Back to Projects
AI AUTOMATIONCOMPLETE

Ms. Pac-Man AI Agent

Autonomous game-playing AI using decision trees and ghost prediction for Ms. Pac-Man.

SOLO DEVELOPER  ·  Apr 1, 2025

Ms. Pac-Man AI Agent

An autonomous AI agent for Ms. Pac-Man built during Full Sail AI coursework. The agent uses decision tree logic with ghost prediction and pellet-path optimization to navigate the maze, avoid ghosts, and maximize score. Demonstrates real-time AI decision-making under time pressure.

// TECH STACK

JavaSYSTEMAIGame PlayingDecision TreesHeuristics

Contributions

  • 01Implemented decision tree for real-time ghost avoidance and pellet collection
  • 02Built ghost trajectory prediction using movement pattern analysis
  • 03Designed scoring heuristic balancing risk (ghost proximity) and reward (pellet density)
  • 04Optimized decision cycle to run within per-frame time budget

Technical Highlights

  • ▸Decision tree with ghost-distance and power-pellet state evaluation
  • ▸Predictive ghost movement model for safer path planning
  • ▸Score maximization heuristic tuned through iterative playtesting