WO2012017120A2

WO2012017120A2 - Method for classifying vehicular traffic

Info

Publication number: WO2012017120A2
Application number: PCT/ES2011/070566
Authority: WO
Inventors: Marta UGARTE SUÁREZ; Susana Briz Pacheco
Original assignee: Universidad Carlos Iii De Madrid; Universidad Europea De Madrid, S.L.U.
Priority date: 2010-08-02
Filing date: 2011-08-01
Publication date: 2012-02-09
Also published as: WO2012017120A3; ES2377613A1; ES2377613B1

Abstract

The invention relates to method enabling the classification of vehicular traffic along a road having one or more lanes. The method enables the determination of when a vehicle has passed by and the determination of the parameters of said vehicle such as the speed or type thereof, wherein said measurements are not affected by factors that could affect the sensors, such as temperature changes or interference from nearby sensors, among others.

Description

METHOD FOR CHARACTERIZING THE ROLLED TRAFFIC

DESCRIPTION

OBJECT OF THE INVENTION

The present invention is directed to a method that allows the characterization of road traffic along a road that contains one or more lanes. The method allows to determine when a vehicle has passed and parameters about the vehicle that has passed such as speed or type of vehicle, where these measures are not affected by disturbances that act on the sensors such as those due to temperature changes or interference between nearby sensors, among others. In order to carry out these measurements, the device according to an example of embodiment that can carry out this method, has as many pairs of measurement modules as rails to be characterized and has been preferably developed to be implemented on a bridge, a step elevated or on any elevated platform that may be arranged on the track. In this way the system can be installed and removed simply without interrupting traffic. Portability, low cost and ease of use make this system an ideal tool for the timely monitoring of road sections, aimed at efficient traffic management or environmental studies, although the system can also be applied to studies in prolonged periods of time.

BACKGROUND OF THE INVENTION

Within the variety of techniques that can be used for vehicle counting, the most frequently used are those that use sensors located on the pavement. Examples of which are pneumatic hoses, inductive loops, piezoelectric sensors, sensors based on fiber optic cables, etc.

These systems have several drawbacks, including the difficulty of their installation (such as inductive loops that must be installed under the pavement), the safety of the personnel performing the installation or the installation. Need to interrupt traffic for installation in the case of hoses, piezoelectric sensors and fiber optic cables.

Another of its drawbacks is the deterioration suffered by these devices due to the passage of vehicles. The effectiveness of these systems may also depend on the characteristics of the road (several lanes) and adverse weather conditions.

A first alternative that avoids installation directly on the track or under the road surface is to use non-intrusive techniques. The non-intrusive techniques that can be used in vehicle counting are very diverse among which are those that make use of infrared (IR) sensors, magnetic sensors, acoustic sensors, video, radar, ultrasound, etc. Although the technique based on the analysis of the images recorded on video provides a large amount of data (number, category and speed of the vehicles) and is easy to implement, it has been found that there are many factors that affect the performance of these systems: stationary and mobile shadows, direct sun, light-dark transitions, errors in software operation etc.

Regarding acoustic systems, these present problems in the detection of vehicles at rest and in situations of high traffic density. Radar is also commonly used but has the disadvantage of its high cost and of requiring frequent, also expensive, calibrations.

There are also devices based on the use of two ultrasonic sensors tuned to two different frequencies and located on individual high supports. The sensors detect the passage of vehicles and classify them according to their profile. This device is designed to detect and classify vehicles that access garages and building entrances.

Although these types of devices are suitable for low speed conditions and well-established parameters at the time of assembly, they are not suitable in situations where the speed of the vehicles is high (fast tracks) since it is not possible to draw the Precisely vehicle profile, the presence of disturbances due to temperature changes, influence of high noise levels and Interferences between nearby sensors make it not appropriate for this kind of demanding application.

The present invention establishes a method that makes use of ultrasonic distance sensors such that, although the sensors are affected by disturbances such as those mentioned, said method allows to discriminate when a vehicle has passed, what speed it carries and some parameters that characterize its typology

DESCRIPTION OF THE INVENTION

The present invention comprises a method that allows it to be carried out by means of an installable system, for example, on a portico of those located on the road in such a way that its installation is simple and does not imply interrupting traffic during installation or uninstallation.

