U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
The purpose of this chapter is to further describe intrusive and non-intrusive data collection equipment used by state DOTs. The discussion identifies the limitations, advantages, and evaluation results of the various data collection equipment. The descriptions are intended to provide a basic guide to technology selection.
Equipment used to count traffic volumes and classify vehicles is very similar. In many cases, the only differences are the layout of the sensors on the roadway and user-selectable inputs in the data collection electronics unit. The following sections identify intrusive and non-intrusive detection technologies that agencies typically use to count and classify vehicles. For HPMS purposes, there must be not only a count of total vehicles but a classification of vehicles according to the prescribed classification scheme. Perhaps the most common scenario for states is to maintain continuous count stations that provide year-round counts from automated systems and apply factors from short-term classification counts to estimate the number of vehicles by type.
Agencies typically use portable traffic volume counters for short-term data collection where a single-axle sensor will suffice. These devices can count all traffic on a roadway or an individual lane, depending on how the installer configures the sensors. The road component may consist of pneumatic tubes or other types of sensors (i.e., piezoelectric film or cable, tape switches, inductive loops, and magnetometers).
For the most part, vehicle classification systems currently fit the "intrusive" category, and they can be either permanent or portable. They typically utilize inductive loops, piezoelectric sensors, or a combination of the two sensor types (AASHTO, 1992). In any case, a minimum of two sensors sends detections to a data collection and storage unit at the roadside. Most classifier systems generate their most accurate data by using a combination of both piezoelectric (or other axle sensor) and inductive loop detectors. This means either two piezoelectric sensors and one inductive loop (preferred) or two inductive loops and one piezoelectric sensor. The standard FHWA classification scheme (Scheme F) measures axle spacing, which requires an axle sensor, with inductive loops providing vehicle presence. Automatic vehicle classification (AVC) sites store vehicle classification information for specific lanes (e.g., Long Term Pavement Performance [LTPP] sites) or for each lane of an entire roadway.
All states interviewed rely on a combination of intrusive permanent counting equipment (primarily loops plus piezoelectric sensors) and pneumatic road tubes for short-term counts. The primary method for short-term data collection is road tubes and inductive loops for permanent counts. All the states interviewed have similar issues with using road tubes on high-volume locations, including safety of data collection crew, securing road tubes, and classification errors.
The following are the common problems identified by the states for traffic data collection on high-volume routes:
Pneumatic tubes are hollow rubber tubes stretched across the portion of the roadway for collecting vehicle count and/or speed data. One end of the tube connects to a traffic counter/ classifier with the other end plugged to prevent air leakage as a vehicle crosses the tube. As a vehicle passes over the tube, its tires compress the tube, actuating an air pressure transducer on the classifier. This means that pneumatic tubes operate in pulse mode only.
Although there are several problems associated with them, these tubes are the most common device used by states for short-term counts. Tubes are relatively inexpensive, and installation is quick and easy. These tubes, typically 0.5 inch in diameter, are relatively accurate for light traffic flows, but they damage easily. The safety of traffic personnel installing road-tubes in high-volume roads is also a concern.
The inductive loop consists of one or more turns of insulated loop wire installed in a shallow slot that is sawed in the pavement, a lead-in cable, and a detector electronic unit. Electrical induction consists of a detector unit that passes a current through the stranded loop wire, thereby creating an electromagnetic field around the wire. Moving a conductive metal object, such as a vehicle, through this field disturbs the electromagnetic field, producing a change in energy level. As the vehicle enters the electromagnetic field of the loop, it causes a decrease in the inductance of the loop and an increase in the oscillation frequency. The inductive loop detector, which was introduced in the 1960s, continues today as the most commonly used form of detector, even though its weaknesses are widely recognized.
