From UBC Wiki

Multi Agent System (ACO)


Frances Chong, Jacobus Philip Haupt

A multi-agent system is a system of simple interaction (but ultimately independent) between “intelligent” agents (that make decision based on rules on their interaction with the environment and other agents). Although each agent might have a simple goal/program, the emergent problem solving can be more sophisticated.

What is the problem?

To develop multi-agent planning framework/module that can be used to solve problems in a distributed fashion. Certain problem domains call for such limited agents; we intend to implement these many interacting agents in Prolog. We will be implementing travelling salesman problem using ant colony optimization (ACO).

What is something extra?

Displayed the pheromone levels through a graph representation between each city after the ants finished touring all the cities.

What did we learn from doing this?

Learned how to think logically. Propositions acted as a set of rules for the program. We also learned how to program in Prolog making us think outside of what we're used to. A benefit of Prolog is that it's a great language to write a program that has sets of rules. As well, pattern matching was huge thing we learned about. It's also quite fast. One thing that Prolog lacked was documentation for visualizations. We tried to find a library to would help us visualize our results, but found that most of the posts online were outdated and couldn't find many resources for visualization.

Links to code etc