Yani Julian

Pong A.I

Developed an A.I to play pong. The A.I was implemented using q-learning, a reinforcment learning algorithm. The A.I was trained to play two versions of Pong, against an opponent and against itself (solo mode). The A.I randomly choses an action (up, down, nothing) and is rewarded based on the outcome (win, loss, bounce). The A.I then remembers this to make future decisions.

This A.I was trained on one million training games which achieves over a 50% win rate against an opponent and has an average over ten bounces in solo mode with up to over 70 bounces in a single match.

3D Flight Simlator

Created a 3D flight simluator using WebGL and glMatrix. The terrain is randomly generated and each vertex is colored in the vertex shader based on the vertex height. There are four different intervals of vertice height to color them either white, brown, green or blue. The terrain is generated from a flat plane which is then randomly split at different points to increase/decrease the vertice height. The plane controls are able to pitch and roll along with increasing/decreasing the plane speed.