The method for characterizing road traffic according to the invention is applicable to traffic that runs along a traffic lane of one or more lanes and comprises the following steps:

• provide a first ultrasonic distance sensor module and a second ultrasonic distance sensor module on the same lane of the traffic lane, distanced in track height, where the second ultrasonic distance sensor module is separated by a water distance below the first ultrasonic distance sensor module; and both are oriented in such a way that they measure the distance to the track,

The pair of sensors is located on a lane of the traffic lane, both sensors facing the surface where the vehicles run so that each of them carries out an independent reading. If there is more than one lane, each lane can have a pair of sensors.

Ultrasonic distance sensors offer a high degree of sensitivity to external disturbances such as temperature variations, background noise due to traffic, noise due to the presence of the other sensors; and others. The rest of the stages are designed so that the results derived from their reading are reliable despite the presence of this noise in the reading signal. The presence of two sensors makes it possible to ensure that the passage of a vehicle is first detected in a sensor module and then in the other in a known order. This order allows to establish the search of patterns in the signal of one and another sensor module that determine the passage of a vehicle.

• establish a measurement time period T,

• carry out a reading of the smi signal of the first ultrasonic distance sensor module and a reading of the smo signal of the second ultrasonic distance sensor module during the same time interval T,

The measurement time T is a period that in practice is long since it corresponds, for example, to the time that traffic on the road is to be studied.

In this period of time 7 the reading provided by the first and second sensor module is recorded. This signal can be a continuous signal although in the preferred embodiment the reading is carried out by means of a temporary sampling that gives rise to a discrete function.

The result of this reading is the definition of two functions defined in a time interval T, a signal corresponding to the first module and another signal corresponding to the second module.

When a vehicle passes under the sensor pair, the first sensor will be the first to record a variation of the signal. If there is no vehicle, the signal is the value of the distance to the track and if there is a vehicle, the distance is smaller since the height is reduced to the position of the sensor module.

If the vehicle that passes underneath does not change lanes, some time later it will also be registered by the second module. In practice, the signal shows multiple fluctuations. The method according to the present invention allows to distinguish between fluctuations due to noise and fluctuations due to the passage of a vehicle.

Both signals will show a peak of distance variation. The distance decreases since the distance from the height at which the sensor module is installed to the ground is greater than the distance from the same reference point to the upper surface of the vehicle.

Both signals, as discussed, also due to noise conditions (some of these sources have already been cited) show fluctuations that in practice can be high. • establish an average smi _med reference value for the smi signal of the first ultrasonic distance sensor module,

• set an average smo _med reference value for the second smo signal

ultrasonic distance sensor module,

The reference values smi _med and smo _med are the values taken as the signal level that corresponds to the distance from the location of the sensor module and the road surface.

These reference values make it possible to discern the relative amplitude of the fluctuations around said value, either by the presence of noise or because a vehicle passes under the sensor module that reduces the distance measurement. Signal fluctuations, seen as a function, give rise to local extremes (either local minimums of the function or local maximums of the function).

The extremes that correspond to the passage of a vehicle are minimal because the distance is reduced. However, an intermediate stage that changes the sign of the signal and a possible displacement would result in a maximum, hence the effective protection is obtained considering the general case given by a local end and this type of sign change or possible displacement particular ways of carrying out the invention are considered.

The local extremes are points and allow to identify the peaks of the function. When the fluctuation that gives rise to a local end is called a peak, not only will the local end be considered, but also the function section that determines the width of said peak. In the discrete case, the local end is represented by a single sample and the peak by a set of samples containing the previous one.

• set a first tolerance value ε ,,

• determine the set of local extremes, maximum or minimum, _{m m} with m = l..M, which exceed the tolerance value e¡ in absolute value in the signal smi with respect to the average reference value smi _med ,

Noise signals offer a very large number of fluctuations. Considering only the fluctuations that exceed a tolerance value, in this case ε ,, allows to significantly reduce the number of peaks that are to be considered candidates to be the response to the passage of a vehicle.

Throughout the entire interval 7, those peaks are distinguished (for the moment those local extremes) such that they are distanced from the va- The reference value is an amount greater than the tolerance value ε ,. This tolerance value must be calibrated.

According to the expression given by this technical rule, exceeding in absolute value the tolerance value must be interpreted as that the difference between the signal and the reference value, calculated in absolute value, is greater than ε ,. Throughout the description it will be understood that tolerance values are always positive values.

