December 12, 2011
Source Files: Attach:dhunt_final_deliverable.zip
My video game "A-Star" is a strategy game that involves getting your autonomous robot to the goal at the center before your opponent robots get there. You can move blocks around, attack, and most importantly change your opponent's search algorithm heuristic to generate sub-optimal paths.
- A* search is used by all agents in the game to calculate the shortest path to the goal at the center.
- Adversarial Game AI is used by the computers to try and stop other robots from advancing.
My game focuses around search algorithms. It is innovative because of the idea of gaining an advantage through algorithm modification. Strong opponents can only be defeated by mastering the search heuristics.
Evaluation of Results
In creating this project, I have made an interesting and fun game. The concept of racing to a goal is fairly straightforward, but changing an opponent's path through search heuristics forces the best players to learn the search patterns. This will teach players about search algorithms while creating a fun experience. I will be evaluating the number of downloads over the next few weeks. If it receives a fair amount of downloads then I will consider it a success. My goal is to give people a fun playing experience with challenging AI while teaching the A* algorithm.
Creating a full game in three weeks is no easy task. If you value your social life, I don't recommend it. With that said, here's a shameless plug for my previous game which took over a year and a half. It uses much more adversarial AI than A-Star. Agents react to your moves and attack based on your location and current heading: