Electronic device and method for recognising one or more obstacles on a railway track, associated computer program and system

Abstract

The invention relates to a method for recognising one or more obstacles on a railway track (18) in front of a vehicle (14), the method is implemented by an electronic recognition device (30) intended to be installed on-board the vehicle. The method comprises acquiring at least one image of a region located in front of the vehicle, the region including the railway track; and processing the at least one image. This processing comprises identifying the rails of the railway track within each acquired image; and then searching for a possible rail discontinuity. The method further comprises triggering at least one action if at least one rail discontinuity of length greater than a predefined minimum length is detected, each action consisting in sending a braking command to a braking system of the vehicle (14) or transmitting an alert to an electronic supervision system (20) external to the vehicle (14).

Description

[0001] Electronic device and method for recognizing obstacle(s) on a railway track, associated computer program and system

[0002] The present invention relates to a method for recognizing obstacle(s) on a railway track in front of a vehicle according to the direction of movement of the vehicle on the railway track, the railway track comprising two lines of rails.

[0003] The invention also relates to a computer program comprising software instructions which, when executed by a computer, implement such a method of recognizing obstacle(s).

[0004] The invention also relates to an electronic device for recognizing obstacle(s) on the railway track, as well as a system for detecting obstacle(s) on the railway track.

[0005] This method makes it possible to identify one or more obstacles in front of the vehicle that could prevent it from moving along the railway track. Examples of obstacles include a fallen tree or a mudslide covering the railway tracks. This method allows the vehicle to move in safer conditions, or even in autonomous mode without a driver.

[0006] It is known to use sensors and a computer on board the vehicle to detect obstacles in front of the vehicle. Usually, the computer uses an algorithm configured to analyze the images taken by each sensor and to determine if an object is present in front of the vehicle, the presence of an object being equivalent to obstacle detection. The algorithm is for example a neural network previously trained on an image database, from which the neural network has learned to recognize objects appearing in front of the vehicle.

[0007] However, such a process is complex to implement and the accuracy of such an algorithm can be low. In particular, the algorithm tends to predict false positives, i.e. to detect an object that does not actually represent an obstacle for the vehicle. For example, the algorithm will detect an obstacle and then order the vehicle to stop if a cloud of midges is present in front of the vehicle, but the latter does not hinder the vehicle's movement. In addition, stopping a vehicle causes many complications, particularly in the case where several vehicles are traveling on the same railway. In addition, some obstacles may not be detected. For example, a mudslide with a relatively low thickness is likely not to be detected as an object by the algorithm, while the latter can prevent the vehicle from passing.The aim of the invention is therefore to propose a method for recognizing obstacle(s) on a railway track making it possible to determine more reliably and more efficiently the presence of obstacle(s) likely to prevent the vehicle from moving on the railway track.

[0008] To this end, the invention relates to a method for recognizing obstacle(s) on a railway track in front of a vehicle according to the direction of movement of the vehicle on the railway track, the railway track comprising two lines of rails, the method being implemented by an electronic recognition device intended to be mounted on the vehicle, the method comprising the following steps:

[0009] - acquisition of at least one image of a region located in front of the vehicle, said region comprising the railway track;

[0010] - processing of at least one image; the processing step comprising an identification of the rails of the railway track within each acquired image; then a search for a possible rail discontinuity; and

[0011] - triggering at least one action if at least one rail discontinuity is detected and has a length greater than a predefined minimum length in a direction of extension of the rail, each rail discontinuity of length greater than the predefined minimum length being assimilated to a potential obstacle, each action being chosen from the group consisting of: sending a braking order to a braking system of the vehicle; and sending an alert to an electronic supervision system, external to the vehicle.

[0012] With the recognition method according to the invention, each rail discontinuity of length greater than the predefined minimum length is assimilated to a potential obstacle, such a search for rail discontinuity(ies) making it possible to determine more reliably and more efficiently the presence of obstacle(s) potentially hindering the movement of the vehicle on the railway.

[0013] Such an obstacle is understood in the broad sense as any object preventing the passage of the vehicle on the track. Such an obstacle is, for example, a material obstacle, such as a tree, a rock, another vehicle, or a stray animal; a landslide; a collapse of the ground; or even a broken rail, etc.

[0014] The transmission of an alert to the supervision system aims, for example, to prevent other vehicles from using the area of ​​the railway track comprising the obstacle that has been detected, and / or to warn other vehicles following the vehicle that has transmitted this alert, with a view to slowing down or stopping these other vehicles. According to other advantageous aspects of the invention, the method for recognizing obstacle(s) comprises one or more of the following characteristics, taken individually or in all technically possible combinations:

[0015] - the predefined minimum length is all the lower as the number of successive images acquired comprising said discontinuity is high;

[0016] - the rail discontinuity corresponds to a series of successive pixels along the direction of extension of the rail which each have a different value, at more than a predefined distance, from an average value of the pixels of the identified rail, the length of the discontinuity corresponding to the number of pixels in the series; the predefined distance preferably being a predefined value or a predefined percentage of said average value;

[0017] - the predefined minimum length corresponds to a predefined minimum number of pixels; the predefined minimum number of pixels preferably being greater than or equal to 2 pixels;

[0018] - the predefined minimum number of pixels is equal to a first value if the triggering of the at least one action is carried out from a single acquired image comprising a rail discontinuity; and the predefined minimum number of pixels is equal to a second value, lower than the first value, if the triggering of the at least one action is carried out only following several acquired images comprising the same rail discontinuity; the first value preferably being equal to 5 pixels; the second value preferably being equal to 2 or 3 pixels;

[0019] - rail identification is performed using additional vehicle position and track mapping; and

[0020] - the braking order sent to the vehicle's braking system is an order to stop the vehicle.

[0021] The invention also relates to a computer program comprising software instructions which, when executed by a computer, implement a method for recognizing obstacle(s), as defined above.

[0022] The invention also relates to an electronic device for recognizing obstacle(s) on a railway track in front of a vehicle according to the direction of movement of the vehicle on the railway track, the railway track comprising two lines of rails, the recognition device being intended to be mounted in the vehicle and comprising:

[0023] - an acquisition module configured to acquire at least one image of a region located in front of the vehicle, said region comprising the railway track; - an image processing module; the image processing module being configured to identify rails of the railway track within each acquired image; then to search for a possible rail discontinuity; and

[0024] - a trigger module configured to trigger at least one action if at least one rail discontinuity is detected and has a length greater than a predefined minimum length in a direction of extension of the rail, each rail discontinuity of length greater than the predefined minimum length being assimilated to a potential obstacle, each action being chosen from the group consisting of: sending a braking order to a braking system of the vehicle; and sending an alert to an electronic supervision system, external to the vehicle.

[0025] The invention also relates to a system for detecting obstacle(s) on a railway track in front of a vehicle according to the direction of movement of the vehicle on the railway track, the railway track comprising two lines of rails, the system being intended to be mounted on the vehicle, the system comprising: an electronic device for recognizing obstacle(s) on the railway track in front of the vehicle, the electronic recognition device being as defined above; and a sensor configured to take at least one image of a region located in front of the vehicle, said region comprising the railway track, and to deliver each image taken to the recognition device.

[0026] The invention will appear more clearly on reading the description which follows, given solely by way of non-limiting example, and made with reference to the drawings in which:

[0027] - Figure 1 is a functional diagram of an obstacle detection system according to the invention, comprising an electronic device for recognizing obstacle(s) on a railway track in front of a vehicle and a sensor configured to take at least one image of a region located in front of the vehicle;

[0028] - Figure 2 is a schematic representation of the vehicle traveling on the railway track without an obstacle, the railway track comprising two lines of rails; of the vehicle traveling on the railway track comprising an obstacle on both lines of rails; and of the vehicle traveling on the railway track comprising an obstacle on one of the two lines of rails;

[0029] - Figure 3 is a schematic representation of the vehicle traveling on the railway track comprising an obstacle and of the region of the railway track visible by the sensor of the device of Figure 1 as a function of the positioning of the vehicle; and - Figure 4 is a flowchart of a method for recognizing obstacle(s) according to the invention, the method being implemented by the recognition device of Figure 1.

[0030] In the remainder of the description, the expression “substantially equal to” defines a relationship of equality to plus or minus 20%, preferably to plus or minus 10%, more preferably to plus or minus 5%.

[0031] In Figure 1, a transport installation 10 comprises a fleet 12 of vehicles 14 capable of moving on a network 16 of railway track(s) 18 and an electronic system 20 for supervising the fleet 12 of vehicles 14.

[0032] The transport facility 10 is advantageously a public transport facility, or a collective transport facility, that is to say an facility allowing several people to be transported together on the same journey.

[0033] The network 16 comprises one or more railway tracks 18, each railway track 18 comprising one or more successive sections.

[0034] Each railway track 18 is advantageously a disused track, that is to say a track which is no longer used for the regular circulation of trains or a track which is no longer operated by the railway infrastructure manager of the territory concerned, such as Réseau Ferré de France (RFF) for French territory.

[0035] Additionally or alternatively, each railway track 18 is an operational railway track used for the regular circulation of trains, but without the presence of other rolling stock at the same time as the fleet 12 of vehicles 14, that is to say in the absence of rolling stock other than the fleet 12 of vehicles 14 when the fleet 12 of vehicles 14 runs on said operational railway track(s).

[0036] Each railway track 18 is then typically a railway track that no longer allows a conventional railway vehicle to run (e.g. TER for regional express train), or a railway track used exceptionally by freight trains (e.g. once a week), or a railway track where public passenger transport is no longer economically viable with conventional railway vehicles.

[0037] Each railway line 18 follows a defined trajectory and is listed on a network map 16.

[0038] With reference to Figure 2, each railway track 18 comprises two lines of rail 22 parallel to each other and divided into sections.

[0039] Each rail line 22 is formed by a plurality of rails assembled end to end along the path of the railway track 18. The fleet 12 comprises a plurality of vehicles 14 moving on the network 16 of railway tracks 18. The fleet 12 of vehicles 14 is configured to preferably transport passengers. Alternatively or additionally, the fleet of vehicles 14 is configured to transport goods, or both passengers and goods.

[0040] Each vehicle 14 is advantageously configured to travel both on the network 16 of railways 18 and on a road network, not shown. Each vehicle 14 is then, for example, of the Ferromobile type developed by the Société d'ingénierie, de Construction et d'Exploitation de la Ferromobile (SICEF). The Ferromobile is a vehicle equipped with a rail-road system allowing it to travel both on road and rail tracks. The rail-road system includes mixed wheels or additional axles at the front and rear of the vehicle 14 allowing it to travel on the road and then on the rails, or vice versa, within the same journey.

[0041] Each vehicle 14 comprises, for example, a positioning device, not shown, configured to provide a position of the vehicle.

[0042] The positioning device is typically a satellite positioning device, also called a GNSS device (for Geolocation and Navigation by a Satellite System, or Global Navigation Satellite System), comprising a satellite positioning receiver and an antenna. This GNSS device is optionally equipped with one or more other additional positioning assistance sensors, such as in particular an inertial unit, a Doppler sensor, etc. The GNSS device uses a constellation of satellites and makes it possible to provide the vehicle 14, via the sensor(s) constituting it, with its 3D position, its 3D speed and the time. The GNSS device is, for example, a GPS device (Global Positioning System), a Galileo device, a Glonass device, or a Beidou device.

