Consider a collaborative task carried out by two autonomous agents that are able to communicate over a noisy channel. Each agent is only aware of its own state, while the accomplishment of the task depends on the value of the joint state of both agents. As an example, both agents must simultaneously reach a certain location of the environment, while only being aware of their respective positions. Assuming the presence of feedback in the form of a common reward to the agents, a conventional approach would apply separately: (i) an off-the-shelf coding and decoding scheme in order to enhance the reliability of the communication of the state of one agent to the other; and (ii) a standard multi-agent reinforcement learning strategy to learn how to act in the resulting environment. In this work, it is demonstrated that the performance of the collaborative task can be improved if the agents learn jointly how to communicate and to act, even in the presence of a delay in the communication channel.
View on arXiv