Drone piloting procedure.

Abstract

The invention relates to a method for piloting a drone (100) equipped with a surface treatment device, a three-dimensional position sensor, and a distance sensor, comprising the steps of: - Flying the drone (100) along the flight plan (400), at a distance from a structural element comprising a surface to be treated; - The flight plan (400) of the drone (100) being defined within a georeferenced point cloud (300) representing the surface to be treated; - The flight plan (400) comprising a flight path in which a set of successive positions that the drone (100) must adopt are defined, the flight path being such that the drone (100) passes within the treatment distance of the surface to be treated, the coordinates of the points of which are known; - Defining the position of the drone (100) during the flight and selectively activating the surface treatment device according to the position of the drone (100). Figure for the abstract: Fig. 3

Description

Title of the invention: Method for piloting a drone.

[0001] The present invention relates to the field of drone piloting, in particular in the field of surface treatment.

[0002] In this respect, and in this area, the applicant has already filed application FR2305773 as well as application FR2107477.

[0003] In the field of surface treatment, the use of drones is now well known, for example for spraying cleaning products. However, it is common for this operation to be carried out manually.

[0004] This human aspect is problematic for example in the treatment of complex surfaces where slopes and overhangs can hinder the operator's vision.

[0005] It is therefore desirable to automate the flight of the drone.

[0006] In this context, the present invention relates, according to a first of its objects, to a method for piloting a drone equipped with a surface treatment device, a three-dimensional position sensor and an assembly of at least one distance sensor.

[0007] It is essentially characterized in that it comprises steps consisting of: - To fly the autonomous drone according to the predetermined flight plan, at a flight distance from a set of at least one construction element comprising at least one surface to be treated; • The drone flight plan being defined in a topographic map comprising a georeferenced point cloud, the map representing at least part of said set of at least one construction element; • the flight plan including a flight trajectory in which are defined a set of successive positions that the drone must adopt, the flight trajectory being such that the drone passes within the processing distance of the surface to be treated, whose coordinates of the points are known; - Define the drone's position during flight and selectively activate the surface treatment device based on the drone's position.

[0008] It can be foreseen that the topographic mapping is provided by a preliminary reconnaissance step comprising steps consisting of: - To fly the drone at a processing distance from said assembly of at least one building element, so that the assembly of at least one distance sensor of the drone can detect the distance between the drone and the surface to be treated of said building element; - Record the three-dimensional coordinates of the drone for a set of its positions during the flight; - For at least one of these positions, record the value of the distance between the drone and the surface of said building element, and - deduce the position of the points of said construction element.

[0009] It can be foreseen that the step of recording the three-dimensional coordinates of the drone in flight is implemented by a simultaneous localization and mapping (SLAM) algorithm embedded in the drone.

[0010] It can be anticipated that the surface treatment device is configured to implement, when activated: • An optical recognition step, visible or IR, of the surface to be treated.

[0011] A further comparison step can be envisaged, consisting of: • Compare the images from the optical recognition stage, visible or IR, of the surface to be treated to reference images for the same three-dimensional coordinates of the drone.

[0012] It can be anticipated that the surface treatment device is configured to implement, when activated: • Spraying a treatment fluid onto the surface to be treated.

[0013] At least one of the following steps can be foreseen: • A step to calculate the drone's offset / drift in flight; • a drift compensation step during flight, so as to maintain the treatment distance substantially constant, which allows following any curvature of the surface to be treated.

[0014] It can be foreseen that the selective activation step of the surface treatment device is implemented according to the three-dimensional position of the drone, and its yaw angle relative to the normal to the surface to be treated.

[0015] According to another of its objects, the invention relates to a drone comprising a computer and a memory in which is stored a computer program comprising program code instructions for the implementation of the method according to the invention when said program is executed on said computer.

[0016] It can be foreseen that the drone includes a controller comprising a PID loop configured to control the position of the drone to that defined in the flight plan.

[0017] Other features and advantages of the present invention will become more apparent upon reading the following description, given by way of illustrative and non-limiting example and with reference to the accompanying figures in which: - Figure 1 illustrates a drone on a wind turbine blade for 3D mapping. of this one, - Figure [Fig. 2] illustrates a scatter plot according to the invention of a wind turbine equipped with blades, - Figure 3 illustrates georeferenced points according to the invention of a mast of a wind turbine and a flight plan according to the invention. Detailed description

[0018] According to the invention, a drone 100 comprises a set of at least one three-dimensional position sensor, typically a GPS sensor.

