Autonomous driving robot cars in a small-scale city model

Photo: Archive UL FRI

Autonomous driving robots, equiped with camera and sensors have become reality. Meet Duckiebots, which are the result of the reasearch of doc. dr. Octavian Machidon from the Faculty of Computer and Information Science (FRI).

He has recently implemented a demonstration project centered on autonomous driving robots, integrating a spectrum of cutting-edge technologies, including computer vision, sensors, artificial intelligence, machine learning, and robotics. Within this project, dr. Machidon introduced the Duckietown, an innovative robotics and AI ecosystem tailored for both educational and research purposes, at FRI.

At its core are scale-model self-driving cars dubbed Duckiebots. These vehicles, equipped with an NVIDIA Jetson Nano mini-computer, cameras, and sensors, navigate autonomously within the miniature metropolis of Duckietown, which features roads, obstacles, traffic signals, and intersections. The Duckietown platform serves as a resource hub for researchers and students alike.

Duckietown (photo: archive UL FRI)

Dr. Machidon plans to use this platform to assess and advance his research focus on context-aware, energy-efficient, on-device deep learning techniques. This direction builds upon his prior exploration, conducted last year on drones for smart agriculture applications within his AgriAdapt European project.

For students, the Duckietown offers a hands-on environment to learn robot autonomy hands-on: how to make real robots take their own decisions and experiment with various self-driving car technologies, such as lane detection and tracking, vehicle recognition, obstacle avoidance, traffic sign recognition, and route planning. Through this innovative initiative, researchers and students can actively engage in practical implementations and empirical investigations in the realm of autonomous transportation.