Tic-Tac-Toe Playing Robot
Project Links
- GitHub Repository: View Code
- Video Demo: Watch Demo
Overview
Created a fully automated Tic-Tac-Toe playing robot using the Dobot Magician Lite that plays against human opponents without intervention. The system uses computer vision to detect human moves, Minimax algorithm to calculate optimal counter-moves, and robotic control to physically place game pieces.
Key Features
- Computer Vision Detection: Real-time game state analysis using OpenCV with HSV color detection
- Minimax AI: Perfect strategic gameplay - never loses, always wins or forces draw
- Autonomous Operation: Complete sense-think-act cycle without human intervention
- Interactive Gameplay: Player chooses color and turn order for customized experience
- Error Prevention: Color validation system prevents invalid moves
Technical Implementation
- Vision System: HSV-based color segmentation for robust block detection, perspective warping for accurate board analysis
- Game AI: Minimax algorithm with recursive tree search, evaluates all possible move sequences, guarantees optimal play
- Robot Control: Pre-calibrated coordinate system for 9 grid cells and block pallets, suction cup end-effector for reliable pick-and-place
- Hardware: Dobot Magician Lite, USB camera (top-down view), colored blocks (red/blue)
Bonus Features
- Player choice: Human or robot can go first
- Move validation: Detects and prevents color mismatches
- Turn-by-turn logging with game outcome announcement
Results
- 100% game completion rate without errors
- Perfect AI performance - never makes strategic mistakes
- Reliable detection under controlled lighting
- Smooth gameplay with consistent pick-and-place accuracy
Project Duration: October 2025
Course: RAS 545 - Robotics and Autonomous Systems (Midterm 1)
Institution: Arizona State University