Roadway Lighting and Visibility. Driver Issues. Motor Vehicle Crashes Magnitude of. the Problem. Motor Vehicle Crashes Magnitude of the Problem

Transcription

1 Roadway Lighting and Visibility Dr. Ronald B. Gibbons Group Leader, Lighting and Infrastructure Technology R A N K Infants Top 10 Leading Causes of Death in the United States for 2001, by Motor Vehicle Magnitude of Age Group 1 Years Cause and Number of Deaths of Young Young Other Adults Life the Problem Toddlers Children Children Youth Adults Elderly All Ages Lost 2 National Center for Statistics and Analysis Under Perinatal Congenital MV Traffic MV Traffic MV Traffic MV Traffic MV Traffic Malignant Malignant Heart Heart Malignant Period Anomalies Neoplasms Neoplasms Disease Disease Neoplasms 13, ,546 5,979 4,136 6,759 16, , , ,142 23%(8,614,131) Congenital MV Traffic Malignant Malignant Homicide Homicide Homicide Heart Heart Malignant Malignant Heart Disease Anomalies Neoplasms Neoplasms Disease Disease Neoplasms Neoplasms 5, ,414 2,738 5,204 13,326 98, , ,768 22%(8,110,571) Heart Accidental Exposure to Suicide Suicide Suicide Suicide MV Traffic Stroke Stroke Stroke MV Traffic Disease Drowning Smoke/Fire ,879 1,924 5,070 6,891 15, , ,538 5%(1,700,952) Homicide Homicide Congenital Homicide Malignant Accidental Malignant Suicide Diabetes Chronic Lwr. Chronic Lwr. Stroke Anomalies Neoplasms Poisoning Neoplasms Resp. Dis. Resp. Dis ,994 6,635 14, , ,013 5%(1,687,683) Septicemia Malignant Accidental Congenital Accidental Malignant Heart HIV Chronic Lwr. Influenza/ Diabetes Chronic Lwr. Neoplasms Drowning Anomalies Poisoning Neoplasms Disease Resp. Dis. Pneumonia Resp. Dis ,160 5,867 14,490 55,518 71,372 4%(1,444,745) Influenza/ Heart Homicide Accidental Heart Heart Accidental Accidental Chronic Liver Diabetes Influenza/ Suicide Pneumonia Disease Drowning Disease Disease Poisoning Poisoning Disease Pneumonia ,507 5,036 13,009 53,707 62,034 3%(1,079,822) MV Traffic Exposure to Heart Heart Accidental Accidental HIV Homicide Suicide Alzheimer s Alzheimer s Perinatal Smoke/Fire 170 Disease 82 Disease 273 Drowning 326 Drowning 211 2,101 4,268 9,259 53,245 53,852 Period 3%(1,070,154) Nephritis/ Septicemia MV NonTraffic Exposure to Congenital Congenital Stroke Chronic Liver MV Traffic Nephritis/ MV Traffic Diabetes 8 Nephrosis Smoke/Fire 140 Anomalies 244 Anomalies Disease 3,336 8,750 Nephrosis 33,121 42,443 3%(1,014,201) Stroke Influenza/ Benign MV NonTraffic Accidental HIV Diabetes Stroke HIV Septicemia Nephritis/ Homicide Pneumonia 92 Neoplasms Falls ,491 5,437 25,418 Nephrosis 39,480 3%(924,263) Meningitis Perinatal Septicemia Chr. Lwr. Acc. Dischg. Accidental Congenital Diabetes Nephritis/ Hypertension Septicemia Chronic Liver Period Resp. Dis. 102 Of Firearms 114 Falls 134 Anomalies 458 1,958 Nephrosis 5,106 Renal Dis. 16,397 32,238 Disease 2%(623,998) 3 ALL 27,568 4,288 2,703 6,672 15,851 14,940 41,683 91, ,204 1,798,420 2,416,425 All Causes 100%(36,866,317) 1 When ranked by specific ages, motor vehicle crashes are the leading cause of death for age 2 and every age 4 through Number of years calculated based on remaining life expectancy at time of death; percents calculated as a proportion of total years of life lost due to all causes of death. 3 Not a total of top 10 causes of death. Source: National Center for Health Statistics (NCHS) CDC, Mortality Data 2001 Note: The cause of death classification is based on the National Center for Statistics and Analysis (NCSA) Revised 68 Cause of Death Listing. This listing differs from the one used by the NCHS for its reports on leading causes of death by separating out unintentional injuries into separate causes of death, i.e., motor vehicle traffic crashes, accidental falls, motor vehicle nontraffic crashes, etc. Accordingly, the rank of some causes of death will differ from those reported by the NCHS. This difference will mostly be observed for minor causes of death in smaller age groupings. Motor Vehicle Magnitude of the Problem Distribution of global injury mortality by cause: World Report on Road Traffic Injury Prevention Road traffic injuries, Suicide, 16.9% World Health Organization, Geneva, % In low-income and middle-income countries, the phenomenon of pedestrians and vehicles not being properly visible is frequently a serious problem. In these Drowning, places, 7.3% there are fewer roads with adequate illuminationi and some may not be lit at all. War, 3.4% Fires, 6.2% Violence, 10.8% Other intentional From Chapter 3: Risk Factors injuries, (pg 0.2% 86) Driver Issues Accident Statistics show that more than 50% of fatal accidents occur during the night hours while only 25% of the vehicle miles are driven. Night driving has been described as a situation for which humans have not evolved, leaving our visual system inadequate and inefficient for certain tasks (Rumar, 1990). Determining the nature of the driving ing task is critical Falls, 7.5% Other unintentional Poisoning, 6.7% injuries, 18.1% Source: WHO Global Burden of Disease project, 2002, Version 1. 1