[0019] Thanks to this three-dimensional position sensor, it is possible to know in real time the position of the drone 100 in space.

[0020] This position is advantageously recorded, as explained later.

[0021] The drone 100 also includes an assembly of at least one distance sensor, or more generally a photogrammetry device, for example an optical sensor, in particular a 3D LiDAR, a stereo camera in the visible, or a sound sensor, in particular a SONAR, which emits a beam 110 or an ultrasonic sensor.

[0022] For the sake of brevity, "a set of at least one distance sensor" will be referred to hereafter as "a" or "the" distance sensor.

[0023] Thanks to the distance sensor, it is possible to know the distance between the drone 100 and a surface, in this case, a surface to be treated.

[0024] For example the surface to be treated is a wind turbine 200 which includes a mast 210 and blades 220.

[0025] The treatment can be physico-chemical or optical, as described later.

[0026] The distance value between the drone 100 and the surface to be treated is advantageously recorded and used to obtain a georeferenced point cloud 300, as described later. It is thus possible to obtain a 3D representation, or topographic map, of a surface to be treated, also called an area of ​​interest.

[0027] According to the invention, the drone 100 is equipped with a memory in which is recorded a simultaneous localization and mapping algorithm, known by the anglicism SLAM (simultaneous localization and mapping) or CML (concurrent mapping and localization), which consists, for a robot or autonomous vehicle, of simultaneously building or improving a map of its environment and localizing itself within it.

[0028] In this case, a SLAM algorithm based on 3D point clouds obtained by LiDAR is planned, for example as described at the following address: https: / / fr.mathworks.com / help / vision / ug / build-a-map-from-lidar-data-using-slam.html.

[0029] Compared to conventionally obtained 3D point clouds, here we have a georeferenced 3D point cloud of 300 points, meaning that the 3D coordinates of each point in the point cloud are known, in this case thanks to the drone 100, whose position The 3D surface is known, and its distance to the surface to be processed is also known. Therefore, we can deduce the coordinates of each point 310 on the 3D surface to be processed.

[0030] In more detail, a step of providing a 3D map is planned first, which includes a cloud of 300 georeferenced points.

[0031] This mapping is in this case constructed by the drone 100 using a SLAM algorithm, but it can of course be pre-existing.

[0032] To provide a 3D map, it is planned, for example, to fly the drone 100 at a distance, referred to as the "processing distance," from a set of at least one building element, comprising a set of at least one surface to be processed. The processing distance is, for example, a predefined distance, preferably configurable between two successive flights; or the minimum distance between: - The maximum sensitivity distance of the sensor, and - The maximum effective distance of the spray nozzle or any other active material.

[0033] The processing distance is such that the set of at least one distance sensor of the drone 100 detects the distance between the drone 100 and the surface to be treated of said construction element.

[0034] The three-dimensional coordinates of the drone 100 are recorded during the flight, for example by means of an integrated GPS sensor, which generates a set of georeferenced points for a set of positions of the drone, as a sampling.

[0035] For each georeferenced point 410 of the drone 100 flight path, the distance between the drone 100 and the surface to be treated of said construction element is recorded. Since the viewing angle of the distance sensor is known, the position of the drone 100 and the distance separating it from the surface to be treated are therefore known; the position of the points on the surface to be treated of said construction element can thus be deduced.

[0036] Thus for each georeferenced point 410 of the flight of the drone 100, we obtain a set of at least one georeferenced point 310 of the surface to be treated of said construction element.

[0037] This gives us a topographic map of the surface to be treated of said construction element comprising a cloud of 300 georeferenced points.

[0038] With this topographic mapping, we can define a flight plan 400, which is the trajectory that the drone 100 must follow, possibly supplemented by a predetermined flight speed, in relation to the surface to be treated of the construction element.

[0039] The building element can be of various types. The building element has dimensions in which at least one of the components among the width, height, and depth is greater than 2m, which justifies the use of a drone. For example, the building element is a residential building (detached house, apartment building, etc.), a building (factory, barn, etc.), a ship, an airplane, a train, a 200 wind turbine, etc.

[0040] Depending on the context, the mapping may only include one surface to be processed.

[0041] Mapping may also include several surfaces to be treated, of the same building element, or of several building elements close to each other.

