FR3082013A1 - ELECTRONIC DEVICE FOR DRIVING A DRONE, DRONE, DRIVING METHOD AND COMPUTER PROGRAM THEREOF - Google Patents

Abstract

This electronic device (14) for piloting a drone (10) comprises a module (30) for acquiring at least one image taken by an image sensor, in particular a reference image of a scene including the area take-off, taken at a predefined altitude after take-off of the drone; a module (32) for calculating at least one secondary image from the acquired reference image, each secondary image corresponding to an altitude greater than the predefined altitude. It comprises a module (34) for detecting, in at least one secondary image, a set of characteristic portion (s); and a module (36) for determining a position of the drone relative to the take-off zone, by detecting in the current image acquired a current set of characteristic portion (s), then by comparing the current set of characteristic portion (s) with at least a portion of the detected set of characteristic portion (s) associated with the secondary image (s).

Description

The present invention relates to an electronic device for piloting a drone.

The invention also relates to a drone comprising such an electronic piloting device.

The invention also relates to a method for piloting the drone.

The invention also relates to a computer program comprising software instructions which, when executed by a computer, implement such a control method.

The invention relates to the field of drones, that is to say flying motorized devices controlled remotely. The invention relates in particular to rotary-wing drones capable of moving in the air by means of at least one rotor actuated by at least one motor, both single-rotor drones and multi-rotor drones, in particular quadricopters with four rotors. The invention also relates to fixed-wing drones.

An electronic piloting device is known comprising an acquisition module configured to acquire reference images of a scene, taken at one or more predefined altitudes, typically from 6 to 10 meters, after takeoff of the drone from a takeoff area. , said scene including the take-off area.

The piloting device comprises a detection module configured to detect a set of characteristic portion (s) in the reference images, and a guidance module configured to, during landing, guide the drone towards the area of initial takeoff, by detecting in a current acquired image a current set of characteristic portion (s), then by comparing the current set of characteristic portion (s) with the set of characteristic portion (s) ( s) detected in the reference images.

However, with such a piloting device, the return of the drone near its takeoff area is not optimal.

The object of the invention is then to propose an electronic piloting device making it possible to improve the return of the drone near its takeoff zone, while offering easier implementation.

To this end, the invention relates to an electronic device for piloting a drone, the device comprising:

an acquisition module configured to acquire at least one image taken by an image sensor fitted to the drone, the acquisition module being configured to acquire a reference image of a scene, taken at a predefined altitude, after the takeoff of the drone from a takeoff area, said scene including the takeoff area,

- a calculation module configured to calculate at least one secondary image from the acquired reference image, each secondary image corresponding to an altitude greater than the predefined altitude,

a detection module configured to detect, in at least one secondary image, a set of characteristic portion (s), and

- a determination module configured to determine a position of the drone relative to the take-off zone, by detecting in the current image acquired a current set of portion (s) characteristic (s), then by comparing the current set of portion (s) characteristic (s) with at least a portion of the detected set of characteristic portion (s) associated with the secondary image (s).

With the electronic control device according to the invention, the calculation module calculates at least one secondary image from the acquired reference image, each secondary image corresponding to an altitude greater than the predefined altitude, and the predefined altitude to which the reference image is taken may be weak, for example of the order of one meter.

With the electronic device for piloting the state of the art, the predefined altitude at which the reference image is taken is on the contrary high, generally greater than 5 meters, or even 10 meters, which is restrictive. This high predefined altitude is necessary so that the guidance module can then, when the drone lands from a hover at an altitude generally between 10 m and 20 m, find in the current image acquired one or more characteristic portions corresponding to those previously detected in the reference image.

According to other advantageous aspects of the invention, the electronic control device comprises one or more of the following characteristics, taken in isolation or in any technically possible combination:

- The device further comprises a guidance module configured to, on receipt of a landing instruction, guide the drone to a landing zone, depending on the position of the drone determined by the determination module;

the device also comprises a second acquisition module configured to acquire the selection of a return mode near the takeoff zone, and if said return mode is selected, the guidance module is then configured to, on reception of a landing instruction, guide the drone towards the take-off zone, according to the position of the drone determined by the determination module;

- the detection module is configured to detect, in each image among the reference image and the secondary image or images, a respective set of characteristic portion (s);

- the determination module is configured to compare the current set of characteristic portion (s) with each detected set of characteristic portion (s) for the reference image and the secondary image (s);

- the determination module is configured to compare the current set of characteristic portion (s) with the detected set of characteristic portion (s) corresponding to the image which among the reference image and the or the secondary images is associated with the altitude closest to a current altitude of the drone; and

the determination module is further configured to modify an orientation of the current image as a function of an orientation of the reference image, prior to detection in the current image acquired of the current portion assembly (s ) characteristics).

The invention also relates to a drone comprising an electronic piloting device, in which the electronic piloting device is as defined above.

