Method for controlling observation by means of a space tracking system and associated device

Abstract

Method for controlling observation by a tracking system of a space and associated device. The present invention relates to a method for controlling observation by a tracking system (10) of a space, the method being implemented by a control module (22) forming part of the tracking system (10) and comprising: - a step of forming thumbnails, each thumbnail gathering a set of data accessible to the tracking system (10) for a respective area and a predefined time interval, the areas associated with each thumbnail tiling the space observed by the tracking system (10) and the set of predefined time intervals covering an observation time interval, and - a step of controlling observation by a tracking system (10) of the space corresponding to at least one thumbnail by applying a control function to the set of data of at least one thumbnail. Figure for the abstract: Figure 2

Description

Title of the invention: Method for controlling observation by a space tracking system and associated device

[0001] The present invention relates to a method for controlling observation by a space tracking system. The invention also relates to an associated control module and tracking system.

[0002] The proposed invention lies in the field of air traffic control.

[0003] Airspace control is ensured by monitoring all aircraft. It This involves supervising air traffic to prevent collisions between aircraft and controlling traffic, both in cruising flight and around airports - takeoff and landing.

[0004] To implement such control, air traffic controllers use an air traffic monitoring system based on aircraft tracking, which is called a tracking system. The tracking system is capable in real time of estimating from data coming from sensors the best possible estimate of the position, heading and speed of each aircraft in a given flight information region.

[0005] The flight information region is often designated by the acronym FIR referring to the corresponding English name of “Flight Information Region”.

[0006] The set of estimated data forms a track.

[0007] The quality of the track estimation (tracking) varies depending on the configuration of the tracking system, the quality of the data acquired by the sensors or even environmental data such as the topology of the terrain or the weather.

[0008] In this respect, it is desirable for an air traffic controller to know the quality of the estimation of the tracks reconstructed by the tracking system.

[0009] For this, it is known to use the requirements of the ESASSP standard which is detailed in the document entitled “EUROCONTROL Specification for ATM Surveillance System Performance” (Volume 1) whose ISBN is 978-2-87497-022-1 and which was published in March 2012.

[0010] More specifically, a set of requirements defined by this ESASSP standard is evaluated by comparing the outputs of the tracking system with an estimate of the ideal trajectory of the aircraft.

[0011] The ideal trajectory is therefore a reconstructed trajectory that can only be calculated offline.

[0012] Therefore, such an evaluation only provides an a posteriori evaluation of the quality of the evaluation of the tracks by the tracking system on the flight information region (FIR).

[0013] There is therefore a need for a method for evaluating the quality of tracking carried out by a tracking system.

[0014] To this end, the description describes a method for controlling observation by a tracking system of a space, the method being implemented by a control module forming part of the tracking system, the control method comprising:

[0015] - a step of forming thumbnails, each thumbnail bringing together a set of data accessible to the tracking system over a respective area and a predefined time interval, the areas associated with each tile paving the space observed by the tracking system and all of the predefined time intervals covering an observation time interval, the data comprising at least sensor data, and

[0016] - a step of controlling the observation by a cor space tracking system corresponding to at least one vignette by applying a control function to all the data of the at least one vignette.

[0017] According to particular embodiments, the control method has one or more of the following characteristics, taken in isolation or in all technically possible combinations:

[0018] - the data set of each vignette includes a trajectory re built.

[0019] - the tracking system outputs calculated data, all of the data of each thumbnail including the calculated data.

[0020] - the control function is a function of detecting the possible presence of a anomaly in the vignette.

[0021] - the anomaly detection function is obtained by a learning procedure, the learning procedure including:

[0022] - learning a vector representation of the thumbnails, and

[0023] - obtaining an anomaly detection function from the representation learned vector.

[0024] - the learning step comprises learning a first sub-function to starting from a labeled dataset to obtain a first learned sub-function suitable for implementing a pretext task, the first learned sub-function being a neural network comprising a plurality of layers of neurons, the vector representation being the penultimate layer of the first learned sub-function.

[0025] - the first subfunction is a residual neural network.

[0026] - the obtaining step is implemented using support vector machines at single class.

[0027] - the obtaining step is implemented using a neural network adapted for measure the distance from a tile to a set of tiles considered as normal.

[0028] - the control method comprises a step of identifying a possible cause of the presence of an anomaly by applying an identification function to the vignette(s) in which the presence of an anomaly was detected at the implementation stage, the identification stage being carried out by the anomaly correction module.

