Achievement

IGERT fellow, Leonard Kinniard-Heether, together with his advisor, Robert Reynolds, and their REU student, Tracy Lai, developed an extension of the Cultural Algorithm Toolkit (CAT 3.0 system). They used the CAT 3.0 system as the basis for training a Neural network controller to play the game of SuperMario. The Super Mario entry generated by the group won first place in the IEEE Super Mario Competition last summer (IEEE World Congress on Computational Intelligence, Barcelona Spain, July 18-23, 2010). Since then they have gone on to use the CAT 3.0 system to allow large herds of caribou to do goal-based optimal path-planning in virtual worlds designed to simulate the environment of ancient hunters 10,000 years ago in Michigan. The team of archeologists, oceanographic engineers, and computers scientists has already made meaningful discoveries because of this work.