Proper installation of the loop in the road surface is important to ensure the reliability of the system. Some pavement surfaces, such as bridge decks, preclude the saw cutting necessary to install permanent inductive loop detectors. A primary disadvantage of inductive loop detectors is the expense of relocating or repairing loops after installation. This procedure requires extensive traffic control and results in congestion and motorist delay (Tyburski, 1989). Detector "cross-talk" and increased pavement stress are two additional disadvantages of inductive loop detector systems. There are also several adverse conditions that affect the operation of inductive loops, including high voltage power lines under the pavement, a pavement subsurface with a high iron content, and unstable pavement conditions. Underground wires, conduit, and pull boxes are susceptible to being damaged by utility work. Modern detection electronics can overcome the first two conditions, but changing or unstable pavement conditions result in increased inductive loop maintenance costs (TTI, 1992). One advantage of inductive loop systems over some of the non-intrusive alternatives is their ability to maintain accuracy in all weather and lighting conditions (ITE, 1991).
Opinions differ on the reliability of inductive loop systems. Some agencies believe that inductive loop technology is the best available, while others have experienced high failure rates (TTI, 1992). Studies on inductive loops revealed that several installation processes needed revision to improve the inductive loop detectors' reliability. Improper saw-cutting techniques, loop-wire splicing, and inadequate loop-sealant bonding resulted in loop wire breakage (Labell and May, 1990).
Given the widespread use of inductive loops throughout the United States, it is logical to fully utilize their capabilities and even to further enhance these capabilities. Inductive loops detect "presence" of vehicles. In its typical use, the inductive loop is basically an on-off device, or a contact closure, indicating that a vehicle is either present or not. In conjunction with its companion electronics, a single loop can provide vehicle counts and occupancies, whereas dual loops (often referred to as "traps") can provide speeds and vehicle classification (by length). However, other useful information is available from inductive loops by adding the appropriate hardware and software. These new concepts need to be considered because they add a new dimension to a state or local agency's capabilities in traffic monitoring.
The previous two sub-sections discussed traffic-detection equipment. Another component of traffic detection relates to the classifiers used to translate axle-presence detection to vehicle volumes and classes. There are many different classifiers in the market today that use the spacing between axle hits to determine classification based on previously determined class tables.
The Georgia Tech Research Institute and Georgia DOT performed a series of field tests on several vehicle classification devices that are currently used in order to determine accuracy and adequacy of the equipment. The field test location was on IH-20 in the metropolitan Atlanta area, and the test included two 48-hour tests for detailed vehicle-by-vehicle analysis and one seven-day test for longer term accuracy statistics (Harvey and Champion, 1996).
Published results were in a format that provided anonymity to participating companies and to specific equipment to avoid the appearance of competitiveness (Harvey and Champion, 1996). Documentation of results compared actual vehicle classification to system classification and the overall classification accuracy. The analysis of results found that the most common classification errors involved the differentiation of class 2 (Passenger Cars) and class 3 (Other Two-Axle, Four-Tire, Single Unit Vehicles) vehicles by test equipment. The results also found that the most accurately classified vehicles were large trucks, which comprise classes 8 through 12. The test team also found that there is a strong correlation between the accuracy of a classifier and the reliability of the axle sensor used to collect the data, and that axle-sensor error accounts for a large number of the overall classification errors. The increased accuracy regarding trucks is attributed to the distinct separation in the number and spacing of truck axles (Harvey and Champion, 1996).
Virginia DOT uses the following equipment and strategies:
A magnetometer typically consists of an intrusive sensor about the size and shape of a small can, a lead-in cable, and an amplifier. The cylinder portion of the magnetometer contains sensor coils that operate similarly to inductive loops. These coils are installed in a small circular hole in the center of each lane and communicate with the roadside by wires or radio link. Magnetometers function by detecting increased density of vertical flux lines of the earth's magnetic field caused by the passage of a mass of ferrous metals, such as a motorized vehicle. They operate in either presence or pulse modes and are embedded in the pavement. Magnetometers require less cutting of the pavement than inductive loop sensors, are easier to install, and can be installed underneath bridge decks without damage to the deck. The disadvantages of magnetometers are similar to those of inductive loop detector systems, in that they sometimes double count trucks and are less likely to detect motorcycles due to the vehicle's small detection zone (Labell and May, 1990).
Illinois DOT has had great success in using Numetric Hi-Star sensors. These sensors use Vehicle Magnetic Imaging (VMI) technology and are capable of the volume, speed, and length classification of vehicles plus road surface temperature, wet/dry surface condition, and roadway occupancy. IDOT finds these sensors easy to install and found them to be excellent for traffic volume data for highways carrying less than 75,000 AADT. While high-volume routes exist in Illinois (especially in the Chicago area), IDOT does not use these sensors in such locations but gets the data from the Chicago Area Transportation Study (CATS). This equipment also performs well for length based classification which Illinois is a big proponent of.
