By: Taek M. Kwon, Ph.D., Ryan Weidemann and Victor Lund, P.E.
In Minnesota, the majority of intersection-related crashes occur at rural, two-way-stop intersections.
Sight restrictions caused by vertical and horizontal curves increase the risk of an intersection crash by making it difficult for drivers to safely accept a gap in the traffic stream. Right-angle crashes account for the largest percentage of crashes at these intersections and the vast majority of them are gap-recognition-related. Rural crashes also are generally more severe than urban crashes due to factors such as speed, and availability and response time of emergency services.
Static intersection warning signs are commonly used in advance of intersections to warn drivers on the major road that an intersection is ahead. These warning signs appear to be ineffective at changing driver behavior. Various intersection countermeasures such as realigning intersection approaches, correcting approach grades, clearing sight triangles and even installing roundabouts have been shown to be effective at improving intersection safety performance. However, these countermeasures are expensive and cannot be readily justified at low-volume, rural intersections, especially for local-road authorities such as counties and townships with limited budgets.
This article presents the results of a two-year research project performed by the University of Minnesota-Duluth in collaboration with St. Louis County, Minn., and sponsored by the Minnesota Local Road Research Board (LRRB). The project developed a low-cost, dynamic intersection warning system that provides information to drivers based upon the presence of traffic at rural, two-way-stop intersections, referred to as the ALERT system. This acronym was derived from the original title of the research project, Advanced LED Warning Signs for Rural IntersecTions Powered by Renewable Energy.
Being ALERT
The ALERT system was designed as a low-cost, easy-to-install warning system to replace the static warning signs that are commonly used at rural, two-way-stop intersections. The purpose of the ALERT system is to provide additional information to drivers so they can safely navigate the intersection. The ALERT system consists of nonintrusive vehicle detectors that detect vehicles on each approach and efficient LED blinker signs that blink according to the messages received from the detectors. The warning system is powered entirely by photovoltaic (PV) panels (solar panels), requiring no connection to a local power grid, increasing the range of intersections where this system can be installed. All communications between vehicle detectors and blinker signs are transmitted wirelessly, which eliminates the problems associated with buried wires.
The ALERT system was installed and evaluated at the intersection of West Tischer Road and Eagle Lake Road in Duluth. This rural, two-way-stop intersection was chosen because of a severe vertical curve on the east approach of the major road. The curve significantly reduces the available intersection sight distance for vehicles stopped on either the north or south approaches of the minor road. Conversely, drivers traveling westbound on the major road cannot see cross traffic at the intersection until they are nearly in the intersection.
The basic operation is described using the system diagram illustrated in Figure 1. The blinker sign for the major road, S1, displays the message “CROSS TRAFFIC WHEN FLASHING” for westbound traffic. There are two radar detectors (D2, D3) installed on the top of the existing STOP signs at the intersection. When a vehicle is detected at either STOP sign, a wireless signal is transmitted to S1 and the LEDs blink for as long as a vehicle is detected at either STOP sign. The activation of S1 warns vehicles traveling westbound on the major road that there is cross traffic detected at the intersection ahead.
The blinker signs for the minor road, S2 and S3, display the message “VEHICLE APPROACHING WHEN FLASHING” for northbound or southbound traffic. These signs were installed on the northeast and southwest quadrants of the intersection. Two radar detectors are located on the major road (D1 and D4). Detector D1 only detects vehicles traveling westbound and detector D4 only detects vehicles traveling eastbound on the major road. When a vehicle is detected by either radar detector (D1 or D4), a wireless activation signal is transmitted to S2 and S3 and the LEDs blink for a fixed time period of 10 seconds. This flash time was calculated as the expected time for a westbound vehicle traveling at the posted speed limit (55 mph) to pass through the intersection. Signs S2 and S3 provide a warning to vehicles stopped at either STOP sign that a vehicle was detected approaching the intersection from the east or west on the major road.
Slower reaction
A before-and-after crash analysis was not performed for this study. The experimental site is a low-volume intersection and there was only one crash between 1998 and the present. The analysis time period of the ALERT system was one year. As such, a statistical crash analysis would have had no value. Instead, this study observed changes in driver behavior as a surrogate for the safety effects of the ALERT system.
Video data showing vehicle movements was recorded at the intersection and analyzed before and after the installation of the ALERT system. An on-site video monitoring system was installed at the intersection, consisting of a digital video recorder and two color-video cameras. The first camera recorded vehicles traveling westbound on the major road toward the intersection. The second camera recorded vehicle movements at the intersection. Video recording was limited to periods of vehicle activity (triggered by a motion detection) to reduce the total amount of data.
The analysis of the video data consisted of three traffic parameters: speeds of westbound vehicles on the major road, time a vehicle was stopped at a STOP sign, also known as wait time, and vehicles that rolled through a STOP sign, also known as roll-throughs.
To measure vehicle speeds, two transverse white lines, referred to as speed markers and separated by a distance of 150 ft, were installed on the pavement surface. Each video file was analyzed by a computer program, which counted the number of frames for a vehicle to travel the distance between the speed markers. Using this frame count, vehicle speeds were calculated to within a 1-mph accuracy.
Vehicle speeds were analyzed in four different categories. First, average speeds for vehicles traveling at daytime or nighttime. Second, average speeds for vehicles traveling during the morning peak-traffic periods (7 a.m.–9 p.m.), afternoon peak-traffic periods (3 p.m.–5 p.m.) and off-peak traffic periods (all other times). Third, average speeds for vehicles traveling either during the weekday or weekend. A weekday was defined as Tuesday, Wednesday or Thursday. A weekend was defined as Saturday or Sunday. And fourth, average speeds for vehicles on a monthly basis. Holidays were eliminated from each average-speed comparison.
The comparison of average speeds before and after the installation of the ALERT system found average vehicle speeds decreased only for the nighttime period after installation. It was presumed that this reduction in the average vehicle speed was due to higher visibility of the LED blinker signs at night.
Vehicle speeds were then analyzed after the installation of the ALERT system when sign S1 was blinking, indicating a vehicle conflict, or not blinking, indicating no vehicle conflict. A conflict was defined as the time when a driver on the major road did observe the sign blinking. No conflict was defined as when a driver on the major road did not observe the sign blinking. The results are summarized in Table 1.
In all cases, vehicle speeds decreased for the conflict condition with an average decrease of 4.5 mph. This decrease in speed translates to an increase of 1.2 seconds in time from the moment the driver sees the blinking sign and the vehicle enters the intersection. This increases the reaction time of the driver on the major road as well as increases the gap time for a vehicle on the minor road. This change in behavior suggests that the driver on the major road is well informed that a vehicle on the minor road is present at the intersection. This is a critical observation given that static intersection warning signs appear to be ineffective at changing this driver behavior.
A comparison of average wait times after installation of the ALERT system was performed when the signs S2 and S3 were blinking, indicating a vehicle conflict, and when they were not blinking, indicating no vehicle conflict. A conflict was defined as the time when a driver on the minor road was waiting for a vehicle on the major road to pass through the intersection so they could complete their turning maneuver. No conflict was defined as the time when a driver on the minor road could complete their turning maneuver after coming to a complete stop at the STOP sign. The average wait time increased when signs S2 and S3 were blinking by an average of 5 seconds. The ability of the driver on the minor approach to identify acceptable gaps improved as indicated by a longer wait time for vehicles on the minor road.
A comparison of roll-throughs of vehicles on the minor road was performed before and after the installation of the ALERT system. There were no roll-throughs observed when signs S2 and S3 were blinking. This suggests the ALERT system eliminated roll-throughs whenever a conflict between opposing vehicles existed. However, roll-throughs actually increased after installation of the warning system (from 13% to 24%) when signs S2 and S3 were not blinking. There appears to be a driver adaptation effect where drivers come to rely on the warning system rather than obey the STOP sign and use their own judgment when entering the intersection. A similar driver behavior was observed in past studies. Although a disturbing effect, this behavior suggests that drivers do understand and use information provided by the warning system.
Customer feedback
A mail-in survey was performed to supplement the video-data analysis. The survey was sent to local residents living within a half-mile radius of the intersection and was printed on pre-stamped postcards to encourage those residents to respond. On the back of the postcard was a short explanation of the purpose of the mail-in survey with four short questions.
There were 96 mail-in surveys distributed. A total of 46 (47.9%) surveys were completed and returned. From the responses received, 77% frequent the intersection at least once a day. A strong majority, 89%, either strongly agreed or agreed that the warning system was easy to understand. Another strong majority, 81%, either strongly agreed or agreed that the warning system has improved the safety of the intersection. A total of 83% strongly agreed or agreed that the warning system could be used at other intersections. And 86% rated the overall effectiveness of the warning system as either excellent, good or fair.
A common theme of the mail-in survey comments was that residents were concerned that motorists will treat the warning system like a traffic signal and disregard the STOP sign and roll-through the intersection when signs S2 and S3 are not blinking. This concern was validated by the intersection wait-time and roll-through data. The majority of residents believe the warning system improves safety, but at the same time were concerned of potential crashes caused by roll-throughs.
Adding to the solution
As a low-cost safety countermeasure for rural, two-way-stop intersection crashes, the ALERT system appears to be effective at positively changing driver behavior by reducing vehicle speeds on the major road, increasing the wait time and altogether eliminating roll-throughs for vehicles on the minor road when a conflict exists at the intersection. When no conflict exists at the intersection, the only observed change in driver behavior was an increase in roll-throughs for vehicles on the minor road.
The following are recommendations for future research. First, integrate the STOP signs on the minor road with blinking LEDs. The STOP signs could blink when the warning signs for the minor road are not blinking, thereby emphasizing the message to drivers on the minor road to obey the STOP sign and use care when completing a turning maneuver at the intersection. Second, install standard-width or wide stop bars. Stop bars were not used at the intersection and could have contributed to the roll-through behavior of drivers. Third, place signs S2 and S3 at an alternate location that would require minor-road drivers to come to a complete stop before comprehending whether the warning signs were blinking. The original design concept of the ALERT system placed the warning signs on the east and west legs of the major road. This alternate location of the warning signs was not implemented due to a concern about standard sign placement. Finally, the ALERT system could be installed at multiple locations to support a more robust crash analysis. If the negative effect of roll-throughs is successfully abated and the positive effect on driver behavior is maintained, the ALERT sSystem could be proven effective as a low-cost system safety countermeasure for rural two-way-stop intersections. R&B
About The Author: Kwon is a professor in the Department of Electrical and Computer Engineering at the University of Minnesota, Duluth. Weidemann is a research associate in the Department of Electrical and Computer Engineering at the University of Minnesota, Duluth. Lund is
