Traffic Monitoring Guide

Chapter 2. TRAFFIC DATA COLLECTION TECHNOLOGY AND EQUIPMENT

2.1 MONITORING TECHNOLOGY AND EQUIPMENT

The following sections include commonly used technologies for vehicle and micromobility detection. Additional information is available in the FHWA's Traffic Detector Handbook (FHWA 2006) and in Chapter 3 of the AASHTO report Guidelines for Traffic Data Programs (AASHTO 2009). The most up-to-date information on specific equipment is available from the manufacturer. The FHWA WIM Pocket Guide (FHWA 2018) and FHWA's Travel Monitoring and Traffic Volume website also contains a wealth of information.

2.1.1 Types of Traffic Data Collected

Table 2-1 lists typical types of traffic data collected by traffic monitoring devices for motorized and micromobility data collection programs. See Chapter 4 for a detailed listing of data types that could be provided by traffic monitoring devices.

**Table 2-1. Basic Data Types Reported by Traffic Monitoring Devices**
Motorized Vehicles	Micromobility
Individual vehicle records Vehicle volumes Vehicle classification Vehicle length Vehicle speed Axle spacing Axle weights (GVW) Gap Headway Lane occupancy	Pedestrian volumes Micromobility device volumes Micromobility device classification

2.1.2 Detection Methods

There are two general methods for detecting passing vehicles and micromobility users: automatic and manual.

Automatic methods – refer to the collection of vehicular and pedestrian data using automatic equipment. The equipment is designed to continuously record the presence, distribution, and variation of traffic flow. The data are recorded and reported for each individual detected observation or summarized (binned) by discrete time periods (e.g., by 5 min., 15 min., hour of the day, daily counts). Automatic methods include permanently installed or portable (typically, for less than 7 days) equipment. Automated methods are further classified as intrusive (i.e., the traffic detection sensor is placed on or under the road surface) or non-intrusive (i.e., there is no sensor on or under the road).

Manual methods – refer to those methods that involve a human observer determining and recording the numbers of vehicles and micromobility users. The total counts are reported by time interval, or the observed objects are sorted by a human observer and reported by vehicle classification, vehicle occupancy, turning movement counts, etc. While these methods have limited application due to their limited use and high cost, they serve an important function for verifying automated vehicle detection equipment performance and accuracy.

2.1.3 Equipment Types

The common names for traffic monitoring equipment are traffic counters and vehicle detectors. The equipment for specialized traffic data collection has additional names, including Continuous Vehicle Classifiers (CVC), Weigh-In-Motion (WIM) systems, and micromobility detectors. These devices can be portable and referred to as

Portable Traffic Recorder (PTR) or permanently installed at Continuous Count Stations (CCSs). The TMG Glossary contains the detailed definitions for the above-mentioned equipment types.

2.1.4 Equipment Components

Although vehicle detectors sometimes differ in function and application, based on whether permanent or portable, these systems have common major components. These systems consist of (1) sensor(s) installed in, under, alongside, or above the roadway; (2) an electronic data collection device (controller), typically stored in a roadside cabinet that processes the signals from the sensor; (3) a power source supporting data collection device;

(4) telecommunication equipment; and (5) additional support infrastructure such as wires, pull boxes, conduits, etc. A complete traffic monitoring site layout is shown in Figure 2-1. The equipment shown is a typical configuration used to collect vehicle count, vehicle classification, vehicle length, axle spacing, axle and gross vehicle weight, and vehicle speed.

FHWA recommends using 8 feet wide by 6 feet long (along the wheel path) loops and full lane width pneumatic road tubes or axle sensors for lanes wider than 10 feet.

Source: Federal Highway Administration.

Figure 2-1. Example of Traffic Monitoring Equipment and Site Layout

The technology used to sense the passing traffic stream determines what each data collection device physically counts. The signals collected from the sensors are converted and processed by the controller (typically using a proprietary method specific to that equipment's vendor) into vehicular and traffic flow information. This information is stored in the data storage unit of the traffic controller device for later download. Figure 2-2 shows an example of the signals collected by the axle sensors and the inductive loop being interpreted into vehicle classification data.

Source: Virginia Department of Transportation

Figure 2-2. Example of Sensor Output Conversion to Vehicle Classification Data

2.1.5 Sensor Types Overview

The theory and operation of vehicle sensors is discussed in detail in the sensor technology chapter of the Traffic Detector Handbook (FHWA 2006). The following types of vehicle presence technologies are used for detecting moving vehicles and/or Micromobility users:

Inductive Loop Detector – detects vehicle or micromobility device passage and presence. This sensor uses a low voltage alternating electrical current through a formed wire coil embedded in the pavement. The wire coil loop sometimes is placed in the aggregate base below the pavement. The current creates lines of flux field above the formed wire coil, which is disturbed by a passing conductive object of detecting sufficient conductive material (e.g., car, truck, and bike). If this disruption meets predetermined threshold criteria, detection occurs, and the object is counted by a data logger or computer. Advanced signal processing (such as recording and interpretation of loop signatures) is used to derive certain vehicle class characteristics. The sensor consists of four parts, namely 3 to 4 turns (depending on the loop size) of wire embedded in the pavement, a lead-in wire running from the wire loop in the pavement to the pull box, and a lead-in cable spliced to the lead-in wire at the pull box, which connects to the inductive loop detector electronic circuit on a card or device within the equipment cabinet or traffic counter.
Magnetic Sensor (Induction or Search Coil Magnetometer) – detects the presence of a ferrous metal object through the perturbation (known as a magnetic anomaly) it causes in the Earth's magnetic field. It is placed under or in the roadway to detect the passage of a vehicle over the sensor. Its output is connected to an electronics unit. The two types of magnetic sensors are fluxgate magnetometers and induction magnetometers, referred to as magnetic detectors as described in the Traffic Detector Handbook.
Microwave Sensor – transmits electromagnetic energy from an antenna toward vehicles traveling the roadway. When a vehicle passes through the antenna beam, a portion of the transmitted energy is reflected towards the antenna. The energy then enters a receiver where the detection is made and traffic flow data, such as volume count, speed, and vehicle length, are calculated. Microwave sensors that utilize the Doppler principle analyze the frequency of the received signal. The frequency is decreased by a vehicle moving away from the radar and increased by a vehicle moving toward the radar. Vehicle passage or count is denoted by the presence of the frequency shift. Vehicle presence cannot be measured with the constant frequency waveform since only moving vehicles are detected.
Passive Infrared Sensor – detects energy from two sources: 1) energy emitted from vehicles, road surfaces, and other objects in the field of view; and 2) energy emitted by the atmosphere and reflected by vehicles, road surfaces, or other objects into the sensor aperture. The energy captured by passive infrared sensors is focused by an optical system on an infrared-sensitive material mounted at the focal plane of the optics. The infrared sensitive element converts the reflected and emitted energy into electrical signals. Real-time signal processing is used to analyze the signals for the presence of a vehicle. The sensors are mounted overhead or in a side-looking configuration to view approaching or departing traffic. Infrared sensors are used for signal control; volume, speed, and class measurement; detection of pedestrians in crosswalks; and transmission of traffic information to motorists.
Passive Acoustic Array Sensors – measures vehicle passage, presence, and speed by detecting acoustic energy or audible sounds produced by vehicular traffic from a variety of sources within each vehicle and from the interaction of a vehicle's tires with the road. When a vehicle passes through the detection zone, an increase in sound energy is recognized by the signal-processing algorithm and a vehicle presence signal is generated. When the vehicle leaves the detection zone, the sound energy level drops below the detection threshold, and the vehicle presence signal is terminated.
Ultrasonic Sensor – transmits pressure waves of sound energy at a frequency between 25 and 50 kHz, which is above the human audible range. Most ultrasonic sensors operate with pulse waveforms and provide vehicle count, presence, and occupancy information.
Laser Radar Sensor – transmit energy in the near infrared spectrum. Models are available that scan infrared beams over multiple lanes or use multiple laser diode sources to emit a number of fixed beams that cover the desired lane width. Laser radars provide vehicle presence at traffic signals, volume, speed, length assessment, queue measurement, and classification.
Video Detection System – typically consists of one or more cameras, a microprocessor-based computer for digitizing and analyzing the imagery, and software for interpreting the images and converting them into traffic flow data.
Pneumatic Tube – uses an air switch connected to the end of the rubber road tubes to detect short burst(s) of air from a vehicle that rolls over the tube. The data logger then uses pre-defined criteria (e.g., axle spacing, speed) to determine whether a valid vehicle type has passed over the tubes. This detector is widely used for portable traffic monitoring. When two tubes are placed parallel on the roadway, the signal results provide information sufficient to classify vehicles using detected axle spacing under free flow traffic conditions.
Piezoelectric Sensor – detects a vehicle axle's passage using a change in the sensor's voltage that is directly proportional to the pressure applied by the vehicle wheel on the sensor. Piezoelectric sensors are used independently to detect and classify vehicles or in conjunction with loop inductors to increase classification accuracy by providing overall (loop) length. Piezoelectric sensors, in combination with passive infrared technologies, are capable of detecting bicycles in mixed pedestrian and bicycle traffic. Piezo sensors are also used for WIM sensing.
Fiberoptic Sensor – detects changes of a tightened optical fiber when a vehicle passes over the sensor embedded in the road surface. This sensor is capable of vehicle detection, classification, and weighing.
Bending Plate for WIM – utilizes strain gauges bonded to the underside of the plate to collect loading data. As vehicle axles pass over the bending plate, the system measures the strain on the plate at highway speeds and calculates the load required to induce that level of strain.
Load Cell for WIM – utilizes two scales to detect an axle and to weigh both the right and left side of the axle at highway speeds per axle passage. The load cell sensor utilizes transducer(s), which creates an electrical signal whose amplitude is directly proportional to the force being measured. The system records the weights measured by each scale and sums them to obtain the axle weight.
Strain Gauge Strip Sensor for WIM – uses strain gauge load cell technology. As a vehicle passes over the WIM sensor, the system measures the vertical strain placed on the sensor by the weight of the wheel. The resultant change in the electronic properties of the strain gauge load cells is translated into the dynamic load that is used by the software to estimate wheel, axle, and vehicle weight.
Bridge WIM (B-WIM) – uses strain transducers mounted to the underside of culverts, bridge decks, or bridge structural members. These deformations of bridge structural members detected by the strain sensors are analyzed to measure the axle loads of passing traffic.
Seismic sensors – operates by detecting the passage of energy waves through the ground caused by feet, bicycle tires, or wheels on micromobility devices. These sensors are most common on unpaved trails or paths.
Pressure plate – detects vehicles, micromobility devices, or pedestrians by detecting changes in force (i.e., weight), much like an electronic bathroom scale. These sensors are most common on unpaved trails or paths.