The invention also relates to a method for piloting a drone, the method being implemented by an electronic device and comprising the steps consisting in:

acquire at least one image taken by an image sensor fitted to the drone, said step comprising the acquisition of a reference image of a scene, taken at a predefined altitude after takeoff of the drone from a takeoff area, said scene including the takeoff area,

- calculate at least one secondary image from the acquired reference image, each secondary image corresponding to an altitude greater than the predefined altitude,

- detecting, in at least one secondary image, a set of characteristic portion (s), and

- determine a position of the drone relative to the take-off zone, by detecting in the current image acquired a current set of characteristic portion (s), then by comparing the current set of characteristic portion (s) ) with at least a portion of the detected set of characteristic portion (s) associated with the secondary image (s).

The invention also relates to a computer program comprising software instructions which, when executed by a computer, implement a control method as defined above.

These characteristics and advantages of the invention will appear more clearly on reading the description which follows, given solely by way of nonlimiting example, and made with reference to the appended drawings, in which:

- Figure 1 is a schematic representation of a drone comprising an electronic piloting device, the electronic piloting device comprising in particular an image acquisition module (s) taken (s) by an image sensor equipping the drone , in particular of a reference image of a scene, taken at a predefined altitude, after takeoff of the drone from a takeoff area, a module for calculating at least one secondary image from the reference image acquired, each secondary image corresponding to an altitude greater than the predefined altitude, a detection module, in at least one secondary image, of a set of characteristic portion (s), and a module for determining a position of the drone relative to the take-off zone, by detecting in a current acquired image a current set of characteristic portion (s), then by comparing the current set of characteristic portion (s) eu (s) with at least a portion of the detected set of characteristic portion (s) associated with the secondary image (s);

- Figures 2 and 3 illustrate the calculation of secondary images from the reference image acquired for different altitudes;

- Figure 4 is a flowchart of a method of piloting the drone according to the invention;

- Figure 5 is a representation of an example of state machine used to perform the control method according to the invention;

- Figure 6 is a comparison of positions of the drone determined by the piloting device, represented in the form of points, with those determined by an inertial unit equipping the drone, represented in the form of a curve;

- Figure 7 is a view of an acquired reference image;

- Figure 8 is a view of a current acquired image; and

FIG. 9 illustrates the detection in the current image acquired of characteristic portion (s) corresponding to the characteristic portion (s) associated with the secondary image (s), and the determination of '' an altitude of the drone from this detection.

In the remainder of the description, the expression “substantially equal to” is understood to mean an equality relationship of plus or minus 10%, that is to say with a variation of at most 10%, of more preferably as an equal relation at plus or minus 5%, that is to say with a variation of at most 5%. Likewise, the expression "substantially parallel to"

is understood as a relation of parallelism at plus or minus 10 °, more preferably still as a relation of parallelism at plus or minus 5 °.

In FIG. 1, a drone 10, that is to say an unmanned aircraft on board, comprises a camera 12 and an electronic piloting device 14.

The drone 10 is a remote-controlled motorized flying machine, in particular via a joystick 16 equipped with a display screen 18, allowing the user to enter his flight commands.

The drone 10 is for example a rotary wing drone, comprising at least one rotor 20. In FIG. 1, the drone 10 comprises a plurality of rotors 20, and is then called a multirotor drone. The number of rotors 20 is for example equal to 4, and the drone 10 is then a quadrotor drone. Alternatively, not shown, the drone 10 is a fixed-wing drone.

The drone 10 comprises a transmission module 22 configured to exchange data, preferably by radio waves, with one or more electronic equipment, in particular with the joystick 16, or even with other electronic equipment for the transmission of the acquired image or images. by the camera 12 and / or other information relating to the drone 10, such as an altitude, an inclination, a speed, a flight status, a quality of GPS or Gallileo geolocation, a charge of a battery equipping the drone 10.

The camera 12 is for example a camera mounted on a nacelle (from English gimbal), preferably stabilizer, not shown. As a variant, the camera 12 is a front view camera making it possible to obtain an image of a scene towards which the drone is oriented. As a further variant, the camera 12 is a vertical view camera, not shown, pointing towards the low and configured to capture successive images of a terrain overflown by the drone 10.

The camera 12 includes at least one image sensor (s) and a lens associated with the image sensor (s), not shown. The image sensor (s) comprises for example a matrix photodetector comprising a plurality of photosites, each photosite corresponding to a respective pixel of the image taken by the sensor. The image sensor (s) then extends along an extension plane, and the extension plane corresponds to the plane of the matrix photodetector. The objective is for example a fixed focal objective, such as a hemispherical objective of hypergone type (from the English fisheye), that is to say covering a visual field having an angle of view, also called field of view. sight and noted FOV (from the English Field Of View), of high value, of the order of 180 ° or more. The lens is positioned in front of the image sensor (s) so that the image sensor (s) picks up images through the lens. Alternatively, the lens is a lens with a variable focal length.

