Obstacle detection method implemented by an aircraft-mounted detection system and associated obstacle detection system

Abstract

Obstacle detection method implemented by an aircraft-mounted detection system and associated obstacle detection system. The present invention relates to an obstacle detection method implemented by an aircraft-mounted obstacle detection system, the aircraft comprising at least one on-board camera having an associated field of view, configured to acquire full-field digital images.The method comprises the steps of: a) receiving (50) at least one full-field digital image captured by said at least one on-board camera, b) determining (54) the spatial position of a region of interest of predetermined size in said full-field digital image, and extracting (56) said region of interest from said full-field digital image, c) implementing an obstacle detection module (58) on said extracted region of interest, implementing a neural network previously trained to detect and classify obstacles in images having said predetermined size, each detected obstacle being located in said region of interest. Figure for the abstract: Figure 4.

Description

Title of the invention: Method for detecting obstacles implemented by a detection system on board an aircraft and associated obstacle detection system

[0001] The present invention relates to a method for detecting obstacles implemented by a detection system on board an aircraft, the aircraft comprising at least one on-board camera having an associated field of vision. It also relates to an associated obstacle detection system, and an associated computer program.

[0002] The invention lies in the field of avionics, and applies more particularly in the field of low-altitude flying aircraft, such as for example helicopters, drones, surveillance aircraft (forest fires, borders, etc.).

[0003] The term low altitude generally refers to an altitude less than or equal to 900 meters above sea level or AMSL (“Above Mean Sea Level”), or 300 meters above ground level if the ground is at an altitude greater than 900 meters above sea level.

[0004] It is necessary, for an aircraft in general, and more particularly in low altitude flight, to detect objects which could constitute possible obstacles, fixed or moving, in order to avoid possible collisions.

[0005] Hereinafter, an obstacle for an aircraft will be called an object of which at least a portion exceeds a predetermined altitude, the object being fixed or mobile. For example, fixed obstacles include poles, pylons, cables suspended between pylons, buildings, towers. Mobile obstacles include in particular other aircraft, for example unmanned drones on board.

[0006] Obstacle detection is also of general interest for the geo-referenced mapping of fixed obstacles at various altitudes, e.g. skyscrapers, pylons, wind turbines, thus enabling the implementation of obstacle avoidance strategies based on such mapping.

[0007] Some low-flying aircraft, for example drones or helicopters, are not equipped with obstacle detection solutions, because existing solutions, based on optronic sensors, are expensive.

[0008] Furthermore, aircraft have limited computing resources and energy resources, and it is of interest to develop solutions minimizing electrical energy consumption in particular.

[0009] The object of the invention is to overcome the drawbacks of known methods, by providing an on-board obstacle detection solution compatible with existing on-board sensors and with low energy consumption.

[0010] For this purpose, the invention relates to an obstacle detection method implemented by an obstacle detection system on board an aircraft, the aircraft comprising at least one on-board camera having an associated field of vision, configured to acquire full-field digital images. This method is implemented by a processor of a computing platform and comprises steps of:

[0011] a)-reception of at least one full-field digital image captured by said at least one on-board camera,

[0012] b)-determination of a spatial position of an area of ​​interest of predetermined size in said full-field digital image, and extraction of said area of ​​interest from said full-field digital image,

[0013] c)-implementation of an obstacle detection module on said extracted area of ​​interest, implementing a previously trained neural network to detect and classify obstacles in images having said predetermined size, each detected obstacle being located in said area of ​​interest.

[0014] Advantageously, the method of the invention makes it possible to limit the consumption of computational and energy resources, while providing precise obstacle detection thanks to the use of an obstacle detection module implementing a previously trained neural network.

[0015] The obstacle detection method according to the invention may also have one or more of the characteristics below, taken independently or in any technically conceivable combination.

[0016] This method further comprises a post-processing step comprising a calculation of a distance between the aircraft and the or each obstacle detected and / or a storage of a geo-referenced position, in a fixed terrestrial reference frame, for each obstacle detected belonging to a class of fixed obstacles.

[0017] This method comprises repeating the determination of a spatial position of an area of ​​interest in the same wide-field digital image, making it possible to obtain a plurality of areas of interest in said wide-field digital image.

[0018] Before step b) of determining a spatial position of an area of ​​interest, the method comprises a step of acquiring at least one piece of avionics information relating to a movement parameter of the aircraft, and the determination of a spatial position is a function of at least one piece of avionics information.

[0019] The avionics information comprises a trajectory vector of the aircraft, the area of ​​interest being centered on a point indicated by the direction of said trajectory vector.

[0020] Determining a spatial position of an area of ​​interest comprises receiving a spatial position of the area of ​​interest by means of a communication interface with an external system.

[0021] Determining a spatial position of an area of ​​interest involves receiving from a spatial position of the area of ​​interest of another sensor on board the aircraft.

[0022] The camera being configured to acquire a succession of full-field digital images forming a video, steps a) to c) are implemented on a subset of acquired digital images spaced temporally by a given time step, the method further comprising temporal tracking of the detected obstacles.

[0023] According to another aspect, the invention relates to an obstacle detection system adapted to be mounted on an aircraft, the aircraft comprising at least one on-board camera having an associated field of vision, configured to acquire full-field digital images, the obstacle detection system comprising a computing platform comprising at least one processor configured to implement:

[0024] - a module for receiving at least one full-field digital image captured by said at least one on-board camera,

[0025] - a module for determining a spatial position of a zone of interest of size predetermined in said full-field digital image, and extracting said area of ​​interest from said full-field digital image,

[0026] -an obstacle detection module, taking said area of ​​interest as input and implementing a previously trained neural network to detect and classify obstacles in images having said predetermined size, each detected obstacle being located in said area of ​​interest.

[0027] According to an advantageous aspect, the obstacle detection system further comprises a post-processing module, configured to calculate a distance between the aircraft and the or each obstacle detected and / or store a geo-referenced position, in a fixed terrestrial reference frame, for each obstacle detected belonging to a class of fixed obstacles.

[0028] According to another aspect, the invention relates to a computer program comprising software instructions which, when executed by a computer, implement an obstacle detection method as defined above.

[0029] Other characteristics and advantages of the invention will emerge from the description given below, for information purposes only and in no way limiting, with reference to the appended figures, among which:

[0030] [Fig-1] [Fig.l] is a block diagram of an obstacle detection system in a implementation example;

[0031] [Fig.2] [Fig.2] schematically represents a first example of an image wide-field digital with two areas of interest;

[0032] [Fig.3] [Fig.3] schematically represents a second example of image wide-field digital with an area of ​​interest;

[0033] [Fig.4] [Fig.4] is a flowchart of the main steps of a detection method of obstacles according to one embodiment.

[0034] [Fig.l] schematically represents an aircraft 2, equipped with an on-board obstacle detection system 4.

[0035] The aircraft 2 is equipped with a plurality of avionics instruments 6, for example a sensor 6A configured to provide a geo-referenced position of the aircraft, for example a GPS sensor, an inertial sensor 6B, and an aerodynamic data computer 6C, also called ADC for “Air Data Computer”, configured to provide flight data from a plurality of external sensors, the external sensors being for example pressure probes, temperature probes or incidence probes.

[0036] The avionics instruments 6A, 6B, 6C are given as an example, any other known avionics instrument may be present, depending on the type of aircraft 2.

[0037] Avionics instruments are instruments certified by avionics safety standards and provide avionics information, in particular information relating to the movement parameters of the aircraft, in particular flight path vectors or FPV vectors (for "Flight Path Vector"), speed vectors, information relating to the attitude of the aircraft, positioning information in a fixed terrestrial reference frame, also called geo-referenced, for example GPS coordinates.

[0038] The aircraft is also equipped with a camera 8 having an associated field of vision.

[0039] For example, camera 8 is an optical camera with a wide field of view, eg a field of vision of at least 120°, for example providing successive digital images called full-field digital images, of large size. The size of a digital image is defined by its number of rows and its number of columns, in pixels. For example, full-field digital images are Ultra High Definition (or UHD) digital images, also called 4K images, for example of size 3840x2160 pixels. Alternatively, a larger size is considered.

