How does Lane Departure Warning Effectiveness The first regards the references (articles/papers) that you should Transportation Research Board | Main This
How does Lane Departure Warning Effectiveness The first regards the references (articles/papers) that you should Transportation Research Board | Main This is the Transportation Research , then read and take notes. As you have time and as this should directly assist your development of a bibliography and related Literature Review and associated references, articles, read them, and take notes following these instructions: Provide the reference information for the article/paper – i.e., authors, title, journal, and such. Provide the abstract – verbatim is fine – and comment on/discuss whether the abstract accurately portrays the content of the article/paper Provide the conclusions and comment on/discuss whether these conclusions are supported by the contents of the article/paper Take notes in the form of bullet points throughout the contents of the article/paper including the literature review, data, method, analysis, and results Consider the references for each article/paper and note commonalities of these references – i.e., if several or every article/paper you read references the same prior articles/papers, then these prior articles/papers are likely extremely pertinent and should be obtained, if possible, and reviewed as well … thanks Proceedings of the Human Factors and Ergonomics Society 2017 Annual Meeting
1929
How does Lane Departure Warning Effectiveness vary by
Severity of Departure
1
Timothy L. Brown1, John Gaspar1, Dawn Marshall1, John D. Lee2
National Advanced Driving Simulator, University of Iowa, Iowa City, IA, USA
2
University of Wisconsin, Madison, WI, USA
Lane departures are a significant traffic safety problem. They can be attributed to a variety of types of
impairment including the increasing prevalence of distracted driving. Research to date has focused on how
drivers respond to imminent lane departure warnings, often with short time to line crossing at the onset of
the alert. This paper examines how this effectiveness changes over a range of severities associated with
various times to line crossing. Although warning systems are generally beneficial in reducing the severity
of lane departures, time to line crossing had mixed effects on lane departures. Longer time to line crossing
was associated with a greater likelihood of departing the lane but with less severe lane departures.
Additionally, an active warning that provides steering torque to help correct lane departures showed
decreased effectiveness with increasing time to line crossing in terms of likelihood of lane departure and
duration and area of lane departure. These results point to the importance of considering the range of
situations in which a warning may be issued, not just the most severe cases.
INTRODUCTION
A great deal of research has been done on the potential
safety benefits of lane departure warning systems
(e.g.,Barickman et al., 2007; Gaspar et al., 2016; Kozak et al.,
2006; Lerner et al., 2011; Lerner et al., 2015; Suzuki &
Jansson, 2003; Tijerina, 1999). Much of that research has
focused on alert timing and the modality of the alert.
Copyright 2017 by Human Factors and Ergonomics Society. DOI 10.1177/1541931213601968
Lane Departure Crash Problem
In 2015 more than 32,7440 fatalities and over 2.44
million injuries occurred in automobile accidents (NHTSA,
2016) . With an estimated annual cost of $230.6 billion in the
year 2000, it also has an immense economic impact on our
society (Blincoe et al., 2002). Lane change crashes accounted
for approximately 9 percent of crashes (NHTSA, 2011). Lane
departure crashes can result from drifting out of the lane into
oncoming traffic, in to adjacent traffic or off the roadway.
Single vehicle roadway departure crashes remain a major
traffic safety concern; regularly accounting for over half of
fatal crashes (Barickman et al., 2007). The largest contributor
for lane change crashes is “encroachment from adjacent
lanes”, accounting for 35% of crashes (Fitch et al., 2009;
Najm & Smith, 2002).
Lane departures are also common byproducts of reduced
attention to the forward roadway. This can be attributed to a
variety of impairments including drowsiness, alcohol, and
distraction. The National Highway Traffic Safety
Administration (NHTSA) has identified distracted driving a
major source of crashes and has been working to reduce them.
In the SAVE-IT project, sponsored by NHTSA, Zhang et al.
(2008) found that visual distraction led to increased lane
departures. In work presented to the Michigan Traffic
Safety/Engineering Advisory Committee, McAvoy and Datta
(2005) listed driver distraction as one of the leading causes of
lane departures, In work completed as part of the Crash
Warning Interface Metrics program, a variety of secondary
tasks were shown to result in increased rates of lane departures
(Lerner et al., 2011; Lerner et al., 2015).
Research has shown that in many cases drivers do not
attempt an avoidance maneuver prior to a crash or departing
the roadway (Chovan et al., 1994; Fitch et al., 2009). This
suggests that drivers may benefit from a lane departure
warning system that can make them aware of the impending
crash and provide them the opportunity to return their
attention to the road and respond.
Evaluating Lane Departure Warning Systems
The work to evaluate these systems has often focused on
the most severe of the lane departures in terms of rate of
exiting the lane. These situations require quick responses due
to the imminent potential to crash. Work conducted by
NHTSA focused on situations with a high rate of departure
(Lerner et al., 2011; Lerner et al., 2015). The method used in
this NHTSA research was similar to other methods employed
to present the driver with an unanticipated departure from the
lane (Tijerina et al., 2010). While this standardization allows
for more powerful analysis of the warning systems, it
potentially limits the generalizability to natural or less severe
departures. The impact of warning systems across a range of
severity of lane departures would help better understand their
potential utility and the impact on effectiveness across
severities.