The electronic control device 14 comprises a first acquisition module 30 configured to acquire at least one image taken by the image sensor equipping the drone 10, the first acquisition module 30 being configured to acquire a reference image l _R a scene, taken at a predefined altitude Z _re f, after takeoff of the drone 10 from a takeoff area, said scene including the takeoff area.

The electronic control device 14 comprises a calculation module 32 configured to calculate at least one secondary image Is from the acquired reference image l _R , each secondary image Is corresponding to an altitude greater than the predefined altitude Zref.

The electronic control device 14 also comprises a detection module 34 configured to detect, in at least one secondary image Is, a set of characteristic portion (s), and a determination module 36 configured to determine a position of the drone relative to the take-off area, by detecting in the current image acquired a current set of characteristic portion (s), then by comparing the current set of characteristic portion (s) with at least one portion of the detected set of characteristic portion (s) associated with the secondary image (s) Is.

In optional addition, the electronic piloting device 14 further comprises a guidance module 38 configured to, on receipt of a landing instruction, guide the drone 10 towards a landing zone, as a function of the position of the drone 10 determined by the determination module 36.

As an additional optional element, the electronic piloting device 14 further comprises a second acquisition module 39 configured to acquire the selection of a return mode near the take-off area, and if said return mode is selected, the guidance module 38 is then configured to, on receipt of a landing instruction, guide the drone 10 towards the take-off zone, as a function of the position of the drone 10 determined by the determination module 36. In other words, the module guide 38 is then configured to guide the drone 10 to a landing zone close to the take-off zone.

In the example of FIG. 1, the electronic control device 14 comprises an information processing unit 40, formed for example by a memory 42 and a processor 44, such as a processor of the CPU type (of English Central Processing Unit), of GPU type (from English Graphies Processing Unit), or of VPU type (from English Vision Processing Unit), associated with memory 42.

In the example of FIG. 1, the electronic piloting device 14 is on board the drone 10.

Those skilled in the art will understand that the electronic piloting device 14 is not necessarily on board the drone 10 and that, as a variant, not shown, the electronic piloting device 14 is at least partially external to the drone 10. According to this as a variant, the electronic control device 14 is for example at least partially on board the controller 16, in particular when the controller 16 is a smartphone. The calculation module 32 and the detection module 34 are for example arranged outside of the drone 10, and are distinct from the drone 10, and are then typically embedded in said controller 16.

In another variant, the electronic piloting device 14 is totally, or even entirely, disposed outside of the drone 10. The whole of the electronic piloting device 14 is then separate from the drone 10, and for example on board the joystick 16 , especially when it is in the shape of a computer or an electronic tablet.

The controller 16 is known per se and allows the drone 10 to be piloted. In the example of FIG. 1, the controller 16 is implemented via a computer (smartphone) or an electronic tablet, comprising the screen display 18, preferably tactile.

In a variant not shown or in addition, the lever 16 comprises two grips, each intended to be grasped by a respective hand of the pilot, a plurality of control members, including two joysticks (from the English Joystick) , each being arranged near a respective grip handle and being intended to be actuated by the pilot, preferably by a respective thumb.

The controller 16 comprises a radio antenna and a radio transceiver, not shown, for the exchange of data by radio waves with the drone 10, both in uplink and in downlink.

In the example of FIG. 1, the first acquisition module 30, the calculation module 32, the detection module 34 and the determination module 36, as well as an optional complement the guidance module 38 and the second module 39, are each produced in the form of software executable by the processor 44. The memory 42 of the information processing unit 40 is then able to store a first acquisition software configured to acquire at least one image taken by the image sensor equipping the drone 10, the first acquisition software being configured to acquire the reference image l _R of a scene, taken at the predefined altitude Z _re f, after takeoff of the drone 10 from the take-off zone, calculation software configured to calculate at least one secondary image Is from the acquired reference image l _R , each secondary image Is corresponding to an altitude greater than the predefined altitude Z _re f, a detection software configured to detect, in at least one secondary image Is, the set of characteristic portion (s), and determination software configured to determine a position of the drone relative to the take-off zone, by detecting in the current image acquired the current set of characteristic portion (s), then by comparing the current set of characteristic portion (s) with at least one portion of the detected set of portion (s) ) characteristic (s) associated with the secondary image (s) Is. In optional addition, the memory 42 of the information processing unit 40 is capable of storing guidance software configured for, on receipt of an instruction landing, guide the drone 10 towards the landing zone, according to the position of the drone 10 determined by the determination software. In addition, as an optional addition, the memory 42 of the information processing unit 40 is capable of storing a second acquisition software configured to acquire the selection of a return mode near the take-off area, and if said Return mode is selected, the guidance software is then configured to, on receipt of a landing instruction, guide the drone 10 towards the take-off zone, according to the position of the drone 10 determined by the determination software. The processor 44 is then able to execute the first acquisition software, the calculation software, the detection software and the determination software, as well as optional guidance software and the second acquisition software.

In a variant not shown, the first acquisition module 30, the calculation module 32, the detection module 34 and the determination module 36, as well as an optional supplement the guidance module 38 and the second acquisition module 39 , are each produced in the form of a programmable logic component, such as an FPGA (from the English Field Programmable Gate Array), or alternatively in the form of a dedicated integrated circuit, such as an ASIC (from the English Application Specific Integrated Circuit).

When the electronic control device 14 is produced in the form of one or more software, that is to say in the form of a computer program, it is also able to be recorded on a medium, not shown, readable by 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 disc, a magneto-optical disc, 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. On the readable medium is then stored a computer program comprising software instructions.

The calculation module 32 is configured to calculate at least one secondary image Is, preferably several secondary images Is, from the acquired reference image l _R , each secondary image Is corresponding to a respective altitude greater than the predefined altitude Z _re f which is associated with the acquired reference image l _R.

In the example of FIGS. 2 and 3, the secondary images Is calculated correspond to successive octaves of an octave 0 associated with the acquired reference image Ir.

In other words, the calculation module 32 is for example configured to calculate the secondary images Is from the acquired reference image Ir, according to an implementation of pyramidal type, as shown in FIG. 2, the pyramid 50 comprising several levels 52 and the image l ^K associated with a respective level K being an undersampled image of the image l ^K_1 associated with the previous level K-1. K is an index representing the level and between 0 and N, 0 designating the initial level corresponding to the acquired reference image Ir and 1, ..., N designating the following levels each corresponding to a calculated secondary image Is.

The most subsampled image l ^N then corresponds to the final level N, or even to the upper level of the pyramid 50. The image l ° associated with the initial level 0, or even at the lower level of the pyramid 50, is l reference image Ir as acquired by the first acquisition module 30. In other words, the image l ° associated with the initial level 0 is not subsampled. In terms of notations, the image l ° designates the reference image Ir, and the images I ¹ , ..., I ^N designate the secondary images Is calculated successively, these being for example denoted Is ¹ for the first calculated secondary image corresponding to level 1, Is ² for the second calculated secondary image corresponding to level 2 and to the image I ² , ... and ls ^N for the last calculated secondary image corresponding to the final level N and to the image l ^N.

The subsampling factor F from one level K to the next level K + 1 is for example equal to 2. The number of levels 52, equal to N + 1, is, for example, between 2 and 5, and of preferably equal to 4. In other words, N is, for example, between 1 and 4, and preferably equal to 3.

The acquired reference image Ir has a small side A and a large side B, each expressed in number of pixels, ie a resolution B x A. The first calculated secondary image Is ¹ corresponding to level 1 then has an equal resolution (B / F) x (A / F), the second calculated secondary image Is ² corresponding to level 2 then has an equal resolution (B / F ² ) x (A / F ² ), and the last calculated secondary image ls ^N corresponding to final level N then has an equal resolution (B / F ^N ) x (A / F ^N ).

In the example of FIGS. 2 and 3, where F is equal to 2 and where A and B are respectively equal to 240 pixels and 360 pixels, the acquired reference image Ir has a resolution of 320 x 240 pixels, the first image calculated secondary Is ¹ has a resolution of 160 x 120 pixels, the second calculated secondary image Is ² has a resolution of 80 x 60 pixels, and the third calculated secondary image Is ³ has a resolution of 40 x 30 pixels, N being equal to 3 in this example.

The subsampling of the image l ^K associated with a level K with respect to the image l ^K_1 associated with the previous level K-1 is, for example, obtained by calculating each pixel of the image l ^K from F ² pixels of image l ^K_1 , each pixel of image l ^K being for example an average of F ² corresponding pixels of image l ^K_1 , F representing the subsampling factor. In the example of FIGS. 2 and 3, F is equal to 2, and each pixel of the image l ^K then depends on 4 pixels of the image l ^K_1 , for example being an average of these 4 pixels.

The altitude corresponding to each secondary image Is calculated is for example a respective multiple of the predefined altitude Z _re f which is associated with the acquired reference image Ir.

In the example of the pyramid-type implementation of FIG. 2, this respective multiple is a multiple depending on the factor F of sub-sampling from one level K to the next level K + 1.

The altitude corresponding to the first calculated secondary image Is ¹ is then for example equal to F x Zref, that corresponding to the second calculated secondary image Is ² is equal to F ² x Zref, and that corresponding to the last calculated secondary image ls ^N equal to F ^N x Z _re f.

The altitude corresponding to each calculated secondary image Is then checks for example the following equation:

Z _i = F ⁱ -Z _ref (1) where Z represents the altitude corresponding to the i ^th calculated secondary image, i is an integer index between 1 and N, representing the number of the secondary image calculated among the succession calculated secondary images Is,

F represents the subsampling factor of the reference image acquired Ir at the 1 ^st calculated secondary image, or of a secondary image calculated at the following, and

Z _re f represents the predefined altitude.