[0029] - the control method includes a test of a corrective action associated with the cause identified.

[0030] - the tracking system is capable of collecting data from several sensors, the cause being a failure of a sensor and the corrective action being the suppression of the consideration of data from the sensor exhibiting the failure.

[0031] - the identification step comprises:

[0032] - the generation of thumbnails corresponding to the data accessible to several sub- separate sets of sensors,

[0033] - the detection of anomalies in the thumbnails generated by implementing the method detection, and

[0034] - the deduction of the sensor(s) causing the anomaly.

[0035] - the control function is a function for calculating the performance of the system of tracking.

[0036] The description also describes a module for controlling observation by a space tracking system, the control module being suitable for:

[0037] - forming thumbnails, each thumbnail gathering a set of the data ac transferable to the tracking system over a respective area and a predefined time interval, the areas associated with each tile paving the space observed by the tracking system and all of the predefined time intervals covering an observation time interval, the data comprising at least sensor data, and

[0038] - control the observation by a space tracking system corresponding to at at least one thumbnail by applying a control function to all the data of the at least one thumbnail.

[0039] The description also proposes a tracking system provided with a control module as previously described.

[0040] In the present description, the expression “suitable for” means indifferently “adapted for”, “adapted to” or “configured for”.

[0041] Characteristics and advantages of the invention will appear on reading the description which follows, given solely by way of non-limiting example, and made with reference to the appended drawings, in which:

[0042] - [Fig.l] [Fig.l] is a schematic representation of an example of a system of tracking in interaction with a set of sensors,

[0043] - [Fig.2] [Fig.2] is a block diagram of the implementation of an example of a method of detection of anomalies in a space observed by the tracking system,

[0044] - [Fig.3] [Fig.3] is a representation of simulations of implementation of the method anomaly detection according to [Fig.2], and

[0045] - [Fig.4] [Fig.4] is a schematic representation of another example of a system tracking in interaction with a set of sensors,

[0046] [Fig.l] schematically illustrates a tracking system 10 interacting with a set 12 of sensors.

[0047] The tracking system 10 is capable of observing an airspace to determine the trajectories of aircraft passing through an observed space.

[0048] This makes it possible to carry out air traffic control in the area in question.

[0049] For this, the tracking system 10 is capable of collecting data from all of the sensors and analyzing the collected data.

[0050] The tracking system 10 outputs calculated data.

[0051] The speed, position and heading of each aircraft passing through the observed space are examples of calculated data.

[0052] Advantageously, to these values, the tracking system 10 adds data making it possible to associate each new point calculated with an existing trajectory.

[0053] The set 12 of sensors is capable of obtaining data in the observed space.

[0054] According to the example described, the set 12 of sensors comprises an ADS unit 14, a WAM unit 16 and a radar 18.

[0055] The ADS 14 unit is a cooperative surveillance system for air traffic control and other related applications. An aircraft equipped with an ADS 14 unit determines its position by a global positioning system (GPS) and periodically sends this position and other information to ground stations.

[0056] The abbreviation ADS refers to the corresponding English term for “Automatic Dependent Surveillance” literally meaning “automatic dependency monitoring”.

[0057] Such an ADS unit 14 is sometimes also called a B-unit, the abbreviation ADS-B referring to the corresponding English name of “Automatic Dependent Surveillance-Broadcast” literally meaning “automatic dependency surveillance-multicast”.

[0058] A WAM unit 16 uses data from multiple sensors to obtain the location of an aircraft.

[0059] The abbreviation WAM refers to the corresponding English term "Wide Area Multilateration" literally meaning "wide area multilateralization" and designates an aircraft surveillance technology based on the principle of the difference in arrival time which is used at an airport.

[0060] For example, the WAM unit 16 collects data from several ground antennas to apply mathematical calculations to obtain the position of the aircraft.

[0061] A radar 18 makes it possible to detect the presence of aircraft in the sky and to determine their position. The radar 18 emits electromagnetic pulses into the sky, and the detection and location of an aircraft are obtained by analyzing the wave reflected by the aircraft and retransmitted in the direction of the radar 18.

[0062] According to the example of [Fig.l], the tracking system 10 comprises an analysis module 20 and an anomaly detection module 22.

[0063] The analysis module 20 is capable of analyzing the data from the sensors to calculate new data.

[0064] In particular, the analysis module 20 is capable of predicting the trajectory of an aircraft.

[0065] Such an analysis module 20 is often referred to by the English term “tracker”, literally meaning “tracker”.