[0043] Each vehicle 14 also comprises a system 24 for detecting obstacle(s) 26 on the railway track 18 in front of the vehicle 14 according to the direction of movement of the vehicle 14 on the railway track 18.

[0044] Here and in the remainder of the description, the expression "at the front of the vehicle" is equivalent to the expression "in front of the vehicle".

[0045] The term “obstacle” is understood in the broad sense as any object preventing the passage of the vehicle 14 on the railway track 18. Such an obstacle 26 is for example a material obstacle, such as a tree, a rock, another vehicle, etc.; a landslide; a collapse of the ground; or even a broken rail, etc.

[0046] An obstacle 26 covers at least a portion of at least one row of rails 22. In the example shown in the middle of Figure 2, the obstacle 26 covers a portion of the two rows of rails 22. In this particular example, the obstacle 26 is a tree that has fallen onto the railway track 18.

[0047] In the example shown at the bottom of Figure 2, the obstacle 26 covers a portion of only one of the two lines of rails 22. In this particular example, the obstacle 26 is a buoy flow having partially covered a single line of rails 22.

[0048] The detection system 24 is intended to be installed on the vehicle 14.

[0049] With reference to FIG. 1, the detection system 24 comprises a sensor 28 and an electronic device 30 for recognizing obstacle(s) 26, the recognition device 30 being connected to the sensor 28.

[0050] The sensor 28 is configured to take at least one image of a region located in front of the vehicle 14, said region comprising the railway track 18. The region visible by the sensor 28 is represented by a detection cone 34 in FIGS. 1 and 2.

[0051] The sensor 28 is for example a LiDAR (Light Detection And Ranging), a RaDAR (Radio Detection And Ranging), a camera or a photo sensor. Preferably, the sensor 28 is a camera.

[0052] The image taken by the sensor 28 is then understood in the broad sense as a camera image, a lidar image, or even a radar image.

[0053] The sensor 28 is for example configured to take a grayscale or color image of the region located in front of the vehicle 14.

[0054] The sensor 28 is further configured to deliver each image taken to the recognition device 30. Preferably, the sensor 28 is configured to deliver between 10 and 60 images per second to the recognition device 30.

[0055] The recognition device 30 comprises an acquisition module 36, a processing module 38 and a trigger module 40, the acquisition module 36 being connected to the sensor 28.

[0056] In the example of Figure 1, the recognition device 30 comprises an information processing unit 42 formed for example of a memory 44 and a processor 46 associated with the memory 44. The processing unit 42 is connected to the sensor 28.

[0057] In the example of Figure 1, the acquisition module 36, the processing module 38 and the trigger module 40 are each produced in the form of software, or a software brick, executable by the processor 46. The memory 44 of the recognition device 30 is then able to store acquisition software, processing software and trigger software. The processor 46 is then able to execute each of the software among the acquisition software, the processing software and the trigger software. In a variant not shown, the acquisition module 36 and the processing module 38 and the trigger module 40 are each produced in the form of a programmable logic component, such as an FPGA (Field Programmable Gate Array), or in the form of a dedicated integrated circuit, such as an ASIC (Application Specific Integrated Circuit).

[0058] When the recognition device 30 is produced in the form of one or more software programs, that is to say in the form of a computer program, it is also 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. By way of 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 EPROM, EEPROM, FLASH, NVRAM), a magnetic card or an optical card. A computer program comprising software instructions is then stored on the readable medium.

[0059] The acquisition module 36 is configured to acquire at least one image of the region located in front of the vehicle 14, said region comprising the railway track 18.

[0060] For example, the acquisition module 36 is configured to receive each image taken by the sensor 28.

[0061] For example, each acquired image consists of a matrix of pixels. Each pixel is associated with a pixel value. If the image is taken in grayscale, the value of a pixel corresponds to its brightness level and is generally between 0 and 255, where 0 represents black and 255 represents white. If the image is taken in color, the value of a pixel corresponds, for example, to a triplet representing the RGB (Red, Green, Blue) color mode. The triplet generally consists of three numbers, each number being between 0 and 255 and corresponding to the brightness level of a color from the group Red, Green, Blue.

