DEVELOPMENT OF FEED FORWARD NEURAL NETWORK-BASED VEHICLE ACCIDENT PREVENTION SYSTEM

Subject Area: Computer Science

This paper addresses a pressing issue in Nigeria's transportation landscape and proposes an innovative solution using artificial intelligence. The ban on motorcycles due to high crime rates has led to a significant shift towards tricycle transportation, creating new challenges in accident prevention. The absence of existing accident prevention systems tailored to tricycles underscores the need for a specialized approach. The use of a Feed Forward Neural Network (FFNN) for tricycle detection is a promising method, leveraging machine learning to identify these vehicles on the road. Additionally, the incorporation of rule-based systems for accident detection and control ensures adherence to safe distance standards prescribed by the Federal Road Safety Corp (FRSC). By combining these approaches, the proposed system aims to mitigate collisions between tricycles and other vehicles effectively. The evaluation metrics, including Mean Square Error (MSE), Receiver Operator Characteristics (ROC) curve, and confusion matrix, demonstrate the efficacy of the developed model. With a low MSE, high accuracy, and a robust ROC curve, the system shows promise in accurately detecting and preventing accidents involving tricycles. The implementation of this system using Simulink provides a practical framework for deployment, offering real-world applicability. By addressing a critical gap in existing safety measures, this research contributes significantly to enhancing road safety in Nigeria, particularly in the context of the evolving transportation landscape dominated by tricycles