Making Signal Systems Work for Cyclists
A recent FHWA study of inductive loop sensors focused on whether they adequately detect motorcycles and bicycles.
“Imagine a bicyclist or motorcyclist who sits indefinitely at an intersection waiting for the light to change before being able to move forward,” says Carol H. Tan, Ph.D., former program manager for the Pedestrian and Bicycle Safety Research program, Federal Highway Administration (FHWA), and currently project manager for FHWA’s congressionally mandated Motorcycle Crash Causation Study. “Transportation professionals know that the cycle’s wheels have to cross a sensor in the pavement so the traffic signal can detect the vehicle’s presence. However, even if bicyclists and motorcyclists know there is a sensor, the riders may not know exactly where they need to place their vehicles to be detected.”
A large number of control systems for traffic signals detect small vehicles, including bicycles and motorcycles, by relying primarily on inductive loop detectors (ILDs), a type of in-roadway sensor. An inherent problem associated with detecting motorcycles and bicycles is assuring that riders will ride within the loop’s detection zone. Also, pavement deterioration, improper installation, and street and utility repair can impair a loop’s integrity and degrade its performance in detecting any vehicle, much less small vehicles. Thus, problems in detecting cyclists actually can be the result of problems in sensor or pavement design, installation, or maintenance.
A recent study by FHWA focused on identifying whether ILDs, in practice as well as in principle, should trigger actuations when motorcycles and bicycles pass over them. Are the problems with detecting small vehicles such as bicycles and motorcycles likely to lie with a fundamental flaw in the design and installation of the ILDs or with the implementation and operational settings? The study’s results showed that the settings and location where a cycle crosses the loop, not the design and installation, are mainly responsible for the poor detection of small vehicles.
The Electrical Principle Of How ILDs Work
ILDs, which are loops of wire embedded into sawcuts in the pavement, typically are installed 5 to 10 centimeters (2 to 4 inches) below the road surface.
The loops are electromagnetic sensors. A loop amplifier in a traffic signal’s controller cabinet passes a small alternating current through the loop wire in the pavement at a very low frequency. An engineer usually sets the frequency at about 10 to 200 kilohertz (kHz). The current flowing through the loop wire generates a weak magnetic field around the wire.
An electrically conductive metal object, such as a vehicle passing over or stopped within the sensor’s detection area, induces eddy currents in the inductive loop’s magnetic field, opposing the original magnetic field produced by the loop. In other words, the vehicle decreases the loop’s inductance (an electrical property), producing an electrical signal. The signal is transmitted through a curbside junction box (a “pull box”) to an electronics unit housed in the traffic signal’s controller cabinet. The electronics unit analyzes the signal, interpreting it as the presence or passage of a vehicle, and sends an appropriate call to the traffic signal’s controller.
The sensitivity of the loop system is critical. Loop system sensitivity is defined as the smallest change of inductance at the electronics unit terminals that will cause the controller to activate. Many States specify that the electronics unit must respond to a 0.02 percent change in inductance, and typically many departments of transportation (DOTs) set the sensitivity setting at 4 or even lower by observing the flow of traffic and turning the sensitivity down until they stop getting detections and then turning it up a notch. (Note: On digital detectors with alphanumeric readouts, the scale typically goes from 1 to 10.) If no bicycles or motorcycles have gone by, inadvertently they might set the sensitivity too low.
The mass of the vehicle is important only insofar as it increases the conductive area over the loop. Because motorcycle and bicycle hubs are vertical metal circles, when a bicycle or motorcycle travels over a loop wire, eddy currents are induced in the conducting wheel rims and frame. When a bicycle or motorcycle is directly over the loop wire, the coupling between the inductive loop and the cycle is maximized — hence detection of the cycle. Bicycles have far less conductive material than automobiles and therefore represent the worst case for detection.