2.1.6 Sensor Technology Combinations

Many State highway agencies have adopted a policy to "collect data once and use it multiple times." This approach leads to traffic monitoring site designs that combine multiple sensor technologies to achieve more comprehensive traffic data collection satisfying needs of multiple customers/data users.

Various types of sensors are used in combination, for traffic management, law enforcement, and other applications. Examples of sensor technology combinations include ultrasonic-infrared-microwave combinations, inductive loops enhanced with magnetic signature detector card, WIM sensor and inductive loop combinations, and WIM sensors combined with license plate and DOT registration number video readers. A combination of inductive loops with axle sensors or magnetic signature detector card improves quality of vehicle classification.

An example of sensor technology combinations that have been widely used over the years is a combination of inductive loops and axle sensors. That sensor combination provides opportunities to collect the following data at the same site:

Individual vehicle records
Vehicle volumes
Vehicle classification (axle based)
Vehicle length
Vehicle speed
Axle spacing
Axle weights
Gap
Headway
Lane occupancy

When the loop and axle sensor combination is coupled with the roadside camera, they can collect even more information about each passing vehicle, including body type (associated with carrying specifics commodities), license plate, and U.S. DOT vehicle registration number (for commercial trucks).

A combination of WIM and video monitoring equipment allows collection of the extended dataset, providing additional information of interest about heavy vehicles, including license plate, U.S. DOT registration information, body type, cargo trailer configuration, axle arrangement details (such as lifted axles).

Other examples of technological combinations, common for detecting and classifying micromobility, include the following combinations:

Video and infrared sensors
Video and pneumatic road tubes
Infrared sensors and inductance loop
Infrared sensors and piezo axle sensors

2.1.7 Strengths And Weaknesses Of Commercially Available Sensor Technologies

This section describes the kinds of technologies that are available to support traffic monitoring programs, including the general strengths and weaknesses of each of those technologies. An additional source of information on the selection of traffic monitoring equipment is Chapter 3 of the report AASHTO Guidelines for Traffic Data Programs (AASHTO 2009).

Table 2-2 describes the strengths and weaknesses of the types of technology used for motorized vehicle presence detection. Presence detection refers to the ability of a vehicle detector to sense that a vehicle, whether moving or stopped, has appeared in its zone of detection. Table 2-3 describes the strengths and weaknesses of the sensors used for weighing and classifying vehicles in motion.

Table 2-4 describes the strengths and weaknesses of the sensors used for micromobility detection. In addition to sensor capabilities consideration, it is important to consider that even for the selected technology, data accuracy varies significantly based on sensor configuration, site location, pavement condition, installation, calibration, and maintenance practices.

**Table 2-2. Strengths and Weaknesses of Commercially Available Sensor Technologies for Motorized Vehicle Detection and Counting**
Technology	Strengths	Weaknesses
Inductive Loop	Flexible design to satisfy large variety of applications Mature, well-understood technology, long lasting at 20+ years Large experience base Provides basic traffic parameters (e.g., volume, presence, occupancy, speed, headway, and gap) Insensitive to inclement weather such as rain, fog, and snow Provides best accuracy for count data as compared with other commonly used techniques Common standard for obtaining accurate occupancy measurements	Installation requires pavement cut Improper installation decreases pavement life Installation and maintenance require lane closure Wire loops subject to stresses of traffic Multiple loops usually required to monitor a location Detection accuracy decreases when design requires detection of a large variety of vehicle classes (unless loop signature technology is deployed) Does not detect axles in commonly used configurations
Piezo Sensors	High accuracy in vehicle classification Insensitive to inclement weather such as rain, fog, and snow Common standard for obtaining axle count and classification Used to collect weight classification data (WIM); quartz performance comparable to bending plates Mature, well-understood technology	Installation requires pavement cut Improper installation decreases pavement life Installation and maintenance require lane closure Polymer piezo is sensitive to temperature Does not detect vehicle overall length Does not work well in slow or stopped traffic
Pneumatic Tube	Common standard for obtaining axle count and classification in portable applications Mature, well-understood technology Low cost and no damage to the pavement	Installation sometimes requires lane closure Does not detect vehicle overall length Does not work well in very high volume or very slow (5 mph) or stopped traffic
Magnetic (Two-axis Fluxgate Magnetometer, Induction or Search Coil Magnetometer)	Less susceptible than loops to the stresses of traffic Insensitive to inclement weather such as snow, rain, and fog Some models transmit data over wireless radio frequency (RF) link Used where loops are not feasible (e.g., bridge decks) Some models are installed under roadway without need for pavement cuts; however, boring under roadway is required	Installation requires pavement cutting or boring under roadway Improper installation decreases pavement life Installation and maintenance require lane closure Models with small detection zones require multiple units for full lane detection Cannot detect or classify stopped vehicles or axles unless special sensor layouts and signal processing software are used
Microwave	Typically, insensitive to inclement weather at the relatively short ranges encountered in traffic management applications Direct measurement of speed Multiple lane operation available Detects slow-moving vehicles Accommodates changes in lane assignment at a location Non-intrusive installation	Sometimes miss occasional vehicles traveling side-by-side (occlusion) Calibration and sensor position are crucial to proper operation Cannot detect stopped vehicles or individual axles
Active Infrared (Laser Radar)	Transmits multiple beams for accurate measurement of vehicle position, speed, and class Multiple lane operation available Good motorcycle detection Non-intrusive installation Installed on one side or both sides of the roadway depending on the system used	Operation is affected by fog when visibility is less than ≈20 feet (6 meters) or blowing snow is present Installation and maintenance, including periodic lens cleaning, require lane closure (should not require a lane closure for cleaning and maintenance of a side-fired laser) Side-fire axle detection will not work with roads that have a substantial crown or median obstructions
Passive Infrared	Multizone passive sensors measure speed Good motorcycle detection Non-intrusive installation	Passive sensor has reduced vehicle sensitivity in heavy rain, snow, and dense fog Some models not recommended for presence detection No accurate vehicle length or axle detection (requires periodic lens cleaning)
Ultrasonic	Multiple lane operation available Capable of over-height vehicle detection	Environmental conditions such as temperature change and extreme air turbulence often affect performance; temperature compensation is built into some models Large pulse repetition periods degrade occupancy measurement on freeways with vehicles traveling at moderate to high speeds Cannot detect stopped vehicles or individual axles
Acoustic	Passive detection Insensitive to precipitation Multiple lane operation available in some models	Cold temperatures affect vehicle count accuracy Specific models are not recommended with slow-moving vehicles in stop-and-go traffic Cannot detect stopped vehicles or individual axles
Video Detection System	Monitors multiple lanes and multiple detection zones/lane Easy to add and modify detection zones Rich array of data available Generally, cost effective when many detection zones within the camera field of view or specialized data are required Non-intrusive installation	Installation and maintenance, including periodic lens cleaning, require lane closure when camera is mounted over roadway (lane closure is not required when camera is mounted at side of roadway) Performance affected by inclement weather such as fog, rain, and snow; vehicle shadows; vehicle projection into adjacent lanes; occlusion; day-to-night transition; vehicle/road contrast; and water, salt grime, icicles, and cobwebs on camera lens Reliable nighttime signal actuation requires illumination Requires 30- to 50-ft (9- to 15-m) camera mounting height (in a side-mounting configuration) for optimum presence detection and speed measurement Cannot detect axles Some models susceptible to camera motion caused by strong winds or vibration of camera mounting structure Complicated or expensive data processing required

Source: Adapted from Traffic Detector Handbook, 2006.

**Table 2-3. Strengths and Weaknesses of WIM Technologies for Moving Motorized Vehicle Detection, Counting, Weighing, and Classifying**
Sensor Type	Strengths	Weaknesses
Load Cell	Higher accuracy Provides count, classification, speed, weight, axle spacing, individual vehicle records data Long service life	Use in concrete only High cost: total life-cycle cost is more efficient over long data collection periods High maintenance Installation requires more time, training, and resources
Bending Plate	Higher accuracy Provides count, classification, speed, weight, axle spacing, individual vehicle records data Long service life	Use in concrete only High to Moderate cost; total life-cycle cost is more efficient over long data collection periods Must be regularly maintained to prevent getting loose and causing safety hazard
Quartz Piezo	High accuracy, low maintenance Provides count, classification, speed, weight, axle spacing, individual vehicle records data	Moderate cost
Strain Gauge Strip Sensor	High accuracy, low maintenance Provides count, classification, speed, weight, axle spacing, individual vehicle records data	Limited long-term performance record Implemented with limited number of controllers
Permanent Polymer Piezo (including co-ax)	Low cost, low maintenance Provides count, classification, speed, weight, axle spacing, individual vehicle records data	Temperature sensitive Needs seasonal calibration
Portable Polymer Piezo	Low initial cost, easy to set up, portable Provides count, classification, speed, weight, axle spacing, individual vehicle records data	Low accuracy Temperature sensitivity, needs local calibration
Bridge WIM (Strain Gauge)	Quick to set up, provides portable count, classification, weight, axle spacing, individual vehicle records data	Depending on local bridge structural response, needs local calibration Safety risk & special equipment for tall bridges Accuracy depends on accurate bridge response modeling

Source: Adapted from FHWA Weigh-In-Motion Pocket Guide, 2018.