The predefined altitude Z _re f is for example between 1 meter and 5 meters.

In the example of FIGS. 2 and 3, where F is equal to 2 and N is equal to 3, for a predefined altitude Z _re f equal to 1 meter, the respective altitudes corresponding to the N calculated secondary images Is are then equal to 2 meters, 4 meters and 8 meters. Similarly, for a predefined altitude Z _re f equal to 3 meters, the respective altitudes corresponding to the N calculated secondary images Is are then equal to 6 meters, 12 meters and 24 meters.

The detection module 34 is configured to detect a set of characteristic portion (s) in at least one secondary image Is, preferably in each image from the reference image l _R and the secondary image (s) Is.

The detection module 34 is configured to first detect a set of characteristic point (s) in the corresponding image Is, Ir via the implementation of a characteristic point (s) detection algorithm ), then to characterize, via the implementation of a characterization algorithm, a portion of the image associated with each characteristic point detected.

Those skilled in the art will understand that the portion of the image associated with each characteristic point detected corresponds to a neighborhood of this point, also called a descriptor, such a portion then being more likely to be identified by comparison in a new image that 'a simple point.

The characteristic point (s) detection algorithm is, for example, a characteristic algorithm from accelerated segment tests, also called FAST (from the English Features from Accelerated Segment Test), or a detection algorithm Harris-type corners, also called Harris Corner (from the English Harris Corner Detector).

The characterization algorithm is, for example, a BRIEF algorithm (from the English Binary Robust Independent Elementary Features), or an ORB algorithm which is a combination of the FAST and BRIEF algorithms (from the English Oriented FAST and rotated BRIEF), or a BRISK algorithm (from the English Binary Robust Invariant Scalable Keypoints), or a FREAK algorithm (from the English Fast REtinA Keypoint).

As a variant, the algorithm implemented by the detection module 34 is for example an algorithm for transforming visual characteristics invariant to scale, also called SIFT (from the English Scale-Invariant Feature Transform), which is times a characteristic point (s) detection and characterization algorithm, by providing as output the characteristic portion (s) detected, in the form of SIFT descriptors.

In another variant, the algorithm implemented by the detection module 34 is for example an algorithm of accelerated robust characteristics, also called SURF (from English Speeded Up Robust Features), which is both a detection algorithm characteristic and characterization point (s), by providing as output the characteristic portion or portions detected, in the form of SURF descriptors.

The determination module 36 is configured to determine the position of the drone 10 relative to the take-off zone, by detecting in the current image acquired a current set of characteristic portion (s).

The determination module 36 is preferably configured to then compare the current set of characteristic portion (s) with each detected set of characteristic portion (s) for the reference image l _R and the one or more secondary images Is.

The determination module 36 is preferably further configured to compare the current set of characteristic portion (s) with the detected set of characteristic portion (s) corresponding to the image which among the image of reference l _R and the secondary image or images Is is associated with the altitude closest to a current altitude of the drone 10.

The current altitude is for example estimated via an inertial unit, not shown, or even via a geolocation device, such as a satellite positioning device, for example GPS or Gallileo, equipping the drone 10.

By way of example, assuming the current altitude of the drone 10 substantially equal to 20 meters and the altitudes corresponding to the N calculated secondary images Is respectively equal to 6 meters, 12 meters and 24 meters with the previous example of the altitude predefined Z _re f equal to 3 meters, then the determination module 36 is configured to compare in this case the current set of characteristic portion (s) with the detected set of associated characteristic portion (s) at the 3 ^rd secondary image calculated, the current altitude of 20 meters being closer to 24 meters than to 12 meters.

The determination module 36 is further configured to modify an orientation of the current image as a function of an orientation of the reference image l _R , prior to detection in the current image acquired of the current portion assembly. (s) characteristic (s).

The determination module 36 is then preferably configured to modify the orientation of the current image so that it is identical to that of the reference image l _R , by applying to the current image a rotation d angle equal to the angular difference between the orientation of the reference image l _R and the orientation of the current image before modification. This allows the image taken into account then to have the same orientation as the reference image l _R.

The determination module 36 is then configured to establish a correspondence between one or more characteristic portions of the current set of characteristic portions associated with the current image and one or more characteristic portions of the detected set of characteristic portions associated with the or to the secondary images Is or even to the reference image l _R. The determination module 36 is for example configured to establish this correspondence via a RANSAC method (from the English RANdom SAmple Consensus), and the correspondences between characteristic portions are deduced from the "inliers" obtained via the RANSAC method, ie - say relevant data, as opposed to "outliers" corresponding to non-relevant data.