For traffic engineers, the engineering problem with detecting motorcycles and bicycles is that they are very small vehicles. Thus, they have significantly less cross section and conductive metal to be picked up by a sensor than do cars and trucks. Making matters worse for the engineer, many high-tech cycles have significant portions made of plastics. Many engineers and technicians are reluctant to use higher sensitivity levels because the settings may cause detection of highly conductive vehicles in adjacent lanes and may cause cross talk or interference between loops. If cross talk issues arise, engineers may consider upgrading to detector amplifiers that allow use of a wider selection of frequencies to prevent cross talk. Depending on the circumstances, the engineers may wish to tolerate some false calls, as failure to call a green for a vehicle may induce dangerous driving behaviors while an unneeded call to a phase without traffic will cause unwanted delay and stops. FHWA will be investigating the Type Q and Type D loops in further research to see how well they preclude these problems.
The Study’s Parameters
The FHWA researchers examined two of the most common ILD layouts: the 1.8- by 1.8-meter (6- by 6-foot) square and 1.8-meter (6-foot)-diameter circular loops. Both types of loops were installed at a test intersection at the Turner-Fairbank Highway Research Center (TFHRC) in McLean, VA. The TFHRC Intelligent Intersection is a four-legged intersection with one leg being the exit from the building’s parking lot.
In this preliminary set of experiments, FHWA researchers conducted the loop tests with a touring-class motorcycle and an aluminum frame bicycle.
The study’s researchers also conducted a literature review and drew primarily on: (1) FHWA’s Traffic Detector Handbook: Third Edition — Volume I, 2006, which describes the electrical principles and the practical experience of State and local DOTs in utilizing traffic sensors; (2) Design Considerations for Detecting Bicycles with Inductive Loop Detectors by Kidarsa, Pande, Vanjari, Krogmeier, and Bullock (conducted at Purdue University and published in 2006 by the Transportation Research Board of the National Academies), which discusses systematic inductance tests with wheel rims; and (3) Detection of Bicycles by Quadrupole Loops at Demand-Actuated Traffic Signals, published in 2003 on the Internet by Steven G. Goodridge, Ph.D., which discusses the electrical principles of quadrupole loops as related to bicycle detection.
First Experiments: Bicycle Actuation or No Actuation?
In the first set of experiments, an FHWA engineer rode a 10-speed bicycle with an aluminum frame across four square loops on the Fairbank entrance approach and one circular loop on the Turkey Run Farm Road entrance approach. The passes were 0.3 meter (1 foot) to the left of the left loop edge, on the left loop edge, 0.3 meter (1 foot) to the right of the left loop edge, across the middle (center) of the loop, 0.3 meter (1 foot) to the right of the middle (center of the loop), 0.3 meter (1 foot) to the left of the right loop edge, on the right loop edge, and 0.3 meter (1 foot) to the right of the right loop edge.
The researcher made two passes, one with the loop amplifier set to a sensitivity level of 6 and the other at a sensitivity level of 4. At the sensitivity level of 6, all passes yielded an actuation at the loop amplifier. When the sensitivity level was reduced to 4, only the passes made over the left and right edges of the loop set off an actuation.