**Table 2-4. Strengths, Weaknesses, and Applications of Micromobility Detection Technologies**
Technology	Typical Applications	Strengths	Weaknesses
Inductance Loop	Permanent counts Bicyclists and some micromobility device users (e-bikes)	Accurate when properly installed and configured Uses traditional motor vehicle counting technology	Capable of counting bicyclists only Requires saw cuts in existing pavement or pre-formed loops in new pavement construction Often have higher error with groups
Pressure Plate Sensor/Pressure Mats	Permanent counts Typically on unpaved trails or paths Bicyclists, micromobility device users, and pedestrians combined	Some equipment able to distinguish bicyclists and pedestrians	Expensive/disruptive for installation under asphalt or concrete pavement
Seismic Sensor	Short-term counts on unpaved trails Bicyclists, micromobility device users, and pedestrians combined	Equipment is hidden from view	Commercially available, off-the-shelf products for counting are limited
Radar Sensor	Short-term or permanent counts Bicyclists, micromobility device users, and pedestrians combined	Capable of counting bicyclists in dedicated bike lanes or bikeways	Commercially available, off-the-shelf products for counting are limited
Video Imaging (Lidar) – Automated	Short-term or permanent counts Bicyclists, micromobility device users, and pedestrians separately	Potential accuracy in dense, high-traffic areas	Typically, more expensive for exclusive installations Algorithm development still maturing
Infrared – Active	Short-term or permanent counts of micromobility device users and pedestrians combined	Relatively portable Low profile, unobtrusive appearance	Cannot distinguish between bicyclists and pedestrians unless combined with another bicycle detection technology Very difficult to use for bike lanes and shared lanes Often have higher error with groups
Infrared – Passive	Short-term or permanent counts micromobility device users and pedestrians combined	Very portable with easy setup Low profile, unobtrusive appearance	Cannot distinguish between bicyclists and pedestrians unless combined with another bicycle detector Difficult to use for bike lanes and shared lanes, requires careful site selection and configuration Have higher error when ambient air temperature approaches body temperature range Often have higher error with groups Direct sunlight on sensor may create false counts
Pneumatic Tube	Short-term counts of 2-axle micromobility device users	Relatively portable, low-cost Possible to use existing motor vehicle counting technology and equipment	Capable of counting bicyclists only Tubes pose hazard to trail users Greater risk of vandalism Special care needed when installing and configuring for counting bikes in bike or shared lanes
Video Imaging – Manual Reduction	Short-term counts of micromobility devices and pedestrians separately	Cost is lower when existing video cameras are already installed	Limited to short-term use Manual video reduction is labor-intensive Weather and lighting reduce the accuracy Video image processing has the highest equipment costs
Manual Observer	Short-term counts of micromobility devices and pedestrians separately	Very portable Also used for automated equipment validation	Expensive and possibly inaccurate for longer duration counts

2.2 SELECTING MOTORIZED TRAFFIC MONITORING TECHNOLOGIES

2.2.1 Vehicle Detection Equipment Capabilities by Traffic Data Types

A good way to categorize traffic monitoring devices is based on the type of data they collect. Table 2-5 provides a comparison of sensor capabilities for motorized vehicle detection by key traffic attributes. The following sub-sections describe technologies for collecting the different types of traffic data.

**Table 2-5. Motorized Vehicle Detection Sensor Comparison**
Sensor Technology	Vehicle Count	Vehicle Presence	Speed	Vehicle Classificationⁱ	Weight	Multiple Lane, Multiple Detection Zone Data	Sensor Purchase Cost
Inductive Loop	X	X	X^a	X^b		X	Low^h
Magnetometer (2-Axis Fluxgate)	X	X	X^a				Low^h
Magnetic Induction Coil	X	X^c	X^a				Low to moderate^h
Microwave	X	X^d	X	X^d			Low to moderate
Active Infrared	X^g	X	X^e	X		X	Moderate
Passive Infrared	X^g	X	X^e				Low to moderate
Ultrasonic	X	X					Low to >moderate
Acoustic Array	X	X	X			X^f	Moderate
Video Detection System	X	X	X	X		X	Moderate
Laser	X	X				X	Moderate
Contact switches closures (e.g., pneumatic tubes)	X	X	X	X			Very low, only for short-term counts
Fiber optic	X	X	X	X	X	X	Moderate
Load Cell	X	X	X	X	X	X	Very high
Bending Plate	X	X	X	X	X	X	High
Quartz piezo	X	X	X	X	X	X	Moderate to high
Strain Gauge Strip Sensor	X	X	X	X	X	X	Moderate to high
Polymer Piezo	X	X	X	X	X	X	Low
Bridge WIM	X	X	X	X	X	X	Moderate

^a Measured using 2 sensors a known distance apart or estimated from 1 sensor, the effective detection zone, and vehicle length

^b With specialized electronics unit containing embedded firmware that classifies vehicles.

^c With special sensor layouts and signal processing software.

^d With microwave sensors that transmit the proper waveform and have appropriate signal processing.

^e With multi-detection zone passive or active mode infrared sensors.

^f With models that contain appropriate beam-forming and signal processing.

^g Axle counts converted to vehicle counts through factoring.

^h Includes underground sensor and local detector or receiver electronics. Electronics options are available to receive multiple sensors and multiple lane data.

i There are different types of classification schemes (axle-based, length-based and visual schema)

Source: Federal Highway Administration.

Individual Vehicle Records

An Individual Vehicle Record (IVR) contains information about each discrete passing vehicle, such as vehicle type (i.e., vehicle classification), number of axles, axle spacing, and vehicle length. In addition, some IVRs include information about vehicle speed, gross vehicle weight, axle and wheel loads, and magnetic loop signatures. This information is obtained by the equipment using both vehicle presence and axle detection technologies. The most commonly used sensor combinations are inductive loops and axle sensors (e.g., piezo electric, bending plates, load cells, and strip strain gauge sensors). In some cases, a pair of axle sensors used on their own is sufficient but the accuracy of IVR data collection degrades in congested traffic conditions, when the distance between vehicles converges on the distance between the axles of a large vehicles. The use of video cameras, in addition to inductive loops and axle sensors, allows for the collection of additional IVR information, such as vehicle license plate, and vehicle DOT registration numbers.

FHWA recommends collecting data in individual vehicle record (IVR) format whenever possible. For more details on the IVR format, see chapter 4. Considerations for data storage requirements must be made when purchasing equipment or evaluating older equipment because IVR data reporting requires large storage space.

Vehicle Volume

Some vehicle detection technologies count each passing object, where in most cases an object is a vehicle, whether it is a car or multi-unit truck. Others do not detect a vehicle but instead count the axles of those vehicles or actual sensor actuations. Additional information is then used to convert the axle count data into measures of vehicle volume. In many cases, this extra information comes from a second sensor. But for simple, single sensor, axle-based counters, an adjustment factor (the axle correction factor) is applied against the total axle count to provide an estimate of vehicle volume. Table 2-6 summarizes which of the currently available traffic monitoring technologies directly count vehicle volumes, and which count axles requiring conversion of those data to vehicle volume estimates.

This table shows the most common application of the technology. In some cases, specific implementations of the technologies are used in different ways. For example, advanced processing of magnetic signatures of the loop sensors is used to count axles very accurately and even classify vehicles. However, most loop installations are not capable of detecting axles.

Table 2-7 describes which vehicle counting technologies are capable of classifying vehicles.

**Table 2-6. Common Technologies Used for Counting Vehicles Versus Axles**
Presence-Sensing Technologies	Axle-Sensing Technologies
Inductive loops Magnetic Video detection system Acoustic Ultrasonic Microwave Laser radar Passive infrared	Infrared Laser (most) Polymer piezo Quartz piezo Strain gauge strip sensor Fiber optic Capacitance mats Bending plates Load cells Inductive loop signatures Contact switch closures (e.g., pneumatic tubes)

**Table 2-7. Common Technologies for Classifying Motorized Vehicles**
Technologies for Axle-Based Vehicle Classification	Technologies for Length-Based Vehicle Classification
Infrared (passive) Laser radar Polymer piezo Quartz piezo Strain gauge strip sensor Fiber optic Capacitance mats Bending plates Load cells Contact switch closures (e.g., pneumatic tubes) Specialized loop signature systems Any of the above combined with inductive loops	Dual inductive loops Inductive loops Magnetic (magnetometer) Video detection system Microwave

Speed

Most modern traffic monitoring technologies produce a measure of speed as part of their routine traffic monitoring function. Some technologies are particularly well suited for reporting individual vehicle speeds (that is tracking how fast each specific vehicle is moving), while others are designed to provide average facility speed over a given reporting interval. Although both data represent speed information, the usefulness of those data is very different. Traffic monitoring technologies that provide either axle-based vehicle classification or weigh in motion are also capable of providing vehicle speed data.

How the speed data are collected is as much a function of the equipment connected to the sensor as it is of the sensor technology itself. For example, the traditional method for estimating speeds when using a single inductive loop is to measure total sensor on time (lane occupancy) over a set period, along with the total number of vehicle observations during that period. By dividing the lane occupancy by the volume and multiplying by a constant that represents the average vehicle length for that location, average speed for that reporting period is computed and reported. However, more modern electronics often take the same basic single loop signal, and by analyzing that signal, directly calculate vehicle speed from the shape of the loop signature. Another approach to using loop technology is to place two loops in the lane at a known distance apart configured one after the other, thus forming a speed trap. When these loops are properly calibrated, the distance between the leading edge of the two detectors (d₁₂) divided by the difference in time it takes for the passing vehicle to activate the second loop after it activates the first loop (T₂ – T₁) yields the speed of the vehicle (d₁₂ / (T₂ – T₁)).

Speed data are collected by axle detectors using the same timing principle as described for inductive loops, based on the known distance between the axle sensors and the amount of time between the activation of the first axle sensor and the activation of the second axle sensor by a particular axle. Axle spacing is computed by multiplying the calculated speed by the time between hits for a particular axle on the first and second axle sensors. Therefore, since both speed and axle spacing measurements are related, if the detector is measuring axle spacings accurately, it is also measuring speed accurately. If a true axle spacing is known (e.g., obtained by manual measurement), the known true axle spacing value can be compared with the value estimated by the detector. The error in the axle spacing could be used to flag the error in speed measurement as well.