[0062] The acquisition module 36 is further configured to transmit each acquired image to the processing module 38.

[0063] The image processing module 38 is configured to process each acquired image.

[0064] First, the processing module 38 is configured to identify rails of the railway track 18 within each acquired image.

[0065] For example, the processing module 38 is configured to estimate the position of the two lines of rails 22 on each acquired image.

[0066] The identification of the rails is for example carried out via an image processing algorithm capable of recognizing the shape of the line(s) of rails 22 in each acquired image. The identification of the rails is advantageously carried out by further using the position of the vehicle 14 determined by the positioning device and the mapping of the network 16. The trajectory of the railway track 18 being known thanks to the mapping, the processing module 38 determines a direction of extension of the rails from the position of the vehicle 14 and thus also determines the region located in front of the vehicle 14, taken by the sensor 28.

[0067] Alternatively or additionally, the identification of the rails is for example carried out by also using information concerning the sensor 28. Such information includes for example the angle, the aperture and / or the magnification of the sensor 28. The information concerning the sensor 28 makes it possible for example to determine the portion of the railway track 18 included in each acquired image.

[0068] The processing module 38 is typically configured to identify the pixels of each acquired image corresponding to the estimated position of the lines of rails 22.

[0069] Second, the processing module 38 is further configured to search for a possible discontinuity 48 on at least one rail line 22.

[0070] The discontinuity 48 corresponds for example to a series of successive pixels along the direction of extension of the line of rails 22 which have a different appearance from the pixels of the rail, while being located inside a zone corresponding to a respective line of rails 22 and delimited by the edges of said line of rails 22.

[0071] A different appearance is for example defined by a different pixel value, at a predefined distance from a reference value of the pixels of the rail line 22.

[0072] In other words, a pixel has a different appearance from the rail if the difference, i.e. the gap, between the value of said pixel and the reference value is greater than the predefined gap.

[0073] The difference between the value of said pixel and the reference value corresponds, for example, to the Euclidean distance in the case of a color image, or to the absolute value of the difference between the value of said pixel and the reference value for a grayscale image.

[0074] The reference value corresponds, for example, to the average of the pixel values ​​of the identified line of rails 22.

[0075] Preferably, the predefined deviation is a predefined value or a percentage of the reference value. The predefined deviation is, for example, equal to 10% of the average value of the rail pixels.

[0076] Alternatively or additionally, the processing module 38 is configured to search for the possible discontinuity 48 via the use of predefined patterns. As a further variant, the processing module 38 is configured to search for the possible discontinuity 48 via the implementation of an artificial intelligence algorithm, the training of the artificial intelligence algorithm then being carried out with, on the one hand, images of rail(s) exhibiting a discontinuity, and on the other hand, images of rail(s) not exhibiting a discontinuity. The artificial intelligence algorithm is for example a neural network.

[0077] In addition, the processing module 38 is configured to determine the position of said discontinuity 48 on the railway track 18.

[0078] Those skilled in the art will note that the same discontinuity 48 may appear on several images at different positions.

[0079] Indeed, depending on the speed of the vehicle 14 and the frequency of image capture, a discontinuity 48 appears on one or more images.

[0080] The processing module 38 is then configured to determine on how many successive images the same discontinuity 48 appears, based on the speed of movement of the vehicle 14 on the railway track 18 and the frequency of image capture by the sensor 28.

[0081] In the example of Figure 3, a discontinuity 48, represented schematically in the form of a rectangle with a cross filling, appears on three successive images: on a first image taken at a first instant T0, a second image taken at a second instant T0+dt1 and a third image taken at a third instant T0+dt2. The three instants are successive and defined such that T0 < T0+dt1 < T0+dt2.

