LambertSAE20060113

33 %
67 %
Information about LambertSAE20060113
Education

Published on February 8, 2008

Author: Doride

Source: authorstream.com

Passive Sensing of Driver Intoxication:  Passive Sensing of Driver Intoxication David K. Lambert, Mark E. Myers, Jr., Larry Oberdier, Michel F. Sultan, and Christopher M. Thrush Delphi Research Labs Tie Li Shanghai Institute of Microsystem and Information Technology Outline :  Outline Introduction Breath concentration in vehicles Infrared sensing of CO2 and alcohol vapor In-vehicle demonstration Possible applications Summary 2006-01-1321 Introduction :  Introduction Drunk driving U.S. annual fatalities: ~12,900 / yr 30% of total U.S. traffic fatalities U.S. annual cost: $40 billion / yr $200 / registered vehicle This paper reports on a passive monitor of driver intoxication. 2006-01-1321 Method :  Method Responds to driver’s breath (diluted by air) Infrared is used to measure ethanol and CO2 2006-01-1321 Ethanol Sensor CO2 Sensor Controller Out Outside Air Cabin Air Legal Alcohol Intoxication :  Legal Alcohol Intoxication Legal limit  Blood alcohol concentration (BAC) BAC = (grams of alcohol) / (100 mL of blood) ~ % alcohol Legal intoxication in the U.S.: BAC > 0.08 g/dL 2006-01-1321 Breath Alcohol and CO2 :  Breath Alcohol and CO2 Breath alcohol concentration  BAC At BAC = 0.08 g/dL, exhaled breath has 189 ± 15 ppm alcohol. Independent of alcohol, exhaled breath has 3.6% CO2. Fresh air has 370 ppm CO2. 2006-01-1321 Breath Concentration in Vehicles :  Breath Concentration in Vehicles Vehicle experiments used breath CO2 from sober subjects to simulate breath alcohol from intoxicated subjects. Estimates of alcohol vapor concentration assume subject is at threshold of legal intoxication. Our sensor also uses breath CO2 as a tracer to obtain breath concentration. Breath CO2 and ethanol vapor are diluted similarly. 2006-01-1321 CO2 Study of Breath Dilution:  CO2 Study of Breath Dilution Variables: Time in vehicle HVAC mode (heater, vent, defrost) Fan speed Vent orientation Sensor location (steering wheel, ceiling,…) Person location Vehicle type Test subject 2006-01-1321 Example of Measured Dilution :  Example of Measured Dilution 2006-01-1321 Calculated threshold to detect drunk driver Dilution from driver to steering wheel Minivan Get in Person-to-Person Breathing Variability :  Person-to-Person Breathing Variability 2006-01-1321 Volumetric breathing rate data from W. C. Adams, CARB Rep. A033-205 (1993). Breathing Rate [L/min] Alcohol Concentration in Cabin Air :  Alcohol Concentration in Cabin Air Even with same driver, BAC= 0.08 g/dL, alcohol concentration in cabin air varies by a factor of 30 Highest, 5 min with fan off, ~10 ppm Lowest, 5 min with fan at max, ~0.3 ppm Individual difference: 23% more variation If breath is fully mixed in the cabin, driver and passenger can have similar effect 2006-01-1321 Method :  Method Responds to driver’s breath (diluted by air) Infrared is used to measure ethanol and CO2 2006-01-1321 Ethanol Sensor CO2 Sensor Controller Out Outside Air Cabin Air Infrared Sensing of Alcohol and CO2 :  Infrared Sensing of Alcohol and CO2 2006-01-1321 Conditions: Path length: 10 cm Water: 80% humidity, 25 °C CO2: background (370 ppm) Ethanol: 500 ppm Ethanol: 500 ppm Infrared Sensing of Alcohol and CO2 :  With 10 cm path length Simple to detect ambient CO2 (370 ppm) Difficult to detect lowest alcohol concentration (0.3 ppm) A vapor concentrator first amplifies the alcohol concentration by a factor of 50 The sensitivity improvement may be 1000 2006-01-1321 Infrared Sensing of Alcohol and CO2 Infrared Sensing of Alcohol and CO2 :  Measure transmitted IR intensity IR source is a hot filament IR detected by two thermopiles Response to alcohol and CO2 is predictable 2006-01-1321 Infrared Sensing of Alcohol and CO2 Sensor Implementation :  Sensor Implementation 2006-01-1321 Adsorber collects alcohol, flow is stopped, adsorber is heated, cycle repeats Each BAC measurement takes about 1 min Adsorber Temperature IR 4.4 ppm Alcohol Vapor Measurement in Lab :  Alcohol Vapor Measurement in Lab Used known concentrations Mass flow controllers used to mix gases Linear response to alcohol vapor Performance: 0.05 ppm residual std. error 2006-01-1321 CO2 Measurement in Lab :  CO2 Measurement in Lab Ratio intensities transmitted in CO2 and ethanol bands Non-linear response to CO2 (as expected) Performance: 2.8 ppm residual std. error 2006-01-1321 In-Vehicle Demonstration :  In-Vehicle Demonstration Driver at threshold of intoxication is simulated by respirator supplied by gas cylinder Equivalent BAC: 0.092 g/dL Average flow rate: 16 L/min 2006-01-1321 Respirator Passive Sensor Compressed Simulated Breath Air Sampling Simulated Breath Vehicle Tests:  Vehicle Tests 2006-01-1321 Test was performed in sedan As expected, alcohol was detected with respirator, but not with sober person Response is delayed after the driver gets in: 2 min max Indicated BAC of sober driver < 0.006 g/dL Fan-Speed Effect :  Fan-Speed Effect Limiting alcohol concentration is highest with the fan off As fan speed increases, less delay before valid output Largest deviation: 0.028 g/dL 2006-01-1321 Limitations of Passive Sensor :  Limitations of Passive Sensor Too slow to use as an ignition interlock Some situations cannot be classified Windows down Intoxicated passengers However, system could identify situations that cannot be accurately classified 2006-01-1321 Drunk Driving Countermeasures :  Drunk Driving Countermeasures Law enforcement Annual U.S. drunk driving arrests: 1.5 million Breath sensor with ignition interlock Imposed after drunk driving conviction Only effective while mandated Efforts to change societal norms 2006-01-1321 Causes of Drunk Driving Fatalities :  Causes of Drunk Driving Fatalities Unsafe driver choices Delayed reaction times and poor coordination Impaired vision Teen peers distract teen driver Only 25% of fatally injured drunk drivers were wearing seat belts Drinking far beyond legal limit — 63% of fatally injured drunk drivers had BAC > 0.15 g/dL 2006-01-1321 Potential Passive Sensor Applications :  Potential Passive Sensor Applications Safety system countermeasures Example: increase headway distance for ACC Alert parents of teen drivers Substitute for breath interlock Trigger use of active sensor on future trips 2006-01-1321 Summary :  Summary It is feasible to passively monitor driver intoxication There are limitations Too slow to use as ignition interlock Passenger breath mixes with driver breath Need a closed cabin Potentially, 1000’s of lives / yr could be saved Implementation depends on public policy Paper can be freely downloaded from: http://www.delphi.com/pdf/techpapers/2006-01-1321.pdf 2006-01-1321

Add a comment

Related presentations