[0040] For example, the camera 8 is fixed to the aircraft and oriented outwards, so that the field of vision of the camera encompasses an area located in front of the aircraft, with an angular orientation relative to a vertical axis, chosen so as to capture a part of the ground flown over when the aircraft is in flight.

[0041] Thus, the detection system 4 is capable of detecting both obstacles located on the ground and aerial obstacles when the aircraft 2 is in flight. The detection system 4 can be activated, for example by command, preferably when the aircraft is in flight.

[0042] Of course, other variants of positioning of the camera 8 are conceivable.

[0043] According to a variant, the aircraft 2 carries several optical cameras 8, positioned at various locations, for example with different orientations towards the exterior of the aircraft.

[0044] As an optional addition, one or more other cameras 10, of a type different from the optical cameras 8, are also on board, for example an infrared camera 10.

[0045] The aircraft comprises, in a known manner, an avionics system referenced 5, configured to implement avionics functionalities of a safety level certified by avionics safety standards, in particular configured to implement a flight management system FMS (from the English Flight Management System), a guidance system, a flight control system, etc. The obstacle detection system 4 comprises a peripheral computing platform 12.Unlike the avionics system 5 and more generally the on-board avionics instruments, the peripheral computing platform 12 comprises, in addition to a first communication interface 14, configured to communicate with the avionics instruments 6, the avionics system 5 and more generally with all the on-board equipment certified according to avionics safety standards, a second communication interface 16, allowing communication, by a wireless communication system, with external systems 18, 20, not subject to avionics safety standards.

[0046] For example, the first communication interface 14 is configured to communicate with the certified on-board equipment by wired communication or by wireless communication, for example Wifi or Bluetooth.

[0047] The second communication interface 16 is for example configured to communicate with non-certified external systems 18, 20, by a radio communication protocol, for example a standard 4G or 5G mobile telephony protocol, the SATCOM satellite communication protocol, the Datalink AOC / ATC protocol (for “Aeronautical Operation Control” / “Air Traffic Control”).

[0048] For example, the external system 18 is another aircraft, notably comprising a communication interface compatible with the communication interface 16.

[0049] For example, the external system 20 is a remote computing system located on the ground, for example a control center.

[0050] According to variants, the peripheral computing platform 12 comprises several second radio communication interfaces 16, according to several radio communication protocols, making it possible to communicate with external systems implementing different radio communication protocols.

[0051] The peripheral computing platform 12 is, in one embodiment, an electronic computing device further comprising a processing unit 22, comprising one or more processors (CPU or GPU), at least one electronic memory unit 24 and a human-machine interface 26, these elements being adapted to communicate via a communication bus (not shown).

[0052] In the example of [Fig.l] only one processing unit 22 and one unit of electronic memory 24 are represented.

[0053] The electronic memory unit 24 includes in particular memories of the RAM, ROM type, any type of non-volatile memory (for example FLASH, NVRAM).

[0054] The set of processing unit(s) 22 and electronic memory unit(s) 24 forms the local computing resources of the peripheral computing platform 12.

[0055] In one embodiment, the peripheral computing platform 12 is modular, with multiple processing and memory units being able to be added, allowing for dynamic increase in available resources, as needed.

[0056] The processing unit 22 is configured to execute: - a module 28 for receiving at least one full-field digital image captured by the on-board camera 8, and for receiving at least one piece of avionics information relating to a movement parameter of the aircraft; - a module 30 for determining a spatial position of an area of ​​interest of predetermined size in said full-field digital image, and for extracting said area of ​​interest from said full-field digital image, - an obstacle detection module 32 implementing a neural network previously trained to detect and classify obstacles, applied to the area of ​​interest of predetermined size; - a post-processing module 34.

[0057] For example, the obstacle detection module 32 implements a neural network previously trained by machine learning, for example a multi-layer neural network, of the convolutional neural network (or CNN) type or a recurrent neural network (RNN) trained by deep learning.

[0058] For example, the post-processing module 34 implements a calculation of a distance between the aircraft and the or each obstacle detected and / or a storage of a geo-referenced position for each obstacle detected belonging to a class of fixed obstacles.