[0082] The durations dt1 and dt2 vary depending on the speed of the vehicle 14 and the frequency of image capture by the sensor 28.

[0083] Furthermore, to facilitate reading of Figure 3, an arrow 49 corresponding to a particular observation axis of the sensor 28 is shown in bold and consequently points to an element which will appear at a certain point in the image, the position of the element pointed to in the image then evolving from one image taken to the next.

[0084] The person skilled in the art will then observe that at the first instant T0, the arrow 49 does not reach the discontinuity 48. At the second instant T0+dt1, the tip of the arrow 49 is included in the discontinuity 48 and aims at the start of the latter. At the third instant T0+dt2, the tip of the arrow 49 is included in the discontinuity 48 and aims at the end of the latter.

[0085] A person skilled in the art will deduce that the position of the discontinuity 48 on each acquired image varies depending on the distance traveled by the vehicle between two times.

[0086] The processing module 38 is then configured to monitor a discontinuity 48 between at least two successive images acquired by the acquisition module 36. The processing module 38 is for example configured to determine the variation in position of the discontinuity 48 between two successive images acquired by the acquisition module 36, then to check whether said variation in position is consistent with the movement of the vehicle, i.e. the distance traveled by the vehicle, between two time instants of taking said two images.

[0087] Furthermore, the number of pixels in each series corresponding to discontinuity 48 in the three images described above differs from one image to another. In fact, the closer discontinuity 48 is to the vehicle, the larger discontinuity 48 will be in the image taken by sensor 28.

[0088] The processing module 38 is then also configured to take this variation in length into account when determining the number of images comprising said discontinuity 48.

[0089] The trigger module 40 is configured to trigger at least one action if at least one rail discontinuity 48 is detected and has a length greater than a predefined minimum length according to the direction of extension of the rail.

[0090] Each rail discontinuity 48 of a length greater than the predefined minimum length is considered a potential obstacle 26. For example, the length of a discontinuity 48 corresponds to the number of pixels forming the series of pixels corresponding to said discontinuity 48.

[0091] Indeed, certain discontinuities 48 detected do not actually correspond to obstacles 26 preventing the movement of the vehicle 14 on the railway track 18. For example, a discontinuity 48 which would result from an imperfection, such as an artifact when taking the image, from an object crossing the railway track, or even from a small expansion gap between two rails arranged end to end does not correspond to an obstacle 26 preventing the movement of the vehicle 14.

[0092] In this example, the predefined minimum length corresponds to a predefined minimum number of pixels. Preferably, the predefined minimum number of pixels is greater than or equal to 2 pixels.

[0093] Advantageously, the predefined minimum length is all the lower as the number of successive images acquired comprising said discontinuity 48 is high.

[0094] According to this advantageous aspect, the minimum number of pixels is for example equal to a first value if the triggering of the at least one action is carried out from a single acquired image comprising a rail discontinuity; and the predefined minimum number of pixels is equal to a second value, strictly less than the first value, if the triggering of the at least one action is carried out only following several acquired images comprising the same rail discontinuity 48.

[0095] For example, the first value is equal to 5 pixels. For example, the second value is equal to 2 pixels. Alternatively, the minimum number of pixels decreases linearly as a function of the number of images comprising the same rail discontinuity 48. For example, for a single acquired image comprising the discontinuity 48, the minimum number of pixels is 5 pixels; for two images, the minimum number of pixels is 4 pixels, and so on until a plateau of 2 pixels is reached.

[0096] Advantageously, the predefined minimum length is all the smaller the further said discontinuity 48 is from the sensor 28, and therefore from the vehicle 14, the sensor 28 being on board the vehicle 14. Conversely, the predefined minimum length is all the greater the closer said discontinuity 48 is to the sensor 28.

[0097] As an optional addition, the minimum number of pixels is then weighted according to the distance between the discontinuity 48 and the sensor 28. In fact, said distance influences the number of pixels corresponding to said discontinuity 48.

