MDBR: Mobile Driving Behavior Recognition Using Smartphone Sensors

The driving behavior is interested approach in human life service provider, special using various smartphone sensors. We proposed an efficient framework for recognizing driving behavior using smartphone sensors. It names Mobile Driving Behavior Recognition Systems (MDBRS). The system implement while users put and change their smartphones dynamic due to their trips. The synchronous Practice Swarm Optimization (PSO) is used to auto select suitable features extracted from sensor data. The online user activity is predict by classification algorithms via only accelerometer signal. Hence, the system recognizes online behavior base on training data set by Artificial Neural Network (ANN). It auto predicts abnormal behavior from seven activity such as stop, moving, acceleration, deceleration, turn left, turn right and U-turn. MDBRS experiment on walking, bicycle, motorbike, bus and car and announce safety or unsafe behavior by abnormal behaviors predicted. The system also allowing update data training set by user confirmation from feedback module and achieve higher results with 86.71% accuracy.