Monday , October 22 2018

Fuzzy Algorithm for Human Drowsiness Detection Devices

Simona DZITAC
Faculty of Energy Engineering, University of Oradea
1 Universităţii Street, Oradea, Romania

Tiberiu VESSELENYI
Faculty of Energy Engineering, University of Oradea
1 Universităţii Street, Oradea, Romania

Laurenţiu POPPER
Faculty of Energy Engineering, University of Oradea
1 Universităţii Street, Oradea, Romania

Ioan MOGA
Faculty of Energy Engineering, University of Oradea
1 Universităţii Street, Oradea, Romania

Călin D. SECUI
Faculty of Energy Engineering, University of Oradea
1 Universităţii Street, Oradea, Romania

Abstract: There are several human activities where the awareness and conscious control is a very important factor: vehicle driving, heavy equipment operation, hazardous materials manipulation. In these cases drowsiness can be the cause of injury or even death. For example car driver drowsiness is one of the causes of serious traffic accidents, which makes this an area of a significant importance. Continuous monitoring of driver’s or operator’s drowsiness is of great importance if we want to reduce accidents due to operator’s fault. If drowsiness is detected in time, a significant part of these accidents could be successfully prevented. In the last years various methods were tested, based on the use of: heart rate variability, video monitoring of the eyes, EEG, EMG and ECG signals.

Our research is based on the study of EEG and EMG signals and aims to develop algorithms capable to detect features specific to the drowsiness state and decide the moment in which the driver or operator should be alerted.

Keywords: Drowsiness alert, EEG, EMG, fuzzy decision algorithm.

>Full text
CITE THIS PAPER AS:
Simona DZITAC,  Tiberiu VESSELENYI, Laurenţiu POPPER, Ioan MOGA, Călin D. SECUI, Demo Portal of Information and Documentation in Science and Technology, Studies in Informatics and Control, ISSN 1220-1766, vol. 19 (2), pp. 419-426, 2010.

1. Introduction

As it was reported in several studies 60% of adult drivers have driven while felling drowsy in the past years, and 37% have actually fallen asleep while driving. For this reason, a technique that can detect the driver’s drowsiness in real-time is very important in order to prevent accidents caused by drowsiness. In cases of equipment operation there are no statistics published but the danger of accidents is also present. If drowsiness could be detected, incidents can be prevented by countermeasures in order to awake the driver or the operator.

In order to analyze the state of awareness it is important to know the phases of sleep. Sleep cycle is divided into Non-Rapid Eye Movement (NREM) sleep and Rapid Eye Movement (REM) sleep, and the NREM sleep is further divided into four stages. Drowsiness is the first stage of sleep in the NREM domain [1, 2]. So the detection of awareness can be based, among others, on the distinction between first stage of NREM state and awake state.

In the last years a large number of methods had been tested, based on the use of: heart rate variability, video monitoring of the eye movement and facial expression, EEG, EMG and ECG signals.

The video monitoring approach analyzes the images captured by cameras to detect physical changes of driver’s image, such as eyelid movement, eye gaze, and head nodding. Some systems are using cameras and imaging processing techniques to measure the percentage of eyelid closure over time. Although this vision based method is not invasive and is not causing annoyance to drivers, the drowsiness detection accuracy is severely affected by the environmental backgrounds, driving or operating conditions and also requires the camera to focus on a relatively small area.

The physiological signal detection approach is to measure the changes of driver or operator biological signals, such as the electroencephalography (EEG) and electrocardiogram (ECG). Since the sleep rhythm is strongly correlated with brain and heart activities (brain rhythms correlated with NREM first stage), these physiological signals can give more accurate drowsiness detection than video monitoring. The drawback of this method is that the electrode contacts and wires can make a discomfort for the driver or operator and also that the electrodes must be placed on the skull using a conductive gel which ensures a good contact with the skin. These difficulties can be overcome by using dry-contact, low-noise EEG sensors, as it is described in [3]. The dry-contact EEG electrodes were created by the authors [3] with micro-electrical-mechanical system (MEMS) technology. Each channel of the analog signal processing front-end comes on a custom-built, small-sized circuit board which contains an amplifier, filters, and analog-to-digital conversion. As the authors describe daisy-chain configuration between boards with bit-serial output reduces the wiring needed and the system is capable to detect alpha-band rhythms and eye-blink signals.