[0098] According to this optional addition, the minimum number of pixels is in particular higher the closer the detected discontinuity 48 is to the sensor 28. According to this optional addition, the minimum number of pixels is then higher if the detected discontinuity 48 is in the foreground than if the detected discontinuity 48 is in the background.

[0099] The action triggered by the trigger module 40 is, for example, sending a braking order to a braking system of the vehicle 14 and / or sending an alert to the supervision system 20, external to the vehicle 14.

[0100] For example, the braking command sent to the braking system of vehicle 14 is a command to stop vehicle 14.

[0101] For example, the alert to the supervision system 20 aims to request an intervention on the railway track 18 to remove the obstacle 26 and thus allow the vehicle 14 to circulate again.

[0102] According to another example, the alert to the supervision system 20 aims to warn all of the vehicles 14 of the fleet 12 of the presence of the obstacle 26 detected on the railway track 18 and of the position of said obstacle 26.

[0103] For this purpose, the supervision system 20 is configured to communicate via a radio link 50, shown in FIG. 1, with each vehicle 14 of the fleet 12.

[0104] The supervision system 20 is for example configured to communicate a braking order to the braking system of each vehicle 14 traveling on the railway track 18 where the obstacle 26 has been detected. The braking order is for example a limitation of the speed of each vehicle 14, or an order to stop each vehicle 14. Furthermore, the supervision system 20 is for example configured to command, if possible, a change of direction to at least one vehicle 14 via an appropriate switch, in order to avoid the obstacle 26.

[0105] A method for recognizing obstacle(s) 26 on the railway track 18 in front of the vehicle 14 implemented by the recognition device 30 will now be described with reference to FIG. 4.

[0106] In an initial step 100, the acquisition module 36 acquires at least one image of the region located in front of the vehicle 14, said region comprising the railway track 18.

[0107] For example, each acquired image was transmitted to the acquisition module 36 by the sensor 28.

[0108] During a following step 110, the processing module 38 processes the at least one image acquired during the acquisition step 100.

[0109] The processing step 110 comprises a first sub-step 120 during which the processing module 38 identifies the rails of the railway track 18 within each acquired image.

[0110] This first sub-step 120 is for example carried out from the position of vehicle 14 determined by the positioning device and the mapping of the network 16 or of the railway track 18.

[0111] The processing step 110 further comprises a second sub-step 130 during which the processing module 38 searches for a possible rail discontinuity 48 on at least one of the lines of rails 22 on each acquired image.

[0112] As an optional addition, if a discontinuity 48 is detected, a warning is issued indicating a possible obstacle 26, even if the discontinuity is not yet associated with a real obstacle 26.

[0113] At the end of the processing step 110, during a following step 140, the trigger module 40 triggers at least one action only if at least one obstacle 26 is considered recognized.

[0114] A respective obstacle 26 is considered recognized if a detected rail discontinuity 48 has a length greater than the predefined minimum length according to the direction of extension of the rail.

[0115] Each action triggered during the triggering step 140 is typically the sending of a braking order to the vehicle braking system 14, or the sending of an alert to the supervision system 20.

[0116] As an optional addition, the method further comprises an additional step, not shown, during which the supervision system 20 communicates to the vehicles 14 of the fleet 12 the presence of the obstacle 26 and adapts the circulation of the fleet 12 accordingly.

[0117] It is thus understood that the recognition method and the recognition device 30 according to the invention make it possible to determine more reliably and more effectively the presence of obstacle(s) 26 likely to prevent the movement of the vehicle 14 on the railway track 18.

