Road Departure Crash Warning System Field Operational Test in the US

From FOT-Net WIKI
Jump to: navigation, search
Road Departure Crash Warning System Field Operational Test in the US
General information
Type: Field operational test
Tested system/service: Autonomous Systems
Countries: USA 78 test users
5 partners 11 vehicles
Active from 12/2001 to 01/2006
Contact
Final Report (PDF)
James R. Sayer
jimsayer@umich.edu
University of Michigan Transportation Research Institute
Catalogue entries
Data catalogue Tools catalogue
Data sets used in this FOT:

No data set is
linked to this FOT

Click here to create a new
data set or here to browse
existing data sets to link
them to this FOT

The following tools
were used in this FOT:

No tool is linked
to this FOT

Click here to create a
new tool or here to
browse
existing tools
to link them to this FOT

This project was conducted under a cooperative agreement between the U.S. Dept. of Transportation and the University of Michigan Transportation Research Institute, along with its partners, Visteon Corporation and AssistWare Technologies. Road departure crashes account for 15,000 fatalities annually in the U.S. This project developed, validated, and field-tested a set of technologies intended to warn drivers in real time when the driver was drifting from their lane, and a curve-speed warning system designed to provide alerts to help driver slow down when approaching a curve too fast to safely negotiate the curve.

Two areas were addressed: safety-related changes in driver performance including behaviour that may be attributed to the system, and levels of driver acceptance in key areas. Testing used 11 passenger sedans equipped with RDCW and a data acquisition system that compiled a massive set of numerical, video, and audio data. 78 drivers each drove a test vehicle, unsupervised, for four weeks. The resulting data set captured 83,000 miles of driving, with over 400 signals captured at 10 Hz or faster.

Details of Field Operational Test

Start date and duration of FOT execution

This testing occurred within a 10-month window including summer, fall, and winter weather. Each driver was driving the vehicle for a period of 4 weeks.

Geographical Coverage

Drivers were recruited from the southeast Michigan area, including Detroit and surrounding suburbs and rural areas.

Link with other related Field Operational Tests

The RDCW FOT project was created as part of the Intelligent Vehicle Initiative (IVI). The IVI effort itself was a continuation of an ongoing ensemble of inter-related programs within US DOT on crash avoidance technology that grew in scope in the early 1990s (USDOT 2005).

Objectives

The purpose of the Road Departure Crash Warning System Field Operational Test (RDCW FOT) was to gain insight into the suitability of road departure crash warning systems for widespread deployment within the U.S. passenger vehicle fleet. This was done by developing and field-testing a set of automotive crash warning functionalities – the RDCW system [LDW and CSW] – and observing a set of lay drivers as they used an equipped test vehicle as their own personal travel vehicle for four weeks.

The project sought to determine the suitability of introducing such a system into the U.S. light vehicle fleet.

The two critical objectives while analyzing the field test data were:

  • an assessment of potential safety impacts of the RDCW system, and
  • an assessment of the driver acceptance of such a system.

Results

Analysis of the data shows that with the RDCW system active, relative to the baseline condition, drivers improved lane keeping by remaining closer to the lane center and reducing the number of excursions near or beyond the lane edges. In addition, turn signal use increased dramatically. The data, however, were unable to confirm a change in driver’s curve taking behaviours that could have been attributed to the curve speed warning system. Driver acceptance was generally positive in relation to the lateral drift component of the system, with reactions to the curve speed warning system being rather mixed. Many additional results and insights are documented in the report.

