The project files here are all used in my thesis, "Methods and implementations for coordinated multi-agent learning". In another repository "reinforcement-learning", the implementations for popular single agent and multi-agents reinforcement learning methods are shown. Here, multi-agent actor-critic method is used in an agent-based system. For each folder inside this repository:

It is a Simulink block implemented in MATLAB for tracking the locations of rolling robots--Spheros.

It is MATLAB simulation for getting the parameter vectors for controlling the Spheros.

It is an agent-based system simulating the communication among agents in the model that I am dealing with. In this project, the agents are situated in one container--the main conrainer, in my PC.

It is also an agent-based system the same as above, but, the agents are situated in different containers: the broker agent is in the main container in my PC, while the two robot agents are on the other two containers in two Raspberry Pi.

It is the final implementation for an embedded design of an agent based system. And the moving of the Sphero is controlled by reinforcement learning method--MAAC.

The video is also uploaded in this repository, for viewing it online, visit the link: https://www.youtube.com/watch?v=7po-JgCF4vE