Project Summary
Healthcare and Driver monitoring system are becoming increasingly important due to increasing of older drivers. Therefore, although several monitoring systems are being developed, there are no verified systems in actual vehicles.
Therefore, in this project, we develop a Bio-signal monitoring system and Removal logic of motion noise for drivers.
Figure 1. Number and percentage of drivers 65 years of age or older in the U.S. driver population. Source: Federal Highway Administration statistics and predictions (www.fhwa.dot.gov).
Project Name : Research of Healthcare system for driver accident prevention
Project Duration : Mar. 2018 ~ Jan. 2019
Project Host : HYUNDAI MOTOR GROUP and GIST
Goals
Development of driver's bio-information acquisition technology and implementation of monitoring system
Introduce technology to real vehicles by acquiring and analyzing signals from virtual drivers in a vehicle simulator environment
Result
Construction of multiple bio-signal acquisition & analysis system
Removal of bio-signal motion noise in driving situations
Data Acquisition & Analysis
1. Force Sensing Resistor (FSR) - Driving posture estimation
Figure 2. Flow of FSR experiment.
In the FSR experiment, the pressure signal for each posture of the driver in the vehicle seat is measured using the FSR sensor. The acquired signal is normalized to remove motion noise, and the driver's posture is classified through an optimal AI classifier.
The postures to be classified are five that drivers can easily do and they are the same as the postures shown in Figure 3.
The previous five postures were classified as Gaussian SVM in this experiment, and result values in Figure 4 are proved that.
Figure 3. Five postures to be classified
Figure 4. Result of Classification for five postures in multiple AI model
2. BallistoCardioGraphy (BCG) - Unobtrusive Heart rate estimation
Figure 5. Flow of BCG experiment
The goal of the BCG experiment is to estimate the driver's Heart Rate in unobtrusive way. Ballistocardiography is a signal that acquires vibrations on the surface of the human body caused by heartbeat. Therefore, it is the most optimal and unobtrusive signal for estimating the heart rate.
As shown in Figure 6, the BCG signal has several peaks representing the detailed oscillation of the heartbeat. In the BCG experiment, we detect the J peak and estimate the Heart Rate using interval of J peaks.
In the heart rate estimation using the BCG signal, the heart rate is estimated by frequency analysis and the performance is verified by comparing it with the ECG signal. The result is shown in Figure 7, and it can be seen that the heart rate is estimated with more than 90% accuracy even in driving situations.
Figure 6. Several peaks of BCG signal
Figure 7. BCG signal analysis and display
Comentarios