Lessons learned

  1. Improve the CSW false alarm rate
    • Based on analysis of FOT data, the consumer acceptance of CSW can be improved significantly by a relatively small software change that reduces the number of false alarms.
    • Based on vehicle level testing of the CSW software change, implemented on an RDCW development vehicle, a substantial reduction of 50-90% of the false alarms was obtained.
  2. Improve LDW availability
    • Availability of LDW will be increased in the next generation by utilizing a high dynamic range CMOS camera.
  3. Reduce System latency
    • Common communication protocol
    • Leads to reduction of missed alarms or other unintended side effects
  4. Design System Visualization/Diagnostic Tools
    • Tools for individual subsystems have limited usefulness
    • Work off the DAS model to minimize resources
    • It will make system development, refinement and testing much easier
  5. FOT methodology has merit
    • The FOT methodology can be effective at discovering and quantifying significant safety-relevant changes in driver behaviour

Main events

Financing

Summary, type of funding and budget

Cooperation partners and contact persons

  • Public Authorities: U.S. Dept. of Transportation
  • Industry
    • Vehicle Manufacturer:
    • Supplier: Visteon Corporation, AssistWare Technologies
  • Users:
  • Universities: University of Michigan Transportation Research Institute
  • Research Institutes:
  • Others (specify): Volpe Center (Independent evaluator)

Applications and equipment

Applications tested

The RDCW system targeted crashes involving vehicles that drift off the road edge or into occupied adjacent lanes, as well as those involving vehicles travelling too quickly into turns for the driver to maintain control. Included in the RDCW package were two warning functions.

The lateral drift warning system (LDW) was intended to help drivers avoid drifting off the road by providing a set of driver-alert cues when the vehicle was observed to be moving over either dashed or solid lane edge boundaries. The driver was expected to assess the situations and consider steering the vehicle back into the original travel lane if the drift was unintentional. The crashes addressed by LDWD are often associated with driver inattention, intoxication, and drowsiness. The LDW system used a camera to observe visual features that delineate lane and road edges, such as painted lane boundaries. Furthermore, a set of onboard 2 radars was used to modulate the warnings when potentially dangerous objects were sensed alongside the edge of the lane or road.

The curve speed warning system (CSW) was intended to help drivers slow down to a safe speed before entering an upcoming curve. The desired driver response to a CSW alert was for the driver to consider applying the brakes to slow the vehicle and reduce the lateral acceleration in the curve ahead. The CSW system relied on GPS and a digital map to anticipate the curve location and radius. Measurements of recent driver control actions, such as changing lanes or applying turn signals, were considered in CSW’s decision to issue an alert. Both the CSW and the LDW used a set of visual, audible, and haptic cues to alert the driver at two levels per system.

Vehicle

A fleet of 11 vehicles (Nissan Altima 3.5SE sedans (model year 2003) purchased by the project team ) was equipped with the RDCW system and a data collection system in order to conduct the field operational test.

Prototype vehicles have been used

Equipment carried by test users

Infrastructure

Road marking lanes and edges of the road were used by the system. Furthermore, one of the primary sensors for CSW was GPS, in combination with a digital map database.

Test equipment

The DAS was developed and managed by UMTRI and collected data from the RDCW CAN bus, two radar buses, two video streams, audio stream and several other instruments installed to monitor other aspects of the experiment.

These sensors monitored two axes of vehicle acceleration, vehicle location (via differential GPS separate from the RDCW GPS), steering wheel angle, pitch angle, roll angle, roll angle rate, instrumentation space temperature, and outputs from a cellular phone antenna.

Methodology

Pre-simulation / Piloting of the FOT

In order to identify and mitigate RDCW and data system issues, fine-tune system functionality, and explore preliminary driver perceptions, a multi-stage sequence of pilot tests was conducted. The first two stages (Stages 1 and 1.5) involved laypersons who drove an RDCW-equipped vehicle along a predetermined route while accompanied by UMTRI research staff. Using the results of these tests, minor modifications to the RDCW system were made in preparation for a short FOT-style pilot test (Stage 2). Stage 2 pilot testing involved laypersons who drove RDCW equipped vehicles for a 12-day unsupervised period. For the first four days of driving, the RDCW system was disabled. Beginning on the fifth day, the drivers experienced the RDCW system for a total of eight days before returning to UMTRI and completing a post-drive questionnaire.

Method for the baseline

Each driver was trained briefly on the RDCW system and then asked to drive the vehicle where – and how – they normally would during their four weeks of vehicle use. In order to account for variations between drivers, the RDCW alerts were not displayed in the first week of use, which provided a baseline data set for each driver. The RDCW system displays were then presented to the driver during the subsequent three-week period.

