Rash Driving Detection Using IoT and ML
摘要
This paper discusses an integrated approach towards driver behavioural analysis and rash driving detection system using reverse geocoding and Multiplexed sensor system connected through serial communication via HC-05 Bluetooth module, NodeMCU (esp8266) Wi-Fi module. The system built looks to create a model that provides a continuous evaluation of various driving patterns followed by drivers with the help of GPS, Accelerometer and Gyroscope. By creating this standalone system set up inside vehicles, we look to solve the problem of negligence and lack of responsibility among drivers. With limited requirement of human intervention, the system would upload the collected data points directly to a cloud server with the help of a Wi-Fi module inserted into the vehicle. The collected data points are then used to understand the behaviour and driving patterns of drivers. The Carla Simulator Platform and 6-axis virtual Inertial Measurement Unit (IMU) sensors are used to collect this data. The data collector environment has been set up with Carla where we choose a simulated city model along with a user defined car model. Having let multiple subjects drive around in a defined track with multiple high and low speed turns, we try to understand their driving pattern and with the help of cross-correlation, it is then trained by different classification algorithms to check this obtained data for accuracy. This system can be used for any motor vehicle with minor changes to the setup and the Sensor Network. This low-cost system can go a long way in solving the ever-existent problem of rash driving.