The determination module 36 is then, by virtue of this or these correspondences, able to identify the take-off zone in the current image, then to determine the position of the drone 10 relative to the take-off zone as a function of the location of the representation of the take-off area identified inside the current image.

The operation of the electronic piloting device 14 according to the invention will now be explained with reference to FIG. 4 showing a flow diagram of the method for piloting the drone 10 according to the invention.

During each step 100, the piloting device 14 acquires, via its first acquisition module 30, at least one image taken by the image sensor equipping the drone 10.

Initially, when the drone 10 is substantially at the predefined altitude Z _re f after takeoff of the drone 10, the piloting device 14 acquires, via its first acquisition module 30, the reference image l _R of a scene including the take-off area.

The control device 14 calculates, then during the next step 110 and via its calculation module 32, at least one secondary image Is from the reference image l _R acquired, each secondary image Is corresponding to a higher altitude at the predefined altitude Z _re f.

The control device 14 preferably calculates several secondary images Is from the acquired reference image l _R , according to an implementation of the pyramidal type, as shown in FIG. 2.

The altitude associated with each calculated secondary image Is verifies for example the previous equation (1).

During the next step 120, the control device 14 detects, via its detection module 34, a set of characteristic portion (s) in at least one secondary image Is, and preferably in each image among the reference image l _R and the secondary image (s) Is.

This detection includes the implementation of a characteristic point (s) detection algorithm, such as FAST or Harris Corners; followed by the implementation of a characterization algorithm, such as ORB, BRISK, FREADK, or even BRIEF. Those skilled in the art will observe that, as a variant, the characteristic point (s) detection algorithm and the characterization algorithm are a single algorithm, such as the SIFT algorithm or even the SURF algorithm.

The piloting device 14 then determines, during the next step 130 and via its determination module 36, the position of the drone 10 relative to the take-off zone, by detecting the current assembly of characteristic portion (s). ) in the current image acquired, then by comparing said current set of characteristic portion (s) with at least a portion of the detected set of characteristic portion (s) associated with the secondary image (s) Is see the reference image l _R.

During the next step 140, the electronic piloting device 14 determines, via its second acquisition module 39, whether the return mode near the takeoff area has been selected by the user or predefined according to the type of drone 10, and if necessary acquires this selection via its second acquisition module 39.

If, during step 140, no selection of the return mode near the take-off area is acquired, the electronic piloting device 14 then returns to step 100 to acquire a new current image.

When a selection of the return mode near the takeoff zone has been acquired during step 140, the electronic piloting device 14 then goes to step 150 to guide, via its guidance module 38, the drone 10 to a landing zone as close as possible to the take-off zone, as a function of the position of the drone 10 determined by the determination module 36.

At the end of step 150, the electronic control device 14 returns to step 100 to acquire a new current image.

Those skilled in the art will observe that steps 110 and 120 are carried out only once after the initial acquisition of the reference image l _R during step 100, and that when the electronic control device 14 then returns to l step 100 to acquire a current image or a new current image, it goes directly from the acquisition step 100 to step 130 of determining the position of the drone 10 relative to the take-off zone.

In FIG. 5, representing an example of a state machine implemented to carry out the piloting method according to the invention, during an inactive state 200, the drone 10 is placed, that is to say in contact from the ground, pending a take-off instruction.

The piloting device 14 remains in this inactive state 200 while looping according to arrow F1, as long as it has not received a take-off instruction. In this inactive state 200, the piloting device 14 acquires, via a measurement device, such as the inertial unit, a position of the drone 10 in this state, that is to say a take-off position.

After receiving a take-off instruction, the piloting device 14 changes, according to arrow F2, from the inactive state 200 to an initial state 210.

In this initial state 210, the piloting device 14 acquires, via the measuring device, a new position of the drone, also called the initial position, and compares it to the take-off position of the drone 10 to determine whether conditions for passing through a learning state 220 are fulfilled or not.

The conditions for passing into learning state 220 are, for example, a deviation of less than 0.5 meters along the longitudinal X and transverse Y axes of the terrestrial frame of reference between the initial position and the take-off position, and an altitude in the initial position. greater than or equal to the predefined altitude Z _re f with respect to the ground, i.e. a deviation greater than the predefined altitude Z _re f along the vertical axis Z of the terrestrial frame of reference between the initial position and the position takeoff.

If during the transition from the inactive state 200 to the initial state 210, the drone 10 has shifted too much laterally with respect to the take-off position, and the difference between the initial position and the take-off position is greater than 0.5 meters along one of the longitudinal X and transverse axes Y of the terrestrial frame of reference, then the control device 14 returns to the inactive state 200 according to arrow F3. The landing aid of the drone 10 via the detection of the take-off area in the current image is therefore not implemented for this flight.

From the initial state 210, the control device 14 also returns to the inactive state 200 according to arrow F3, in the event of reception of a landing instruction from the user, of an emergency instruction, by example in the event of insufficient remaining electric power, or else an urgent landing instruction, for example in the event of too strong wind or collision with an obstacle.