[0066] For this, the analysis module 20 uses a data fusion technique generally involving a Kalman filter.

[0067] The fusion technique depends on parameters representative of the environment, such as sensor noise.

[0068] These parameters are set according to the environment usually encountered. For example, if the sensor considered most often presents Gaussian noise, parameters corresponding to such noise will be set.

[0069] The anomaly detection module 22 is capable of implementing the steps of a method for detecting anomalies in the observed space.

[0070] For this, according to the proposed example, the anomaly detection module 22 comprises a training unit 24 and a detection unit 26 whose roles will appear in the remainder of the description.

[0071] An example of implementation of the anomaly detection method is now described with reference to [Fig.2].

[0072] The anomaly detection method aims to detect anomalies in the space observed by the tracking system 10.

[0073] Detection is here to be taken in a broad sense in the sense that it can simply be determined a probability that there is an anomaly in a part of the observed space.

[0074] Such a probability will be expressed here in the form of a normality score, it is a value representative of the probability of the presence of an anomaly.

[0075] An anomaly corresponds to a degradation of the quality of tracking by the tracking system 10.

[0076] Therefore, the normality score can also be interpreted as a score assessment of the quality of tracking carried out by the tracking system 10.

[0077] The causes of such degradation of the tracking system 10 can be multiple. This can come from a meteorological or external problem such as a very strong storm, a frozen radome of a radar at altitude, a solar flare.

[0078] It can also be a sensor-related problem, such as an O-ring problem on a rotating radar (creates an azimuth shift on the positions sent by the radar) or an abnormal noise problem on the data from a sensor.

[0079] The cause of the degradation may be a cyber attack in which false sensor data is sent.

[0080] According to yet another example, the problem may arise from a malfunction of the analysis module 20.

[0081] Of course, a degradation may correspond to the presence of several of the examples of causes mentioned above.

[0082] The detection method comprises a training step and a detection step.

[0083] During the thumbnail formation step, the detection module 22, and more precisely the formation unit 24, forms thumbnails.

[0084] This step is shown diagrammatically by a block 30 in [Fig.2].

[0085] A vignette brings together a set of data accessible to the tracking system 10 over a respective area and a predefined time interval.

[0086] The data comprises at least data from one or more sensors of the set 12 of sensors.

[0087] It is assumed for the following that the data of the vignette are only the data of the ADS unit 14, the WAM unit 16 and the radar 18, which allows a rapid collection of these values.

[0088] However, the data set may include other elements.

[0089] According to one example, the data set of each vignette comprises one or more reconstructed trajectories.

[0090] According to another example, all of the data of a vignette includes the data calculated by the tracking system 10.

[0091] Alternatively, the set of data of a vignette comprises the data from the sensors, the data reconstructed by the tracking system 10 and the data calculated by the tracking system 10.

[0092] The set of vignettes covers the space observed in time and space.

[0093] Thus, the zones associated with each thumbnail pave the space observed by the system tracking 10 and the set of predefined time intervals covers one observation time interval.

[0094] By way of example and without this being limiting, the sticker may correspond to an area of ​​a few dozen kilometers and a time interval of one hour.

[0095] A thumbnail thus corresponds to a limited volume of space (cell) over a finite time period. A thumbnail is therefore a spatially and temporally local vision of the sensor data and possibly of the tracking carried out by the tracking system 10.

[0096] It is also possible to limit a vignette to a two-dimensional space since aircraft fly at altitudes fixed by air corridors.

[0097] In such a case, only latitude and longitude are taken into account.

[0098] At the end of the training step, the training unit 24 thus has a set of thumbnails.

[0099] During the detection step, the detection module 22 detects the possible presence of anomalies in the vignette.

[0100] For this, the detection unit 26 of the detection module 22 applies an anomaly detection function to all of the data in the vignette.

[0101] Such an anomaly detection function is shown schematically in [Fig.2],

[0102] This anomaly detection function is obtained by a learning procedure.

[0103] The learning procedure comprises a step of learning a vector representation of the thumbnails.

[0104] In this context, a vector representation corresponds to the projection of elements of (space of images having c colors, n pixels in their length and m pixels in their width) into a space of R"' (vector space of dimension d).

[0105] Such a representation makes it possible to reduce the number of variables necessary to express the information and to be able to manipulate this data as vectors, thus giving access to known mathematical tools such as distance calculation. This results in a simplified implementation. The function giving the representation corresponds to block 32 in [Fig.2] and the learning is shown diagrammatically by part 34 in [Fig.2].