Figure 2-3 demonstrates how changes in speed and axle spacing can be used to identify the need for equipment calibration. Dashed horizontal lines show low and upper boundaries for a typical average tractor-tandem axle spacing for class 9 vehicles (i.e., spacing between 2^nd and 3^rd axles). Data points outside these boundaries signal that axle spacing measurements are out of calibration. The example shows that both the average monthly tractor-tandem axle spacing and the average monthly speed in are low in year 2015 and the first part of 2016, as compared to later months.

The key to collecting speed data is that the agency needs to understand both what use they need from the data, and the capabilities of their available equipment.

Speed data are also being obtained from other sources. One such source is vehicle probe data. Data collected from vehicle probes, within the overall roadway performance-monitoring program of an agency, should be loaded into the overall traffic monitoring database using the FHWA speed format. The FHWA speed format provides flexibility with a minimum of 15 speed bins to a maximum of 25 speed bins (all in 5 mph increments).

Axle and Gross Vehicle Weight

A specific subset of traffic monitoring devices is capable of weighing vehicles while they travel down the road. These devices are commonly referred to as weigh-in-motion (WIM) scales. The sensors used are designed to not only detect the presence of an axle, but to measure the force being applied by that axle during the duration of the time the axle is in contact with the axle sensor. Sophisticated analysis is then applied to the signal produced by each sensor to estimate the static weight of each passing axle. Weights for all axles associated with a given vehicle are then combined to estimate the gross vehicle weight (GVW). Axle spacings are also recorded.

GVW is gross vehicle weight. This value is reported by the WIM devices as part of the traffic monitoring program. GVWR is gross vehicle rate. It is the maximum rating a vehicle can have for its gross weight. WIM devices do not provide the GVWR values.

The WIM technologies used in the U.S. include piezo-electric (polymer piezo and quartz piezo), bending plates, fiber optic cables, load cells (both hydraulic and mechanical), capacitance mats, and strain gauge strips. Bridges and culverts instrumented with strain gauges are also being used as weight sensors.

In almost all cases, secondary sensors (e.g., inductive loop detectors) are used in combination with the primary axle and weight sensors to provide information on presence and vehicle length. Combining vehicle speed and presence information with the time between axle weight measurements allows the WIM system to correctly assign specific axles to specific vehicles and to group the axles correctly (that is to determine if the observed axles are single axles, tandem axles, tridems, quads, pentads or even larger groups of axles), and thus correctly classify each vehicle and compute its total weight. It is important to note that WIM measures the dynamic axle weights, and these are different from static axle weights; therefore, an adjustment algorithm is used by the WIM system to produce an estimate of static weight.

Motorcycle Counting

The relatively small amount of metal in many motorcycles combined with the fact that many motorcyclists ride near lane lines to give themselves more time to avoid cars moving into their lanes means that inductive loop detectors and half lane axle sensors often undercount motorcycles. When motorcycles ride in closely spaced groups, the closely spaced axles and cycles often confuse available traffic monitoring equipment, which have not been designed to identify the resulting pattern of closely spaced axles and vehicles.

Four aspects of traffic counting can be changed to improve accuracy of counting motorcycles:

Use of full lane width axle sensors
Use of wide loops of (8-foot-wide) in the lane for motorcycle counting
Counting by wheel path (see Montana example in Appendix I)
Video Detection (might be limited to detection during daytime only)

To improve accuracy of counting motorcycles, use full lane width axle sensors and 8-foot-wide loops.

Lane Occupancy

Inductive loop detectors and other devices, such as video image-based counters, can produce lane occupancy statistics that describe the percentage of time a vehicle occupies the detection zone. Many urban freeway and arterial performance monitoring programs use lane occupancy measurements to describe the onset and duration of congested conditions.

Headway and Time Gap

Traffic monitoring devices that time stamp the passage of either individual vehicles or the axles of individual vehicles are called event recorders. The collected detailed data not only describe the traffic volume and often vehicle classification, but they explicitly measure the headway between vehicles and thus the vehicle gaps in the traffic stream. Traffic monitoring systems—such as WIM scales—routinely collect time-stamped vehicle records that can be used to report the headway between vehicles and/or the time gap between vehicles. When States collect data using the FHWA IVR format, they should consider reporting lane occupancy, headway, and gap data.

2.2.2 Equipment Selection Considerations

In addition to the basic functional requirements discussed above, roadway agencies should consider a variety of other functions when selecting traffic monitoring technologies, including:

Number of data collection lanes performed by any one piece of equipment and/or sensor.
Cost of the equipment (initial cost, placement cost, operating cost, and expected maintenance costs).
Expected life of the sensors and data collection electronics.
Warranties supplied by the manufacturer/vendor.
Environmental conditions under which the equipment is expected to operate relative to the strengths and weaknesses of each specific technology.
Whether the agency staff has the required knowledge and equipment for placing, annual calibrating, and maintaining the equipment.
Available communications capabilities (i.e., what options does the agency have for retrieving data from the data collection electronics, and how do those options fit within the agency's current or planned traffic data processing procedures?).
Ability of the vendor to supply data outputs in a format that works seamlessly with the agency's existing or planned data processing system (ability to integrate data into a centralized system and utilize information to calculate and summarize statewide year-end statistics).
Vendor agreement for software support, equipment maintenance, warranty work.
Pavement condition for surface sensor like piezo and WIM sensors.
Installation materials and methods.

The first four of these issues provide the reviewer with the ability to trade off cost and performance. Of particular importance is the warranty provided by the vendor, as it provides an important level of assurance regarding the first three issues.

The fifth topic relates to the fact that some technologies work better in some specific environmental and traffic conditions than others. Some equipment might work very well in specific instances while working poorly in other circumstances. For example, pneumatic road tubes generally work well for short duration counts (48 hours) on lower volume, rural roadways. However, they do not work effectively on higher volume, multi-lane urban roadways. While vendors can create product modifications/versions to help technologies function in conditions for which they are generally not suited, when selecting technologies, agencies should be very aware of the increased likelihood of count issues/failures from those technologies in those conditions.

The answer to the sixth topic determines whether the agency needs to purchase additional equipment to place, operate, and maintain new technologies, as well as have staff undergo new training to perform those tasks.

Finally, the last six topics describe how efficiently and reliably the vendor's implementation of the selected technology will work within the existing or planned data processing system of the roadway agency. Collection, calibration, processing, reporting, accessing, and storing traffic monitoring data is resource intensive and roadway agencies should consider how much effort is required for any given device.

Selecting Intrusive Versus Nonintrusive Sensors

When deciding about selecting intrusive or non-intrusive sensors, site characteristics need to be carefully considered. See the Sensor and Equipment Location Considerations section later in this chapter for additional details.

Non-intrusive sensors are further divided into overhead-mounted sensors and side-fired sensors. Side-fired sensors have the advantage of being mounted beside the road. This makes it easy to install, access, and maintain. The drawback is that on multi-lane roadways, traffic using the roadway lanes farthest away from the sensor location can be obscured from the side-fired sensors by vehicles (and particularly trucks) traveling in the lanes closer to the sensor. This is called occlusion. Occlusion results in undercounting of total volume and biased speed estimates if the traffic on the inside of the roadway is traveling at a different speed than traffic on the outside lanes.

Generally, the higher above the roadway the non-intrusive sensor is placed, the smaller the problem with occlusion. However, raising the sensor vertically 1) increases the cost of installation and maintenance; 2) decreases the resolution with which the sensor detects vehicles in the road; and 3) creates movement in the sensor (as the pole on which the sensor sits sways) leading to other forms of accuracy degradation. Manufacturer-recommended installation heights need to be followed.

Mounting the sensor directly above the lane of travel is one way of significantly reducing the opportunity for occlusion to occur. Thus, overhead mounted sensors tend to be more accurate than side-fired sensors of that same technology. The disadvantage of overhead mounted sensors is that the lane of travel must normally be shut down for the sensors to be installed and then each time maintenance is performed because of fears of material dropping onto the roadway during those activities. Closed lanes create potential congestion on high-volume roadways.

The most common reasons for choosing non-intrusive sensors over intrusive are as follows:

Unsafe conditions for placing or maintaining intrusive sensors, such as high-volume multi-lane facilities.
Expensive traffic control to place and/or maintain intrusive sensors.
Poor pavement condition.
Disruption of traffic occurring with the placing of sensors introduces safety as well as sensor performance issues.
It is difficult to close the lane because of high traffic volumes. Intrusive sensors perform best under the following conditions:
- Even traffic flow at constant speed.
- Traffic follows good lane discipline.
- Straight and flat road section.
- Strong pavement with good surface condition.
- Safe access to the sensors for maintenance using common traffic control protocols.

The performance of in-roadway sensors such as inductive loops and magnetometer sensors is based, in part, on their close location to the vehicle. Thus, in-road sensors are insensitive to inclement weather due to a high signal-to-noise ratio. In traffic monitoring applications, in-road sensors effectively differentiate vehicle characteristics (e.g., axle spacing, class, length) on a lane-by-lane basis, without being subject to errors introduced by multiple vehicles simultaneously in the field of view of the sensor. Axle weights are the one form of data that cannot be collected non-intrusively. (Some bridge WIM systems are designed to operate without sensors being placed in the lane of travel. These systems only work on a limited set of bridges.)

The main disadvantage is the in-roadway installation, necessitating physical changes in the roadway as part of the installation process. Over-roadway sensors often provide data not available from in-roadway sensors and some monitor multiple lanes with one unit.

Table 2-8 describes which sensors are intrusive and which are non-intrusive.