2 100 Car Naturalistic Approach Data collection in a naturalistic setting to obtain crash/pre-crash/near-crash/conflict data as well as distributions of driver performance 100 drivers in their own (or leased) vehicles with specialized instrumentation, on public roads, as close to unobserved as possible. Subjects use instrumented vehicles for an extended period (up to 13 months) without an experimenter present. Subjects are not coached or instructed to perform any specific actions other than drive as they normally do. Instrumentation is unobtrusive and inconspicuous to other drivers, but not invisible. Naturalistic Data Collection Approach Highly capable instrumentation (well beyond EDRs) Five channels of digital, compressed video Four radar sensors front, rear (for all 100 vehicles), and side (for 20 vehicles) Machine vision-based lane tracker Many other sensors: GPS, glare, RF, acceleration, yaw rate, controls, etc. Cell phone, wireless internet, or hardwire download Tie into vehicle network to obtain other sensor information Rugged, crash tested, all solid state Crash detection, Fault detection Remote Access The Naturalistic 100 Car Driving Study: Database Statistics Driver Behavior - Spinny 42,300 hours of driving data collected 82 and collisions Defined as any contact between the subject vehicle and another vehicle, fixed object, pedestrian pedacyclist, animal. 761 Near crashes Defined as a conflict situation requiring a rapid, severe evasive maneuver to avoid a crash Critical incidents Conflict requiring an evasive maneuver, but of less magnitude than a near crash. 2

3 Driver Behavior Silly Driver Behavior Dopey Driver Behavior - Sleepy Data Analysis 3

4 Results The primary cause of all of incidents is secondary task distraction Doing other things in the vehicle Any visual task must overcome the distraction of the driver in the vehicle Lighting and Driver Safety What we know Box [1972] showed that the night/day accident ratio was 66% higher on unlighted freeways than on lit ones. 0.5 lux appeared to be the illuminance level which provided the lowest accident rate Osner [1973] and Nishimori[1973] both showed a 56% reduction in accidents when lighting was added to a roadway. CIE Pub. N o 93 Road Lighting as an Accident Countermeasure rigorously analyzed 62 lighting and accident studies from 15 countries. (S)tatistically significant results show reductions (in nighttime accidents) of between 13 and 75 percent. Motor Vehicle Implications of Darkness Motor Vehicle Implications of Darkness Vehicle occupant deaths, FARS, Pedestrian deaths, FARS,

5 Evaluating Impact of Light Pedestrian Fatalities Fall PM Return to Standard Time Light Dark Pedestrian Run-off-road Weeks Before and After Return to Standard Time Visual Activity The Eye As drivers, the visual task is a very complex activity Detection of road hazards Lane keeping Wayfinding Monitoring of the instrument panel Observing other drivers Pedestrian Detection Sightseeing? Minding the other occupants in the vehicle We distribute our visual resources between all of these activities We allot attention to the task which seems most demanding Not necessarily the most important 5

6 The Retina Captures photons and send nerve impulses to the brain Two important regions of the retina are: Fovea The central and most sensitive part of the visual field Highest Acuity Almost entirely Cones Periphery Low Acuity High range of sensitivity Almost entirely Rods Visual Process The visual process is partially automatic and partially conscious Lane keeping is automatic Visual vection process Object Detection and Wayfinding is conscious Object Detection Process Object Visibility Visual Search We have a standard search pattern as we drive Looking for objects Looking at signage Following the road path Detection Through the visual search, we find an object of interest This detection can be peripheral Spotted to the side as a result of motion or through high conspicuity This detection can be foveal Found through the visual search pattern Recognition We attend to the objects of interest This is a foveal task Reaction We decide what the appropriate course of action Braking, steering etc. 6

