Autonomous vehicle using artificial intelligence and deep learning

This paper proposes a model for a self-driving car, transforming a 1/10 scale RC car into an autonomous vehicle using advanced AI and hardware components. The system utilizes a convolutional neural network (CNN) algorithm, Nvidia Jetson Nano Developer Kit, Leopard Imaging 136F0V wide-angle camera, Edimax Wi-Fi adapter, ResNet, YOLO, Jupyter Notebook, and OpenCV. The Leopard Imaging camera, connected to the Jetson Nano, collects real-time images and sensor data, processed by a server. By using input images to provide steering predictions, CNN makes lane identification possible. Monocular vision algorithms determine the distances to stop signs and traffic lights, while a Haar Cascade Classifier recognizes them. An image sensor facilitates front collision avoidance, stopping the car at safe distances from obstacles. Navigation commands (right, left, forward, stop) are transmitted to the car via the Jetson Nano, enabling autonomous driving neural network-based predictions and predefined rules. The model achieves self-navigation and collision avoidance cost-effectively, integrating hardcoded regulations and techniques for avoiding obstacles, and sophisticated object identification to ensure compliance with traffic regulations. AI and deep learning algorithms have transformed the automotive industry, paving the way for self-driving cars. This paper explores the implementation of an autonomous driving system using AI, image processing techniques, and deep learning algorithms, specifically CNNs with YOLO and ResNet, leveraging the Jetson Nano Developer Kit, a compact and powerful platform for AI applications. This approach promises safer and more efficient transportation through innovative use of AI and machine learning.