**Table 2-8. Intrusive and Non-Intrusive Technologies**
Intrusive	Non-Intrusive
Inductive loops Polymer Piezo Quartz Piezo Fiber optic Magnetometer (most sensor designs) Tape Switches (pneumatic tubes) WIM systems (bending plates, load cells, and strip strain gauge)	Infrared (passive, active) Video detection system Microwave (overhead or side mounted) CW Doppler sensor Acoustic Ultrasonic Laser radar

2.2.3 Equipment Selection for Short-Term Temporary Counts

The short-term temporary counts are designed to provide wide geographic coverage at low cost. Most highway agencies perform many short-term counts, with the data collection staff working frequently within the roadway right-of-way to place and retrieve data collection sensors and equipment. This leads to additional priorities when selecting the appropriate technologies and equipment for performing short-term counts (in addition to the issues discussed in the previous sections, and the accuracy and price of the equipment). The technologies used for short-term counts should:

Be easy and quick to put in place and calibrate (because this saves large amounts of staff time when even small savings are multiplied by a large number of counts).
Allow placement of the traffic sensors safely.
The data collection electronics should contain a sufficient power source to allow the device to operate until it is retrieved.
The data collection sensors should stay in place, and operate correctly for the duration of short count, sometimes extending to one week.
The data collection equipment should be theft and vandalism resistant. (Traditionally, this has meant that the data collection electronics have been stored in a rugged case that can be chained to a permanent fixture to prevent theft.)
All data collection equipment should have clear markings for who owns the devices and a contact number to call should the equipment get lost, stolen, or become misplaced.
The data collection electronics should have a robust mechanism for transferring data from the data collection electronics to the central traffic data repository.
Software should be user-friendly and easy to check for proper functionality before the technician leaves the site.

The speed of sensor placement, electronic equipment set up, and calibration are important. The faster sensors and data collection equipment are placed, the more count locations a given staff member can set up in a day and the lower the costs of short-term data collecting are. At the same time, the placement (and pick up) of the sensors must not endanger the staff placing those sensors. This need to safeguard data collection staff, without incurring the high cost of full-scale traffic control, is one of the reasons so much effort has been spent on exploring non-intrusive traffic data collection technologies. However, on lower-volume roads where low-cost road-tube axle sensors can be easily placed without endangering the data collection staff, intrusive sensors are still commonly used.

Where intrusive sensors cannot be safely and easily placed, agencies either use non-intrusive sensors or accept the cost of substantial traffic control each time a sensor must be placed in the roadway. Where short counts are used routinely at such locations, but the agency is not interested in investing in permanent equipment, another option is to place permanent intrusive sensors in the roadway but not to connect those sensors to permanent power and communications. Instead, to use these sensors, the agency simply connects portable data collection electronics to the sensor leads, which are stored in a weatherproof enclosure until they are needed again. This approach saves the agency both the initial capital expenditure of bringing power and communications to the site, and the cost of traffic control during subsequent site visits. However, this approach is only cost effective if the permanent sensors are long lived, and if that location is to be visited for data collection routinely.

If non-intrusive sensors are the desired option for short-term counts, the most common approach is to affix the non-intrusive sensors to an extendable pole, which is placed on a trailer. The trailer is then towed to the roadside locations, placed in a safe position behind a barrier, and the pole raised. The sensors are then calibrated, and the entire trailer system left in place for the duration of the count.

Short-term counts are commonly used to collect volume, volume by classification of vehicle (including micromobility) and speed. Portable road-top WIM sensors for vehicle and axle weight estimation are not recommended due to low weight data accuracy. Weight data are collected with a high degree of accuracy only using WIM sensors installed flush with the road surface. This means that most roadway agencies collect axle weight data from permanent data collection sites. Table 2-9 describes the sensor technologies that are commonly used for short-term counts and the motorized vehicle attributes they routinely collect.

**Table 2-9. Sensors for Motorized Vehicle Data Collection for Short-Term Counts**
Technology	No. Sensors Needed for Speed Data Collection	Types of Vehicle Classifications Collected	Number of Lanes of Data Collected by Each Sensor	Environmental Issues/Concerns	Other Issues/Concerns
Pneumatic Tubes – traditional	2 (One lane only)	Axle Based (FHWA 13+)	1 per pair of sensors (Only lanes bordering shoulders)	Not suited to snowy conditions	Accuracy limitations under very heavy traffic volumes or stop-and-go conditions
Pneumatic Tubes – multi-lane design	2 per lane	Axle Based (FHWA 13+)	1 per pair of sensors	Not suited to snowy conditions	Accuracy limitations under very heavy traffic volumes or stop-and-go conditions
Tape Switches	2 per lane	Axle Based (FHWA 13+)	1 per pair of sensors	Placement difficulties in wet conditions	Needs protection of lead wires if placed on lanes not adjacent to shoulders
Magnetometer (several variations exist)	2 per lane	Length-Based (for most sensor designs)	1 per sensor (2 sensors per lane if speed or vehicle length is needed)	Most magnetic technology sensors require a short lane closure for sensor placement	Some magnetic sensors are placed in the pavement, others on the pavement, and others under the pavement
Video Detection System	1 camera	Length or Axle Based	Multiple	Does not work well in snow, fog, or dust storms	Short-term counter is mounted on an extendible pole on a trailer pulled to the count site; generally slow to set up
Piezo Polymer sensors (piezo-film, piezo-cable)	2 per lane	Axle Based (FHWA 13+)	1 per pair of sensors	Very cold weather often affects performance	Needs protection of lead wires if placed on lanes not adjacent to shoulders
Infrared	1 (transmitter + receptor)	Axle Based (FHWA 13+)	Multiple	Fog and heavy snow degrade performance and large crown in road will block beams; Occlusion	Infrared has an issue of heavy snow accumulation blocks the sensor.
Microwave	1 per direction (side-fired), 1 per lane (overhead)	Length-Based	Multiple	Occlusion issues with heavy or stop-and-go traffic and multiple lanes	Short-term counter is mounted on an extensible pole on a trailer pulled to the count site. Count is not reliable in congested low-speed conditions.
Acoustic	1 sensor	None	Multiple	Background noise/sound often interferes	Short-term counter is mounted on an extensible pole on a trailer; the sensor should be mounted higher than 25 feet
Laser Radar	1 sensor	13+	Maximum of 4	Snow, fog, heavy rain	No in-road installation required
Paste-Down Loop	2 per lane	(FHWA 13+) if ILS-based or Length-Based is 2 loops are used without ILS	For 3 or more lanes, home run of the other lanes needs to be taken care of	It must be above freezing to install, unless the marmac tape rated for lower temps is used. Can't be installed in rain or with snow or dew on the roadway	Requires clean, debris free surface to use a paste down loop. Can be peeled back up or left on the roadway after data collection is done.

Source: Based on Hallenbeck and Weinblatt, 2004.

ILS = inductive loop signature card

2.2.4 Equipment Selection for Continuous Counts

Continuous counts are collected using permanently installed equipment. Permanently installed, continually operating traffic monitoring equipment provides both current measures of traffic flow and a time series record of traffic flow attributes that describe how traffic flow changes over time at that location.

Permanent sensors represent both a large financial investment and a large data resource. As a result, the selection, installation, and calibration of that equipment is particularly important. Sensors that are poorly installed, inadequately calibrated, or that fail quickly because of poor design, installation, construction, or poor pavement conditions not only do not generate useful data, but they also waste resources (both money and staff time) that are needed for other data collection tasks. In part, this is because the funds spent on equipment and installation could be used elsewhere, but also because it requires considerable staff time to determine that the data being provided by poorly performing sensors are not an accurate representation of the traffic stream.

Permanent traffic monitoring locations should have:

Sensors that withstand the harsh roadway environment and operate for multiple years.
Power sources (either electrical power or solar power with battery backup).
Communications (land lines or cellular communications).
Environmental protection (traffic cabinet) for electronic data logger or controller (temperature, moisture, dirt, electrical surge protection on all sensors, power and communications lines, and protection against animal and insect infestation).
Safe pullout for maintenance vehicles and safe area for technician to access traffic equipment.
Straight and flat road segment with good quality smooth pavement (at least 300 ft before and 100 feet after the sensor).
Selected road segment should have even traffic flow during all hours of the day with minimal stop and go traffic queues, not prone to lane changes, mergers, or weaving maneuvers.
For WIM sites, the turnaround time (one loop of the test vehicle) for calibration truck should be 30 minutes or less.

Unlike short-term counts, the speed of sensor installation is much less of an issue for selecting permanent count technology and equipment. The real key for permanent count technologies is that once installed, the equipment should operate accurately for long periods, with only a modest level of maintenance. In high-volume locations, maintenance of an intrusive sensor is challenging unless lane closures are planned for some other purpose. (In a growing number of urban areas, traffic lane closures are very limited due to the size and scope of traffic congestion those closures cause, even at night.)

One of the great advantages of non-intrusive detectors as permanent traffic monitoring devices is the ability of the roadway agency to safely work on that equipment when it starts to have performance problems. Of course, not all non-intrusive equipment has this advantage. Non-intrusive equipment placed above the roadway requires lane closures for maintenance. In other instances, side-fired equipment positions come with penalties to the accuracy with which the devices work.

Table 2-10 summarizes the technologies currently used for permanent vehicular traffic monitoring. Most measure count, presence, and occupancy. Some single detection zone sensors, such as the range-measuring ultrasonic sensor and some infrared sensors do not measure speed. Continuous wave Doppler radar sensors do not detect stopped or slow-moving vehicles.