Motorcycle Delta L/L Values as Vehicle Goes Over Each Loop on Approach
|First Trial Cycle Enters Loop Area||Left of Left Loop Edge||Left Edge||Inside Left Edge||Middle||Right of Middle||Left of Right Wire||Right Edge||Right of Right Loop Edge|
|Square 5.3||Bad Loop||Bad Loop||Bad Loop||Bad Loop||Bad Loop||Bad Loop||Bad Loop||Bad Loop|
|Bar 5.2||Not Measured||0.057||Not Measured||0.121||Not Measured||Not Measured||0.121||Not Measured|
|Quad 5||No signal||No Signal||No Signal||0.221||Not Measured||Not Measured||0.067||No Signal|
|Quad 5 appears to be wired incorrectly||e.g., defective||0.213 at stop bar||0.066 at stop bar for Quad|
|Cycle Exits Loop Area|
The researchers collected the data shown on this table on October 4, 2007. They set the amplifier at a sensitivity level of 4 for the first trial, and the measured vehicle was the touring motorcycle. Square loops numbered 5.6 through 5.1 in the controller cabinet were three-turn, 1.8- by 1.8-meter (6- by 6-foot), square sawcut loops with a depth of 10 centimeters (4 inches), with diagonal sawcuts for corners. The left inside and right inside were 0.3 meter (1 foot) inside the corresponding loop edge. The left of loop and right of loop were 0.3 meter (1 foot) outside the corresponding loop edge. The left edge and right edge were taken in the direction of traffic. The Bar Loop was a 1.8-meter by 0.15-meter (6-foot by 6-inch) loop across the roadway with amplifier number 5.2 in the controller cabinet. Circle 5.1 was a 1.8-meter (6-foot)-diameter circular loop attached to amplifier number 5.1 in the cabinet. The researchers found the Quad 5, which was a 1.8-meter by 6.5-mete r (6-foot by 20-foot) quadupole loop numbered 5 in the controller cabinet, to be defective during the testing. Source: FHWA.
Second Experiments: Motorcycle Actuation
Motorcycles have considerably more conductive material than bicycles have. After the first experiments, the researchers learned to set the loop amplifier to display the change in loop inductance, or delta L over L value, caused by the presence of conductive material in the motorcycle/bicycle.
An FHWA engineer volunteered to ride his new touring motorcycle over the loops. A touring cycle offers a best case for detection because it has a significant amount of conductive material. The researchers therefore set the loop amplifier to a low sensitivity of 4 to see whether it would detect the motorcycle at that setting. The motorcycle was detected consistently by three square loops and the one circular loop on the Turkey Run Farm entrance approach. The motorcycle experiment verified that two other loops (one square and one quadrupole, which failed to detect the motorcycle) also were marked as failed by the loop amplifier diagnostics.
The data from FHWA’s study reinforce the sensor field intensity mapping study conducted at Purdue University by Kidarsa, Pande, Vanjari, Krogmeier, and Bullock. The contours measured by Purdue reflect the change in delta L over L as the bicycle rim passes over different portions of the loop. These contours thus relate directly to the potential sensitivity of the loop to the bicycle rim. These are effectively the same results as FHWA’s less rigorous study — the closer the cyclist is to the wire edges in the direction of travel, the more likely the sensor is to detect the presence of the cyclist and trigger an actuation to call for a green phase for the cyclist.
If a DOT does not have an appropriate testing tool in house, various traffic signal manufacturers offer inductive loop testers for purchase, and several vendors offer loop amplifiers with liquid crystal displays (LCDs) that can report directly the delta L over L values and whether the loop has failed.
When a motorcycle does not actuate a loop, the traffic engineer should ask a technician to adjust the sensitivity setting higher to medium or high (on a low, medium, high loop amplifier) or to 6 or above on a digital display amplifier, as needed. The engineer also should have the technician run the diagnostics recommended in FHWA’s Traffic Detector Handbook to verify that the inductive loop is functioning correctly.
What Is Next?
The next series of experiments to be conducted need to examine long quadrupole loops, that is, loops that have four magnetic poles. Quadrupole loops commonly are used for presence detection and will keep recording a call so long as a vehicle is passing over any portion of the loop. Long loops also are used to keep traffic-actuated controllers displaying a green indication while there is significant traffic flow on the approach serviced by the loop. Quadrupole loops are wired such that they have twice the number of wires in the middle of the loop than they do in the edges.