[0106] The learning step here consists of learning a first sub-function which is a neural network.

[0107] The neural network comprises an ordered succession of layers of neurons, each of which takes its inputs from the outputs of the previous layer.

[0108] More precisely, each layer comprises neurons taking their inputs from the outputs of the neurons of the previous layer, or from the input variables for the first layer.

[0109] Alternatively, more complex neural network structures can be envisaged with a layer that can be connected to a layer further away than the immediately preceding layer.

[0110] Each neuron is also associated with an operation, i.e. a type of processing, to be carried out by said neuron within the corresponding processing layer.

[0111] Each layer is connected to the other layers by a plurality of synapses. A synaptic weight is associated with each synapse, and each synapse forms a connection between two neurons. It is often a real number, which takes both positive and negative values. In some cases, the synaptic weight is a complex number.

[0112] Each neuron is capable of performing a weighted sum of the value(s) received from the neurons of the previous layer, each value then being multiplied by the respective synaptic weight of each synapse, or link, between said neuron and the neurons of the previous layer, then applying an activation function, typically a non-linear function, to said weighted sum, and delivering at the output of said neuron, in particular to the neurons of the following layer which are connected to it, the value resulting from the application of the activation function. The activation function makes it possible to introduce a non-linearity into the processing carried out by each neuron. The sigmoid function, the hyperbolic tangent function, the Heaviside function are examples of activation functions.

[0113] As an optional addition, each neuron is also capable of applying, in addition, a multiplicative factor, also called bias, to the output of the activation function, and the value delivered at the output of said neuron is then the product of the bias value and the value from the activation function.

[0114] As a specific example, the first subfunction here is a residual neural network.

[0115] A residual neural network is a neural network for which at least one neuron of a layer interacts with a neuron of a non-neighboring layer.

[0116] Such a network is often referred to by the corresponding abbreviated English name, namely ResNet.

[0117] In this case, the learning of the first sub-function is implemented from a labeled data set to obtain a first learned sub-function capable of implementing a predefined task.

[0118] The vector representation is then the penultimate layer of the first sub-function learned.

[0119] Thus, a predefined task is used that has no connection with the task of determining the presence of anomalies. The predefined task is only used to enable the learning of the free parameters of the neural network that are defined through the learning operation. In this sense, the predefined task can be referred to as a pretext task, a term that will be used in the following.

[0120] In this sense, it can be considered that training on the pretext task is a pre-

[0121]

[0122]

[0123]

[0124]

[0125]

[0126]

[0127]

[0128]

[0129]

[0130]

[0131]

[0132]

[0133]

[0134]

[0135] workout to achieve pre-trained weights. In the described example, the first sub-function was trained on a classification task. Typically, the first sub-function is trained to recognize one or more elements in an annotated database such as an ImageNet-type image database. According to a first example corresponding to an unsupervised framework, the weights of the first sub-function thus pre-trained are kept as is. According to a second example corresponding to a supervised framework, the weights of the first sub-function thus pre-trained are refined by learning on a specialization task. In this case, a specialization task is a task specific to the field of air traffic control. What has just been described can be formulated more formally as follows. The first sub-function is a neural network according to the ResNet architecture which is the composition of k linear operations parameterized by the learning weights of the neurons 0 = (W? b^ as well as k non-linear operations (activation function). It comes like this: V i fi(zi_^ = 'tp[wiziA + bi+ziS v = ïUv) = ft° ... » / / ... « / ,(4 Or: • i is an integer • fj is a parametrized linear operation, • 4-1 is the vector of neurons of layer i - 1, • Wj is the weight matrix of the linear operation of layer i, ®= .....J • bi is the bias of the linear operation of layer i, • y is the output of the neural network, and • 'î7© is the function parameterized by ® formed by the neural network, and • 0 denotes the mathematical operation of composition. The first sub-function is trained on a classification task using a dataset labeled [ / \ ] by minimizing a criterion. D-yl H P z 7 d The chosen criterion is, for example, the following: 0 = tnin^^ (yT / x))

[0136] where L is a cost function aimed at estimating the distance between the predicted class and the labeled class.

[0137] The criterion is minimized by updating the weights at each iteration by back-propagation.

[0138] As previously indicated, once this task is learned, the vector representation of an image is obtained by removing the last layer of the network and freezing the weights of the network.

