Improvement of Driverless Cars' Passengers on Board Health and Safety, using Low-Cost Real-Time Heart Rate Monitoring System

摘要

In this work, a real-time unobtrusive heart rate monitoring system is proposed and implemented. The proposed system aims to monitor the heart rate of the passengers by using a low-cost camera, which can be readily embedded in the car's rear-view mirror. Additionally, we integrate this system with the main system of our test driverless car, and we propose how driverless cars should act in response to serious medical emergency situations. Moreover, we investigate how this system can benefit from the promising features of Google I/O and Google AI. Our approach is based on Remote Photoplethysmography (rPPG), in which the heart rate is extracted from the subtle tiny changes occurring in the skin color of the face during every pulsation. The face is automatically detected and tracked, then the raw signal is calculated from each frame over a 10-seconds sliding window. After that, a series of signal processing techniques are implemented on the raw signals to recover the heart rate frequency. Finally, the resultant heart rate measurements are processed and stored, then we compare it with ground truth measurements values obtained using pulse oximeter.