Method and system for measuring the speed of a vehicle, and recording medium for carrying out same

0001Method and system for measuring the speed of a vehicle, and recording medium for their use. The invention relates to a method and a system for measuring the speed of a vehicle, and to a recording medium for their implementation. More specifically, the invention relates to a method of measuring the speed of a vehicle using acoustic transducers aligned parallel to the same common axis, this method comprising a step of measuring noise so that each transducer generates an acoustic signal representative of the sound wave generated by the vehicle during its movement. Such a process is, for example, described in French patent application FR 2 812 402. The method described in FR 2 812 402 consists in calculating an intercorrelation function between two acoustic signals measured respectively by a first and a second acoustic transducer. The intercorrelation function makes it possible to estimate the delay time between the signal measured by the first transducer and that measured by the second transducer. This delay time is representative of the angular position of the vehicle with respect to the axis passing through the two acoustic transducers. More precisely, this delay time represents the value of the angle between a first straight line passing through the middle of the two transducers and the position of the vehicle and a second straight line perpendicular to the axis on which these two acoustic transducers are aligned. Therefore, to calculate the linear speed of the vehicle from the evolution of its angular position it is necessary to know the distance D which separates the vehicle traffic lane from the axis on which the two transducers are aligned. The speed thus measured can only be obtained with precision if, on the one hand, the alignment axis of the two transducers is strictly parallel to the traffic lane and if, on the other hand, the distance D is known with precision . During installation, these two conditions require positioning the two transducers relative to the traffic lane with great precision and measuring the distance D. In practice, this therefore complicates the installation of the transducers. The invention aims to remedy this drawback by proposing a method for measuring the speed of a vehicle using acoustic transducers which can be implemented without the distance separating the traffic lane from the acoustic transducers being known. The subject of the invention is therefore a method of measuring speed as described above, characterized: - in that at least three acoustic transducers are used, these transducers being grouped together to form at least a first and a second pair of transducers, the first and second pairs of transducers respectively defining a first and a second median plane perpendicular to the common axis, the first median plane being located midway between the two transducers of the first pair, and the second median plane being located midway between the two transducers of the second pair, and - in that the method comprises: a step of detecting a first and a second moment of passage at which the vehicle respectively crosses the first and second median planes from the noise measured respectively by the first and the second pair of transducers, and a step d 'establishment of the vehicle speed at least from the time interval between the first and second instants of passage and the distance between the first and second corresponding median planes. Since the median planes implemented in the above method are parallel, the time interval between the moment when the vehicle crosses the first plane and that when the vehicle crosses the second plane, is independent of the distance separating the vehicle automotive acoustic transducers. Therefore, this method can be implemented without knowing the distance D separating the vehicle traffic lane from the acoustic transducers. According to other additional characteristics of the measurement method according to the invention, it is characterized in that: - It includes a step of multiplying between them the two acoustic signals measured by each pair of transducers so as to obtain a plane signal representative of the noise power in the median plane defined by this pair, and in that the step of detection further consists in identifying the instant at which each plane signal is maximum, these instants when the plane signals are maximum corresponding to the instants of passage of the vehicle in the median planes; - It includes a step of filtering the acoustic signals used during the detection step to eliminate from these signals the frequency components having a wavelength greater than four times the distance between two acoustic transducers; the step of establishing the speed of the vehicle comprises an operation of establishing a regression line whose slope is proportional or inversely proportional to the speed of the vehicle; - It includes a step of identifying one or more pure sinusoids in one or more of the acoustic signals measured so as to detect the use of an audible warning device; - It includes a step of calculating the power of one or more of the acoustic signals measured and a step of comparing the or each calculated power to a preset threshold to detect the noise of a collision of the vehicle against an obstacle. - It includes a step of incrementing a vehicle counter when the speed of this vehicle is measured. The subject of the invention is also an information recording medium, characterized in that it includes instructions for implementing a method for measuring the speed of a vehicle according to the invention, when these instructions are executed by an electronic computer. The subject of the invention is also a system for measuring the speed of a vehicle capable of carrying out a method for measuring this speed according to any one of the preceding claims, this system comprising: - acoustic transducers aligned parallel to a common axis to measure the noise generated