• set a second tolerance value ε ₀ ,

The signal of the second module can have a different tolerance value, it is enough that the reading is slightly scaled by the variability of the sensitivity in the manufacturing of the sensor module.

• set a maximum tracking time period T _ms for the second smo signal,

When a vehicle passes under a second sensor module, the step is always carried out after having passed through a first sensor module since the direction of the vehicle's travel along the road so imposes.

This implies that if in the first sensor module an end has been located that is taken as a candidate to be the response to the passage of a vehicle, it only makes sense to look for ends that correspond to the response to the passage of the same vehicle through a second sensor module in later moments and in a reasonable time. The reasonable time is defined by the maximum tracking time T _ms and will be determined by the minimum speed of passage of the vehicle to be considered.

• for each local end mi _m with m = l ... M determine the set of local extremes, maximum or minimum, mo _n with n = l ... N, which exceed the tolerance value ε ₀ in the smo signal, and that are located in the time interval T _ms which extends from the instant of time where the local end mi _m is located;

According to this technical rule, once a local end is taken in the signal of the first sensor module as a candidate to be the response to the passage of a vehicle, all ends that are in the signal of the second sensor module will be searched for in the period of time defined by the maximum tracking time T _ms and taken from the instant of time at which the local end is located in the signal of the first sensor module. It only remains to be seen that the peak associated with the signal end of the first sensor module is also reproduced in the signal of the second sensor module. where it is considered that the local end mi _m in the signal smi corresponds to the passage of a vehicle (V) if there is a correlation between the peak of the function associated with said end with at least one peak associated with the end m or _n of the set of local ends m _n with n = 1 ... N of the signal smo determined in the tracking time interval T _ms .

As indicated before, local extremes are only candidates. The local ends will be responses to the passage of a vehicle if the peak pattern corresponding to the local end of the first sensor module is reproduced in at least one peak of the signal of the second sensor module in the interval defined by T _ms . That both patterns are equal is determined because the correlation between the shape of one and the other peak is equal. Equality is also considered with a margin of tolerance and correlation can be assessed using one of the statistical correlation formulas.

The method according to the invention prevents false readings even in conditions of a lot of noise in the reading signal since the passage of the vehicle is only considered when peaks that do not exceed a threshold have been ruled out and have been corroborated by correlation with the second peak sensor.

Particular ways of carrying out the method according to any of the combinations given by dependent claims 2 to 13 are considered incorporated by reference to this description.

In particular, dependent claims 5 and 6 are considered of interest wherein the method of the invention is applied to tracks having several rails. In this case, several pairs of sensor modules are used, one pair per lane.

Each of the couples is able to determine the passage of vehicles passing through their lane. However, it is possible for a vehicle to change lanes just below the porch. In this case, the method according to claim 6 relates the peaks of the signals of pairs of sensors located in adjacent lanes. If a function peak of a first module taken as a candidate is not correlated with a signal function peak of the second module of the same pair then the correlation is sought in the signal of the second modules that are adjacent. If so, it is considered that a vehicle has passed and has done so by changing lanes.

It is also considered incorporated by reference to this description the system constituted by the physical devices that allow to carry out the method of according to claim 14. The most common way of carrying out this method by means of a system formed by a set of devices is by executing a computer program in a process unit. The computer program adapted to carry out any of the methods according to claims 1 to 13 and according to claim 15 is considered part of the invention.

DESCRIPTION OF THE DRAWINGS

These and other features and advantages of the invention will become more clearly apparent from the detailed description that follows in a preferred embodiment, given only by way of illustrative and non-limiting example, with reference to the accompanying figures. .

An exemplary embodiment of the invention is shown in this figure. The exemplary embodiment is represented by an elevation of a three-lane track section on which a gantry is placed in which three pairs of ultrasonic sensor modules are installed. The passage of a car without and with lane change is shown by two dashed arrows.

This figure shows an exemplary embodiment in schematic representation of a portico according to an elevation.

Figure 3 and 4 Figure 3 corresponds to a reading of the signal in a pair of ultrasonic sensor modules. The thin line is the smi signal of the first sensor module, the first one that detects the passage of a vehicle, and the thick line is the smo signal of the second sensor module. It is in figure 4 where an extension of a short interval taken from the representation of figure 3 is shown to explain a case of positive detection of the passage of a vehicle.

