Design of an Intelligent Driving Behavior Monitoring System for Vehicle Drivers Based on CNN-RNN*

With the rapid development of the social economy and the sharp increase in the number of motor vehicles, the threat posed by drivers' improper behaviors to road traffic safety has become increasingly prominent. In particular, behaviors such as fatigue driving, distracted attention, single-hand operation, answering phone calls while driving, and smoking directly lead to a significant increase in the incidence of traffic accidents. This paper proposes an intelligent driving behavior monitoring system for automobile drivers based on deep learning and multi-modal sensor fusion. At the sensing level, a high-definition camera, a non-contact infrared temperature sensor, and a photoresistor are integrated, and the continuous driving duration is recorded through the DS3231 real-time clock. At the data level, the processing flow is deployed on the NVIDIA Jetson Nano edge computing platform: Firstly, preprocessing operations such as denoising, gamma correction, illumination normalization, and face alignment are performed on the collected images; the MobileNetV2 lightweight convolutional neural network is used to extract the spatial features of each frame, and the feature sequence is input into the Long Short-Term Memory network (LSTM) to capture the temporal information, and complete the classification and recognition of five typical illegal behaviors. Finally, at the decision-making level, the DF Player Mini voice module is used to broadcast risk warnings in real time; at the communication level, the ESP32 Wi-Fi/4G module is used to encrypt key events with RSA and upload them to the cloud, supporting remote monitoring and big data analysis. This system has the characteristics of low cost, easy integration, high real-time performance, and scalability, and can provide effective safety guarantees for intelligent connected vehicles and Advanced Driver Assistance Systems (ADAS).