By Carl Tenenbaum David Haynes Philip Pham Rachel Wakim.

INTEGRATED SENSOR TECHNOLOGIES PREVENTING

ACCIDENTS DUE TO DRIVER FATIGUE

By

Carl Tenenbaum

David Haynes

Philip Pham

Rachel Wakim

History of Driver Safety 1930s- Seat Belt first introduced 1949- Safety Cage and Padded Dashboard 1966- National Transportation Safety Board 1978- Child’s Booster Seat 1979- Car Crash Testing 1981- Airbag Introduced 1984- NY Enforced Seat Belt Use 2004- Rollover Risk Test

Causes of Car Accidents

1. Distracted Drivers (12% was Driver Fatigue)

2. Driver Fatigue

3. Drunk Driving

4. Speeding

5. Aggressive Driving

6. Weather

* According to Sixwise.com

Driver Fatigue Results

The National Highway Traffic Safety Administration Yearly Statistics

100,000 police-reported crashes 1,550 deaths 71,000 injuries $12.5 billion in monetary losses.

It is difficult to attribute crashes to sleepiness

SENSOR TECHNOLOGY AND

APPLICATIONS

To be attractive, a vehicle sensor system should be:

Fairly inexpensive, Accurate, with a quick response time, Integrated with the car design, or at

least “plug and play”, Noninvasive, Discreet, and non-distracting, Adaptable to different user conditions:

i.e., sunglasses, gloves.

Head Position Detection

Detect Head Angle

Is Head Tilted?

Audio Alarm

Sense changes in Head Position Tilt Gives off a warning if the Head Tilt is facing a downward

angle. Does Not detect head backwards or turned. Head Position Down is the Last Stage of Sleep Onset.

Usually too late and no warning to Driver.

Reed Switch Device

Speaker/Buzzer

Battery

ReedSwitch

Voice Detection Sense changes in Discrete Voice Parameters such as pitch,

frequency, latency and amplitude. A complex detection algorithm compares normal voice to

sample of potential fatigued voice Can be integrated in GPS or command oriented car

systems

Voice Channel

Types of Voice Sounds Voiced Nasal Fricative Plosive

(Easiest to detect Fatigue)

Behavioral Detection Sense Erratic Driving Behavior Stores Profile of Person’s Driving

Behavior Compares Profile such as Driver’s

Steering and Braking Reaction Time

Behaviors Detected

Steering Wheel Angle Steadiness of Wheel Lane Departure Proximity Braking Reaction Acceleration Reaction

Steering Angle Sensors

Use Mechanical (potentiometers) or Optical (contact-free) technologies to collect data or apply correction

Mount on steering shafts Cover up to 1080o (3x steering wheel

rotations) Angle resolution of 0.1o

Lane Departure Warning

Use video, laser, and infrared to monitor the lane markings

Activate Vehicle Stability Control (Infiniti), Electric Power Steering (Lexus), etc. to maintain lane position

Driving Behavior (Steering Angle)

Driving Behavior (Gas Pedal)

Driving Behavior (Center Lane Distance)

Current Behavioral Sensors Mercedes E-Class, Volvo, Lexus,

Nissan, Infiniti, Volkswagen Aftermarket- 3Q(2011)

AudioVox ($600)

*Daimler Chrysler Website

Optical Detection A camera or system of cameras monitor the driver’s facial

features for signs of drowsiness. Computer algorithms analyze blink rate and duration.

Infrared LEDs are used to enhance pupil detection. Yawning and sudden head nods are also detected.

Head/eye Camera Measure head tilting/eye closing/yawning as

signs of fatigue or drowsiness. Non-invasive, no need for user

interface. Can be thwarted by sunglasses or hats.

Driver movement may confuse the camera. 1/5 people do not show eye closure as a

warning sign. [US Dept. of Transportation]

Pupil Detection on Grayscale Image

Facial Feature Detection

Possible Camera Locations

Current Optical Systems

Nap Alarm (LS888)

DD850 Driver Fatigue Monitor

Biometric Detection

EKG and EEG Blood pressure Skin conductivity (“GSR” – Galvanic

Skin Response) Skin temperature Breathing rate Grip force

All shown with correlation to relative drowsiness

Electrocardiogram (EKG) Get information about user’s heart rhythm from at

least two electrical contacts on skin. By removing common mode noise and amplifying

the signal, a system can “read” the user’s heart rate, the distance between successive “R” peaks

Drowsiness has been shown to be linked to decreasing heart activity and changes in heart rate variability (HRV)

Minimum EKG System

As long as there are at least two contact points, sensor should be able to extract and isolate the signal

Can put these on wheel, seat, or both

Wheel sensor Use sensors on steering wheel to measure

skin temperature and conductivity, pulse, etc. Estimate heart rate variability – can detect

drowsiness. Combines many different

metrics to get an overall assessment of the user’s state.