[0139] Noting V the vector representation operator of a vignette x, it comes with the previous notations:

[0140] V ^Rd x = fk{° ... °

[0141] d here denotes a parameter that can vary. For example, a value of 2048 was used by the applicant.

[0142] The learning procedure also includes a step of obtaining an anomaly detection function from the learned vector representation.

[0143] The detection function is capable of providing a normality score from the vector representation of a thumbnail.

[0144] In the obtaining step, the anomaly detection function is learned by a learning technique.

[0145] According to a first example of implementation, a second sub-function is trained on the vector representation of a subset of thumbnails without anomaly.

[0146] According to the example of [Fig.2], the second subfunction is a single-class support vector machine and corresponds to block 36.

[0147] Such an element is more often referred to as “One-class SVM” referring to the corresponding English name of “One-class Support-Vector Machine”.

[0148] A specific example of a technique for obtaining a normality score is now described.

[0149] This technique seeks to project the observed points into a higher-dimensional space and find a hypersphere that contains almost all the points. Points inside the sphere are considered normal and those outside are considered anomalies.

[0150] The representation corresponding to the image of projector 0 and the hypersphere are obtained by solving the following optimization problem:

[0151] mjn r2+ J- Vd 11 sc ^xeD. i^n, || ¢(^(^)) -c || " <r 2 +

[0152] Where: • <p'.Rd^F est l’application de projection (projecteur), d est la dimension de l’espace d’entrée (dans notre cas, la dimension de la représentation vectorielle des vignettes V) et F l’image du projecteur (sa dimension est vue comme un hyper paramètre). On impose seulement qu’un produit scalaire d’éléments de F puisse être exprimé sous la forme d’un noyau sur les éléments de R67. • r is the radius of the hypersphere, • v is a VG weighting [0,1] considered as a hyper parameter. • sc- denotes the expression under the constraint of, • c is the position of the center of the hypersphere, • £ is a relaxation variable of the boundary of the hypersphere allowing to accept points outside the hypersphere but very close to its border.

[0153] This problem can be reformulated through the dual Lagrange function, by introducing a, P as Lagrange coefficient:

[0154] g^ p} = mfL(r, f, c, a, P) = mf r2 + E.aJ || -c || 2-r2-f.}+ L^^p^

[0155] By deriving with respect to the parameters, the following three relations are obtained: [01561 ÿ = 2r(l-£,«,) =0 arP=0 = 0< ¢=-2 La,(¢(¾) - c) =0 » c = L .■ <l< • '

[0157] Thus, the dual Lagrange function is written:

[0158] g(a, P) = ^at II ¢(^)- E / zjOfxJ || 1if 0< = 1 - oo otherwise

[0159] The dual problem is then written: [OiôO] maxE.a, { ) S£- = 1 ^0< ai^

[0161] A normality score can then easily be calculated from the decision function f written as:

[0162] y(O(x) ) = j ( >

[0163] The decision function is the function defined by / . It allows to identify anomalies (more often referred to by the corresponding English term "outliers"): The points with the highest values ​​are considered as anomalies. This function can be normalized between 0 and 1 to make its interpretation more intuitive.

[0164] Such a first example corresponds to an unsupervised learning technique.

[0165] In fact, from a database assumed to be normal (without anomalies), a detection function is derived which will evaluate for a new vignette whether this vignette resembles those already observed or if it differs from them.

[0166] According to a second example of implementation of the obtaining step, the obtaining step is implemented using a neural network adapted to measure the distance of a thumbnail to a set of thumbnails considered to be normal.

[0167] Such a set of thumbnails can be obtained by evaluating the ESASSP criteria or by performing manual annotation by an expert.

[0168] As an illustration, the detection function outputs a normality score which corresponds to the projection value of the vector representation of the thumbnails in a learned space.

[0169] The learned space can in particular be obtained by a contrastive approach, that is to say by a comparison of examples sharing certain properties (so-called positive examples) with a set of examples which do not share these properties (so-called negative examples).

[0170] The normality score is then defined as the inverse of the distance from the center of a set of thumbnails considered by an expert to be free of anomalies.

[0171] In such a second example, the learning technique is semi-supervised due to the initial provision of the set of vignettes considered as normal.

[0172] The results obtained by implementing the method described in a simulation are now illustrated with reference to [Fig.3].

[0173] For this, three synthetic data sets corresponding to cases 1, 2 and 3 in [Fig.3] are used.

[0174] Case 1 corresponds to a dataset simulating data corresponding to realistic data from the three sensors. In this sense, the dataset of case 1 corresponds to a normal dataset.