The 3M system consisted of three components: Canoga Model 702 Non-Invasive microloop probes, Canoga C800 series vehicle detectors, and 3M ITS Link Suite application software. The microloop probes can monitor traffic from a three-inch non-metallic conduit 18 to 36 inches below the road surface or from underneath a bridge structure. Installers must use a magnetometer underneath bridges to determine proper placement of the probes; otherwise, optimum performance requires trial-and-error. Probes installed in a "lead" and "lag" configuration under pavements or bridges can monitor speeds by creating speed traps in each lane. One of the requirements of this system is that the probes remain relatively vertical, so keeping the horizontal bores straight is critical. Probes placed in a non-vertical orientation can lead to speed errors. MnDOT tests under pavement indicated excellent volume and speed results. The absolute percent volume difference between sensor and baseline was under 2.5 percent, which is within the accuracy capability of the baseline loop system. For speeds, the test system generated 24-hour test data with absolute percent difference of average speed between baseline and test system from 1.4 to 4.8 percent for all three lanes (Minnesota DOT, 2002).
At a relatively low-to-moderate volume site in College Station, Texas, TTI found that, for a six-day count period, 3M microloops were almost always within 5 percent of baseline counts. In the right lane, all except two 15-minute intervals out of the 330 total intervals were within 5 percent of baseline counts. The remaining two were within 10 percent of baseline counts. Therefore, microloop counts were within 5 percent of baseline counts 99.4 percent of the time in the right lane (dual probes). In the left lane (single probes), 94.5 percent of the 15-minute intervals were within 5 percent, 4.5 percent were between 5 and 10 percent, and 1.0 percent were more than 10 percent from the baseline (Middleton and Parker, 2000).
NYSDOT tested 3M Microloops for bridge deck applications. NYSDOT also tested SAS-1 Acoustic sensors due to their advantages of low-power requirements and low cost. The main advantage stated by New York is the safety of traffic personnel.
A number of non-intrusive technologies also can be used for counting traffic volumes and for classifying vehicles. The use of non-intrusive data collection equipment for traffic data collection has been investigated by various states. While some of the states are experimenting and testing some types of non-intrusive equipment, other states are now beginning to review that option. This category of vehicle detectors includes active and passive infrared sensing systems, passive acoustic detectors, ultrasonic detectors, microwave and radar detection systems, automatic vehicle identification systems, and video detection systems. Some of the potential advantages of non-intrusive devices include ease of repair and ability to do so off the roadway. Several potential disadvantages were identified, including:
Illinois DOT is a strong proponent of length-based classification and has worked with FHWA to report length-based classification for HPMS. The use of length-based classifications encourages the use of non-intrusive detectors. Often the inability of such devices to classify vehicles into 13 vehicle categories is mentioned as a major impediment to their increased use.
The following paragraphs describe each of these systems and discuss advantages and disadvantages of system equipment.
Active infrared sensors operate by focusing a narrow beam of energy and either measuring the reflected energy or measuring the direct energy disruption by an infrared-sensitive cell. In the first case, one device both sends and receives energy, and interprets the reflected pattern. In the second, energy disruption represents vehicle presence so that detections occur when vehicles pass through the beam and interrupt the signal. The infrared beam can be transmitted from overhead or from one side of the road to the other. Infrared systems can provide information on vehicle height, width, and length, in addition to simple passage of vehicles.
Preliminary testing of active infrared detectors by public agencies indicates very promising results for monitoring vehicle speeds and classifications. TTI tested the Autosense II by Schwartz Electro-Optics (SEO) and found it to operate during day/night transitions and other lighting conditions without significant problems. However, its cost of $10,000 per lane may be a deterrent to its use. A second disadvantage of this sensor as compared to most other non-intrusive sensors is the requirement to be placed directly over each lane. This requires lane closure to install and remove the sensor element. Advantages include its ease of setup and generation of data protocols for interpreting its output. Also, it was more accurate in its classification accuracy (based on vehicle dimensions) than another non-intrusive sensors tested (Middleton et al. 1997). Based on information from others, weather conditions that appear to be problematic for this device are heavy fog, heavy dust, and heavy rain. England uses infrared detectors extensively for both pedestrian crosswalks and signal control. The San Francisco-Oakland Bay Bridge uses infrared detection systems to detect presence of vehicles across all five lanes of the upper deck of the bridge (ITE, 1991).
