level_exampleFPSEvolver is a competitive multiplayer FPS game in which a group of players can generate, play and improve levels to t their particular preferences by voting on a selection of evolving levels. Supervised creation by Peter Thorup lsted and Benjamin Ma (Master Students). Released January 2015. Available at:

EvoCommander is a game that allows players to arenainteractively evolve neural networks and then switch between them during battle. Supervised creation by Daniel Jallov (Master Student). Release Fall 2014. Available at:

UnityNEAT a port of SharpNEAT from pure C# 4.0 to Unity 4.x (using Mono 2.6), integrated to work with Unity scenes for evaluation. Supervised creation by Daniel Jallov (Master Student). Released fall 2014. Available at:

DrawCompileEvolve website for drawing image that can then be further evolved interactively. Supervised creation of program by Rasmus Taarnby and Jinhong Zhan (Master Students). Released fall 2014. Available at:

BrainCrafter is an online program that allows users to manually build arti cial neural networks to control a robot. BrainCrafter was designed to study how good humans are at building complex networks for control problems and if collaborating with other users can facilitate this process. Supervised creation of program by Peter Greve and Jan Piskur (Master Students). Released Fall 2014. Available at:

Petalz is a social Facebook game that demonstrates novel AI technology invented at the University of Central Florida. Petalz allows the player to breed an unlimited variety of di erent owers. As the project leader I originated the idea for the game and supervised its creation. Petalz is the rst game that combines unique user generated content with social gaming on Facebook and is the rst product of our company FinchBeak LLC. Publicly available at:

ES-HyperNEAT C# is a public implementation of the evolvable-substrate HyperNEAT (ES-HyperNEAT) algorithm in C#. It is build upon the HyperSharpNEAT-Compatible Multiagent Simulator and Experimental Platform. ES-HyperNEAT is an extension of the original HyperNEAT method for evolving large-scale arti cial neural networks. While in the original HyperNEAT the human user had to decide the placement and number of hidden neurons, ES-HyperNEAT can determine the proper density and position of hidden neurons entirely on its own while still preserving the advances introduced by the original HyperNEAT. Released Spring 2012. My implementation of Evolvable-Substrate HyperNEAT (ES-HyperNEAT), can be directly downloaded from here. More information on ES-HyperNEAT can be found at:

HyperSharpNEAT-Compatible Multiagent Simulator and Experimental Platform is an extensible single- and multiagent experimental platform. It contains an updated implementation of HyperSharpNEAT that is a modi cation of the SharpNEAT package by Colin Green. The package includes an implementation of the room-clearing experiments described in our AAMAS 2010 paper. Released Fall 2010, Eplex Research Group, UCF. Available at: