Technology Adoption Modeling: Integrating Engineering Constraints And System Dynamics

Abstract

In this research, an adoption model for anIntelligent Transportation System (ITS) through the integration of engineering parameters and system dynamics for the detection and assessment of potholes can be developed. The suggested methodology involves the use of the Kaggle-based Potholes Detection You Only Look Once version 8(YOLOv8) dataset that includes annotated images of roads in different environmental conditions. At the first stage, Histogram Equalization (HE) is used during data pre-processing to increase the contrast of road images and make the potholes more visible. The Histogram of Oriented Gradients (HOG) approach can be used for feature extraction. Finally, a novel hybrid optimization model, Bonobo Optimizer and Extreme Gradient Boosting (BO-XGBoost), can be used to classify the potholes on the roads.Evaluations of the developed model have been conducted based on Mean Average Precision (mAP), mAP50, mAP75, and mAP50-95 measures, and it performs better by obtaining 0.989, 0.890, and 0.890 values, respectively. Furthermore, the method outperforms other current models, such asYou Only Look Once version 8(YOLOv8), You Only Look Once version 9(YOLOv9), MobileNet, and You Only Look Once version 11(YOLOv11)-based systems in terms of performance. Moreover, the method is assessed based on computational complexity, infrastructure compatibility, and adaptability. The empirical findings show that the proposed model assists in improving pothole detection efficiency and reduces latencies.