Artificial Intelligence in Modern Board Games

This project implements Lost Cities, a two-player expedition card game, as part of a Summer Science Research Program with Dr. Sean McCulloch at Ohio Wesleyan University. The core game logic is written in Java and wrapped by a Node.js and Socket.IO web interface so the game can be played in the browser.

The AI layer supports greedy, minimax, and alpha-beta agents. The search agents evaluate complete turns by simulating both the outgoing card and incoming draw, then scoring the best reachable expedition outcome from the cards actually available in hand. That keeps the agents adaptable without hard-coded opening thresholds.

Highlights

• Java implementation of full Lost Cities rules and scoring
• Greedy, minimax, and alpha-beta AI modes for human-vs-AI and AI-vs-AI play
• Known-card search evaluation that avoids weak expedition starts without tuned thresholds
• Browser-playable Node.js and Socket.IO wrapper around the Java engine
• Regression self-check for fixed-seed AI-vs-AI termination