by the vehicle when it passes in front of these transducers, and - a computer capable of processing the measured acoustic signals to deduce the speed of the motor vehicle, characterized: - in that the system comprises at least three transducers, these transducers being grouped together to form at least first and second pairs of transducers, the first and second pairs of transducers respectively defining a first and a second median plane perpendicular to the common axis , the first median plane being located midway between the two transducers of the first pair, and the second median plane being located midway between the two transducers of the second pair, and - in that the computer is able to execute: a step of detecting a first and a second moment of passage at which the vehicle respectively crosses the first and second median planes from the noise measured respectively by the first and the second pair of transducers, and a step d establishment of the vehicle speed at least from the time interval between the first and second instants and the distance separating the first and second corresponding median planes. According to another additional characteristic of the measurement system according to the invention, it is characterized in that the transducers are distributed at regular intervals along the common axis. The invention will be better understood on reading the description which follows, given solely by way of example and made with reference to the drawings in which: - Figure 1 is a schematic illustration of the architecture of a system for measuring the speed of a vehicle according to the invention, - Figure 2 is a flow diagram of a method for measuring the speed of a vehicle according to the invention, and - Figure 3 is a graph representing the distance traveled by a vehicle as a function of measured transit times. FIG. 1 represents a system 2 for measuring the speed of a motor vehicle 4 moving on a taxiway 6. The vehicle 4 is a conventional vehicle which when it travels on a taxiway continuously generates a wave sound. The system 2 comprises at least three acoustic transducers connected to a unit 10 for acquiring and processing the acoustic signals measured by these transducers. Here, all the acoustic transducers of the system 2 are aligned on a common axis 12 substantially parallel to the axis of the track 6. In the particular embodiment described here, these acoustic transducers are arranged at regular intervals on this common axis 12. The distance separating two consecutive acoustic transducers is here denoted 2e. By way of illustration, system 2 comprises five acoustic transducers referenced from left to right by the symbols Ci to C₅. Preferably, the distance e is chosen as large as possible so as to improve the resolution of the system 2. However, to limit the size of the system 2, this distance e is chosen here as being between 5 and 15 centimeters. Each transducer Ci is able to measure the sound wave generated by the vehicle 4 during its movement and to transmit an acoustic signal sι (t) corresponding to the unit 10 for processing. In FIG. 1, seven dotted lines referenced from left to right by the symbols Pi to P₇ each represent a median plane between two transducers Cj of a pair of acoustic transducers. More precisely, each pair of transducers Ci defines the position of a median plane perpendicular to the axis 12 and located midway between these two transducers Cj. Here, since the transducers Ci are arranged at regular intervals along the axis 12, the same median plane can correspond to several different pairs of transducers. For example, the midplane P₄ corresponds to that defined by the pair of transducers C₂, VS₄ and the pair of transducers Ci, C₅. The distance between two consecutive parallel mid-planes is equal to the distance e. The unit 10 is able to acquire the various acoustic signals measured by the transducers Ci and to process these signals so as, moreover, to measure the speed of the vehicle 4. For this purpose, the unit 10 comprises an analog-to-digital converter 20 , suitable for converting analog signals Si (t) into digital signals. Typically, the sampling frequency of this converter 20 is of the order of 6000 samples per second and the samples of each acoustic signal Si (t) are measured at the same times so as not to introduce phase shift between these different acoustic signals . Then, the unit 10 includes a high pass filter 22 to eliminate digital acoustic signals, the frequency components which are not very sensitive to the direction of the sound wave generated by the vehicle 4. For this purpose, the cutoff frequency f_vs of this filter is chosen so as to eliminate the frequency components whose quarter of the wavelength is greater than the distance 2e separating two consecutive transducers. So the cutoff frequency f_vsis chosen so as to respect the following relationship: <img file="WO2005073736A2_D0001.tif" /> where it is the speed of sound in the air. Here, the cutoff frequency f_vs is equal to c / 8th. Indeed, the frequency components whose wavelength is greater than four times the distance 2e between transducers always arrive with a phase shift less than τr / 2 whatever the position of the vehicle. They therefore correspond to a product Pι (t), explained below, always positive and therefore form a background noise which deteriorates the precision of the speed measurement if it is not eliminated. The acoustic signal Sj (t) after being filtered by the filter 22 is noted