This paper examines the impact of the severity of the
lane departure at the time the alert is given on warning
effectiveness. The study that provided the data for this
analysis involved the examination of factors that impact the
evaluation of the effectiveness of Lane Departure Warning
Proceedings of the Human Factors and Ergonomics Society 2017 Annual Meeting
(LDW) and Forward Crash Warning (FCW) systems. (Lerner
et al., 2015) This analysis focused only on events that might
trigger lane departures. It was hypothesized that alerts would
be less effective as time to line crossing at alert decreases.
METHOD
The experimental design for this analysis was a mixed
between/within-subject design with between subject factors of
alert onset (12 and 3 inches from lane boundary), and alert
modality (no alert, audio, visual, steering wheel vibration,
directional seat vibration, and steering torque). Additional
measures from the time of the alert that were included in the
analysis were the direction that the vehicle is drifting, the time
to line crossing, and vehicle speed.
1930
Apparatus
Data were collected on the NADS-1 (see Figure 1),
located at the University of Iowa. The simulator configuration
for this study consists of a 24-foot dome in which a 1996
Malibu sedan is mounted. It has 400 square meters of
horizontal and longitudinal travel and ±330 degrees of
rotation. The visual system is comprised of three forward
channels with a resolution of 1600 x 1200; and five rear
channels with a resolution of 1024 x 768. Data were sampled
at 240 Hz.
Two different types of events were included in this
analysis – those where the vehicle might naturally drift from
the lane and those where the driver was “pushed” from the
lane through augmented steering input from the driver within
the vehicle dynamics without the addition of motion cues
associated with the push.
The analysis of the data was completed using the SAS
Logistic Regression and General Linear Models (GLM)
procedures. Speed and time to line crossing were included as
continuous independent variables in the models. Additionally
an interactive effect between time to line crossing and alert
modality was included. Both types of analyses used all of the
independent variables in a full model analysis.
Dependent Measures
The following measures were analyzed for each event where
an LDW alert would have been triggered:
lane departure – binary measure of whether part of
the vehicle departed the lane;
lane departure duration – the amount of time that the
part of the vehicle was out of the lane;
maximum lane exceedance – the maximum lateral
extent of the lane departure; and
area of lane exceedance – a combination measure
represent both the time and lateral extent of the lane
departure.
Figure 1. NADS-1 Driving Simulator
A display for the number recall task located to the right
of the participant’s forward facing position mounted on the
rear of the passenger seat headrest and adjusted for each
participant so that the display was located at an angle of at
least 90 degrees (see Figure 2). One second after receiving the
instruction to begin, a series of five random single-digit
numbers was presented for 472 ms each. The participant was
to repeat them aloud in the correct order to the experimenter
following the task (Forkenbrock et al., 2011).
Participants
Ninety-six out of one-hundred thirteen participants between
the ages of 35 and 55 with a license for at least two years who
drive at least 10,000 miles per year, who are willing to engage
in distracting tasks while driving and who do not drive a
vehicle with advanced safety systems enrolled and completed
all study procedures. Reasons for not completing included
simulator sickness, failure to engage with the secondary task,
failure of events to work correctly and taking medicines
contraindicated for driving. Participants were compensated
$45 for completing all study procedures.
Figure 2. Number Recall
Experimental Procedures
Participants were recruited using an Institutional Review
Board approved registry maintained by the National Advanced
Proceedings of the Human Factors and Ergonomics Society 2017 Annual Meeting
Driving Simulator, postings on our recruitment webpage and
through emails. Interested individuals were screened by
phone to verity eligibility and interest. Upon arrival, informed
consent was obtained. Participants were told that the goal of
the research was to evaluate several new in-vehicle
technologies, when in fact their response to surprise events
was being evaluated.
Participants were trained on the drive, navigation
instruction, number recall task and general procedures, but not
on the alerts. Additionally, participants practiced the number
recall task before going into the simulator to become
comfortable with performing the task until they perform the
task successfully by recalling all five numbers correctly with
no sequence errors.
Scenarios
This drive started in a low-speed urban environment
where the participant became comfortable with driving in the
simulated environment. The main portion of the drive was a
two-lane rural roadway with intersections where the speed
limit alternated between 40 and 55 mph. Throughout the drive,
they were asked to maintain a headway with the lead vehicle.
The drive included 15 occurrences of the number recall task
without a related event, and three occurrences where the task
was coupled with a potential forced LDW event. Additionally
there were two occurrences where the task was coupled with
an FCW event that are not included in this analysis. The drive
was designed so that the LDW events occurred prior to the
FCW events. For the forced LDW events, once the task began,
a lateral push was provided toward the closest lane boundary
until an alert was provided. The push was implemented such
that there was no motion feedback associated with the lateral
push.
RESULTS
The outcomes of the two analyses are summarized based
on the two analyses approaches. The first analysis examines
the impact on keeping the vehicle in the lane. The second
analysis focuses on the impact on the magnitude of the lane
departure and speaks to the potential risk of departing the road
or striking a vehicle in the adjacent lanes.
Departing the Lane
The analysis revealed that the following factors were
associated with a decreased likelihood of lane departure:
earlier warnings;
decreased speed;
departing toward the right; and
increased time to lane crossing.
Additionally, having a steering torque warning with increasing
time to lane crossing increased the likelihood of departing the
lane. Results of the analysis are summarized in Table 1. As
can be seen the direction of departure and the timing of the
alert had the biggest effect on likelihood of a departure.
1931
Table 1. Logistic regression predicting no departure
Parameter
Estimate Pr > ChiSq
Time to Lane
Crossing (TLC)
-0.0908
0.0007
Direction
Left
-0.7342
Purchase answer to see full
attachment