Implementation of a robotic arm with machine vision in chess : integrating computer vision and robotics for intelligent gameplay

Abstract

This thesis presents the design and implementation of a robotic arm system equipped with machine vision to autonomously play chess. The project integrates hardware development, computer vision techniques, and control algorithms into a cohesive framework capable of perceiving, interpreting, and interacting with a physical chessboard. A modular software architecture was developed, comprising a main application that manages gameplay, communication, and the user interface, alongside a vision submodule that provides board recognition and move validation. The robotic arm executes moves using trajectory planning methods, including joint and Cartesian motion, with consideration of singularities and workspace constraints. Machine vision, implemented through Python and OpenCV, enables accurate detection of chess pieces and board states, which are then translated into Forsyth–Edwards Notation (FEN) for compatibility with the Stockfish chess engine. The system demonstrates reliable performance in recognising board configurations, planning robot motion, and executing moves with precision. Results confirm the feasibility of combining robotics and vision for interactive, rule‑based tasks, while highlighting opportunities for future improvements such as enhanced recognition accuracy, adaptive difficulty levels, and broader applications in human–machine collaboration