Summary: | Ensuring the safety of autonomous vehicles requires effective detection and tracking of surrounding objects. This paper proposes the design and development of a driverless transportation system module focused on identifying obstacles around vehicles. By integrating computer vision with deep learning, the system presents a reliable and cost-effective solution for autonomous driving. Utilizing Raspberry Pi 4B and a USB webcam, a compact hardware setup is created for seamless implementation in autonomous vehicles. The algorithm presented in this study enables the detection, classification, and tracking of both moving and stationary objects, including cars, buses, trucks, people, and motorcycles. TensorFlow Lite, a deep-learning network, is employed for efficient object detection and classification. Leveraging Python as the primary programming language, known for its high-level object-oriented features and integrated semantics, the algorithm is tailored for web and application development. Experimental results demonstrate the system’s capability to concurrently detect and identify multiple local objects with an accuracy ranging from 50% to 80% in day and night conditions. These findings underscore the potential of deep learning in advancing autonomous vehicle technology. © The Authors 2023. Published by University Malaysia Pahang Publishing. This is an open access article under the CC BY-NC 4.0 license
|