[0059] The modules 28, 30, 32 and 34 are adapted to cooperate, as described in more detail below, to implement an obstacle detection method as described in more detail below, according to various embodiments.

[0060] In one embodiment, the modules 28, 30, 32 and 34 are produced in the form of software instructions forming a computer program, which, when executed by a computer, implements an obstacle detection method according to the invention.

[0061] In a variant not shown, the modules 28, 30, 32 and 34 are each produced in the form of programmable logic components, such as FPGAs (Field Programmable Gate Arrays) or microprocessors, GPGPU components (General-purpose processing or graphics processing), or even dedicated integrated circuits, such as ASICs (Application Specific Integrated Circuits). )•

[0062] The computer program comprising software instructions is further capable of being recorded on a medium, not shown, that is readable by a computer. The computer-readable medium is, for example, a medium capable of storing the 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 FLASH, NVRAM), a magnetic card or an optical card.

[0063] In one embodiment, the external system 20 comprises or is connected to an operational control center, comprising one or more interconnected programmable computing devices, or the external system 20 is connected to cloud computing equipment. Thanks to the communication link established via the second communication interface, the external system 20 is in bidirectional communication with the peripheral computing platform 12. This advantageously makes it possible to offload part of the calculations to be carried out to the external system 20, and thus to reduce the consumption of on-board computing and electrical resources. In other words, the platform 12 cooperates with the external system 20 for the implementation of the calculations, the results of the calculations carried out by the external system being transmitted to the peripheral computing platform 12 via the bidirectional communication link.

[0064] [Fig.2] schematically illustrates a full-field digital image I, for example a UHD image, of size C columns by L lines, with C=3840 and L=2160 pixels.

[0065] Two areas of interest referenced respectively Zi and Z2 are also represented. These areas of interest are sub-images of the full-field digital image I, of predetermined size N columns, M lines, with N <C et M<L.

[0066] For example, N is between 128 and 1024 and M is between 128 and 1024, for example N=1024 and M=1024 or N=1024 and M=512.

[0067] The areas of interest Zb Z2 represented in [Fig.2] are disjoint.

[0068] Alternatively, the determined areas of interest may partially overlap.

[0069] In addition, detected obstacles 01, 02 are schematically represented in [Fig.2]. The obstacles are schematically represented by the rectangles circumscribed by the objects. For example, the detected objects are pylons carrying electric cables.

[0070] [Fig. 3] also schematically illustrates a full-field digital image I, and an area of ​​interest determined in another embodiment. In the example of [Fig. 3], the position of the area of ​​interest Z is indicated by the aircraft flight trajectory vector V, representative of the direction of the trajectory at a time t. By example the area of ​​interest Z is centered on a point Po indicated by the vector V, which corresponds to the projection of the vector V on the field of vision of the camera generating the full-field digital image I.

[0071] Figures 2 and 3 illustrate exemplary embodiments, it being understood that other embodiments are conceivable, for example: tracking of an area of ​​interest on a plurality of successive digital images forming a video, with a chosen temporal sampling step; positioning of an area of ​​interest as a function of information provided by another on-board camera, for example an infrared camera; positioning of an area of ​​interest as a function of information coming from an external system, for example coming from another aircraft, via the second communication interface 16.

[0072] [Fig.4] is a block diagram of the main steps of an obstacle detection method according to one embodiment.

[0073] The method comprises a step 50 of receiving a digital image I, called a full-field digital image, captured by the camera 8. The steps of the method are then iterated, for example on all the full-field digital images captured by the camera, or on a subset of the captured full-field digital images, spaced apart by a chosen time step, or in other words for a digital image on X, the number X being chosen according to various constraints. In this case, tracking of the detected objects is additionally applied.

[0074] The position of the camera, for example in a spatial reference frame associated with the aircraft, the orientation of the camera and the calibration parameters of the camera are also obtained and stored, for use in ensuring the conformity of the generated images.

[0075] The method also optionally comprises a step 52 of acquiring avionics information from the on-board avionics instruments, comprising at least one item of information relating to a movement parameter of the aircraft, for example the flight trajectory vector. Preferably, other avionics information, for example the speed vector and the attitude of the aircraft are also provided, as well as the geo-referenced position of the aircraft. For example, the avionics information such as the attitudes of the aircraft (magnetic heading, attitude and inclination) come from an inertial unit.