Engineers have designed custom loops optimized for motorcycle and bicycle detection. These custom loop designs include the Type D and Q loops. The wires of the Type D loop are diagonal to the vehicle travel path to maximize interaction. “The short, 6-foot [1.8-meter]-long version of the quadrupole has the reputation for being very good at detecting small vehicles,” says Milton K. “Pete” Mills, an electrical engineer, now retired, from FHWA’s Office of Safety Research and Development, “because the quadrupole loop has wires in the direction of bicycle travel in the middle of the lane and at the loop edges. A typical square loop has wires in the direction of bicycle travel at the loop edges. A bicycle in the middle of the lane will not be detected with this type of loop because the magnetic coupling between the loop wire and bicycle is zero when the loop wire is normal (perpendicular) to the bicycle wheel. For best detection by any loop, the cyclist must travel along the portion of loop wire in the direction of travel. This results in the greatest magnetic coupling between the loop wire and the bicycle or motorcycle.”
In addition, the FHWA researchers would like to repeat the Purdue experiments using the TFHRC loops to verify those findings. They also would like to add bicycles with carbon fiber wheel rims in future experiments to verify initial results that others have found about the detectability of those rims. This verification would enable FHWA to recommend with certainty in the Manual of Uniform Traffic Control Devices (MUTCD) where on the roadway to install signs and markings. These signs and markings would advise cyclists precisely where to ride and/or stop their vehicles to be detected by properly working sensors and actuate a request for a green phase.
The FHWA tests showed that the most important aspects of detection for a working loop are the sensitivity setting of the detector amplifier and the location on the loop where the cycle crosses the loop. The preliminary recommendations are that engineers use a sensitivity setting of 6 for the loop amplifier wherever possible. Whenever there are issues with cyclists not being detected and the loops are adequately sensitive, the DOT should consider deploying MUTCD signs and pavement markings to tell the cyclists where to ride or stop on the pavement to enable the sensor to detect the cyclist and request a green signal indication from the traffic signal controller. (See “For more information” in right column.) In addition, FHWA would recommend to cyclists that, for traveling on signalized streets, they utilize wheels made of conductive metal to maximize their detectability and that they wait on the left or right edges of the loops.
If a DOT engineer can see the sawcut of a 1.8- by 1.8-meter (6- by 6-foot) square or circular loop on a section of pavement, then a motorcycle or bicycle riding over the left or right edge of the loop should actuate the signal. If it does not, then (1) the loop sensitivity setting was not set for motorcycle and bicycle detection and should be raised to 6 or above, (2) the loop is malfunctioning and should be repaired (this indeed was the case for two of the loops FHWA tested), or (3) the vehicle wheel rims are not made of conductive material. The tests at both the TFHRC Intelligent Intersection and Purdue University back this conclusion. Therefore, the conclusion is true with a high degree of probability.
David Gibson, P.E., is a highway research engineer on the Enabling Technologies Team in FHWA’s Office of Operations Research and Development. He has a master’s degree in transportation from Virginia Polytechnic Institute and State University. His areas of interest include traffic sensor technology, traffic control hardware, modeling, and traffic engineering education. He worked on the first two editions of the Traffic Detector Handbook and the original Type 170 traffic signal controller system.
For more information, see FHWA’s Traffic Detector Handbook, Third Edition, Volume I at www.fhwa.dot.gov/publications/research/operations/its/06108/index.cfm and Volume II at www.tfhrc.gov/its /pubs/06139/index.htm. Contact David Gibson at 202–493–3271 or firstname.lastname@example.org.
For details on how to use the bicycle actuation sign and marking, see MUTCD Section 9B.12 Bicycle Actuation Sign (R10-22) at: http://mutcd.fhwa.dot.gov/HTM/2003/part9/part9b.htm and MUTCD Section 9C.05 Bicycle Detector Symbol at: http://mutcd.fhwa.dot.gov/HTM/2003/part9/part9c.htm #section9C05. For bicycle actuation sign layout and pavement marking layout, see Standard Highway Signs at: http://mutcd.fhwa.dot.gov/ser-shs_millennium.htm. For the pavement marking layout, see: http://mutcd.fhwa.dot.gov/HTM/2003/part9/fig9c-07_longdesc.htm. X. Peter Huang, Terry Halkyard, and Randall VanGorder assisted with the data collection process in this study.