I’m about to embark on a new course, teaching Multiagent Systems (MAS).
The MAS field centers on the idea that multiple intelligent agents (aka intelligent systems) can operate independently, yet cooperatively, within some environment. The notion that agents are intelligent means that not only must systems be able to operate autonomously within an environment but they must also have the capability to learn and adapt to changes in their environment. This concept is analogous to humans behavior.
What makes these systems particularly intriguing are the approaches one must take in designing such agents. Even within competitive environments, cooperation is required. After all, if I want to sell a product, I must consider what an individual wants/needs and also what the individual is willing to relinquish to buy my product. Similarly, within cooperative environments, where the end goals of multiple individuals are mutual, a person must still conserve their resources and also decide what tasks to perform that will be in the best interests of the community.
With this basic understanding of MAS, I am currently engulfed in a myriad of textbooks and academic papers exploring MAS in depth. I am still trying to figure out the best approach to teaching this course. In the end, I am leaning towards Michael Wooldridge’s Intro to MAS textbook with supplemental material from Gerhard Weiss’s edited text Multiagent Systems in addition to relevant academic articles. In the end, I feel that Wooldridge’s text will provide students with the required background on MAS to prepare them for designing and programming these systems in the future.
While background and theory of MAS is critical, Computer Science students must also be able to engineer and program MAS components on some level. Two frameworks I am looking into right now is JaCaMo (Jason, Cartago Moise), and Jade (Java Agent Development) , both built in Java. The next step is to find some realworld datasets that will provide students with environments upon which to develop intelligent agents.
Stay Tuned.