Requires use of both hands,

without gloves.

Seat sensor

Two pieces of conductive fabric on the driver’s seat (backrest) can take an ECG

- measurement.

• Or on bottom of seat, with wheel as ground (only needs one hand)

• Needs impedance compensation for the driver’s shirt/coat, etc.

Electroencephalogram (EEG) Use multiple electrodes on scalp

to read brain waves Can very accurately determine

sleep/drowsiness stage this way by measuring amplitude/frequency variation of signal

BUT, very invasive

Other Possible Sensor Locations Blood pressure finger cuff on front seat EKG contacts on left or right armrests EKG sensors on shifter Etc.

Or any combination of these. Theory: the more bio-signs, the better!

Wireless wrist monitor Wristwatch capable of detecting heart rate, skin

temperature and conductance. Example: “Exmovere Empath Watch”: Transmits via Bluetooth to phone which can

signal out; easily extended to cars, many of which already are Bluetooth compatible.

Current design is 3.3” long, 1.7” wide, and 1.3” tall.

Can be bulky, and may

not be appealing enough;

currently being remodeled

[http://www.exmovere.com/healthcare.html]

Current Biometric Detection Systems Currently, there are no systems of these

types in commercial use They all display a high level of accuracy,

but their weak point is their invasiveness and unattractiveness

With future work, some of these can be integrated in a behind-the-scenes manner during manufacturing

DECISION MAKING AND CAR ALERTS

Fuzzy Logic Detection

More Uncorrelated Sensors Detecting Driver Fatigue Will Increase Detection

Probability

Corrective and Prevention Actions

1. Elevated Alarms

a) Provide Visual Alarm (lights, signs, etc.)

b) Provide Audio Alarm (warning tone or voice)

c) Recommend short nap (prevent car to start; studies show 15-minute nap increases alertness to 4-5 hours more)

2. Mechanical and Electronic Stimulations

a) Counteract to the effects (steering wheel turn, lane drifting, speed change, etc.)

b) Apply brake to slow down to safety

c) Dispatch for help if no response

Corrective Flowchart Actions

CURRENT MARKET AND TRENDS

Current Driver Fatigue Products

Products Price Accurate

Non-

Invasive Effective

Overall

Score Company Detection Type

Driver Nap Zapper 25 50% 3 3 5 No Nap Motion

Nap Alarm (LS888)500 80% 5 6 6

Leisure Auto

Security Optical

DD850 Driver Fatigue

Monitor 500 80% 5 6 6 Eye Alert Optical

Exmovere Empath

WristWatch 1000 90% 6 5 6 Exmovere Biometric

Driver Assist Package 3000 90% 7 7 7 Mercedes Behavioral

Undeveloped Market. US Consumer Car GPS Market is $5.1 Billion Market in 2010.

Limitations and Future Work Limitations

Probability of DetectionLack of Effective and Timely AlertsIntegration of Sensors

Future WorkIncrease Probability of DetectionUse of Multiple Sensors to Increase

ProbabilityDevelop Effective and Timely Alerts

References [1] “The 6 Most Common Causes of Automobile Crashes(2010)”. Retrieved February 9th 2011, from

http://www.sixwise.com/newsletters/05/07/20/the_6_most_common_causes_of_automobile_crashes.htm

[2] K. Strohl, J. Blatt, F. Council, K. Georges, J. Kiley, R. Kurrus, A. McCartt, S. Merritt, R.N, A. Pack, S. Rogus, T. Roth, J. Stutts, P. Waller, and D. Willis, “Drowsy Driving and Automobile Crashes” (2010), Retrieved February 21st 2011, from http://www.nhtsa.gov/people/injury/drowsy_driving1/drowsy.html#NCSDR/NHTSA

[3] What causes Fatigue (2010), Retrieved February 21st 2011, from http://unsafetrucks.org/driver_fatigue.htm

[4] H. Greeley, E. Friets,, J. Wilson, S. Raghavan and J. Berg, “Detecting Fatigue From Voice Using Speech Recognition”, 2006 IEEE International Symposium on Signal Processing and Information Technology

