Develop algorithms to determine the status of car drivers using built-in accelerometer and GBDT

Thi Thu Nguyen (Hanoi University of Industry)
Phuc Thinh Doan (Nguyen Tat Thanh University)
Anh-Ngoc Le (Electric Power University)
Kolla Bhanu Prakash (Koneru Lakshmaiah Education Foundation (KLEF))
Subrata Chowdhury (Sri Venkateswara College of Engineering and Technology)
Duc-Nghia Tran (Institute of Information Technology)
Duc-Tan Tran (Phenikaa University)

International Journal of Electrical and Computer Engineering
Vol 12, No 1: February 2022

In this paper, we introduce a mobile application called CarSafe, in which data from the acceleration sensor integrated on smartphones is exploited to come up with an efficient classification algorithm. Two statuses, "Driving" or "Not driving," are monitored in the real-time manner. It enables automatic actions to help the driver safer. Also, from these data, our software can detect the crash situation. The software will then automatically send messages with the user's location to their emergency departments for timely assistance. The application will also issue the same alert if it detects a driver of a vehicle driving too long. The algorithm's quality is assessed through an average accuracy of 96.5%, which is better than the previous work (i.e., 93%).