[0175] The functions used for data generation are as follows:

[0176] ^(a) “14x (aixsi)x + 1.1x7?, y^ix) = (aiXSi)x+1.3 xbj yW ( x ) = 0.899 x ( x ) x + b,

[0177] Where: • i denotes the index of a point in the data set, • n denotes the number of points in the dataset, • vA(x) denotes the ordinate of point i of type 1 of the data set • ~U( [ -1,1 ] ) meaning that the variable 5) follows a uniform law whose values ​​are between -1 and 1, • ( [ -1,4] ) meaning that the variable a> follows a uniform law whose values ​​are between -1 and 4, and • bj~N(0.5) meaning that the variable b, follows a normal value law mean 0 and variance 5.

[0178] To the left of the arrow marked 1, the corresponding thumbnail is shown.

[0179] It appears in this vignette that each data set corresponds to a trajectory rectilinear, each rectilinear trajectory presenting a similar variation (similar slope coefficient and ordinate at the origin).

[0180] Case 2 corresponds to a dataset simulating data corresponding to data where one of the sensors has a much lower signal-to-noise ratio than the other two sensors. It is assumed, however, that the signal from the sensor with a reduced signal-to-noise ratio is actually usable. In this sense, the dataset of case 2 corresponds to a noisy dataset.

[0181] With the same notations as for case 1, the functions used for data generation are then the following:

[0182] = + V i < n ■} (*) = ( üix ) x + 1.3 x bf yW ( x ) = 0.899 x ( üj x ) x + b{

[0183] where N(0,2,5) denotes a number obtained by using a normal value distribution mean 0 and variance 2.5.

[0184] Such a case therefore corresponds to simulating Gaussian noise.

[0185] The resulting thumbnail of this noisy data simulation is shown to the left of the arrow marked 2.

[0186] Case 3 corresponds to a dataset simulating data corresponding to data where one of the sensors has unusable data. In this sense, the dataset of case 3 corresponds to a corrupted dataset.

[0187] With the same notations as for case 1, the functions used for data generation are then the following:

[0188] WU) = (U([-l,l]) xU([-l,4]))x^ Vi <n          y^(x)="(aixsi)x+" l,3xfy y^(x) = 0.899 x (al xs()x + bi

[0189] Where: • U( [ -1,1] ) denotes a number obtained by using a uniform law whose values ​​are between -1 and 1, • U( [ -1,4] ) denotes a number obtained by using a uniform law whose values ​​are between -1 and 4, and • N ( 0,2,5 ) denotes a number obtained by using a normal distribution of mean value 0 and variance 2.5.

[0190] The resulting thumbnail of this noisy data simulation is shown to the left of the arrow marked 3 in [Fig.3].

[0191] The right part of [Fig.3] represents the normality score distributions obtained for all simulated datasets.

[0192] Normality scores are the values ​​taken by the decision function. The lowest scores identify anomalies.

[0193] The distribution of scores is obtained by analyzing the score of each vignette, the set of vignettes being associated with a respective score.

[0194] Comparing the histograms of cases 2 and 3 with that of case 1 shows the existence of a shift of the values ​​to the left.

[0195] This makes it possible to show that it is possible to discriminate case 1 with cases 2 and 3, that is to say to discriminate between normal data and noisy or corrupted data.

[0196] A more pronounced shift can also be observed for case 2 compared to case 3. Thus, the analysis of the histograms makes it possible to discriminate between corrupted data and noisy data, that is to say between unusable data and abnormal but usable data.

[0197] Through this example, it has been shown that the method which has just been described therefore makes it possible to detect an anomaly with good efficiency.

[0198] The present method uses vignettes corresponding to a limited volume of space as well as a finite time period. Therefore, the method is a local approach both in time and space.

[0199] This allows for a more detailed analysis of the situation. In particular, the effects are not averaged over an entire FIR, unlike a technique for evaluating the criteria of the ESASSP standard.

[0200] This fineness of analysis does not prevent it from being possible, if necessary, to provide a global analysis, for example by aggregating the normality scores.

[0201] It could also be noted that the method can be implemented in a completely unsupervised manner. This presents a real advantage in the sense that the method can easily adapt to new situations or new contexts.

[0202] For example, the method can be used in contexts other than ATM and in particular in UTM.

