Journal of Advanced Robotics, Autonomous Systems and Human-Machine Interaction
Vision-Based Robotic Ball Tracking with Hough Transform and Modular 3D Estimation: A ROS2 Framework for Dynamic Environments
Abstract
Oussama Errouji
This paper introduces a ROS2-based robotic system for realtimeball detection, 3D localization, and autonomous tracking. Unlike traditional blob detection methods, the system employsthe Hough Circle Transform for robust 2D detection, coupled with a decoupled architecture where 3D estimation operates independently from motion control. The modular design enables simultaneous applications: reactive tracking via 2D data and 3D spatial analysis for external tasks. Evaluated in Gazebo simulations, the system demonstrates adaptability to varying lighting, ball sizes, and distances. Applications span automated ball collection, industrial sorting, and augmented reality interactions.