**Table 2-10. Sensors for Motorized Vehicle Data Collection for Permanent Count Locations**
Technology	Intrusive or Non-intrusive	Types of Vehicle Classifications Collects WIM?	Number of Lanes of Data Collected by Each Sensor	Environmental and Road Maintenance Issues/Concerns	Other Issues/Concerns
Piezo-Polymer Film or Cable	Intrusive	Axle Based (FHWA 13+) Can be used for WIM-capable	1 per pair of sensors	Susceptible to damage from mill and resurface and oversized trucks. Susceptible to snowplow damage, if not flush with surface	Temperature sensitive (for weight classification); doesn't work well in stop-and-go traffic. Recommended full lane width sensors for motorcycle detection.
Piezo-Quartz Cable	Intrusive	Axle Based (FHWA 13+) A good WIM sensor	1 per set of sensors	Susceptible to damage from mill and resurface and oversized trucks. Susceptible to snowplow damage, if not flush with surface	Sensors are typically ½ lane in width, so a site requires 4 sensors/lane for double threshold WIM; it is possible to instrument only on wheel path, with modest loss of volume and classification accuracy. Doesn't work well in stop-and-go traffic.
Strip Strain Gauge WIM Sensor and other Pressure Sensors	Intrusive	Axle Based (FHWA 13+) WIM-capable	1 per pair of sensors	Susceptible to damage from mill and resurface and oversized trucks. Susceptible to snowplow damage, if not flush with surface	Sensors are typically ½ lane in width, so a site requires 4 sensors/lane for double threshold WIM; it is possible to instrument only on wheel path, with modest loss of volume and classification accuracy. Doesn't work well in stop-and-go traffic.
Inductive Loop (Conventional)	Intrusive	Length-Based	1 per pair of sensors	Susceptible to damage from ditching, mill, and resurface Lifecycle is affected by extreme heat or freeze-thaw cycles	Single loops can be used to collect volume and lane occupancy, from which speed can be estimated. Speed estimation accuracy improves by using loop signatures.
Inductive Loop (Undercarriage Profile)	Intrusive	Various^a	1 per set of sensors	Susceptible to damage from ditching, mill, and resurface Lifecycle is affected by extreme heat or freeze-thaw cycles	New technology, not currently in widespread use
Side-Mounted Microwave	Non-intrusive	Length-Based	Multiple	Occlusion	Not as accurate as overhead-mounted, forward-, or rear-facing radar (getting close)
Overhead Microwave	Non-intrusive	Length-Based	1 per sensor	Occlusion	N/A
Doppler Radar	Non-intrusive	None	Multiple	Occlusion	Generally used only for speed data collection, often not accurate as a volume counting device. Count is not reliable in congested low-speed conditions.
Infrared/Laser	Non-intrusive (transmitter + receptor)	Axle Based (FHWA 13+)	Multiple	Fog and heavy snow often degrade performance	Most common device requires equipment on both sides of the right of way
Magnetometer (3-Axis Flux Gate or Magnetic Imaging)	Intrusive^b	Length-Based	1 lane	Lifecycle is affected by extreme heat or freeze-thaw cycles	N/A
Video Detection System (Object Analysis)	Non-intrusive	Various^c	Multiple	Often affected by heavy fog, snow, glare, dust	Requires proper mounting height; new technology, not currently in widespread use
Ultrasonic	Non-intrusive	Length-Based	1 per pair of sensors	Temperature variation and air turbulence often affect accuracy	New technology, not currently in widespread use
Acoustic	Non-intrusive	None	1 per pair of sensors	Multiple reflective signals	Must be carefully calibrated
Bending Plate	Intrusive	Axle Based (FHWA 13+) A good WIM sensor	1 per set of sensors	Lifecycle is affected by extreme heat or freeze-thaw cycles	Too expensive unless used as a WIM system
Bridge WIM	Non-intrusive	Axle Based (FHWA 13+) a WIM system	1 per set of sensors	Lifecycle is affected by extreme heat or freeze-thaw cycles	Only works on specific types and sizes of bridges and culverts; not really a conventional traffic monitoring device
Load Cell	Intrusive	FHWA 13+	1 per wheel path	Large upfront cost	Maintenance costs are yearly and require lane closure

Source: Based on Hallenbeck and Weinblatt, 2004.

^a There are two basic undercarriage loop classifier technologies. One uses the "signature" from existing loops to determine classification by matching the shape of that loop to expected profiles. The other uses specific types of loops to detect changes in inductance associated with wheels and uses that information to detect and measure axles. This device can classify by "axle," while the other defines classes that relate strongly to axle-based classes but are not specifically based on the number and spacing of axles.

^b Overhead-mounted, non-intrusive detectors require a structure (usually a bridge or gantry) upon which to be mounted. The expense of sensor installation increases dramatically where these do not exist.

^c Video image analysis will define classes based on the features detected by the software. The simplest detection algorithms are based on length. More complex algorithms detect and classify using axle information, provided the camera angles are capable of "seeing" different axles.

2.2.5 Sensor and Equipment Location Considerations

State agencies typically consider "site-centric" or "equipment-centric" approaches for traffic monitoring equipment selection. With the site-centric approach, the count location is more important and selected first, followed by selection of the equipment that works best at the selected location. The competitive bidding is an important part of the equipment procurement process. It is advisable to conduct the site selection first and then use a list of locations in the bidding document to provide opportunities for the bidders to assess site characteristics and propose the equipment that operate successfully under known site conditions.

Oftentimes, it is desirable to have similar equipment installed throughout the State to streamline equipment installation, maintenance, calibration, and operation. With this equipment-centric approach, specific locations are selected along the desired candidate road segments that provide the best data accuracy by addressing the manufacturers' requirements for equipment's operating conditions. Note that selecting a single type of equipment or vendor poses a higher cost and additional risk, if the vendor changes equipment, requires upgrades, or goes out of business.

Physical Characteristics of Traffic Monitoring Site Locations

Traffic monitoring systems accurately operate only when the equipment is in a physical environment that meets specific criteria:

Traffic traveling at a speed that does not change while a vehicle is crossing over a sensor array, not prone to stop-and-go conditions, acceleration, deceleration, lane changes, mergers, frequent vehicle passing in sensor lane (i.e., merge or exit lanes), or weaving maneuvers.
Access to landline power and communications or adequate solar/wind energy and wireless signal strength.

If cellular service is planned, its reliability at site location must be tested. If solar power is planned, conditions for its successful operation should be considered in all seasons (recommended minimum of 7 days of data storage available in the memory)

Safe pullout and parking area for maintenance truck and safe area for technician to access traffic cabinet and equipment for maintenance. Location of electronics outside the clear zone is a must.
Cabinet location provides clear view of the sensors and traveling vehicles from the traffic cabinet location to validate and verify equipment is working while on site by visually observing and classifying vehicles passing in the monitoring lane(s).
Cabinet is in the area with a good drainage and not prone to flooding. Cabinet also provides good views of each roadway lane for viewing traffic for calibration purposes.
No major road or pavement work during the intended data collection period.
Pavement in a good condition, free of cracks, rutting, or faulting. Sensors installed as part of pavement rehabilitation projects typically have a longer service life.

The following site conditions should be considered for certain type of sensors or applications:

Avoid power lines
Avoid water bodies (to prevent water reflection, only if video monitoring is used)
Avoid installation of counters that point toward traffic (only for infrared counters)
Avoid areas where people stop and stand around an area (for pedestrian counts)
Avoid curves (all sensors) – 5700 ft
Avoid hills (all sensors) – less than 2 percent grade
Avoid counting at intersections (all sensors) – within 1200 feet
Avoid traffic queues in the sensor area, if possible (all sensors)
Look for locations along the facility where a poll, tree, or other structure might be able to serve as part of the counter installation (short-term counts only)

Non-intrusive sensors come with many of the same requirements as above.

Finally, selecting locations that allow collection of ramp lanes in addition to the mainline lanes reduces the number of separate lane counts. However, consideration should be given to data quality issues of such locations due to the potential for more congested conditions, slower speeds and lane changing within interchange areas.

Additional Considerations for WIM Site Locations

States should place WIM equipment only in pavements that allow for accurate vehicle weighing. An excellent reference for WIM site requirements is ASTM Standard E-1318-09 Highway Weigh-in-Motion (WIM) Systems with User Requirements and Test Method (ASTM 2017). Another excellent source is the FHWA WIM Pocket Guide (FHWA 2018). In addition to general traffic monitoring site characteristics listed in the above section, all WIM sites should have the following characteristics:

Sufficient truck volumes present on the roadway being monitored.
Proximity of certified static scales(optional).
Avoidance of weigh stations or nearby enforcement activities influencing truck composition.
Straight and flat road segment (in all planes) with cross slope and grade 1% or less and good quality smooth pavement (at least 300 to 400 feet before and 100 feet after the sensor, depending on highway speed); for additional details see WIM Pocket Guide, Part I (FHWA 2018).
Smooth pavement that is in good structural condition with enough strength to adequately support axle weight sensors (it is desirable for the sensor to be installed in the first 1/3 of Asphalt Cement or Portland Cement Concrete layer thickness).
Pavement should not be scheduled for resurfacing or other maintenance or rehabilitation activity before the end of the expected WIM data collection period.
The ability to perform calibration of the scales, including a turnaround (one loop of the test vehicle) for test trucks that takes less than 20-30 minutes to complete.
It is important to have constant vehicle speed for WIM (which limits the use of WIM equipment in many urban and suburban areas where routine congestion occurs) primarily because deceleration and acceleration causes shifts in load from one set of axles to another and creates a dynamic rocking of the vehicle front to back, leading to increased weighing errors.

2.2.6 Sensor Layout and Configuration for Motorized Vehicle Detection

Figure 2-4 through Figure 2-9 provide examples of recommended sensor layouts for intrusive sensors to monitor motorized traffic.

Loop Layout and Configuration

Example of Loop Sensor Layout: A plan view diagram of a four-lane road showing four rectangular inductive loops (L1-L4) with wiring leads running to junction boxes and a central control cabinet.

Source: Federal Highway Administration.

Figure 2-4. Example of Loop Sensor Layout

Line Axle Sensor Layout and Configuration

Source: Federal Highway Administration.

Figure 2-5. Example of Piezo Sensor Layout

Loop and Piezo Layout and Configuration

Source: Federal Highway Administration.

Figure 2-6. Example of Loop and Piezo Sensor Layout

Pneumatic Tube Layout and Configuration

Source: Federal Highway Administration.

Figure 2-7. Example of Portable Pneumatic Tube Layout

Source: FHWA, adopted from Maine DOT practice.

Figure 2-8. Example of Portable Pneumatic Tube Layout with a Knot and Two Counters

Staggered WIM Sensor Configurations for Plate Sensors (Bending Plate, Load Cell)

Source: Federal Highway Administration.

