Procedural methods have long been used for generation of art assets, but procedural generation of scenarios has lagged behind. In particular, training games for emergency rescue workers would benefit from procedural scenario generation guided by pedagogical goals. Users could select what skills they wish to train for, and the system would generate a unique level containing the elements necessary to train those skills. We propose such a system using HTN planning and abstract representations of buildings.
Hullett, K. and Mateas, M. 2009. Scenario generation for emergency rescue training games. In Proceedings of the 4th international Conference on Foundations of Digital Games (Orlando, Florida, April 26 - 30, 2009). FDG '09. ACM, New York, NY, 99-106.
The final version of the paper can be found here.
This project uses SHOP, an off-the-shelf HTN planner created by the University of Maryland's Automated Planning Group. I used JSHOP2, the Java implementation found here, but the Lisp implementation should work as well.
The scenario visualization tool is written in Python using the wxPython gui library.