DETAILED EXHIBITION OF THE INVENTION

A detailed explanation of an embodiment of the invention is carried out with the support of the figures, in particular of Figures 1 and 2. Figure 1 shows a plan view of a track (1) having three rails ( 1.1). A vehicle (v) circulates according to the direction of circulation imposed by the route that in this case is represented upwards according to the position of the figure.

The vehicle (V) can change lanes (1.1) and it is also possible for more than one vehicle (V) to circulate in parallel. For clarity, only one vehicle (V) about to pass under a porch (2) is shown in the scene shown. The system that allows to execute the method according to an embodiment of the invention has three pairs of modules (mi, mo) ultrasonic distance sensors, one pair for each lane (1.1). The first ultrasonic distance sensor module (mi) is represented on the side of the gantry (2) that a vehicle (V) circulating is first, and the second ultrasonic distance sensor module (mo) is represented on the opposite side of the porch (2). The vehicle (V) will pass under this second sensor module (mo) an instant later depending on the vehicle speed. The vehicle speed (V), if it does not change lane (1.1), and under the constant speed hypothesis, it will be calculated as the separation distance (L) between modules (mi, mo) distance sensors divided by time what happens since it is detected in the first module (mi) sensor and the second module (mo) sensor.

If the vehicle (V) changes lanes (1.1) the speed can be estimated by increasing the distance traveled according to the hypotenuse of the triangle that forms the first module (mi) sensor and the second second modules (mo) sensors, one which it is in the same lane (1.1) and the other one is in the lane (1.1) adjacent to which the vehicle (V) has been changed. Figure 2 shows the elevation of the portico (2) schematically represented by lines. In the upper part of the gantry (2) the position of the first sensor module (mi) and the position of the second sensor module (mo) are shown at a distance (.). When there is no vehicle (V) under one of the modules (mi, mo) sensors, the measured distance (A) is maximum. The passage of any vehicle (V) below the module (mi, mo) sensor will reduce the distance (B) and therefore the signal that determines the distance will reduce its value. The passage of a vehicle (V), in the total absence of disturbances and with the use of modules (mi, mo) ultrasonic sensors with a perfect response would result in a descent peak with a configuration that would reproduce the vehicle profile (V) . In normal traffic conditions and with the modules (mi, mo) sensors usually found in the market this ideal situation is not possible. Figure 3 shows the reading of a first module (mi) ultrasonic sensor and a second module (mo) ultrasonic sensor measured in real conditions on a track (1).

The signal has local minimums and also maximums. These crazy ends may correspond to the passage of a vehicle but there are also peaks that correspond to disturbances in the reading by external factors. Just look at Figure 2 to see that the signal obtained in the first sensor module (mi) cannot exceed the value that corresponds to the distance to the track (1). However, in Figure 3, maximums are observed with amplitudes that reach values as high or higher than the amplitudes of peaks that do correspond to the passage of a vehicle.

Given this reality, the application of a cut filter would not solve the problem of how to determine whether a vehicle has passed or not with a reading influenced by noise such as that shown in Figure 3.

Figure 4 shows an extension of the same reading between the moments 19900 and 23900 (in ms). This graph shows peaks, up (maximum) and down (minimum).

In this particular case two pairs of peaks with a similar pattern at first sight are shown. One and the other has stood out by framing them in a rectangle of dashed lines.

The two peaks of the left frame do not correspond to the passage of a vehicle. The first peak, as the time variable advances, corresponds to the signal of the second module (mo). Although it resembles the peak that is next in the signal represented in a thin line, this is from the first module (mi) sensor.

According to the described method, once the reading has been carried out in a time interval T, the one shown in Figure 3, those having a local end that exceeds a predetermined threshold value, the value of tolerance ε ,. These peaks are measured with respect to the average value in both signals (smi, smo). These average values can be taken in several ways, among which count:

• numerically set the signal value corresponding to the actual height between the track and the sensor module,

• establish by calibration the signal value corresponding to the actual height between the track and the sensor module by readings without traffic,

• establish by calibration the signal value that corresponds to the actual height between the track and the sensor module by readings with traffic and calculating the average over the entire period T.

This last option does not require any external calibration intervention and when there is not a lot of traffic density it has been proven experimentally that it is a good estimate.

Once it has been determined which peaks of the signal of the first module (mi) sensor are candidates, the signal of the second module (mo) sensor is searched. In this case, the method will detect the second peak as the candidate, the one represented in a thin line. The comparison with the other signal will be carried out in a tracking period T _m s on your right, so you can never compare the peak with the one on your left. This stage takes into account the direction of travel of the vehicles (V) where vehicles that first pass through the first module (mi) sensor and then through the second module (mo) sensor are necessarily detected.

