The problem...

Often researchers in the field of multi-agent machine learning write one-off programs to run experiments.  Others who want to use that work later often end up debugging poorly documented research code.  Either that, or they give up and start again from scratch.

Wouldn't it be great...

if there were a common format for problem specifications and policy algorithms? How about one that researchers can universally understand and exchange?  And execute in a generic, high-performance POMDP engine?  What if that same engine could run as a cluster and allow user-defined graphics output and logging?  What if it were written in a modern programming language like Scala?

The EnMAS POMDP syntax - lightweight yet powerful

And similar goodness for agents

Drop us a line

Stay tuned for further developments!  We are committed to making the API even more expressive.  If this project interests you we would like very much to hear about it.  Send mail to info@enmas.org.  Also, join the google group and mailing list for issues.