The Development of a Speed Anomaly Detection Utilizing Isolation Forest Algorithm as Evidence for Unforeseen Road Incidents

Overspeeding is a leading cause of road accidents, particularly for motorcycle riders, as it compromises vehicle control and raises crash severity. This project addresses this issue by developing an Arduino-based motorcycle speed detection system, integrated with a mobile application for real-time data logging and anomaly detection. The system utilizes the Isolation Forest algorithm to detect speed-related anomalies, aiming to improve road safety for motorcycle riders. Key features include a speech alarm that activates when the speed limit is breached and an automated system that logs road incidents, providing critical data for evaluating driving behavior. Researchers gathered user feedback from motorcycle riders in Metro Manila, Philippines, using a simple random sampling method and an ISO 25010:2023-based survey questionnaire, focusing on essential software quality characteristics. Responses from 91 participants reflected a positive reception toward the system, with quality metrics showing 65.4% recall for anomaly detection-capturing most instances while missing 34.6% of actual cases. Additionally, 91.4% precision indicated high accuracy in identifying true anomalies with minimal false alarms, and 79.4% overall accuracy suggested reliability in distinguishing both normal and anomalous driving events. While results are promising, further development and testing under varied conditions are recommended to fully refine and enhance the system's capabilities.