Towards Safer Roads: A Review of Hybrid Machine Learning and Vision-Based Approaches for Speed Bump Detection in Intelligent Transportation Systems

Abstract

This review discusses recent advances in detecting speed bumps with emphasis on the integration of vision-based, sensor-based, and machine learning approaches in hybrid models. A comparison of studies published from 2017 to 2024 highlights the performance, versatility, and real-world applicability in multiple detection approaches. Special focus is laid on the benefits of hybrid systems, such as increased robustness in dynamic conditions and reduced false alarms. Challenges like generalizability in data available, real-time processing, and low-cost applications in low-resource environments are also taken into account in this review. The critical voids in present research and the path forward for future development that enable scalable intelligent road safety systems are described.