Now we will look at the box on the right. In this case the same pattern is observed in the two signals but it is in the thin line signal, that of the first module (mi) sensor, where the peak is located on the left. Experimentally it has been seen that this pattern has been searched for in the thick trace signal, to its right, and the correlation has been positive. The method has determined that both peaks correspond to the passage of a vehicle (V) through the lane (1.1) where the sensor pair (mi, mo) is located.

The time elapsed between one peak and another allows the vehicle's speed to be calculated.

It has already been commented that in the examples of realization and in the experimental tests the measurements are discrete with a certain sampling rate. Each of the peaks has a certain number of samples. It has been considered as an additional measure of rejection of candidate peaks in the smi signal those that have a number of samples less than 6 when the sampling rate is 50 ms. With this measure, those peaks with an energy density (area under the function) smaller than a certain amount are rejected.

From the configuration of the peak that has positively determined that a vehicle has passed (V), the average values, the extreme value (maximum if valued in absolute value), the width given by the number of points of the vehicle are used sample, and the standard deviation to characterize the type of vehicle.

A table with such values for different peaks identified in a functional test of the embodiment of the invention is shown below. The input index is the index of the vector that determines the sample number.

The index of the vector representing the discretized signal obtained from the first module (mi) sensor is denoted by index mi, the index for reading the second module (mo) sensor is denoted by index mo, the average is the average of the peak with respect to the reference level, "det. tip." is the standard deviation (as a measure of the dispersion), the height is the value of the tip of the peak in bsolute value, and "num" the number of samples comprising the peak.

The following data is highlighted for the input signal: index my average dev. tip num height

52 4,860 0,265 5,180 9

97 4,430 0,291 4,970 7

161 2,340 0,095 2,460 5

166 3,820 0.618 4.870 14

229 3,920 0.944 4,660 5

301 4,350 0.457 5,080 14

336 4,410 0,307 4,920 7

376 4,920 0.154 5,100 5

438 3,160 0,341 3,960 11

504 4,620 0,297 5,100 6

587 4,170 0,279 4,820 9 The rows that are highlighted with greater thickness are the signal peaks of the first sensor module (mi) that have corresponded with the signal peaks of the second sensor module (mo). The last column is the number of samples the peak contains.

Likewise, the following table shows the relevant data in the second sensor module (mo):

where the rows that have been correlated with the signal of the first module (mi) sensor have also been highlighted. This correlation has been obtained in four cases concluding that at least four vehicles (V) have passed.

The correlation data is shown in the following table for a correlation criterion of 0.95:

In this particular case a classification of the vehicle has been made based on the average height value and also its dispersion (for example measured by the standard deviation) with respect to the reference measures. Three classifications or groups have been used depending on the first measure, the average height value:

A. Heavy-duty vehicles with an average peak height value greater than 3.

B. Light-duty vehicles with an average peak height value in the range [2,3]; Y,

C. Cars: with an average peak height value less than 2.

These criteria have been obtained from a statistical study of the vehicle models currently available in the market.

It is possible to subclassify these groups in turn by using a measure of dispersion such as the standard deviation. For example, the standard deviation allows vehicles of type A to classify them in trucks or buses since the square profile of the buses makes the dispersion values of the samples associated with a vehicle (V) smaller. Its roof is of constant height.

Claims

1. - Method for characterizing road traffic along a traffic lane of one or more lanes comprising the following stages:

• provide a first module (mi) ultrasonic distance sensor and a second module (mo) ultrasonic distance sensor on the same lane (1.1) of the route (l) of circulation, distanced in track height (1) , where the second module (mo) ultrasonic distance sensor is separated a distance (L) downstream from the first module (mi) ultrasonic distance sensor; and both (mi, mo), are oriented in such a way that they measure the distance to the track (1),

• establish a measurement time period T,

• carry out a reading of the smi signal of the first module (mi) ultrasonic distance sensor and a reading of the smo signal of the second module (mi) ultrasonic distance sensor during the same time interval T,

• set an average smi _med reference value for the smi signal of the first module (mi) ultrasonic distance sensor,

• set an average smo _med reference value for the smo signal of the second module (mo) ultrasonic distance sensor,

• set a first tolerance value ε ,,

• set a second tolerance value ε ₀ ,

• set a maximum tracking time period T _ms for the second smo signal,

• for each local end mi _m with m = l ... M determine the set of local extremes, maximum or minimum, mo _n with n = l ... N, which exceed the tolerance value ε ₀ in the smo signal, and that are located in the time interval T _ms which extends from the instant of time where the local end mi _m is located; and, where it is considered that the local end mi _m in the signal smi corresponds to the passage of a vehicle (V) if there is a correlation between the peak of the function associated with said end with at least one peak associated with the end m _n of the set of local ends mo _n with n = l ... N of the signal smo determined in the tracking time interval T _ms .

