The ABC's of AVC
The idea of collecting a fee for using a public thoroughfare has been around for a long time. With that idea came the problem of how to assess tolls in an equitable manner. The fare schedule should allow for as much flexibility as possible. It must be easy to implement and audit. Classification is a method of auditing the fare collection process and thereby must be able to support the fare schedule. As the technology for classification improves, the fare schedule can be made more complex and more equitable.
Vehicle classification has been a part of toll-roads since the earliest toll-roads went into operation. Initially the classes were simple and all classification activities were manually performed by the toll collector. As time progressed, systems were developed to provide a means for automatically determining the class of a vehicle. In America the earliest systems consisted on a treadle and a loop; the purpose of which was to detect the number of axles associated with a vehicle. Since that time significant progress has been made in the development of classification technology. Systems around the world use a variety of measurements to create fare schedules. In America, the number of axles has been a significant factor in most fare schedules. Some schedules have been based upon the weight of the vehicle. In Europe, emphasis has been placed on the physical size of the vehicle.
Automatic classification is the ability to detect various physical characteristics of a vehicle and to separate vehicles into classes based upon this information. Classification systems classify rather than identify a vehicle. Classification of a vehicle depends upon the electronic gathering of information about the vehicle. The accuracy of vehicle classification is directly proportional to the amount of information being processed and the ability to make maximum use of that information. The better and more complete this information is, the greater the accuracy of the classification.
The classification requirements are driven directly by the fare schedule. Recent advances in classification capabilities have allowed users to increase the number and types of classification categories. These include such categories as buses, vans, motorcycles, each with or without accompanying trailers. In order to classify these vehicles, the following information about the vehicle is generally required.
Length of the vehicle.
A height profile of the vehicle.
Number of axles and their locations.
The existence of a hitch (trailer) and its location.
A good classification system will be able to accurately define each of these discriminants. In order to define each of them accurately, the system must be able to accurately track the position of the vehicle during the time that it is within the space where the vehicle is being examined. The system will use sensor information to generate each of the above discriminants.
Older classification systems based their classification solely on the number of axles. A vehicle separator of some sort (light curtain, ultrasonic curtain, entry and exit loops, et cetera.) provided a method for separating individual vehicles. The number of axles detected by an treadle in the pavement provided information for the generation of an axle count. Bi-directional treadles were used to indicate when a vehicle reversed direction over the treadle.
These systems suffered from a variety of problems. Loops, while reliable enough for traffic management, do not provide the reliability required for audit purposes. The locations of the axles could not be determined therefore any classification based upon axle location was impossible. The systems could not provide useful height or length information nor could they determine the existence of trailers. In fact trailers were often classified as separate vehicles. Recently the application of other technologies have given new meaning to the profiling and classification processes.
Profiling versus Classification
A profile in the fare collection industry means a set of information about a vehicle from which a classification may be determined. The ability to classify into more categories is dependent upon the amount of detail included in the profile. Today a good profile should include:
Length of the vehicle
A detailed height profile of the vehicle.
Number of axles and their locations
The existence of a hitch (trailer) and its location.
The accuracy of the distance measurements should be accurate to better than one foot. The height profile should reflect the height of the vehicle at small increments along the entire length of the vehicle. The complete picture of the vehicle provided by the sum of this data is then presented to the classifier. The classifier then separates the vehicles into categories based upon this profile information.
The Importance of Vehicle Location Information
In order to be able to generate an accurate and useful profile, the system must be able to determine where the vehicle is located at all times. Otherwise the distance measurements and the locations of the various height samples will be inaccurate and will lead to bad classifications. To this end TDS added a Doppler radar to the system design. Using the velocity data provided by the radar, the system can very accurately track the location of the vehicle as it passes through the profiling window.
Height Profiling
A sensor providing accurate height indications is required to generate a height profile. Once the position of the vehicle is knows, samples of the height of the vehicle can be taken from the sensor as the vehicle moves through it. Without the knowledge of the position of the vehicle, the output of the sensor can not be assigned to a particular segment of the vehicle, for instance the hood, and an accurate height profile cannot be generated.
Height Sensors
Four types of sensors have been considered for the generation of height information. They include ultra-sonic curtains, light curtains, laser height detectors and radar height detectors.
Ultra-sonic Curtains
Early designs utilized an ultrasonic curtain. Sound waves were broadcast from an overhead transmitter and the time measured until an echo was received. Without a vehicle under the transmitter, the signal was reflected by the ground and echo was received at a delta time based upon the speed of sound (1100 feet per second) and the distance above the ground. When a vehicle was located under the unit, the echo time was reduced by the distance of the vehicle above the ground, thereby giving an indication of height. However since the overall system did not determine the position of the vehicle at all times, the only useful height information gathered was the maximum height of the vehicle. The sensor suffered from several other limitations, namely the beamwidth of the antenna and the time required per sample due to the slow speed of sound wave propagation through the air medium. As a result this sensor has been used on several occasions to separate vehicles for classification by axle count but it has never been used to provide height information from which a profile might be generated..
Light Curtains
Light curtains were initially used in the industry as vehicle separators. A vertical series of very narrow beams provided an excellent method for determining the beginning and ending of a vehicle. If it is periodically sampled as the vehicle passes, it also provides good lateral profile information from which a height profile can be constructed and from which axles and hitches can be detected. Unlike the ultrasonic curtain, the light curtain could be scanned in less than 5 milliseconds and can provide excellent data on both stop and go traffic and high speed targets. Because of the nature of these devices, their use is restricted to single lane operations.
Laser & Radar Height Detectors
An overhead height sensor can provide a good profile of the top of the vehicle but does not provide information from which axles can be detected. However in multilane systems, the overhead profiling technique remains the only useful method for defining vehicle height profiles. Unlike the light curtain, the overhead profiler does not provide a detailed lateral picture of the vehicle, including such things as windows and other complex patterns. These sensors are usually augmented with an axle detector (treadle) in order to provide all of the information necessary to provide for classification in an axle oriented fare structure. The clear advantage is that the unit can be mounted overhead above each lane in a multi-lane situation. The disadvantage is that the overhead profiler does not provide the same level of detailed information about the vehicle. In single lane situations, when overhead sensors are used, they are often augmented with side mounted single beam sensors for certain types of detection associated with an individual fare structure.
Profiling and Classification
The height sensor by itself is not a profiler or a classifier. The data that it provides when combined with position data and analyzed provides the profiling and classification capabilities. TDS pioneered the concept of using a Doppler radar to provide velocity information from which the position of the vehicle can be tracked through the profiler. Once the vehicle position is clearly defined, the system can then create a very accurate vehicle profile as the vehicle travels through the profiling zone. When adequately implemented as part of an overall lane system, it will provide excellent classification capabilities for both single lane and multiple lane situations. The need for loops is eliminated. The combination of the lane controller and the AVC system are all that are required to automatically frame and process the lane transaction.