CardMaster - AI-Powered Omi Playing Robot
Led a five-member team building an autonomous Omi-playing robot with vision, RL, mobile, and embedded systems.
Project Overview
Built an autonomous Omi-playing robotic system that detects cards, makes strategic decisions in real time, and interacts with human players. The system combines YOLO-based card detection, reinforcement learning, Flutter, TensorFlow Lite, ONNX Runtime, and ESP32-CAM hardware.
Project Description
CardMaster is an autonomous Omi-playing robot built as an integrated AI, robotics, mobile, and embedded-systems project. The work focused on turning real card images into reliable game decisions and coordinating those decisions through a physical robot interface.
- Led a five-member team across task allocation, development tracking, integration, testing, and communication.
- Technical approach: integrated YOLO card detection, reinforcement learning decision logic, Flutter control UI, TensorFlow Lite, ONNX Runtime, and ESP32-CAM hardware.
- Trained the vision model on 10,000+ images for real-time card detection.
- Combined edge AI, mobile UI, and embedded hardware into an integrated robot system.
- Outcome: completed system integration and reached an ~85% gameplay win rate in testing.
Technologies Used
YOLOReinforcement LearningFlutterESP32-CAM
Project Links
Project Gallery
