Master Thesis

Distributed control based on evolutionary game theory: Multi-agent experiment

Work default illustration

Information

  • Started: 15/06/2015
  • Finished: 11/05/2016

Description

The objective of the project is to design a platform that may illustrate simple multi-agent algorithms. The mindstorm of LEGO is going to be used to construct each agent, and they are going to be managed through G language taking advantage of the provided toolkits NXT http://www.ni.com/academic/mindstorms/).

With the design virtual instruments, and the platform of the mindstorm LEGO with di erent agents, then several multi-agent experiments may be implemented. The multi-agent approach allows to illustrate the performance of multi-variable control systems for large-scale systems. The objective is to test consensus algorithms to achieve an agreement among all the agents. Moreover, it is desired to test di erent algorithms of game theory to nd an optimal point in a multi-agent situation. Moreover, experiments considering only local information are of interest in the design of distributed and decentralized controllers.

The work is under the scope of the following projects:

  • ECOCIS: EConomic Operation of Critical Infrastructure Systems (web)