How an automatic vehicle classifier works?

Answer 1

Current model vehicle classifiers seldom rely on internal calculations to determine classification. Rather they collect "time tagged event" data that is processed using advanced algorithms in software on a PC. Data is collected using high resolution time based (think milliseconds) event recordings on two or more sensors. Sensors are usually mounted on the pavement surface. The most common type is road tube, but piezo strips may also be used. In either case there must be a measured distance between the sensors for the software to be able to determine axle speeds. The speed of each axle is calculated based on (A.) how far apart the sensors are and (B.) how long it took the axle to cover the distance between them. This gives an axle speed for each axle that passes over the two sensors. Once we know the speed of each axle, we can also calculate how distant any following axles are from the first axle detected based on how much time elapsed between strikes on the lead sensor. When vehicles run over the sensors, all of the axles on any single vehicle are always going very close to the same speed. This allows us to match up axles going in the same direction at the same speed and build a list of axle spacings. Now is where the axle classification scheme (the axle spacing list) comes in. As axles and their speeds are determined the software pairs them up and then compares the distance between them to a list of measurements for known vehicle types. For instance although cars come in lots of different sizes, to be classified as cars, they must have a minimum axle spacing of 72 inches but not exceed a maximum spacing of 118 inches. So, if the second axle of a vehicle is between 72 and 118 inches it is classified as a car. This will be true even it the car had a 1, 2 or even 3 axle trailer being towed behind it. A semi-truck with 5 axles may have a variety of possible spacings but to be classified as a particular class the spacings between the axles must match up with a set of rules in the classification scheme. The most commonly used classification scheme has 13 classifications containing everything from motorcycles to 8 axle semi-trucks. Using a two hose data collector it is possible to accurately collect up to 10,000 vehicles per day in bi-directional traffic. Even more is possible using four hose units.