In contrast to the SEO ASII, which monitors and measures vehicle dimensions, the Autosense IIA counts axles. Installation of the IIA is above and to the side of each lane being monitored so that its field of scan includes a side view of the vehicle and its axles. Early testing by the vendor in November 1998 and during the first quarter of 1999 indicates axle-counting accuracy of 95 percent. The manufacturer anticipates further refinement of system algorithms based on "real world" data and improvement of classification accuracy to the design goal of 99.5 percent. The design used by SEO for this detector allows its firmware to execute the axle-counting algorithm without a dedicated computer to perform post-processing. Vehicle classification and axle count are reported within 25 milliseconds of vehicle passage. The release date for the Autosense IIA to be available to the general public was scheduled for April 1999. The Autosense IIA is the only non-intrusive detector identified by the authors that can classify according to the standard FHWA classification scheme using number of axles and axle spacings.
As noted earlier, ODOT uses EIS RTMS units in five locations to collect traffic volume data. ODOT also owns four Off Road Axle Detection Sensors (ORADS) developed and constructed as part of a research project. In addition, ODOT provided funding for an Ohio University research project on Improved Work Zone Design Guidelines. As part of this study, they will be purchasing 16 mobile trailer units equipped with non-intrusive sensors. ODOT will receive these units once the study is complete. ODOT also has tested video (Autoscope) and acoustic. ODOT feels that the main disadvantages are no classification information and difficult set-up.
Virginia DOT is actively researching several non-intrusive technology devices. To date, only the RTMS sidefire radar has been approved for use. It can be used as a portable detector and has the required accuracy needed. Virginia DOT has reviewed other non-intrusive products, but none has met their current needs. For example,
Caltrans tested RTMS extensively but did not obtain favorable results, including long set-up times and occlusion problems. However, Caltrans recognizes that these technologies have improved since and has developed guidelines/requirements for non-intrusive detectors. The draft guidelines are intended to help California personnel make educated estimates of whether microwave sensors can fulfill their requirements. The document contains checklists of requirements that must be met, test results of various microwave models, technology descriptions, and installation overviews.
VDOT uses a portable customized side fire RTMS device for high-volume freeway. The device needs some training to set up and calibrate but works well for volume counts. TTI tested the accuracy of RTMS at a site on the I-35 in Texas. This site does have stop-and-go traffic sometimes during the peak periods so it provides a good test for non-intrusive sensors. It was noted that the RTMS has to be located a minimum of about 18-ft from the nearest traffic lane to be effective. Detectors located less than 6-ft from the nearest lane did not yield reasonable results for that lane. The results indicate that, RTMS accuracy ranges 0 to 5 percent and that occlusion reduces accuracy (both counts and speeds). Also, slow speeds compromise RTMS accuracy. With regards to setup time, it was observed that it takes about an hour per lane even with trained personnel.
The SmarTek SAS-1 is a passive acoustic detector that monitors vehicular noise (primarily tire noise) as vehicles pass the detection area. The detector can monitor as many as five lanes and the SAS-1 must be oriented in a sidefire position. Precise alignment is not critical because the sensor can cover a wide area. Heights recommended by the vendor range from 25 feet to 40 feet, and the recommended offset range is 10 feet to 20 feet. Higher mounting positions can reduce the effects of occlusion in multiple lane applications.
TTI research found that the SAS-1 predominantly undercounted in both peak and off-peak conditions. The SAS-1 speed estimates were within 5 to 10 mph of baseline during some peak periods but as much as 20 to 25 mph different in others. Free-flow speed estimates were usually within 5 mph of baseline speeds (Middleton and Parker, 2002). TTI has not tested the accuracy of the SAS-1 vehicle classification algorithm.
