Eagle's Wings is an artificial player for the popular video game Angry Birds; it participated and represented University of Waterloo and Zazzle Inc. in the 2016 and the 2017 Angry Birds AI Competitions.

Eagle's Wings is the reigning champion of the competition!

It employs a simple multi-strategy affordance based structural analysis with a manually tuned utility to decide between the strategies. It was develped on top of a fork of the agent from Team DataLab, the champion in 2014. Both the tactic gameplay and the strategic gameplay are coded in java. Major efforts were made to employ machine learning method xgboost and deep reinforcement learning. Insights learned from xgboost models were used to formulate the strategies. Key values and utility are learned from 40k shots. Read the description at Eagle's Wing pdf.

1. Open Chrome on the webpage http://chrome.angrybirds.com and make sure you are using the SD version of the game
2. Currently, http://chrome.angrybirds.com is taken down, please refer to http://aibirds.org/basic-game-playing-software/chrome-issues.html (require team registration and login)
3. java -jar ABServer.jar
4. Open another instance of the terminal
5. java -jar ABSoftware.jar

1. ant jar (make sure you have installed Apache Ant)
2. Follow the steps of the above section (Running the agent)

ABSoftware.jar is all you need for running

Refer to Developer Guide for tuning and strategy outline