

By eliminating centralized processing bottlenecks, the algorithm offers a scalable and resilient solution that has been mathematically proven to converge into stable outcomes. Experts anticipate the technology could significantly boost reliability across decentralized computing, robotics, smart energy grids, municipal traffic management, driverless vehicles, and computational economics, according to Mehr News Agency & TV BRICS.