0002A multiplier 24 is provided in the unit 10 to multiply between them the signals shj (t) coming from the two acoustic transducers Cj and Ck of the same pair of acoustic transducers. At output, the multiplier 24 therefore delivers n (n-1) / 2 signals, here called plane signals P | (t), obtained by the following relation:<img file="WO2005073736A2_D0002.tif" /> n is the number of transducers. The index I corresponds to a number identifying the pair of transducers from which the signal P | (t) is constructed. A low-pass filter 26 is connected to the output of the multiplier 24 so as to smooth the plane signals P | (t) obtained. This filter 26 is capable of delivering at output n (n-1) / 2 filtered plane signals Qι (t) obtained by calculating the sliding average of each plane signal P | (t) over a time interval T. T is for example chosen equal to 5 to 20 times the sampling period. The signal Qι (t) is therefore for example obtained using the following relation: 1 Qι (t) ∑₃P '(t + d) where d is a variable which varies in incremental steps equal to the sampling period from -T / 2 to T / 2. Finally, the unit 10 comprises an electronic computer 28 capable of processing the signals Qι (t) so as to further determine the speed of the vehicle 4 as well as other information, as described with reference to FIG. 2. To this end , the computer 28 is capable of carrying out the method of FIG. 2. The computer 28 is also able to control various peripheral devices associated with the unit 10 such as, for example, a camera 40 and a radio transmitter / receiver 42. The computer 28 is, for example, a conventional programmable computer able to execute instructions recorded in a memory 44. Here, the memory 44 includes the instructions necessary for the execution of the method of FIG. 2. The camera 40 is suitable for photographing the vehicle 4 and in particular its license plate during its passage. The radio transmitter / receiver 42 is intended to transmit, via a wireless network, information acquired and processed by the unit 10 to other equipment. By way of illustration, here, the transmitter / receiver 42 is capable of directly transmitting control instructions to a red light 48 installed near the unit 10. The transmitter / receiver 42 is also capable, here, of transmitting the information acquired and processed to a center 52 for road traffic management. The operation of the system 2 will now be described with reference to FIG. 2. During its operation, each transducer Ci continuously measures, during a step 60, the sound waves coming from the channel 6 and transmits the acoustic signal Sj (t) corresponding to unit 10. The converter 20 digitizes, during a step 62, the set of acoustic signals Si (t) transmitted by the transducers Cj. The digitized acoustic signals are filtered, in a step 64, by the filter 22 and the signals obtained shj (t) are transmitted to the multiplier 24. In a step 66, the multiplier 24 calculates continuously for each pair of transducers Ci the corresponding plane signal Pι (t). Here, the multiplier 24 therefore continuously establishes ten signals P | (t) since there are ten possible pairs of transducers in the system 2. Each signal Pι (t) is then filtered, during a step 68, by the filter low pass 26 so as to obtain the signal Qι (t). Steps 60 to 68 are carried out continuously and permanently for each signal Sj (t) measured. From each signal Qι (t) the computer 28 detects during a step 70, the instant of passage tpi of the vehicle 4 in the median plane defined by the pair of transducers corresponding to the index I. It will be recalled that l 'index I corresponds to the number of a pair of transducers and not to the index of a median plane. The instant of passage in a median plane defined by a pair of transducers Ci corresponds to the instant when the plane signal has a significant power or energy peak. Indeed, when the vehicle 4 crosses a median plane, the sound wave which it generates at this instant takes exactly the same time to reach the two transducers Ci of the pair of transducers which defines this median plane. Thus, the plan signal presents a maximum at this instant there. For example, to determine the instant of passage tpi, the property is used according to which the value of this instant of passage tpi is the value which minimizes the dispersion of the values ​​of the signal Qι (t) over a time window W around this instant . The value of the instant tpi is therefore, for example, given by the following relation:<img file="WO2005073736A2_D0003.tif" /> where: - d is a variable varying in incremental steps equal to the sampling period over the interval [- W / 2; W / 2], W being the amplitude of the time window considered, and - 1 is time. The index I of the instant tpi, is equal to the index I of the corresponding signal Qι (t) and therefore corresponds to a pair of known transducers Ci. The value W for the time window is chosen as a function of the average speed of the vehicles observed on track 6. Note that since in this embodiment, several pairs of transducers Cj can define the same median plane, several instants of passage in the same median plane are calculated from signals Qι (t) different. It is therefore understood that there is a redundancy of the measured instants of passage which increases the robustness of the method described here with regard to parasitic noise such as, for example, with respect to the noise generated by fixed reflecting elements. the sound wave of the vehicle 4. Then, the computer 28 establishes, during a step 72, for each instant tpi a confidence index Sb |. This confidence index is established, for example, using the following relationship:<img file="WO2005073736A2_D0004.tif" /> where B | corresponds to the power of a residual noise. Residual noise B | is the noise measured by the pair of transducers of index I between two motor vehicle passages on track 6. Typically, the power of the residual noise is estimated by establishing the average of the minima of the signals Qι (t) over a longer period than the time window W. For example, this period will be at least 5W. The computer 28 then establishes, during a step 74, the speed of the vehicle 4 from the instants of passage tpi and the distances separating the different median planes. By way of illustration, in the graph in FIG. 3, the round dots represent the position of the vehicle at time tpi. This position is deduced from the index I. In fact, each index I is associated with a pair of transducers Ci and therefore corresponds to a known median plane. Here, the position of each