Researchers have developed a method to train multiple agents such as robots or drones to work together using multi-agent reinforcement learning, a type of artificial intelligence.