Project description:

Multi-agent systems are rapidly finding applications in a variety of domains, including robotics, distributed control, telecommunications, and economics. Reinforcement learning (RL) is a powerful framework that allows an agent to behave near-optimally through a trial and error exploration of the environment. Although originally developed for single-agent settings, RL approaches have been extended to scenarios where multiple agents learn concurrently by interacting with each other. To deal with the strategic interaction between those rational decision-makers, game theory models are necessary. These are the basic theories surrounding the DeepMind AlphaStar project, for example.

On the left, a multi-agent reinforcemnt learning (MARL) system called AlphaStar is playing Starcraft II and was able to achieve grandmaster performance. On the right, real life example application of a MARL systems to be used in the future to enchance autonoumous driving cars capabilities.

Subject of design:

This project aims to analyze whether and how RL and game theory tools are used to solve multi-agent systems problems. We will run numerical simulations to compare the performance of different algorithms.

Field of knowledge:

Game theory, multi-agent systems, reinforcement learning