Published Paper Details:

Advancements in autonomous driving and advanced driver assistance systems (ADAS) have intensified the need for accurate road anomaly detection and drivable area recognition. This paper puts forward the Dynamic Fusion Module, a new approach intended to improve these vital aspects of autonomous navigation. The module combines data from multiple sensors using a dynamic fusion technique to generate a thorough and consistent understanding of the environment. The proposed method aims to address the difficulties associated with road anomaly detection, in particular potholes and other road surface irregularities. This goal will be accomplished by harnessing state-of-the-art Convolutional Neural Networks (CNNs), specifically Yolov8 and ResNet architectures. By addressing the limitations of current systems, our project seeks to transform road anomaly detection, providing enhanced accuracy and robustness across diverse real-world conditions, while harnessing the power of deep learning with Yolov8 and ResNet as our selected backbone architectures. The overall fitness score of 0.96 reflects a well-trained model with the potential for fine-tuning.