7 Visibility with Glare Visibility with Extreme Glare Visibility of Objects We see objects based on their contrast to the background This can either be color contrast or luminance contrast In roadway lighting design, color is not considered The IES lighting design requirements are consensus standards based on experience Lighting design can be performed based on illuminance, luminance or STV The STV is a weighted average of a series of calculated l Visibility Levels for a defined target Current target is a flat 7" square with 50% reflectance which represents the smallest object which will collide with a vehicle C L L act Target Background VL is calculated as: VL = = C Δ L th th IES RP-8 Illuminance Design Criteria Road and Pedestrian Conflict Pavement Classification Area (Minimum Maintained Average Values) Uniformity Veiling Ratio Luminance Road Pedestrian R1 R2 & R3 R4 Ratio Conflict Area Lux/fc Lux/fc Lux/fc Eave/Emin Lvrnax/Lavg Freeway Class A 6.0/ / / Freeway Class B 4.0/ / / High 10.0/ / / Expressway Medium 8.0/ / / Low 6.0/ / / High 12.0/ / / Major Medium 90/09 9.0/ / / Low 6.0/ / / / / / High Collector Medium 6.0/ / / Low 4.0/ / / / / / High Local Medium 5.0/ / / Low 3.0/ / /

8 IES RP-8 Luminance Design Criteria Road and Pedestrian Conflict Average Uniformity Uniformity Veiling Area Luminance Ratio Ratio Luminance Rti Ratio Road Pedestrian L Conflict Lave/Lmin Lmax/Lmin Lvmax/Lavg (Maximum (Maximum (Maximum Allowed) Allowed) Allowed) Area ( cd/m 2 ) Freeway Class A Freeway Class B High Expressway Medium Low High Major Medium Low High Collector Medium Low High Local Medium Low IES RP-8 STV Design Criteria Road and Pedestrian Conflict STV Luminance Criteria Area Criteria Road Pedestrian Weighting Lavg Lavg Uniformity Conflict Area Average Cd/m 2 Cd/m 2 Ratio VL Median Median Lmax/Lmin <7.3m >7.3m (Maximum Allowed) Freeway "A" Freeway"B" Expressway High Major Medium Low High Collector Medium Low High Local Medium Low IES RP-8 The Next Revision Separation of street and roadway lighting Different Visual Tasks Adaptive Possibilities Allowing for changing of road class based on pedestrian and vehicle traffic changes Draft is currently under review and should be completed within 1 year Roadway Lighting and Driver Safety What we don t know The impact of lower lighting levels on driver safety Are we over-lighting? The impact of mesopic lighting levels on driver safety Is White truly better? 8

9 Threshold VL Levels IESNA Specifies a Level of 1.6 to 4.9 for STV Actual threshold values are much higher Higher lighting levels are likely not required Contrast control is required Perfect uniformity may not be the best lighting condition VL Visibility Level At Threshold: White Clothed Objects Cyclist Perpendicular Parallel Static The White Issue White Light might provide equivalent visual task performance at a lower illuminance level than non-white Sources Equivalent Performance? The task is performed with the same speed and accuracy In outdoor lighting this equates to driver safety, pedestrian safety, way finding performance, comfort 0 None Hybrid 3UVA 5UVA None Hybrid 3UVA 5UVA HLB HID HHB HOH HLB- IR-T LP VES Why? So why is there a possible white light benefit? The physiology of the eye lends itself to light sources which have radiation in the entire visual spectrum Remember Our eye was developed to interact with the sun The ultimate full spectrum light source. The effect is a result of the difference in the spectral sensitivity of the various photoreceptors in the retina Photoreceptors Rods Sensitive to low levels of radiation The Spectral Sensitivity of the rod is maximum at 507nm The rods define the pupil size in all viewing types (Photopic, Mesopic and Scotopic) Cones Cone are sensitive at high levels of radiation There are 3 cone types defined by their spectral sensitivities There are 3 cone types defined by their spectral sensitivities Long Medium Short 9

10 Photoreceptor Sensitivity Cone Sensitivity Relative Eye Sensitivities Overall Spectral Sensitivity of the Eye Photopic Vision Cone Vision - Using V(λ) > 3.5 cd/m² adaptation Luminance Scotopic Vision Rod Vision - Using V (λ) < cd/m² Mesopic Vision Mixture of Rod and Cone Vision Transition state between photopic and scotopic 10

