G.O.N.C.

Jesse Burstyn and Patrick McKenna

The genetically optimized neural controller (G.O.N.C.) was Jesse and Patrick's undergraduate honour's thesis project. The aim of G.O.N.C. was to design and train a four legged 3D model to walk on even terrain. Specifically, the main research goals were to design a Elman artifical recurrent neural network to control the joints of the model, and to test the efficacy of various evolutionary algorithms to train the network.