[0076] The method further comprises a step 54 of determining a spatial position of an area of ​​interest in the digital image I received in step 50.

[0077] An area of ​​interest is preferably a rectangular area, of predetermined size, i.e. a number N of rows and a number M of columns, N and M being previously fixed integers. In other words, an area of ​​interest corresponds to a sub-image of NxM size of the digital image I.

[0078] A spatial position of the area of ​​interest in a chosen spatial reference frame is for example indicated by the coordinates of a corner, for example the upper left corner, of the area of ​​interest, in the chosen spatial reference frame. For example, the chosen spatial reference frame is the spatial reference frame associated with the digital image I.

[0079] For example, the upper left corner of the digital image I is the origin of the spatial reference system associated with the digital image I, and the coordinates of the respective selectable points in the digital image I are the associated row and column indices.

[0080] Alternatively, the spatial position of the area of ​​interest is indicated by the coordinates of the center of the rectangle forming the area of ​​interest in the chosen spatial reference frame.

[0081] It should be noted that knowing the position and orientation of the camera, it is easy to move from a spatial reference frame associated with the image to the spatial reference frame associated with the aircraft, then to a fixed (or geo-referenced) terrestrial reference frame knowing the attitudes of the aircraft.

[0082] Several embodiments are envisaged for determining 54 the spatial position of an area of ​​interest.

[0083] In a first embodiment, the spatial position is determined randomly. For example, a pseudo-random draw is implemented to determine the respective coordinates (x,y) of a point P, this point P then being chosen as the upper left corner of the area of ​​interest. This first embodiment is particularly suitable for a fixed obstacle mapping application, as indicated later with reference to steps 60 and 62.

[0084] In a second embodiment, avionics information is used to calculate a projection point of the flight trajectory vector (FPV vector), this point being for example chosen as the center of the area of ​​interest. This avionics information is for example the FPV vector and / or the derivative of the FPV vector, the attitudes of the aircraft, the speed vector of the aircraft, as well as the relative position of the camera 8 with respect to the aircraft. This embodiment is particularly suitable for avoiding possible collisions between the aircraft and an obstacle located on its trajectory.

[0085] In a third embodiment, the coordinates (x,y) of a point P, the point P being for example one of the corners of the area of ​​interest, are received from an external system at the determination step 54, for example from a second aircraft, distinct from the first aircraft which implements the method. This embodiment is particularly useful for a cooperative obstacle detection application, for example in the case where a first aircraft is equipped with more efficient sensors, for example a wide field of view camera, than the second aircraft(s). For example, in this embodiment, the coordinates of point P are expressed in the fixed (or geo-referenced) terrestrial reference frame and then transformed into coordinates in the spatial reference frame associated with the aircraft or into coordinates in the spatial reference frame associated with the digital image.

[0086] In a fourth embodiment, the coordinates (x,y) of a point P, the point P being for example one of the corners of the area of ​​interest, are obtained as a function of information provided by another on-board sensor, for example another on-board camera. This embodiment is particularly useful for an application of detecting obstacles detected in part by another sensor. In this embodiment also, the position information of each camera, for example in a reference frame associated with the aircraft, and of the orientation of each camera, as well as the respective calibration parameters are used to carry out a recalibration of the coordinates of the point P considered in a common spatial reference frame, for example the spatial reference frame associated with the aircraft.

[0087] The various embodiments of the determining step can also be combined, which results in the determination of several areas of interest by several distinct methods.

[0088] The method then comprises a step 56 of extracting the area of ​​interest, this step consisting of extracting the matrix of pixel values ​​corresponding to the area of ​​interest, forming an image of size NxM, sub-image of the digital image I.

[0089] This matrix is ​​then provided as input to an obstacle detection module implementing a neural network previously trained by machine learning to detect and classify obstacles. The obstacle detection module is then executed at obstacle detection step 58.

[0090] For example, several classes of obstacles are predetermined, these classes including in particular electricity pylons, electricity lines, wind turbines, electricity poles, antennas, stadium floodlights.