[0203] The abbreviation ATM refers to the English term “Air Traffic Management” which designates the management of traffic for manned aircraft (generally airliners) while the abbreviation UTM refers to the English term “Unmanned Aircraft System Traffic Management” which designates the management of traffic for drones.

[0204] This independence makes it possible to avoid redefining the functions involved if the use of the tracking system 10 is likely to evolve.

[0205] In this sense, the method is universal in the sense that it can be adapted to any type of tracking system 10.

[0206] The method also has the advantage of providing an evaluation even in very degraded situations in which it is no longer possible to calculate a reconstructed trajectory.

[0207] It can also be noted that for data coming from sensors, the method shifts the computational load onto learning. Therefore, in the inference phase (control step), the calculations can be performed in real time.

[0208] Furthermore, learning does not involve calculating the actual trajectory, which corresponds to learning consuming fewer computational resources than learning for the evaluation of ESASSP criteria.

[0209] The method is also robust to the input data insofar as the data coming from the set 12 of sensors are not necessarily regular and are provided in raw form by the sensors of the set 12 of sensors.

[0210] The method which has just been described therefore constitutes a method for local and unsupervised evaluation in real time of the quality of service of aeronautical tracking.

[0211] According to the example of [Fig.4], the tracking system 10 is equipped with the analysis module 20, the detection module 22 described previously and an additional module which is a correction module 40.

[0212] The correction module 40 is a module for correcting anomalies in the observed space.

[0213] In this respect, the correction module 40 is capable of implementing a step of identifying a possible cause of the presence of an anomaly by applying an identification function to the vignette(s) in which the presence of an anomaly has been detected by the detection module 22.

[0214] Advantageously, the correction module is capable of testing a corrective action associated with the identified cause. This makes it possible to determine the most appropriate corrective action.

[0215] According to a particular example, the cause is a failure of one or more sensors of the set 12 of sensors.

[0216] In such a case, the corrective action is to remove the consideration of data from the sensor exhibiting the fault.

[0217] For such an example, a method for correcting anomalies in the observed space could be implemented in which the identification step is specific.

[0218] More specifically, the identification step comprises a generation operation, a detection operation and a deduction operation.

[0219] During the generation operation, the correction module generates new thumbnails.

[0220] The new tiles correspond to the old tiles formed in which data from one or more sensors are deleted.

[0221] Thus, the new vignettes correspond to the data accessible to several distinct sensor subsets.

[0222] According to one example, all the thumbnails corresponding to all possible configurations could be generated.

[0223] During the detection operation, the correction module obtains from the detection module 22 the detection of the possible presence of anomalies in the generated thumbnails.

[0224] During the deduction operation, by comparing the results obtained during the detection operation, it is possible to deduce the sensor(s) causing the anomaly.

[0225] For example, if all of the normality scores are increased in the absence of a sensor, the correction module deduces that said sensor is faulty and must be discarded to improve the performance of the tracking system 10.

[0226] Alternatively or in addition, it is possible to modify the parameters of the analysis module 20.

[0227] The correction method thus provides a recommendation for action with a view to resolving a detected anomaly.

[0228] In fact, the correction method makes it possible to carry out a causal analysis by determining the influence of several corrective actions on the normality score.

[0229] Thus, the method makes it possible to discriminate fairly quickly between actions which do not allow the restoration of the nominal situation and those which do.

[0230] In this sense, the correction method provides an unambiguous explanation of the cause of the presence of anomalies unlike a technique based on ESASSP criteria for which several different anomalies can lead to similar performance evaluation values.

[0231] Other variants of the methods which have just been described are conceivable.

[0232] In particular, it is possible that instead of a normality score, a performance score is obtained.

[0233] According to one example, the tracking system 10 further comprises an evaluation module.

[0234] The evaluation module is capable of determining the performance of the tracking system 10 as a function of the number of anomalies and the frequency of anomalies detected by the detection module 22.

[0235] In other words, the evaluation module is used to calculate a performance score for tracking system 10.

[0236] It is also possible to consider replacing the detection unit 26 with an evaluation unit whose role is to perform the performance calculation on the vignettes.

[0237] The performance calculation can then be based on criteria independent of anomaly detection.

[0238] In such a case, the detection module becomes an evaluation module comprising the training unit 24 and the evaluation unit.

[0239] More generally, the evaluation module or the detection module 22 possibly supplemented by the correction module 40 forms a control module.

[0240] The control module is capable of controlling the observation by the tracking system 10 of the space corresponding to at least one vignette by applying a control function to all of the data of the at least one vignette.

