Improving Driving Safety by Preventing Driver Distraction

Abstract

This paper presents a novel system designed to detect and prevent driver distraction caused by the use of mobile phones while driving. The Phone-Locating-Unit (PLU) is based on a multi-modal approach that exploits inputs from several sensors. In-cabin cellular activity and Inertial Measurement Unit (IMU) information is supplemented by additional information on wireless in-cabin activity. The challenges we faced when designing the PLU, compared to classical RSSI systems is the near-field multipath impediment, and the small number of antennas in the vehicle. On the other hand, only a binary decision is required, whether the emitter is used by the driver or not. The system also employs Neural Networks (NN) and Machine-Learning (ML) to fuse the IMU information. We detail the propagation model and in-cabin field measurements and show that the PLU can detect and prevent drivers from being distracted without any prior knowledge of the number of smartphones, their incabin locations, or the number of passengers in the vehicle. We demonstrate the effectiveness of the approach to accurately detect in cabin cellular activity, with a clear distinction between driver and passenger activity, despite the unique and challenging characteristics of the in-cabin propagation channel.