median plane is measured from the position of the Ci transducer. Therefore, the median plane PK is at the distance ke from the Ci transducer. If the measurements and calculations were perfect, the round points would be aligned on the same straight line whose slope is equal to the speed of the vehicle 4. However, due to measurement errors and various parasitic phenomena, the round points are never, in reality, perfectly aligned. Consequently, here, to determine the speed of the vehicle 4, the computer 28 determines, during an operation 76, the regression line passing as close as possible to these round points. This regression line corresponds to the square points in FIG. 3. This line is determined by a known method such as, for example, the method of least squares. The slope of this straight line then corresponds to the speed measured by system 2. Here, the measured speed v is established using the following formula: life = 12∑ _kk.tp_k /[(P-Ï)P.(P+Ï)oquer-6.∑ _k t_{P ζ} / [(-1). ] where: - P is the number of planes, - k is an index varying from 1 to P, and -tp is the average of the instants tpi measured for the same plane P. It will be noted that since the distance between the planes is measured by taking the position of the transducer Ci as a benchmark, the speed is positive when the vehicle 4 moves from the transducer Ci towards the transducer C₆and negative otherwise. The computer 28 also performs the calculation, during a step 80, of a confidence index Sb for the speed calculated during the step 74. This confidence index Sb is calculated using the following relation: Sb = ∑_I Sb_l / P where I is an index varying from 1 to P. As the time window W moves, the value of this confidence index Sb reaches a maximum before decreasing. During a step 82, the computer stores the speed of the vehicle 4 calculated, for example, during step 74 only when this confidence index Sb is maximum. During this step 82, the computer also stores the passage of this vehicle and, for example, increments a vehicle counter. Then, during a step 84, the computer compares the stored speed with a speed threshold S_v predefined. If the memorized speed exceeds the threshold S_v then, the computer controls, during a step 86, the camera 40 so as to take a photo of the vehicle 4. At regular intervals, the computer 28 transmits, during a step 88, the various information stored in the center management 52 via the transceiver 42. This computer 28 also uses, if necessary, the information stored to control the activation of the red light 48, during a step 90. Steps 70 to 84 are carried out continuously and continuously. In parallel with these steps, the computer performs a step 100 of detecting an accident such as, for example, the collision of the vehicle 4 against an obstacle. To this end, the computer 28 filters, during an operation 102 one or more signals Sj (t) using a high pass filter intended to allow the shock wave to pass. Then, the instantaneous power of the or each filtered acoustic signal is calculated, during an operation 104, then compared, during an operation 106, with the average M_at of the instantaneous power of the same signal over at least one second. If the instantaneous power of one of the signals is greater than at least twice M_at, then the computer proceeds to an operation 108 of indicating this event. Typically, during operation 108, the information according to which an accident has occurred is transmitted via the transmitter / receiver 42 to the management center 52. Also in parallel with step 100, the computer 28 performs a step 110 of identifying one or more pure sinusoids in the signals Si (t) to detect an audible warning. To this end, the computer 28 calculates, during an operation 112, the correlation, for a time interval greater than one second, between one or more of the acoustic signals Sj (t) and N pure sinusoids whose frequencies are regularly distributed , for example, between 200 and 400 Hertz. For this purpose, the computer 28 uses, for example, the following relationship:^pi (^ω_m) = <img file="WO2005073736A2_D0005.tif" />_t S₇(.cos (ω_mt)} where: - ω_m is the pulsation of the pure sinusoid of index m, - If (t) is the acoustic signal of the transducer of index i - Pi (ω_m) is the value of the correlation between the pulsating sinusoid ω _m and the signal S_j^(t). Then, each signal Pj (û>_m) is compared during an operation 114 to a predetermined threshold S_z. If the acoustic signal is very strongly correlated with one of the N pure sine waves, then the value of the signal Pi (ω_m) is greater than the threshold S_z and the computer proceeds to an operation 116 of discrimination between an audible warning of a priority vehicle and that of any other vehicle. Otherwise, the method returns to operation 112. During operation 116, the computer 28 searches in a history for the occurrence of a detection of a buzzer persistent for S consecutive seconds corresponding to a first pulse ωi alternating with a second pulse ω₂. If so, the computer transmits, during an operation 118, to the management center 52 information according to which the audible warning of a priority vehicle has been detected. Otherwise, it is simply a vehicle horn and the computer 28 then proceeds to a step 120 during which, for example, the information that an audible warning of a motor vehicle has been detected is memorized then transmitted to the management center 52. Since the median planes defined by the transducers Ci are parallel to each other, it is not necessary using the method of Figure 2 to know the distance between the transducers Ci and track 6 to measure the linear speed of the vehicle 4. In addition, unlike known systems, it is not necessary for the axis 12 to be strictly parallel to the axis of the track 6. In fact, a slight alignment error has only slight repercussions. on the measured speed. The table below gives some examples of the influence of an alignment error on the speed measurement. The alignment error is here measured between axis 12 and the axis of track 6.