[0091] The obstacle detection module provides, at the end of step 58, a detection result of zero, one or more obstacles in the processed area of ​​interest, each obstacle having an associated class and a spatial position in the area of ​​interest. This then makes it possible to calculate the corresponding coordinates of the obstacle detected in the spatial reference frame associated with the aircraft and / or in the fixed terrestrial reference frame.

[0092] Preferably, the obstacle detection module implements a neural network.

[0093] The neural network comprises an ordered succession of layers of neurons, each of which takes its inputs from the outputs of the previous layer.

[0094] 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.

[0095] 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.

[0096] 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.

[0097] 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.

[0098] 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.

[0099] 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.

[0100] A convolutional neural network is also sometimes called a convolutional neural network or by the acronym CNN which refers to the English term “Convo-lutional Neural Networks”.

[0101] In a convolutional neural network, each neuron in the same layer has exactly the same connection pattern as its neighboring neurons, but at different input positions. The connection pattern is called a convolution kernel or, more often, "kernel" in reference to the corresponding English term.

[0102] A fully connected layer of neurons is a layer in which the neurons of said layer are each connected to all the neurons of the previous layer.

[0103] Such a type of layer is more often referred to as “fully connected” and sometimes referred to as “dense layer”.

[0104] In one embodiment, the YOLO (You Only Look) algorithms Once”) and SSD (from the English “Single Shot Detector”) are used for obstacle detection.

[0105] In certain embodiments, several areas of interest, which may partially overlap, are determined in step 54. In this case, each area of ​​interest is extracted and provided as input to obstacle detection step 58, which is applied as many times as there are areas of interest to be processed.

[0106] At the output of step 58, a plurality of obstacles detected in the area(s) of interest extracted from the digital image I, Oi,... On are obtained, where appropriate, each obstacle having an associated class, and being located in the area of ​​interest. For example, the position of each obstacle detected in the area of ​​interest is expressed by spatial coordinates in a chosen reference frame.

[0107] The method then comprises a post-processing step 60, which uses the result of obstacle detection step 58.

[0108] From the position of an obstacle O in the area of ​​interest, expressed by first coordinates (xl, yl), it is possible to obtain the position of the obstacle O in the digital image I is obtained expressed by second coordinates (x2, y2), and / or the position of the obstacle O in the spatial reference frame of the aircraft expressed by third coordinates (x3, y3, z3), and / or the position of the obstacle O in the fixed (or geo-referenced) terrestrial reference frame expressed by fourth coordinates (x4, y4, z4). The transition from one spatial reference frame to the other is carried out by affine transformations in a known manner, within the reach of those skilled in the art.

[0109] During post-processing 60, the relative distance between the aircraft and each detected obstacle can be calculated.

[0110] Furthermore, by using avionics information, for example the aircraft's speed of movement and the flight path vector, it is also possible to calculate a time remaining before possible impact if the trajectory direction is maintained.

[0111] Furthermore, by using the avionics information, in particular the geo-referenced position of the aircraft at the time of capture of the digital image I, it is easy to calculate, for each detected obstacle, and in particular for each fixed obstacle, the corresponding coordinates of the obstacle in the geo-referenced reference frame. The position of an obstacle expressed in the geo-referenced reference frame is called the geo-referenced position of the obstacle. The post-processing 60 may further comprise a storage of a geo-referenced position for each detected obstacle belonging to a class of fixed obstacles. This then makes it possible to enrich a map of fixed obstacles.

[0112] The method further comprises, optionally, an additional step 62 of exploiting the obstacle detection, comprising for example a display on the human-machine interface of the platform 12 and / or the transmission of the result of the obstacle detection, using the second communication interface, for example to a ground control center.

[0113] Furthermore, if the obstacle detection has made it possible to detect an obstacle on the trajectory of the aircraft, for example with a remaining time before possible impact Ti less than a predetermined time threshold Ts, an alarm is raised in step 62. Alternatively, an automatic modification of the trajectory to avoid a collision can be envisaged, for example if the aircraft is a drone without a pilot on board.

[0114] Thus, advantageously, the method allows the avoidance of collisions of aircraft in flight.

