ASTRO (Autonomous System for Teaching Robotics) is a compact differential-drive mobile robot developed at CRTA (Regional Centre of Excellence for Robotic Technologies) at the Faculty of Mechanical Engineering and Naval Architecture (FSB), University of Zagreb. It was purpose-built as an educational and research platform that bridges theoretical knowledge in robotics and control engineering with hands-on practical experience.
The platform serves as a testbed for students and researchers working on motion control algorithms, suspension systems, and autonomous navigation. Its approachable form factor makes it equally suited for outreach activities aimed at inspiring interest in STEM fields, while its open architecture supports advanced research into topics such as Model Predictive Control (MPC), indoor localization, and data-driven autonomy.
ASTRO is currently being used for a dedicated autonomous driving curriculum powered by visual data from an Intel RealSense depth camera. Students drive the robot around a physical track, collect image datasets, train neural network classification models, and then deploy the trained models back onto real hardware — completing a end-to-end machine learning pipeline from data acquisition to real-world inference. This end-to-end workflow teaches practical skills in dataset collection, neural network training, and embedded model deployment.