From the initial state 210, the control device 14 switches to the learning state 220 according to the arrow F4 as soon as the aforementioned passage conditions are met.

Finally, if the difference between the initial position and the take-off position is less than 0.5 meters along the longitudinal X and transverse Y axes of the terrestrial frame of reference, but the altitude relative to the ground is not yet at l 'predefined altitude Z _re f, the control device 14 remains in this initial state 210 by looping according to arrow F4', until the altitude relative to the ground is equal to the predefined altitude Z _re f.

In this learning state 220, carried out substantially at the predefined altitude Z _re f, the control device 14 acquires the reference image l _R of the scene including the take-off zone, then calculates at least one secondary image Is, and preferably several secondary images Is, from the acquired reference image l _R , and then detects a set of characteristic portion (s) in at least one secondary image Is, and preferably in each image among l reference image l _R and the secondary image (s) Is.

If during this learning state 220, the control device 14 does not detect any characteristic portion, the reference image l _R containing no particular characteristic, then the control device 14 returns to the inactive state 200 according to the arrow F5.

The landing aid for drone 10 via the detection of the take-off area in the current image is no longer used for this flight.

After having carried out the aforementioned operations in the learning state 220 and detected at least one characteristic portion, goes according to the arrow F6 to the ready state 230, indicating that the control device 14 is ready to implement a function d precision landing, with assistance with the landing of the drone 10 via the detection of the take-off zone in the current image, so that the landing is carried out as close as possible to the take-off zone.

The piloting device 14 remains in this ready state 230 while looping according to arrow F7, as long as it has not received a landing instruction. This ready state 230 corresponds to a main flight phase or to a hover phase before landing.

After receiving a landing instruction, the piloting device 14 then passes, according to arrow F8, from the ready state 230 to a precision landing state 240.

From the ready state 230, the piloting device 14 also returns to the inactive state 200 according to the arrow F9, in the event of receiving an emergency instruction, or even an urgent landing instruction. The landing aid for drone 10 via the detection of the take-off area in the current image is no longer used for this flight.

From the precision landing state 240, the piloting device 14 returns according to arrow F10 to the ready state 230, in the event of receipt of a flight instruction canceling the previously received landing instruction, such as '' a drone climb instruction 10.

The piloting device 14 remains in this precision landing state 240, looping along arrow F11, as long as the landing instruction remains active.

In this precision landing state 240, the piloting device 14 implements the precision landing function, with assistance in landing the drone 10 via the detection of the take-off zone in the current image. To this end, the piloting device 14 determines the position of the drone 10 relative to the take-off zone, by detecting the current set of characteristic portion (s) in the current image acquired, then by comparing said set characteristic portion (s) current with at least a portion of the detected set of characteristic portion (s) associated with the secondary image (s) Is or even with the reference image I _R , then guide progressively the drone 10 to a landing zone as close as possible to the take-off zone, as a function of the position of the drone 10 thus determined.

When the drone 10 is placed, and is then again in contact with the ground, the piloting device 14 finally returns to the inactive state 200 according to arrow F12.

In FIG. 6, the positions of the drone 10 determined by the piloting device 14 are represented successively in time T, indicated on the abscissa, in the form of points 300 and 310 for the coordinates along the axes X and Y of the reference frame considered, and of points 320 for the coordinates along the vertical axis Z of the reference system considered, by a distance D, expressed in meters and indicated on the ordinate, with respect to the take-off position, corresponding to the origin of the reference system considered.

In FIG. 6, the positioning in the high state of an indicator 340 indicates the phases during which the positions of the drone 10 are not determined by the piloting device 14 according to the invention, either because the drone 10 is too distant from the take-off zone and that the latter then no longer appears in the current image acquired, corresponding to the main phase during which the indicator 340 is in the high state; either because the drone 10 is in contact with or in close proximity to the ground, corresponding to the final phase during which the indicator 340 is in the high state.

A curve 350 represents the coordinates, along the vertical axis Z of the reference system considered, of the positions of the drone 10 successively supplied over time T by the inertial unit equipping the drone 10.

Those skilled in the art will then be able to observe that the piloting device 14 according to the invention makes it possible to determine, with satisfactory precision, the position of the drone 10 during the takeoff and landing phases, by comparison with that coming from the power plant. inertial.

During the landing phase, the piloting device 14 according to the invention even provides a position of the drone 10 more precise than that provided by the inertial unit equipping the drone 10, the position coming from the inertial unit joining that determined by the piloting device 14 at the end of the landing phase.

Figures 7 to 9 show examples of images used by the control device 14 according to the invention. FIG. 7 is an example of acquired reference image l _R and FIG. 8 is a view 400 of a current acquired image. FIG. 9 is a view 410 of the current image acquired, with in addition the detection in the current image acquired of characteristic portion (s) corresponding to the associated characteristic portion (s). ) the secondary image (s), making it possible to find the acquired reference image l _R indicated by the rectangular box, then to determine the altitude of the drone 10 from this detection, the determined altitude being equal to 11, 36 m in this example.