Figure 2-9. Example of Staggered WIM Sensor Layout for Plate Sensors

2.3 SELECTING MICROMOBILITY COUNTING EQUIPMENT

Proper selection of technologies to count micromobility (i.e., bicyclists, pedestrians, scooters, e-scooters, e-bikes, e-skateboards, hovercrafts, etc.) traffic volumes at fixed locations is an important consideration. This section differentiates between those technologies best suited to count pedestrians versus micro-powered travelers. The discussion also identifies those technologies that are ideal for short-duration (i.e., portable) count locations and those that are ideal for continuous (i.e., permanent) count locations.

2.3.1 Challenges Unique to Micromobility Counting

Many of the basic technologies for micromobility counting, presented earlier in this chapter in Table 2-4, are similar to those used to count cars and trucks; however, the design/configuration of the sensors and the signal processing methods are often quite different. Therefore, separate equipment typically is used to monitor micromobility. The goal of micromobility counting devices are to accurately count:

Pedestrians only
Pedestrians and micromobility device users combined
Pedestrians and micromobility device users separately

Technological challenges to pedestrians and micromobility device monitoring are as follows:

micromobility users are less confined to fixed lanes or paths of travel than motor vehicles, and sometimes make unpredictable movements. Pedestrians take shortcuts off the sidewalk or cross streets at unmarked crossing locations. Bicyclists sometimes ride on sidewalks or travel outside designated bikeways. They sometimes stop in front of a sensor to talk, wait, or even to examine the sensor. These actions make it difficult to place or aim sensors and decrease the accuracy of the sensor equipment.
micromobility users sometimes travel in closely spaced groups, and some sensors have difficulty differentiating between individuals within the group. In these cases, a group with multiple persons is counted as one person, and the sensor underestimates the actual counts.

Despite these challenges, several technologies are capable of accurately count micromobility traffic. The growing demand for automatic counters, increased competition in the marketplace, and advancing collective knowledge with existing products have resulted in improvements in equipment accuracy and capabilities.

2.3.2 How to Select Micromobility Counting Equipment Type

Commercially available counters use a variety of technologies and features that vary dramatically and affect how, what, where, and how long counts are collected. Even within a specific technology, the accuracy of commercially available products varies significantly based on configuration, installation, and level of use. Cost per data point also varies greatly between counter technologies. Equipment calibration/validation needs must be also considered to ensure that count data are within the bounds of acceptable accuracy. Annual calibration is recommended.

Figure 2-10 presents a simplified flowchart that helps to narrow possible choices based on the two most important aspects of data collection:

What are you counting? Only micromobility devices (bicycles, scooters, hoverboards, e-bikes, etc.), only pedestrians, pedestrians and micromobility devices combined, or pedestrians and micromobility devices separately? The TMG handles all 4 of these types of counts.
How long are you counting? Permanent, temporary, or somewhere in-between?

Consider the following example: a city wishes to monitor a shared-use path and desires to count micromobility devices and pedestrians separately on a permanent basis. Using the simplified flowchart in Figure 2-10, the first decision point is "What are you counting?" In this example, the city wants to count both modes separately (note the circled icons in Figure 2-10). The next question in the decision process is "How long will the counters be collecting data?" In this example, the city wants continuous data from a permanent location, so technologies toward the middle and top of the table are relevant. Several equipment technologies are possible (e.g., pressure sensor and automated video imaging), but only a few have been used in common practice. Manual observers and video imaging with manual reduction are possible but are typically used for short-term data collection.

For a permanent site capable of counting pedestrians and bicyclists separately, two technologies will have to be paired together. Because inductance loops are more commonly used at permanent sites, the city staff selects a combined system that has an infrared pedestrian counter paired with an inductance loop detector for bicyclists. The infrared sensor by itself is not capable of differentiating between people walking or bicycling; however, when combined with the inductance loop detector, the bicyclist counts are automatically subtracted from the infrared sensor counts. Based on budget and commercial availability, a final decision is made about technology to be deployed.

Figure 2-10. Equipment Selection Process

2.3.3 Sensor Layout and Equipment Set-up

Inductance Loop Detectors for Wheeled Micromobility Counting

Inductance loop detectors do not require the presence of ferrous (i.e., iron, steel) bicycle frames; however, large conductive objects (like a car or truck) are more likely to meet the predetermined disruption criteria than smaller conductive or non-ferrous objects (like a motorcycle or bicycle). The sensitivity of an inductance loop should be changed to better detect motorcycles or bicycles, but the increased sensitivity often results in over counting for cars and trucks. For this reason, most agencies typically use dedicated loop detectors for counting bicycles rather than trying to use existing loop detectors to count cars, trucks, and bicycles.

The preferred counting location is where bicycles are free-flowing and/or not likely to stop. Loop detectors for bicycle counting should be placed in lanes primarily used by bicycles. If the loop detectors are placed in lanes shared by motorized traffic and bicycles, special algorithms should be used to distinguish the bicycles from the motorized traffic.

Inductance loop detectors measure the direction of bicyclist travel using at least two possible options:

Installing an inductance loop within each directional travel lane and if all (or a certain percentage) bicyclists in that lane are traveling in the specified direction (e.g., shared-use path or directional bike lane).
Installing two inductance loops in series to infer direction from the timing of detection events for each loop.

The first option is the most used practice to date. For the second option, not all data loggers or controller equipment are capable of interpreting signals from a paired inductance loop sequence.

The most important variables in accurate bicycle detection via a loop detector are:

Loop configuration: Several different wire patterns have been used for counting bicycles, including quadrupole, diagonal quadrupole (also called Type D), chevron, and elongated diamond patterns (see Figure 2-11). Loop configuration depends on travel behavior and road/trail type.
Detector circuit sensitivity: The sensitivity should be high enough to detect non-ferrous bicycle frames but not so high as to detect motor vehicles in adjacent lanes.
Bicycle position over the loop: Pavement stencils should be used to indicate optimal (i.e., most accurate) bicycle position over the loop detector, which is typically directly over the saw cut for the wire coil.
Bicycle size and composition: A large steel frame is more likely to disrupt the loop detector's field than a smaller non-steel frame, but the threshold amount of ferrous metal is not a known quantity and varies based on the above three and other variables. Some inductance loop detectors detect bicycles with non-steel frames due to the presence of conductive material in the wheels or other bicycle components.

Examples of Inductance Loop Detector Shapes for Bicyclist Counting: Technical drawings of various wire patterns used to detect bicycles, including Square, Circle, Rectangle, Modified Chevron, Diamond, and Double Diamond.

Figure 2-11. Examples of Inductance Loop Detector Shapes for Bicyclist Counting

Infrared Sensors

Two types of infrared sensors are used for micromobility detection: active infrared sensors and passive infrared sensors. For portable applications, infrared sensors should be enclosed in a vandal-resistant, lockable box and attached to an existing pole, fence post, or tree. For permanent applications, infrared sensors are often enclosed within wooden fence or other vertical posts.

For micromobility counts, infrared sensors are frequently paired with another technology to improve differentiation between multiple persons in a group (i.e., side-by-side or closely spaced front-to-back and between bicyclists and pedestrians. Typical pairings include loops or axle sensors. In addition, use of an overhead gentry makes it easier to detect micromobility users traveling in a group. For example, Figure 2-12 and Figure 2-13 show a permanent monitoring location that combines a passive infrared sensor with inductance loop detectors.

The passive infrared sensors look for heat differentials and their patterns. These sensors have higher error rates when the ambient air temperature approaches normal body temperature (97°-100° Fahrenheit). The error varies among different brands of passive infrared counters.

Figure 2-14 shows a typical configuration for an active infrared sensor. This example shows an ideal location:

1) primarily used by pedestrians and bicyclists only; 2) the travel area is constrained with the detector pointing across the sidewalk away from the street; and 3) the detection area is well defined in a position where pedestrians and bicyclists will be traveling perpendicular to the sensor.

A cyclist on a park bicycle path passing over an inductive loop in a diamond configuration next to a passive infrared device.

Source: Shawn Turner, TTI.

Figure 2-12. Example a of Passive Infrared Sensor Combined with Inductance Loop Detectors

Cyclists and pedestrians on a busy park bicycle path passing over inductive loops in a diamond configuration next to a passive infrared device.

Source: Louis Queruau, Eco-Counter

Figure 2-13. Example B of Passive Infrared Sensor Combined with Inductance Loop Detectors

A solitary cyclist on a park bicycle path passing through an infrared signal with active infrared devices on both sides of the bicycle path.

Source: Steve Hankey, University of Minnesota (Red horizontal line is provided to visualize the infrared beam.)

Figure 2-14. Typical Configuration for Active Infrared Sensor

Pneumatic Tubes

Pneumatic tubes are a low-cost, portable approach for counting micromobility vehicles (Figure 2-15). The data logger should be set with pre-defined criteria (e.g., axle spacing) and/or algorithms to determine whether a valid vehicle type has passed over the tubes. Pneumatic tubes should be combined with infrared sensors at locations where both bicyclist and pedestrian counts are desired.

Due to their construction (smaller diameter and easier to squeeze due to thinner walls), smaller pneumatic tubes are better at capturing micromobility devices when they are used on dedicated pathways, since they are not also being used to also collect vehicle count data. The placement of pneumatic tubes for bicycles should adequately cover the bicycle travel path while not being exposed to excessive passage by motor vehicles.

When counting bicycles in a bike lane or shared lane, passage and activation by motorized traffic is unavoidable. In these cases, the data logger criteria should be capable of ignoring typical motor vehicle axle spacing. If direction of bicyclist travel is desired, a pair of pneumatic tubes should be used (see Figure 2-15), and travel direction should be inferred from the timing of detection events at each tube. For shared roadways with motorized traffic, the classic tube diameter used for detection is 0.50 inches (15mm) and for greenways that do not encounter motorized traffic a thinner, mini-tube is utilized with a diameter of 0.35 inches (9mm).

A solitary cyclist on an urban downhill slope passing over 2 road tubes. This collects both direction and speed of the cyclist.

Source: Louis Queruau, Eco-Counter.