[5] D. Hu, G. Gong, C. Han, Z. Mu, and X. Zhao, “Modeling research on Driver Fatigue”, 2010 International Conference on Computer Application and System Modeling (ICCASM 2010)

[6]L. Bergasa, J. Nuevo, M. Sotelo, R. Barea, and M. Lopez, “Real-Time System for Monitoring Driver Vigilance”, IEEE Transactions on Intelligent Transportation Systems, Vol. 7, no. 1, March 2006

[7] Z. Zhu, Q. Ji, K. Fujimura, and K. Lee,  “Combining Kalman Filtering and Mean Shift for Real Time Eye Tracking Under Active IR Illumination”,  International Conference on Pattern Recognition, Quebec, Canada, 2002

[8] US Department of Transportation, “An Evaluation of Emerging Driver Fatigue Detection Measures and Technologies”, June 2009

[9] Haisong Gu, Qiang Ji, and Zhiwei Zhu, “Active Facial Tracking for Fatigue Detection” IEEE Workshop on Applications of Computer Vision, Orlando, Florida, 2002.

[10]Y. Jie, Y. DaQuan, W. WeiNa, X. XiaoXia, and W. Hui, “Real-Time Detecting System of the Driver’s Fatigue”, 2006

[11]L. Bergasa, J. Nuevo, M. Sotelo, R. Barea, and M. Lopez, “Real-Time System for Monitoring Driver Vigilance”, IEEE Transactions on Intelligent Transportation Systems, vol. 7, no. 1, March, 2006

References (Continued) [12] S. Deshmukh, D. Radake, K. Hande , “Driver Fatigue Detection Using Sensor Network”,

International Journal of Engineering Science and Technology, NCICT Conference Special Issue, pp 89-92, February 2011

[13] Y. Tanida, H. Hagiwara, “Simple Estimation of the Falling Asleep Period using the Lorenz Plot for Heart Rate Interval”, JSMBE vol. 44, no. 1, pp. 156-162, Nov. 2005.

[14] S. Kar, M. Bhagat, and A. Routray, “EEG signal analysis for the assessment and quantification of driver’s fatigue”, June 2010

[15] L. Servera, M. Fernandez-Chimeno, and M. González, “Study of Sleep Stages By Controlled Inducement and Measurement of Drowsiness Related Biomedical Signals”, 4th International IEEE EMBS Conference on Neural Engineering, April 2009

[16]P. Kithil, R. Jones, and J. MacCuish, “Development of Driver Alertness Detection System Using Overhead Capacitive Sensor Array”, International Driving Symposium on Human Factors in Driver Assessment, Training and Vehicle Design, Aspen, CO, 2001.

[17]X. Yu, “Real-time Nonintrusive Detection of Driver Drowsiness”, May 2009 [18] G. Yang, Y. Lin, and P. Bhattacharya , "A driver fatigue recognition model using fusion of multiple

features" Systems, Man and Cybernetics, 2005 IEEE International Conference on , vol.2, no., pp. 1777- 1784 Vol. 2, 10-12 Oct. 2005

[19]The John Hopkins university Applied Physics Laboratory “Technologies: Drowsy Driver Detection System” http://www.jhuapl.edu/ott/technologies/featuredtech/DDDS/ 

[20]T. Matsuda and M.Makikawa, “ ECG Monitoring of a Car Driver Using Capacitively-Coupled

Electrodes”, 30th Annual International IEEE EMBS Conference ,Vancouver, British Columbia, Canada, August 2008

[21]Y. Lin, H. Leng, G. Yang, and H. Cai, “An intelligent noninvasive sensor for driver pulse wave measurement,” IEEE Sensors J., vol. 7, no. 5, pp. 790–799, May 2007.

[22] M. Bundele, and R. Banerjee, “Design of Early Fatigue Detection Elements of a Wearable Computing System for the Prevention of Road Accidents”, IEEE,International Society of Automation, Vol 1 , pp 136-139, 2010

References (Continued) [23]I. Jeong, S. Jun, D. Lee and H. Yoon, “Development of Bio Signal Measurement System for

Vehicles”, 2007 International Conference on Convergence Information Technology [24]Exmovere Holdings Inc, “The New Biotechnological Frontier: The Empath Watch”. Feb. 2011

http://www.exmovere.com/pdf/Exmovere_Wearable_Sensor_Research.pdf [25] Frost & Sullivan’s, North American GPS Equipment Markets, 2010 (Report A601-22)