0003<img file="WO2005073736A2_D0006.tif" /> System 2 also has the advantage of being insensitive to the height at which it is placed relative to track 6. System 2 has been described in the particular case where only one line of transducers Ci is used so as to reduce clutter. As a variant, the system comprises several lines of transducers Ci. In this variant, the method of FIG. 2 is executed for each line of transducers Cj. The transducers Ci have been described here as being aligned. However, this is not essential to define several median planes parallel to each other. For example, several pairs of transducers Ci can be offset in height with respect to each other while defining each of the median planes parallel to each other. For such an arrangement, the method of FIG. 2 is not modified. System 2 has been described in the particular case where it only comprises five transducers. However, as a variant, in order to increase the resolution and the precision of the speed measurement and to limit the consequences of parasitic noise, the number of transducers Ci is chosen to be greater than 10, 20 or 30. In fact, the more the number of transducers is high, the greater the number of instants of tpi passages calculated and the greater the precision and the insensitivity to parasitic noises of the system. It will be noted in particular that the number of passage times calculated increases exponentially with the number of transducers Cj and not linearly. It is therefore possible to calculate a large number of passage times using a reduced number of acoustic transducers. The system 2 has been described in the particular case where the transducers are distributed at regular intervals along the axis 12. This simplifies the calculations. However, as a variant, other distributions of the acoustic transducers Ci along the axis 12 are possible. It is then necessary to adapt the calculation formulas exposed here to the layout of the chosen acoustic transducers. The system has been described here in the particular case where it comprises a low-pass filter 26. However, as a variant, this low-pass filter 26 is eliminated and the corresponding step 68 of the method of FIG. 2 is also eliminated. The other steps remain unchanged. Alternatively, to suppress echoes due to the reflection of the sound wave generated by the vehicle on fixed elements, the unit 10 includes an additional filter placed between the filter 22 and the multiplier 24. This additional filter is an adaptive filter . In this variant, the method of FIG. 2 comprises an additional step, carried out by the adaptive filter, inserted between steps 64 and 66. During this additional step, the signal shi (t) is filtered by the adaptive filter which implements for this purpose, for example, the following relationship: uh_t (t) = sh_t (t) - ak.uh_t (t - kd) where: - Uhi (t) is the signal delivered by the adaptive filter at the input of the multiplier 24, - ak are coefficients which represent the proportion of reflected sound waves which arrive with a delay equal to k times the sampling period d, and

0004- k is an index which varies between 1 and, where dmax is the cd maximum distance between the transducers and the parasitic reflecting elements to be taken into account. The coefficients ak are calculated by conventional methods which minimize, for example, the quadratic error and using algorithms such as those of Levinson and Durbin, that of the least squares or that of the stochastic gradient. The other steps of the method of FIG. 2 remain unchanged except for the fact that it is the signal uhj (t) which is processed instead of the signal shj (t). As a variant, during step 100, the computer 28 uses a fast Fourier transform to identify one or more pure sinusoids in the signals Sj (t). The system has been described in the particular case of measuring the speed of a motor vehicle. However, system 2 can be applied to the measurement of the speed of any object generating a sound wave during its movement.