[0115] Furthermore, advantageously, the method also makes it possible to carry out mapping of fixed obstacles, thanks to an on-board solution compatible with existing on-board sensors and with low energy consumption.

[0116] The use of an obstacle detection module implementing a neural network makes it possible to obtain good obstacle detection accuracy, with a controlled computational cost thanks to the use of a zone of interest of predetermined size, compatible with rapid and efficient operation of the neural network.

[0117] In addition, extracting areas of interest from the full-field digital image rather than calculatingly reducing the size of the full-field digital image to the predetermined size compatible with the detection module improves the accuracy of obstacle detection.

[0118] It should be noted that the method described above recommends the extraction of areas of interest of predetermined size, the predetermined size corresponding to the size of the input data of the detection module implementing a neural network.

[0119] Alternatively, it is possible to implement several sizes, depending on several levels of qualitative performance of the algorithms implemented by the obstacle detection module.

Claims

Claims

1. A method for detecting obstacles implemented by an obstacle detection system on board an aircraft, the aircraft comprising at least one onboard camera (8, 10) having an associated field of vision, configured to acquire full-field digital images, the method being implemented by a processor (22) of a computing platform (12), and comprising steps of: a)-receiving (50) at least one full-field digital image captured by said at least one onboard camera, b)-determining (54) a spatial position of an area of ​​interest of predetermined size in said full-field digital image, and extracting (56) said area of ​​interest from said full-field digital image, c)-implementing an obstacle detection module (58) on said extracted area of ​​interest, implementing a neural network previously trained to detect and classify obstacles in images having said predetermined size,each detected obstacle being located in said area of ​​interest, further comprising a post-processing step (60) comprising a calculation of a distance between the aircraft and the or each detected obstacle and a storage of a geo-referenced position, in a fixed terrestrial reference frame, for each detected obstacle belonging to a class of fixed obstacles.,

2. Method according to claim 1, comprising repeating the determination (54) of a spatial position of an area of ​​interest in the same wide-field digital image, making it possible to obtain a plurality of areas of interest in said wide-field digital image.

3. Method according to one of claims 1 or 2, further comprising, before step b) of determining (54) a spatial position of an area of ​​interest, a step of acquiring (52) at least one piece of avionics information relating to a movement parameter of the aircraft, and in which the determination (54) of a spatial position is a function of at least one piece of avionics information.

4. A method according to claim 3, wherein said avionics information comprises a trajectory vector of the aircraft, the area of ​​interest being centered on a point indicated by the direction of said trajectory vector.

5. Method according to one of steps 1 to 4, wherein the determination (54) of a spatial position of an area of ​​interest comprises receiving a spatial position of the area of ​​interest by means of a communication interface (16) with an external system (18, 20).

6. Method according to one of claims 1 to 5, in which the determination (54) of a spatial position of an area of ​​interest comprises receiving a spatial position of the area of ​​interest from another sensor on board the aircraft.

7. Method according to any one of claims 1 to 6, the camera being configured to acquire a succession of full-field digital images forming a video, in which steps a) to c) are implemented on a subset of acquired digital images spaced temporally by a given time step, the method further comprising temporal tracking of the detected obstacles.

8. A computer program comprising software instructions which, when executed by a programmable electronic device, implement an obstacle detection method according to claims 1 to 7.

9. Obstacle detection system, adapted to be carried on an aircraft, the aircraft comprising at least one on-board camera (8, 10) having an associated field of vision, configured to acquire full-field digital images, the obstacle detection system comprising a computing platform (12) comprising at least one processor (22) configured to implement: - a module for receiving at least one full-field digital image captured by said at least one on-board camera, - a module for determining a spatial position of an area of ​​interest of predetermined size in said full-field digital image, and for extracting said area of ​​interest from said full-field digital image, - an obstacle detection module, taking said area of ​​interest as input and implementing a previously trained neural network to detect and classify obstacles in images having said predetermined size,each detected obstacle being located in said area of ​​interest, further comprising a post-processing module (34), configured to calculate a distance between the aircraft and the or each detected obstacle and to store a geo-referenced position, in a fixed terrestrial reference frame, for each detected obstacle belonging to a class of fixed obstacles.,

Images