Analysis of Driving Behavior and Classification by Vehicle Sensors and Image Sensors

Recent studies show 75% of crashes result from driving errors with fatigue, sleepiness, and inattention being the main causes. Driving Behavior (DB) is crucial for preventing crashes. This paper reviews current research on DB, including data collection, modelling, analysis, and system assessment. DB and driving habits are interconnected and may be used to enhance fuel efficiency and identify drivers. Engine operations (engine temperature, oxygen, RPM, air flow) in vehicles are continuously monitored and checked using sensors which are sent to the Electronic Control Unit (ECU) for processing via the Internal Vehicle Communication Protocol. The data is converted to digital format for optimal efficiency using inputs from In-car sensors, and driving styles are detected and dangerous driving behavior is recognized through the application of Machine Learning (ML) technology. For instance, engine load, throttle position, vehicle speed, and the engine speed are monitored when driving. Through thorough research, the idea to combine image and In-car sensors to increase DB accuracy is proposed. This combination of ML and image sensors makes it possible for the system to identify dangerous databases with higher precision. After discussing the concept of dangerous driving behavior and the role that technology plays in promoting safe driving, we evaluate relevant work in database identification. This study suggests using the fusion of sensors to enhance DB identification. The method proposed is to enhance driver identification accuracy by integrating in-car and image sensors, and Controller Area Network (CAN) technology. Driving states and reactions are efficiently detected, improving safety, driver training and policy implementation.