International Journal of Science, Engineering and TechnologyAuthors: Ashu Ayush, Anshika Shrivastava, Himanshu Saini, Sanjana Sen, Dr. Md. Iqbal
Abstract: The rapid evolution of autonomous vehicles (AVs) necessitates advanced safety frameworks to mitigate the impact of unforeseen collisions. This paper presents an emerging accident detection system that bridges the gap between raw sensor data and real-time emergency response. By utilizing a multi-sensor fusion pipeline—integrating LiDAR, 3-axis accelerometers, and high-speed cameras—the system identifies critical impact patterns and vehicle orientation anomalies. We propose a hybrid model combining Convolutional Neural Networks (CNN) for visual crash analysis and Support Vector Machines (SVM) for telemetry-based classification. Preprocessing involves data normalization and segmentation of sensor streams into 10-second high-resolution windows. Expected outcomes include a 90%+ detection accuracy in simulated environments, reduced false-alarm rates through sensor cross-validation, and immediate transmission of precise GPS coordinates via IoT-enabled notification modules.