2. Method according to claim 1 characterized in that:

• a confidence value e _{c is established} in the interval (0,1],

• the width of the function peak corresponding to the local end mi _m in the smi signal is defined, • the peak width of the function corresponding to the end of the mo _n smo signal is defined;

• the statistical correlation between both peaks is evaluated numerically and if this numerical value is greater than e _c then it is considered a correlation exists.

3. - Method according to claim 1 characterized in that the reading of the smi and smo signal is a time discretized reading giving rise to two vectors smi _k and smo _k .

4. - Method according to claim 2 and 3 characterized in that the correlation is carried out if the number of samples for each peak is at least 6.

5. - Method according to any one of claims 1 to 4 applicable to a track (1) having a plurality of rails (1.1) characterized in that additionally one or more pairs of modules [mi, mo] sensors are provided. ultrasonic distance, where:

• for each pair [mi¡, mo¡) of modules, j = l. J with J the total number of module pairs, there is a first module [mi¡) ultrasonic distance sensor and a second module [mo _j ) ultrasonic distance sensor on the same lane (1.1) of the route (1) of circulation , distanced in track height, where the second module [mo _j ) ultrasonic distance sensor is separated a distance (.,) downstream from the first module [mi!] ultrasonic distance sensor; Y,

• where the rails (1.1) containing pairs [mi¡, moj) of ultrasonic distance sensor modules are adjacent.

6. - Method according to claim 2 and 5 characterized in that, given the confidence value e _c , the correlation greater than e _c between the local end mi _m in the smi signal and the local end m _n of the smo signal in the interval Tracking time T _ms is searched for the signals obtained in each of the pairs [mi¡, moj) of modules, j = l..J, determining the number of vehicles (V) that pass through the track (1) as at least the number of correlations found that verify to be greater than the trust value e _c .

7. - Method according to claim 6 characterized in that, given a local end mi _m in the smi signal of a lane that does not verify having a correlation greater than e _c with any local end m _n of the smo signal in the time interval flush The distance T _ms of the same lane (1.1) is compared to the local extremes m _n of the signals of the adjacent lanes (1.1) also in the tracking time interval T _ms , so that if there is a correlation with any of such ends a vehicle (V) is considered to have passed and it has changed lanes (1.1).

8. - Method according to claim 7 characterized in that the total number of vehicles (V) that pass through the track (1) is determined as the number of vehicles (V) found that do not change lanes (1.1) plus the number of vehicles (V) found that change lanes (V).

9. - Method according to any of the preceding claims characterized in that for a vehicle (V) detected as having passed through the track (1), its speed is determined as the distance between the first module (mi) sensor the second module (mo ) sensor that has detected it divided by the time between peaks.

10. - Method according to any one of claims 1 to 8, characterized in that the average reference value in the input signal smi _med the average reference value in the output signal smo _med or both are established from the actual measurement between the track (1) and the corresponding sensor module (mi, mo).

11. - Method according to any of claims 1 to 8, characterized in that the average reference value in the input signal smi _med the average reference value in the output signal smo _med or both are set from the average of the signals obtained in each module (mi, mo) sensor for a period of time in which no vehicle circulates.

12. - Method according to any of claims 1 to 8 characterized in that for a vehicle (V) detected as having passed through the track (1), a classification by groups is carried out according to intervals of the average peak height associated to said vehicle (V) to group according to vehicle heights.

13. - Method according to claim 12 characterized in that for each group, a subclassification is established according to a measure of the statistical dispersion to group according to the shape of the vehicles.

14. - System comprising: one or more pairs of sensor modules of Ultrasonic distance and a process unit adapted to carry out any of the methods according to claims 1 to 13.

15. A computer program adapted to carry out any of the methods according to claims 1 to 13.