Figure 2-15. Example of Pneumatic Tube Configuration for Counting Directional Bicyclist Traffic

Bicyclist safety is a concern when pneumatic tubes are installed with pavement nails or other metal fixtures, as they could possibly dislodge from the pavement and puncture a bicycle tire or create a road hazard for bicyclists. Extra care should be taken in installing pneumatic tubes, either by placing metal fixtures outside the bicycle facility or by using tape or other adhesive. Pneumatic tubes pose a tripping hazard, and it is also known that bikes tend to go around them. Therefore, application of this technology is limited. Usage of this technology is not recommended for bike lanes on shared roadways with cars driving over the tubes and getting counted.

2.3.4 Pressure and Seismic Sensors

Pressure plate sensors operate by detecting changes in force (i.e., weight), much like an electronic bathroom scale. Seismic sensors (also sometimes called acoustic sensors) operate by detecting the passage of energy waves through the ground caused by feet, bicycle tires, or other nonmotorized wheels. As with other monitoring technologies, pre-defined criteria should be used to determine a valid detection and therefore a valid user to be counted.

Both pressure plate and seismic sensors require the sensor element to be placed underneath or very near the detection area. Pressure and seismic sensors are most common on unpaved trails or paths (Figure 2-16), where burial of the sensor element is typically low-cost and minimally disruptive. However, pressure plate sensors should be used at curbside pedestrian signal waiting areas, as a supplement to or replacement of a pedestrian crosswalk push button.

Some models of pressure plate and seismic sensors are capable of detecting the difference between pedestrians and bicyclists. Placement and size of the pressure plate sensors (also known as pressure mats) is used to gather directional information. When installed properly, pressure and seismic sensors serve as permanent continuous counters.

(a) Pressure sensor on natural surface trail

Placement of a pressure sensor on a natural surface trail before it is covered up and after it is covered up showing no signs of the pressure sensor being in place.

(b) Pressure sensor on paved surface

A pressure sensor on a paved surface on an entrance to a park.

Source: Jean-Francois Rheault, Eco-Counter.

Figure 2-16. Examples of Pressure Sensors on Natural (A) and Paved (B) Surfaces

2.3.5 Video Image Processing

Video image processing operates by using visual pattern recognition algorithms to identify (and sometimes track) a pedestrian or bicyclist traveling through a video camera's field-of-view. Video image processing has the capability to distinguish pedestrians and/or bicyclists traveling in a group or cluster (see Figure 2-17). These capabilities vary, depending on the algorithm implemented in the commercial products. Weather and lighting often reduce the accuracy of this technology. Finally, video image processing typically has the highest equipment costs. This type of sensing technology is rapidly developing, and improvements are being made to overcome bad weather using Lidar. The equipment should be deployed to many more locations, including signalized intersections.

Pedestrian and bicyclist counts are manually reduced by viewing recorded video from intersection control or surveillance cameras. This manual approach is practical and low-cost for periodic short-term counts, but it is not sustainable for continuous monitoring purposes (due to required labor and associated costs). This approach eliminates equipment installation (and corresponding traffic control), but also requires a low-cost labor force to manually review the video. Several companies offer a portable video recording unit as well as data reduction services. The recorded video may be useful to other agencies or departments that wish to study bicyclist and pedestrian behavior (e.g., in response to safety issues or concerns). Additionally, this recorded video should be used for quality assurance purposes (i.e., for verification/validation of nearby automated counts).

A video capture of bicycles and pedestrians on a crowded urban intersection showing direction and speed.

Source: Louis Queruau, Eco-Counter.

Figure 2-17. Example of Video Capture of Bicycles and Pedestrians

2.4 TRAFFIC MONITORING EQUIPMENT EVALUATION, MAINTENANCE, CALIBRATION, AND QUALITY ASSURANCE CONSIDERATIONS

To assure that the selected equipment performs to the best of its potential and high-quality traffic data are being collected and reported on a continuous basis, the following factors should be carefully and thoroughly addressed: site selection conducive to accurate data collection, use of proven technology, rigorous attention to detail and use of high-quality materials during the installation process, routine equipment maintenance and calibration, and robust data quality assurance process.

2.4.1 Equipment Testing and Technology Evaluation

Traffic monitoring technology is evolving quickly due to a combination of the availability of modern, low-cost computing and communications technology, but is also driven by the need for more timely information. Not all equipment vendors produce equipment of equal quality. Some equipment has been heavily tested and operates very robustly. Even within a single technology, equipment performance varies widely from vendor to vendor based on each vendor's internal software algorithms and the components that make up their equipment.

The reason different vendor's equipment produces different results for any given sensor technology is that the data collection electronics and the software that resides in those electronics perform in different ways. (For example, two different video image-counting devices may produce vastly different results if one uses a robust image-processing algorithm, while the other does not.)

Consequently, as agencies make decisions on what type of hardware and supporting software to purchase, they should continue to consult the available and more detailed literature (such as from pooled funds, FHWA Highway Community Exchange [HCX], and FHWA Long-Term Infrastructure Performance [LTIP]) that describes the performance of specific technologies. They should work cooperatively with their peers to share their working experience with specific equipment. Using these resources effectively is a key to selecting the best data monitoring equipment for each agency's needs. A very good source of additional information on traffic data collection technologies is available on the FHWA's Travel Monitoring Policy website. A variety of other excellent technical resources are included in Appendix D.

It is also important that agencies carefully test equipment before they purchase specific devices from a vendor, and once they have purchased devices that meet their needs, they should routinely calibrate and continue to test the performance of their equipment in the field. The first of these steps ensures that the equipment they purchase performs as advertised. The second step ensures that the equipment they are using is being correctly installed in the field, and that the performance of the sensors and electronics has not degraded over time due to use and changing environmental conditions. Careful site selection, use of high-quality materials, and rigorous attention to detail during the installation process will facilitate the reliable collection of high-quality traffic data on a continuous basis.

2.4.2 Equipment Installation, Maintenance, and Operation

Every agency that owns and/or maintains traffic monitoring sites should perform the following tasks to ensure that the equipment they purchase works to the best of its capability:

The equipment should be tested to assure that it meets the users' data accuracy needs before being placed into service.
The equipment installation should be inspected by a certified agency's agent to ensure that the equipment is being correctly installed in the field.
The equipment should be periodically calibrated (annually or per manufacturer's specification).
The equipment performance should be validated periodically (semiannually or annually) to ensure that it continues to perform as intended and that the performance of the sensors and electronics has not degraded over time due to use and changing environmental conditions.
The collected data should be routinely subjected to quality assurance tests to identify potential equipment malfunction or degradation.
The data should be analyzed and then quickly and routinely supplied to users so that data quality concerns not caught by the primary data quality process are quickly identified by users.
A feedback process should be in place so that the traffic monitoring group obtains this feedback from users, and effectively responds to improve the quality of the data.

2.4.3 Equipment Calibration and Data Monitoring

FHWA recommends annual calibration of all traffic counting equipment for volume counts, classification, weigh-in-motion, volume, speed, and portable hardware.

Calibration ensures that the purchased equipment performs as advertised. Validation checks when equipment is initially installed are an essential first step in that process. After the initial calibration, the equipment should be actively monitored for quality performance, regardless of a vendor's assurances of self-calibration capabilities. The equipment should be periodically calibrated on-site. The FHWA recommends annual calibration of all traffic counting equipment. In addition to on-site calibration, in-office tracking of site information and data comparison and reasonableness checks should occur regularly (daily, monthly, and annually, as needed).

A robust traffic monitoring equipment calibration program should include:

Implementing software tools that help automate the process.
Performing daily diligence activities that ensure checking the quality of data as they are collected/processed/stored in the master (centralized or distributed) traffic database.
Evaluating data using weekly, bi-weekly, monthly, and yearly trends to determine validity and reasonableness (e.g., checking of a specific day like Wednesday each week for trending and issues, making sure the checking is done at the lane level or sensor level).
Collecting manual counts and comparing counts against portable equipment collected counts.
Performing field and electronic calibration of classification, weigh-in-motion, volume, speed, and portable hardware annually.

On-site equipment performance validation and calibration includes performing a variety of tests on equipment to ensure that it functions as intended and correctly collects, processes, and reports the traffic data. The calibration process identifies both major errors (such as failed sensors) and minor errors (such as errors in site set-up, or the wrong classification algorithm installed on a shipment of devices) that result in the collecting, processing, storing, and disseminating of inaccurate traffic statistics. The entire traffic monitoring program credibility is at stake when erroneous data are collected, processed, stored, and disseminated. To avoid the risk of producing and disseminating erroneous data, traffic data programs should calibrate often.

Errors associated with calibration inaccuracies significantly increase the cost and decrease the usability of data from the entire traffic monitoring program. FHWA's WIM Pocket Guide and its Appendix E contain detailed information about WIM equipment calibration. In addition, the TMG Appendix D contains compilation of State practices for equipment calibration.

2.4.4 Quality Assurance Considerations

Data collection quality assurance processes should be established and documented to ensure active management from those collecting and using the data so that the technology performs well. Equipment that is not actively monitored for quality performance eventually goes out of calibration, regardless of a vendor's assurances of self-calibration capabilities. Validation checks when equipment is initially installed are an essential first step in that process. Following a formal quality assurance and field maintenance program and providing resources to fix problems that are identified by that process ensures that funding available for collecting data is spent on collecting valid, useful information.

Once the data have been physically collected, an agency must process (integrate, convert, calculate, QA/QC, store, manage, and provide access, etc.) the data consistently and correctly to convert data into published statistical information. Review of data consistency and data reasonableness should be part of this process, including tests to proactively check for potential equipment issues. Correct and consistent data processing is important for ensuring quality and reliability of AADT estimates. Consistent processes also ensure agency credibility, which allows an agency to easily defend their reported statistics and show through transparent audit processes that their data accurately reflect current traffic conditions. This allows States to pass Federal compliance reviews used to ensure that reported statistics are being accurately reported and assures decision-makers the data deliverables are appropriate for critical decisions. A compendium of data quality control criteria implemented by State highway agencies is provided in Appendix C.

Return to top

Page last modified on May 18, 2026