11 A lumen is defined as: Scotopic Lumen P(λ) is the spectral power distribution (SPD) of the light source V(λ) is the spectral sensitivity of the eye k = 683 lumens/watt Φ = Luminous Flux in Lumens A scoptic c lumen represents ese the luminous flux evident to the rods. It is calculated using V (λ) instead of V(λ). The S/P Ratio is the ratio of Scotopic Lumens to Photopic Lumens Mesopic Lumens The Mesopic lumen is calculated based on using a Mesopic sensitivity curve rather than a Photopic Difficulty: The Mesopic Sensitivity changes with the adaptation luminance There are 2 models LRC Move Currently being accepted by the CIE Mesopic Lumens are not uniform across the retina What is the Actual Benefit? Better Visual Acuity Higher Equivalent Luminance Better Visual Acuity Some researchers suggest that light sources with a higher S/P Ratio are more beneficial as they force the pupil to be smaller, allowing the observer to have better visual acuity and visual performance Issue for outdoor lighting: This research was only performed for indoor lighting levels and may not apply to outdoor applications 11

12 Higher Perceived Luminance Light sources with a blue content create a higher response in the cones and therefore a higher perceived luminance Higher Mesopic Luminance Higher S/P ratio Issue: It is important to remember that this change is only evident in the periphery of the eye and not in the fovea Rods must be present to create the effect For outdoor lighting, this is the most critical aspect of white light. How big is the benefit? There is conflicting research LRC has shown that an object which appears at 20º peripherally is perceived at a lower level in white light than in amber Lewis has shown similar effects for peripheral targets It has been proposed that Luminance Equivalence Multipliers be applied to lighting levels to account for the increased benefit of white light. These values are based on the Mesopic lumen rating of the light source The applicability value of the scaling factors is a topic of debate Important: The use of these scaling factors has not been approved by the IESNA for Lighting Applications Proposed LEMs for White Light Source 1 fc Illuminance Metal Halide 1 1 Incandescent Mercury Vapor High Pressure Sodium Low Pressure Sodium fc Illuminance Conflicting Research VTTI Crosswalk Lighting Results Metal Halide under performs HPS in a object detection task in a roadway environment For the same vertical illuminance on the object, detection distances were longer under HPS than MH for Black Clothed Objects and Equivalent for Denim Clothed Objects 12

13 Crosswalk Lighting Pedestrian Detection Crosswalk Lighting Black Clothed Impact of Pedestrian Clothing On Pedestrian Detection 900 Impact Of Lamp Type Pedestrian Detection For Black Clothed Pedestrians Distance (ft) Black No Glare Denim No Glare Surrogate No Glare White No Glare Distance (ft) D HPS No Glare MH Lamp Level Lamp Level Distance (ft) Crosswalk Lighting Denim Clothes 0 Lamp Type and Denim Clothed Pedestrians Lamp Level HPS No Glare MH No Glare Why the difference in the Research Results? Experimental Methods differences Many experiments uses fixed geometry to determine the effects The results are then interpreted t to apply across the entire visual field This is invalid as the visual field is not uniform across the retina Crosswalk Lighting Investigation Use non fixed geometry Free Driving Most objects appeared foveally No Mesopic White Light effect is evident Older Drivers 65+ years old The lens yellows with age and may impact performance for a non-yellow source Neither of these methods truly represent what is happening in a vehicle 13

14 Other White Light Issues High Color Temperatures with a high blue content can cause greater sky scatter and sky yglow Rayleigh scattering is spectrally sensitive Collision of light with atmospheric molecules Which is why the sky is blue Low angle blue content light scatters more We have evolved under the moonlight We believe the color temperature of the moonlight is about 4200K Research Goal In order to fully assess the impact of lighting on the driver, a visual model must be developed which is accounts for all of these issues Peripheral vs. foveal issues Color impacts Spectral effects Object movement Visual Search parameters VTTI Research VTTI is currently working on a project to establish this visual model Using eye tracking to determine visual search characteristics A wide variety of visual tasks, light sources and driving conditions Free driving will be the basis of the experimentation No fixed geometry Naturalistic data collection We collected: Illuminance Luminance Color Observer Input GPS The Vehicle Our experiment here provides validation to some of the work performed in the experiment Color 4 different target colors Gray, Green, Blue, Red Light Sources 4200k LED and induction 14

15 Collecting RBG of road view Assists in determining conflicts Color Camera Collecting Photometrically Accurate Road Characteristics Luminance Camera Illuminance 5 illuminance meters 4 measuring horizontal illuminance on the roof of the vehicle Wheel path and centerline of the vehicle 2 along centerline to measure gradient 1 measuring vertical illuminance at the windshield Estimates of glare impact Where the Results Go We will be analyzing the detection distances and the luminance in each area. A short report will be prepared The data will add to our entire database of data. 15