Battleground: An AI Challenge

Pick available AI
Pick available AI
Game Start
Upload a new AI
Game Finish

Inspiration

We love hacking, as well as Artificial Intelligence and Machine Learning. Hence we thought: why not combine both of them? That's how we came out with this idea of an AI Challenge Platform.

What it does

Battleground: An AI Challenge is a platform whose world consists of a hexagonal grid. There are 2 players, and each of them starts with a "Hive" which will spawn unit every turn. There are also obstacles and free Hives that can be captured by players. The objective of this game is to either wipe out your enemy, or control as many Hives as possible throughout the game. In case of a tie, number of units maintained in the map throughout the game will be the tiebreaker.

Players have to create an AI program (in the form of Python script whose skeleton is provided) to control the movement of all the units. Units will automatically engage in battle with the enemy units when they come close enough to enter the units' attack radius. The battle resolution mechanism provides wide range of strategy that can be explored by players for their AI. As example, players can form some formations for their units to gain advantage in battle.

"Line of Sight" system is also implemented to enable more strategic options. It means that players only can get the vision of the area around their units. This feature makes scouting and map control become important in winning the game.

We hope that players can have fun exploring different AI approach in winning this game. Have fun!

How we built it

This system is built with Python with Tornado as the framework to achieve fast prototyping. The battleground is created with HTML5 Canvas.

Challenges we ran into

Deciding the game mechanics was one of the toughest phase, because there are so many possibilities and game plays that can be deployed. Hence we tried our best to fine tune each element in this platform. In addition, time constraint also limits us. After finishing the platform, we still need to think of some AI algorithms to try out the system. It means that we had to solve 2 problems in one hackathon (It was challenging, but we enjoyed it!). The last problem is we didn't have enough time to deploy this system to the real server due to some implementation constraints and also time constraints.

Accomplishments that we're proud of

We are very happy (and proud of ourselves) to be able to realize this system. We really hope that this platform can help people to learn more about AI and have fun at the same time. At the same time, we felt that we tackled two problems (creating the system and creating some AIs to test out the system) in this hackathon. We really pushed ourselves to the limits because we believe sky is the limit!

What we learned

We learned that AI is really a sophisticated field that can provide us countless possibilities. Creating an AI to control some simple units is not as easy as we though.

What's next for Battleground

We will try to make more tuning to achieve better game play balance. We also will push the system to the real server so that more people can try this system.