Course: Robotics Motion and Control Apr 2025 – Jun 2025
Overview
Engineered an end-to-end chess robot in ROS, integrating a KINOVA Gen3 Lite arm and RealSense camera for a full perception-decision-actuation loop, enabling a robot to play Chinese Chess against human players.
Key Highlights
- Full System Integration: Built a complete perception-decision-actuation pipeline in ROS, integrating a KINOVA Gen3 Lite arm and Intel RealSense camera.
- Computer Vision & AI: Deployed a custom-trained YOLOv8s model for real-time piece recognition and designed a queue-based filter to stabilize perception against motion-induced errors.
- Robotic Motion Planning: Implemented a custom, optimization-based inverse kinematics (IK) solver and cubic polynomial trajectory planner to achieve high-precision manipulation with positional errors under $10^{-5}$ m (Computational Error Limit).
Gallery
System Setup
Chessboard Localization
Chess pieces randomly located
Chess Piece Detection by YoloV8s