[0241] According to the example of [Fig.2], the control function is a function for detecting the possible presence of an anomaly in the vignette.

[0242] In the case of the evaluation, the control function is a function for calculating the performance of the tracking system 10.

[0243] In each of the examples described, each module or sub-module is each produced in the form of software, or a software brick.

[0244] The set of modules is then produced, that is to say in the form of a computer program, also called a computer program product, it is furthermore capable of being recorded on a medium, not shown, readable by a computer. The computer-readable medium is for example a medium capable of storing electronic instructions and of being coupled to a bus of a computer system. For example, the readable medium is an optical disk, a magneto-optical disk, a ROM memory, a RAM memory, any type of non-volatile memory (for example FLASH or NVRAM) or a magnetic card. On the readable medium is then stored a computer program comprising software instructions, stored in a memory executable by a processor.

[0245] In a variant not shown, each module or sub-module is produced in the form of a programmable logic component, such as an FPGA (Field Programmable Gate Array) or an integrated circuit, such as an ASIC (Application Specific Integrated Circuit).

[0246] The invention relates to any technically possible combination of the previously described embodiments.< / n>

Claims

Claims

1. A method for controlling the observation by a tracking system (10) of a space, the method being implemented by a control module (22, 40) forming part of the tracking system (10), the control method comprising: - a step of forming thumbnails, each thumbnail gathering a set of data accessible to the tracking system (10) over a respective area and a predefined time interval, the areas associated with each thumbnail paving the space observed by the tracking system (10) and the set of predefined time intervals covering an observation time interval, the data comprising at least sensor data, and - a step of controlling the observation by a tracking system (10) of the space corresponding to at least one thumbnail by applying a control function to all the data of the at least one thumbnail.

2. A control method according to claim 1, wherein the set of data of each vignette comprises a reconstructed trajectory.

3. A control method according to claim 1 or 2, wherein the tracking system (10) outputs calculated data, the set of data for each sticker comprising the calculated data.

4. Control method according to any one of claims 1 to 3, in which the control function is a function of detecting the possible presence of an anomaly in the sticker.

5. Control method according to claim 4, in which the anomaly detection function is obtained by a learning procedure, the learning procedure comprising: - learning a vector representation of the thumbnails, and - obtaining an anomaly detection function from the learned vector representation.

6. A control method according to claim 5, wherein the learning step comprises learning a first sub-function from a labeled data set to obtain a first learned sub-function capable of implementing a pretext task, the first learned sub-function being a neural network comprising a plurality of layers of neurons, the vector representation being the penultimate layer of the first learned subfunction.

7. A control method according to claim 6, wherein the first sub-function is a residual neural network.

8. A control method according to any one of claims 5 to 7, wherein the obtaining step is implemented using single-class support vector machines.

9. Control method according to any one of claims 5 to 7, in which the obtaining step is implemented using a neural network adapted to measure the distance of a thumbnail to a set of thumbnails considered as normal.

10. Control method according to any one of claims 4 to 9, in which the control method comprises a step of identifying a possible cause of the presence of an anomaly by applying an identification function to the vignette(s) in which the presence of an anomaly was detected in the implementation step, the identification step being carried out by the anomaly correction module.

11. A control method according to claim 10, wherein the control method comprises testing a corrective action associated with the identified cause.

12. A control method according to claim 10 or 11, wherein the tracking system (10) is adapted to collect data from several sensors, the cause being a failure of a sensor and the corrective action being the suppression of the consideration of the data from the sensor exhibiting the failure.

13. Control method according to claim 12, in which the identification step comprises: - the generation of thumbnails corresponding to the data accessible to several subsets of distinct sensors, - the detection of anomalies in the thumbnails generated by implementing the detection method, and - the deduction of the sensor(s) causing the anomaly.

14. A control method according to any one of claims 1 to 3, wherein the control function is a function for calculating the performance of the tracking system (10).

15. Control module (22, 40) for the observation by a tracking system (10) of a space, the control module (22, 40) being capable of: - forming thumbnails, each thumbnail bringing together a set of data accessible to the tracking system (10) over a respective area and a predefined time interval, the areas associated with each thumbnail paving the space observed by the tracking system (10) and all of the predefined time intervals covering an observation time interval, the data comprising at least sensor data, and - controlling the observation by a tracking system (10) of the space corresponding to at least one vignette by applying a control function to all the data of the at least one vignette.

16. Tracking system (10) provided with a control module (22, 40) according to claim 15.

Images