Thus, the electronic control device 14 according to the invention makes it possible to avoid the acquisition of the reference image l _R at a predefined altitude Z _re f which is too high, by calculating at least one secondary image from the image. reference acquired, each secondary image corresponding to an altitude greater than the predefined altitude. The predefined altitude Z _re f at which the reference image is taken is then preferably low, for example of the order of one meter, which also makes it easier to have a reference image l _R which is more representative of the take-off area.

With the electronic device for piloting the state of the art, the predefined altitude at which the reference image is taken is, on the contrary, high, generally greater than 5 meters, or even 10 meters, which is restrictive. This high predefined altitude is, with the electronic piloting device of the state of the art, necessary so that the guidance module can then, when the drone lands from a hovering flight at an altitude generally between 10 m and 20 m, find in 10 the current acquired image one or more characteristic portions corresponding to those previously detected in the reference image.

It can thus be seen that the electronic piloting device 14 and the piloting method according to the invention make it possible to improve the return of the drone 10 near its takeoff zone, while offering easier implementation.

Claims (10)

1. Electronic device (14) for piloting a drone (10), the device (14) comprising: - an acquisition module (30) configured to acquire at least one image taken by an image sensor fitted to the drone (10), the acquisition module (30) being configured to acquire a reference image (lr) d a scene, taken at a predefined altitude (Z _re f), after takeoff of the drone (10) from a takeoff area, said scene including the takeoff area, - a calculation module (32) configured to calculate at least one secondary image (ls) from the acquired reference image (Ir), each secondary image (Is) corresponding to an altitude (F 'x Z _re f) higher than the predefined altitude (Z _re f), a detection module (34) configured to detect, in at least one secondary image (Is), a set of characteristic portion (s), and - a determination module (36) configured to determine a position of the drone (10) relative to the take-off zone, by detecting in the current image acquired a current set of characteristic portion (s), then by comparing the current set of characteristic portion (s) with at least one portion of the detected set of characteristic portion (s) associated with the secondary image (s). 2. Device (14) according to claim 1, wherein the device (14) further comprises a guidance module (38) configured to, on receipt of a landing instruction, guide the drone (10) towards an area landing, according to the position of the drone (10) determined by the determination module (36). 3. Device (14) according to claim 2, in which the device (14) further comprises a second acquisition module (39) configured to acquire the selection of a return mode near the take-off area, and if said return mode is selected, the guidance module (38) is then configured to, on receipt of a landing instruction, guide the drone (10) towards the take-off area, as a function of the position of the drone ( 10) determined by the determination module (36). 4. Device (14) according to any one of the preceding claims, in which the detection module (34) is configured to detect, in each image among the reference image (Ir) and the secondary image or images (Is) , a respective set of characteristic portion (s). 5. Device (14) according to any one of the preceding claims, in which the determination module (36) is configured to compare the current set of characteristic portion (s) with each detected set of portion (s) characteristic (s) for the reference image (l _R ) and the secondary image (s) (Is). 6. Device (14) according to any one of claims 1 to 4, in which the determination module (36) is configured to compare the current set of portion (s) characteristic (s) with the detected set of portion (s) characteristic (s) corresponding to the image which among the reference image (l _R ) and the secondary image (s) (Is) is associated with the altitude closest to a current altitude of the drone (10 ). 7. Device (14) according to any one of the preceding claims, in which the determination module (36) is further configured to modify an orientation of the current image as a function of an orientation of the reference image ( l _R ), prior to detection in the current image acquired of the current set of characteristic portion (s). 8. Drone (10) comprising an electronic piloting device (14), characterized in that the electronic piloting device (14) is according to any one of the preceding claims. 9. Method of piloting a drone (10), the method being implemented by an electronic device (14) and comprising the steps consisting in: - acquire (100) at least one image taken by an image sensor fitted to the drone (10), said step (100) comprising the acquisition of a reference image (l _R ) of a scene, taken at a predefined altitude (Z _re f) after takeoff of the drone (10) from a takeoff area, said scene including the takeoff area, - calculate (110) at least one secondary image (Is) from the acquired reference image (l _R ), each secondary image (Is) corresponding to an altitude (F 'x Z _re f) greater than the altitude predefined (Z _re f), - detect (120), in at least one secondary image (Is), a set of characteristic portion (s), and - determining (130) a position of the drone (10) relative to the take-off zone, by detecting in the current image acquired a current set of portion (s) characteristic (s), then by comparing the current set of portion (s) characteristic (s) with at least a portion of the detected set of characteristic portion (s) associated with the secondary image (s). 5 10. Computer program comprising software instructions which, when executed by a computer, implement a method according to the preceding claim.