Development of Analytical Tools to Evaluate Road Departure Crashes Using Naturalistic Driving Study Data Safety Research Symposium Strategic Highway Research Program 2 Transportation Research Board July 26-27, 2007 Research Team: The Center for Transportation Research and Education (CTRE) at Iowa State University and the University of Iowa Shauna Hallmark, Tom Maze, Linda Boyle, Reginald Souleyrette, Neal Hawkins, Tom McDonald, Omar Smadi, Alicia Carriquiry
Iowa s Roadway-Related Fatal Crashes 52% of Iowa s fatalities are related to Lane Departure 39% of Iowa s fatal crashes are single-vehicle Run-Off-the-Road (ROR) crashes
Single-Vehicle, Run-off-the-Road Crashes, 2- Lane Highways, SE Iowa, 1996-2000
Research Goal 1) Use data from existing naturalistic driving studies and other sources to better under the factors that result in road departure crashes by mapping the sequence of events leading to road departure incidents and crashes and quantify how roadway, environmental, vehicle, and human factors influence whether an incident occurs in the first place and how those factors affect subsequent events and final outcome. Understanding why a crash didn t occur may be as relevant as why one did. 2) Provide suggestions to improve full scale naturalistic driving study data collection and analysis and mobile mapping data collection so that road departures can be fully addressed.
Objectives Formulate specific research questions related to roadway departure that will guide the development of analytical tools Identify data needs to address the research questions Conduct a cursory evaluation of existing driving studies and other datasets and determine whether they are suitable to use in testing the analytical tools Outline the methodologies and study parameters to develop the analytical tools and answer the research questions Acquire existing datasets and determine compatibility with proposed methodology Evaluate and quantify the relationship between roadway, environmental, vehicle, and human factors and pre- and post-road departures Report results Identify reasonable modifications that could be made to the SHRP II full scale field driving study in order to apply this dataset to answer the research questions posed in this research Identify data elements that should be collected in the SHRP II mobile mapping study
Existing Datasets University of Iowa s Quasi-Naturalistic Driving Study VTTI 100-Car Naturalistic Study UMTRI Field Tests Michigan road database Michigan crash database University of Iowa Naturalistic Study of Teenage Drivers Iowa DOT Crash Database Iowa DOT Geographic Information Management System (GIMS) Roadway Database FARS, CDS, GES
Basic research question What roadway, vehicle, driver, and environmental factors lead to a lane/road departure and why do some lane/road departures result in a crash while other have a more positive outcome road departure event where a vehicle drifts off on a 4-lane roadway with paved shoulder safe recovery (unrecorded event in crash databases) road departure event where a vehicle drifts off on a 2-lane roadway with narrow unpaved shoulders and several inches of drop-off crash Difference to a certain extent is roadway features
Research Question 1 What key driver, vehicle, roadway, and environmental factors affect lane keeping which may result in a road departure? Research Question 2 What kinematic variables can be used to determine when a road departure is likely or imminent? 0.04 0.02 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5-0.02-0.04-0.06-0.08-0.1 Drift_lateral Normal_lateral -0.12-0.14 Lateral Acceleration Signatures for a Drift-off Road and Normal Lane Change Event
Research Question 3 What environmental factors influence whether a vehicle actually departs the roadway once a road departure is precipitated? How frequently do road departures occur given a specific set of environmental variables? i.e. does roadway lighting result in fewer nighttime road departures? Dry Wet Ice Snow Slush Sand/mud/dirt/oil/gravel Water (standing/moving) Other/unknow Rural single vehicle run-off-road crashes in Iowa (2005 crash data) Clear Partly cloudy Cloudy Fog/smoke Mist Rain Sleet/hail/freezing rain Snow Severe winds Blowing sand/soil/dirt/snow Other (explain in narrative)
Research Question 4 To what extent do roadway features influence whether a vehicle actually departs the roadway once a roadway departure is precipitated? How frequently do road departures occur given a specific set of roadway variables? i.e. are drivers more likely to lane keep on roadways with edge line rumble strips? Research Question 5 Once a road departure occurs, what are the next most common sequence of events and outcomes (i.e. safe recovery and return to roadway, minor conflict with safe return, near miss with safe return, property damage accident, injury accident, etc)? Typical sequence for ROR-crash Run-off-road-right, overcorrect, cross centerline, collision with other vehicle