Also for driver awareness detection a number of methods were proposed which are based on the technique of embedding biosensors into steering wheel or in the seat of the vehicle in order to measure heart beat pulse signals [4]. Time series of heart beat pulse signal can be used to calculate the heart rate variability. As it is shown in [4], the frequency domain spectral analysis of heart rate variability shows that typical it has three main frequency bands: high frequency band, low frequency band and very low frequency. A number of psycho-physiological researches have found that the low and high frequency power spectral density ratio decreases when a person changes from waking into drowsiness/sleep stage, while the high frequency power increases associated with this status change. The authors in [4] show that this variability can be an effective method for the detection of driver drowsiness. The problem with this method is that the drowsiness detection equipment is not portable and it must be embedded in the vehicle parts which will limit its applicability.

Analyzing the above mentioned approaches we decided to study algorithms which can be used for drowsiness/awareness detection when applying techniques based on EEG and EMG signals measurement. There are also a large number of studies which uses EEG signals for brain – computer interface applications [5, 6, 7, 8]. If the EEG device is built in the manner described in [3] this method has the advantage that is independent from the equipment, vehicle or situ where it is used, and can be applied both for drivers and equipment operators too.

References:

  1. Carlson, N. R., Sleep and Biological Rhythms. in Physiology of Behavior, 7th ed. Boston: Allyn & Bacon Publishers, 2001.
  2. Sleep Research Society, Basics of Sleep Behavior Syllabus, WebSciences International and Sleep Research Society, 1997.
  3. Sullivan, T. J., S. R. Deiss, T.-P. Jung, G. Cauwenberghs, A Brain-Machine Interface using Dry-Contact, Low-Noise EEG Sensors, 2008 IEEE International Symposium on Circuits and Systems 18-21 May 2008, Washington, US
  4. X. Yu, Real-time Non-intrusive Detection of Driver Drowsiness, Intelligent Transportation Systems Institute Center for Transportation Studies, University of Minnesota, May 2009.
  5. Daly, J. J., J. R. Wolpaw, Brain-Computer Interfaces in Neurological Rehabilitation, Volume 7, Issue 11, November 2008, Pages 1032-1043
  6. Ting, J.-A., A. D’Souza, K. Yamamoto, T. Yoshioka, D. Hoffman, S. Kakeif, L. Sergio, J. Kalaska, M. Kawato, P. Strick, S. Schaal, Variational Bayesian Least Squares: An Application to Brain-machine Interface Data, Neural Networks, Volume 21, Issue 8, October 2008, Pages 1112-1131.
  7. Cvetkovic, D., E. D. Übeyli, I. Cosic, Wavelet Transform Feature Extraction from Human PPG, ECG, and EEG Signal Responses to ELF PEMF Exposures: A Pilot Study, Digital Signal Processing, Volume 18, Issue 5, September 2008, Pages 861-874.
  8. Ting, W., Y. Guo-zheng, Y. Bang-hua, S. Hong, EEG Feature Extraction Based on Wavelet Packet Decomposition for Brain Computer Interface, Measurement, Volume 41, Issue 6, July 2008, Pages 618-62.
  9. Georgescu, V., Fuzzy Time Series Estimation and Prediction: Criticism, Suitable New Methods and Experimental Evidence, Studies in Informatics and control, Volume 19 , Issue 3, 2010.
  10. Abid, H., M. Chtourou, A. Toumi, Robust Fuzzy Sliding Mode Controller for Discrete Nonlinear Systems, International Journal of Computers Communications & Control Volume III No.1, 2008.

https://doi.org/10.24846/v19i4y201010