Claims

CLAIMS 1. Method for recognizing obstacle(s) (26) on a railway track (18) in front of a vehicle (14) according to the direction of movement of the vehicle (14) on the railway track (18), the railway track (18) comprising two lines of rails (22), the method being implemented by an electronic recognition device (30) intended to be mounted on the vehicle (14), the method comprising the following steps: - acquisition (100) of at least one image of a region located in front of the vehicle (14), said region comprising the railway track (18); - processing (110) of the at least one image; characterized in that the processing step (110) comprises an identification (120) of the rails of the railway track (18) within each acquired image; then a search (130) for a possible discontinuity (48) of rail; and in that the method further comprises the following step (140): - triggering at least one action if at least one rail discontinuity (48) is detected and has a length greater than a predefined minimum length in a direction of extension of the rail, each rail discontinuity (48) of length greater than the predefined minimum length being assimilated to a potential obstacle (26), each action being chosen from the group consisting of: sending a braking order to a braking system of the vehicle (14); and sending an alert to an electronic supervision system (20), external to the vehicle (14).

2. Method according to claim 1, in which the predefined minimum length is all the lower the higher the number of successive images acquired comprising said discontinuity (48).

3. Method according to claim 1 or 2, in which the rail discontinuity (48) corresponds to a series of successive pixels in the direction of extension of the rail which each have a different value, at more than a predefined deviation, from an average value of the pixels of the identified rail, the length of the discontinuity corresponding to the number of pixels in the series; the predefined deviation preferably being a predefined value or a predefined percentage of said average value.

4. The method of claim 3, wherein the predefined minimum length corresponds to a predefined minimum number of pixels; the predefined minimum number of pixels preferably being greater than or equal to 2 pixels.

5. Method according to claim 4, in which the predefined minimum number of pixels is equal to a first value if the triggering of the at least one action is carried out from a single acquired image comprising a rail discontinuity (48); and the predefined minimum number of pixels is equal to a second value, lower than the first value, if the triggering of the at least one action is carried out only following several acquired images comprising the same rail discontinuity (48); the first value preferably being equal to 5 pixels; the second value preferably being equal to 2 or 3 pixels.

6. A method according to any preceding claim, wherein the identification of the rails is performed further using a vehicle position and a mapping of the railway track.

7. Method according to any one of the preceding claims, in which the braking command sent to the braking system of the vehicle (14) is a command to stop the vehicle (14).

8. Computer program comprising software instructions which, when executed by a computer, implement a method for recognizing obstacle(s) according to any one of the preceding claims.

9. Electronic device (30) for recognizing obstacle(s) (26) on a railway track (18) in front of a vehicle (14) according to the direction of movement of the vehicle (14) on the railway track (18), the railway track (18) comprising two lines of rails (22), the recognition device (30) being intended to be mounted in the vehicle (14) and comprising: - an acquisition module (36) configured to acquire at least one image of a region located in front of the vehicle (14), said region comprising the railway track (18); - an image processing module (38); characterized in that the image processing module (38) is configured to identify rails of the railway track (18) within each acquired image; then to search for a possible rail discontinuity (48); and in that the device (30) further comprises a trigger module (40) configured to trigger at least one action if at least one rail discontinuity (48) is detected and has a length greater than a predefined minimum length according to a direction of extension of the rail, each discontinuity (48) of rail of length greater than the predefined minimum length being assimilated to a potential obstacle (26), each action being chosen from the group consisting of: sending a braking order to a braking system of the vehicle (14); and sending an alert to an electronic supervision system (20), external to the vehicle (14).

10. System (24) for detecting obstacle(s) (26) on a railway track (18) in front of a vehicle (14) according to the direction of movement of the vehicle (14) on the railway track (18), the railway track (18) comprising two lines of rails (22), the system (24) being intended to be mounted on the vehicle (14), the system (24) comprising: - an electronic device (30) for recognizing obstacle(s) (26) on the railway track (18) at the front of the vehicle (14); and - a sensor (28) configured to take at least one image of a region located in front of the vehicle (14), said region comprising the railway track (18), and to deliver each image taken to the recognition device (30), characterized in that the recognition device (30) is according to the preceding claim.

Images