The Traffic Monitoring Unit of the New York State Department of Transportation has successfully developed a permanent acoustic traffic monitoring site. This type of site was developed in-house by NYSDOT personnel to support non-intrusive sensor technology with applications in data collection and ITS activities. The conceptual priority for use of this type of site was installation on facilities where the cost of in-pavement sensors was not justified due to roadway and traffic conditions that greatly limited sensor service life. Use of this type of site greatly reduces data collection costs, but still meets the needs of the Department. Each site consists of a Smartek SAS-1 acoustic sensor mounted on an existing light pole or sign structure at a height of 30 to 40 feet, structure dependant. A small cabinet mounted at the base houses Smartek electronic and communication interfaces as well as power management electronics. The platform is supported by a 12 volt electrical system with one 50 watt Kyocera solar module charging two 75 Ah deep-cycle batteries to supply power. A Trafinfo.com Trafmate digital pager is used to download archived data via telemetry.
In addition to using the acoustic sensors as permanent stations, NYSDOT also has four mobile platforms equipped with the sensor for portable counts including coverage counts, special counts and some ITS design applications. Four Mobile Traffic Monitoring Platforms have been built to date. Each is used to collect volume data on high-speed, high-volume multi-lane facilities where safety concerns or equipment limitations prevent use of typical collection methods. Each platform supports a Smartek SAS-1 acoustic sensor extended on a 35-foot telescoping mast. The platform weighs approximately 1000 pounds, is easily transportable, and can be erected outside the traveled way and operational in approximately 30 minutes. The cost of each platform fully outfitted with solar power, deep-cycle batteries, a telescoping 35-foot mast, acoustic sensor, and supporting electronics is approximately $7,000. A somewhat similar commercial version of the platform is available for approximately $28,000. However, that setup uses a different type of sensor with a high power consumption rate. It requires generator-supplied power and has no communications capability. The in-house research, development, and construction of this project represent an initial cost savings to NYSDOT of approximately $21,000 for each platform. The anticipated life span of the clean, maintenance-free, solar cell-charged deep-cycle batteries is five years with no additional fuel costs. The batteries are recycled at the end of their useful life. The average cost of construction of one three-to-six-lane count site with loop sensors that is typically used for only a few weeks during the life span of the loops is approximately $30,000. Each count taken utilizing the platform at each location will save the Department $30,000 each time. Assuming two trailers will be used to take a minimum of ten scheduled counts each year on facilities with three or more lanes, the benefit cost ratio for such a device was estimated to be 21:4.
A video image detection (VID) system consists of one or more cameras providing a clear view of the area, a microprocessor-based system to process the video image, and a module to interpret the processed images. Advanced VID systems can collect, analyze, and record traditional traffic data; detect and verify incidents; classify vehicles by length; and monitor intersections. The ability of VID systems to classify vehicles is generally limited to daylight hours unless street lighting is bright enough for the VID's daytime algorithm. Their nighttime detection algorithms depend on detection of headlights, and the systems cannot distinguish between the various headlights of individual vehicle classes. It should also be noted that video systems on the market today provide only three to five vehicle length classifications. Therefore, these systems cannot be used to classify by axles as required by the FHWA classification scheme unless approved by FHWA. The most recent Texas Transportation Institute (TTI) tests indicate some very promising features of one VID system, the Autoscope Solo Pro, but its classification accuracy was not included in the tests.
While there have been rapid advances in vehicle detection technology, inductive loops and piezo electric sensors are considered by states as the most efficient way to collect traffic data. Improvements in loop installations and vehicle counters have greatly reduced the problems associated with inductive loops. Advanced vehicle counters with loop signatures-based detection and classifications promise to build upon the improvements. However, the use of loops continues to be cumbersome due to its inherent requirements such as pavement cutting, traffic control and lane closures, and maintenance problems. Pneumatic tubes are the preferred technology for short-term counts.
Non-intrusive detectors provide an alternative to minimize or eliminate some of the safety and maintenance issues with loops and tubes. These technologies include infrared-, acoustic-, microwave-, and video-based sensors. Various tests have shown that these sensors currently meet requirements as far as volume monitoring is concerned but fall short on classification of vehicles.
|<< Previous||Contents||Next >>|