Techniques for measurement and data collection

The output of the DAS was composed of (1) the set of complete data files, stored onboard the vehicle and later off-loaded at UMTRI, that contained all the numeric, audio, and video data, and (2) smaller data files, transferred to UMTRI via cellular modem each time the ignition was turned off, that contained summary and diagnostic data gathered during the previous trip. These later files, and the cellular transfer mechanism allowed near-real-time monitoring of the use and the health of the RDCW vehicles.

Objective (logged data…):

Approximately 400 data signals were collected at a rate of 10 Hz or higher (up to 50Hz), with video capturing the forward scene and driver’s face at various frame rates throughout testing (Video and audio data were captured).

Subjective (questionnaires, focus groups…):

At the conclusion of their 26-day RDCW driving experience, drivers returned the test vehicle to UMTRI. During a two-hour debriefing session, drivers completed an extensive questionnaire investigating their experiences with and impressions of the RDCW system. While the driver completed the questionnaire, a researcher prepared to show the driver video from a sample of alerts from their time with the RDCW vehicle. Once drivers had completed the questionnaire, the researcher discussed their responses with them and drivers were provided an opportunity to offer further amplification and clarification where necessary. Lastly, the alert-event videos were replayed for the driver. Detailed feedback concerning the usefulness of each of the alerts was elicited via a five-point utility rating scale.

Upon completion of their debriefing, drivers were invited to participate in one of four focus group sessions. The focus groups provided drivers with the opportunity to expand on their answers to the detailed questionnaire and provide additional information about their experience with the RDCW system.

Recruitment goals and methods

78 drivers distributed evenly by gender and within three age cells. Time of exposure was 4 weeks (including 1 week for baseline).

With the exception of the first ten drivers who were recruited through local newspaper ads, drivers were recruited with the assistance of the Michigan Secretary of State’s office. Six thousand licensed drivers were selected at random for possible participation in the FOT. From this randomly selected pool of 6,000 drivers, smaller random samples of names were selected to receive informational postcards. A minimum-annual mileage threshold was required for a driver to qualify as an experience of at least 2 years of driving.

Methods for the liaison with the drivers during the FOT execution

During the FOT, two researchers carried pagers which shared a common number. Researchers were available 24 hours per day. Drivers were instructed to contact a researcher if they were involved in a crash, had mechanical or RDCW system problems, or simply had questions about the RDCW system. A cell phone was placed in each test vehicle so that drivers could conveniently contact researchers.

On a limited number of occasions, UMTRI researchers had to initiate contact with drivers: when a RDCW system component failure was detected by remote monitoring, when system software upgrades were required and when a lack of activity was detected.

Methods for data analysis, evaluation, synthesis and conclusions

An analysis of the trip data was performed to remove trips that were either problematic due to a faulty sub-system or sensor or were deemed invalid for some other clear reason (e.g., video indicated that someone other than the subject was driving the car). Also, included in this group of invalid trips, were trips with zero distance travelled.

The Van der Laan scale was used to measure acceptance of both LDW and CSW, as well as of the RDCW system as a whole. The scale was integrated into the post-drive questionnaire near the end of each subsystem section (i.e., the scale appeared three times within the questionnaire).

Sources of information

Final Report: http://deepblue.lib.umich.edu/bitstream/2027.42/49242/1/99788.pdf

RDCW Final Research Report (UMTRI-2006-9-1)

Presentation “An IVI Road Departure Crash Warning System Field Operational Test”

CompanyUniversity of Michigan Transportation Research Institute +
ContactJames R. Sayer +
CountryUSA +
EndedJanuary 2006 +
Is type ofField operational test +
NameRoad Departure Crash Warning System Field Operational Test in the US +
Number of partners5 +
Number of test users78 +
Number of vehicles11 +
StartedDecember 2001 +
Tested system or serviceAutonomous Systems +
Websitehttp://deepblue.lib.umich.edu/bitstream/2027.42/49242/1/99788.pdf +
Email
This property is a special property in this wiki.
jimsayer@umich.edu +