SV ROR 1 st Event ROR-right (1,920) 2 nd Event 3 rd Event 4 th Event fixed object (766) ROR-left (1) ROR-right (1) crossed CL (5) ROR-right (1) ROR-left (1) fixed object (1) overturn (183) collision/other (1) fire/exp/imms (2) evasive action (4) fixed object (2) overturn (1) collision w/veh (2) fixed object (1) collision w/other (6) overturn (495) ROR-left (2) fixed object (84) collision w/other (3) fire/exp/imm (2) evasive action (273) fixed object (104) overturn (32) ROR-left (1) overturn (72) fixed object (17) crossed CL (1) ROR-left (1) collision/other (1) ROR-right 4 th : overturn (2) ROR-left (43) fixed object (17) overturn (24) ROR-right (1) collision/veh (1) jack-knife (2) overturn (1) collision/veh (7) fixed object (2) crossed CL (39) fixed object (11) overturn (16) ROR-right (1) ROR-left (3) collision/veh (5) collision/other (1) jack-knife (4) fire/exp/imm (7) vehicle defect (2) Iowa DOT Crash Data, 1 st or 2 nd event ROR, 2005 data fire/exp (1)
Research Question 6 Once a road departure occurs, what kinematic variables can be used to define the different outcomes (road departure crash, near-crash, conflict, or incident)? Research Question 7 How do driver behavior and response influence subsequent events and outcomes after a vehicle initially leaves the roadway? Driver distraction: Conversation Grooming Cell phone Other/multiple behaviors Eating Drinking Smoking Hands-free cell phone Time into trip percent time or number of time driver glances away, amount of time or number of times driver engages in non-driving behaviors Aggressiveness: Percent of time driver exceeds speed limit by a certain threshold Number of times driver engages in hard braking or hard acceleration Headway example: percent time spent following at certain distance Aggressiveness indices from driving questionnaires
Research Question 8 How do vehicle characteristics, such as size, braking capabilities, center of gravity, affect subsequent events and outcome after a vehicle initially leaves the roadway? Research Question 9 What roadway and roadside characteristics influence subsequent events and outcome after a vehicle initially leaves the roadway? Research Question 10 What environmental characteristics influence subsequent events and outcome after a vehicle initially leaves the roadway? Research Question 11 Which non-crash incidents can be used as crash surrogates to assess risk for road departure crashes? Research Question 12 What exposure variables are available and which exposure measures provide the best measure of risk? Possible exposure variables include traffic volume, traffic density, driver sub-populations, VMT.
Necessary data streams from instrumented vehicle studies Identified data elements necessary to answer road departure research questions Identified existing and desirable data streams for full scale instrumented vehicle study Forward facing or other outward facing video Necessary for a number of data elements Highest resolution possible to distinguish data elements (i.e. snow cover from concrete pavement surface) Color desirable (but cost and data storage are prohibitive)
Low resolution versus higher resolution imagery for the same road type
Color versus B&W notice ability to distinguish rumble strips
Color versus B&W notice ability to distinguish shoulder type and presence of centerline
Necessary data streams from instrumented vehicle studies Driver face video Accelerometer GPS Forward and side radar Images of passengers (# of passengers is risk factor for young drivers, distraction for other ages) On-board vehicle data extraction Include ABS data if possible Light meter (detect presence and amount of artificial overhead light at night)
General suggestions Retain data Ability to detect lane position is critical to understanding run-offroad crashes Driving study should include data streams and integrated or post processing ability to detect lane position Image source: Iteris
General suggestions Ability to detect presence of curves is critical to understanding run-off-road crashes May not be available in roadway or mobile mapping datasets
Data Elements Identified data elements necessary to answer road departure research questions Literature review of elements most likely to influence What is currently collected in naturalistic driving studies or in roadway database What can realistically be collected prioritized
Suggested data elements from naturalistic driving study
Suggested data elements from mobile mapping van or roadway databases
Area wide data to consider for pilot study locations Comprehensive crash data for all public roadways Spatially located Includes attributes such as sequence of events, driver action, roadway contributing circumstances, etc Compressive roadway database for all public roadways Can spatially linked to crash database Includes attributes such as speed limit, number of lanes, shoulder type, etc.
Analysis Plan Define surrogate crash measures Select exposure based risk measures Identify data elements (independent variables) Develop data extraction methodology Develop analytical tool framework Assess limitations Suggest modifications for full scale SHRP II field study