[0042] Each surface to be processed is defined as an area of ​​interest. They can be defined on the images obtained in several ways, manually or automatically, for example according to at least one of the possibilities below: - using a human-machine interface and a pointer; - thanks to edge detection software, - using gradient calculation software, - etc.

[0043] Surface treatment can be physico-chemical, including defrosting, painting, cleaning, including high-pressure cleaning, etc. of the surface to be treated.

[0044] For example, the drone 100 can be equipped with spray nozzles and a treatment fluid can be sprayed onto the surface to be treated when the position of the drone 100 corresponds to that of the surface to be treated.

[0045] The treatment fluid is for example water or a mixture comprising water and a treatment fluid (degreaser, solvent or other), paint, de-icer, etc.

[0046] Surface treatment can also be optical treatment, for example for the detection or tracking of surface defects, for example cracks, paint crazing, etc.

[0047] For this purpose, it can be provided that the drone 100 is equipped with a set of at least one optical sensor, allowing optical, visible or IR recognition of the surface to be treated.

[0048] It is also possible to compare the images obtained during the optical recognition step, visible or IR, of the surface to be treated with reference images obtained previously for the same three-dimensional coordinates of the drone 100.

[0049] This comparison step allows in particular the monitoring of surface defects.

[0050] It is possible to provide for the emission of an alarm signal, for example, if the length of a detected defect is greater than a predetermined threshold value, which may indicate, for example, the appearance of a surface defect.

[0051] Similarly, if the difference between the length of a defect detected during the recognition step and the length of said defect detected during a previous recognition step, which can serve as a reference image, is greater than a predetermined threshold value; which can reflect, for example, the rapid increase in the size of a surface defect, and therefore the rapid degradation of the surface.

[0052] Preferably, selective activation of the surface treatment device is provided.

[0053] It is implemented according to the three-dimensional position of the drone 100, and preferably also according to the yaw angle of the drone 100 with respect to the normal to the surface to be treated.

[0054] Preferably, the drone 100 is equipped with a controller including a PID loop allowing to calculate a drift or offset of the drone 100 in flight and to compensate for this drift during flight.

[0055] Thanks to this feature, it is possible to maintain the treatment distance substantially constant, which makes it possible to follow any curvature of the surface to be treated.

[0056] If the drift calculation is such that the calculated drift is greater than a threshold value stored in memory, it is possible to provide for the emission of an alarm signal or inhibition of the processing, or even a forced return to the ground; for example in the event of an excessively strong gust of wind which would cause the drone 100 to drift.

[0057] The possible drift of the drone 100 can also be determined by images obtained by the optical sensors.

[0058] To this end, an image processing step is provided, consisting of defining on the images a set of remarkable points, also called key points. These remarkable points are, for example, calculated using edge detectors, SIFT descriptors or other methods, or based on contrast or brightness gradients.

[0059] The key points of the surface to be treated are, by definition, georeferenced. In flight plan 400, the relative position between the position of drone 100 and the position of the key points is therefore known. During the flight, the position of drone 100 is known, and the position of the key points is calculated. If a difference in relative position is observed between the actual flight and flight plan 400, then drone 100 is drifting. This difference from the flight reference is known and can therefore be compensated.

[0060] Alternatively, or in addition, the flight plan 400 can be provided for to include not only a flight trajectory but also a flight speed. Thus, the position of the drone 100 should be known at any given moment.

[0061] During the flight, the actual position of the drone 100 can be compared with the position it should have in the flight plan 400, and any discrepancy can be compensated for.

[0062] Nomenclature

[0063] 100 drone

[0064] 110 beam of the distance sensor

[0065] 200 wind turbine

[0066] 210 wind turbine mast

[0067]

[0068]

[0069]

[0070]

[0071] 220 wind turbine blade 300 point cloud 310 georeferenced points of the surface to be treated 400 flight plan 410 georeferenced points of the flight plan

Claims

Demands

1. A method for piloting a drone (100) equipped with a surface treatment device, a three-dimensional position sensor, and an assembly of at least one distance sensor, characterized in that it comprises steps consisting of: - To fly the autonomous drone (100) according to the predetermined flight plan (400), at a flight distance from an assembly of at least one construction element comprising at least one surface to be treated; • The flight plan (400) of the drone (100) being defined in a topographic map comprising a georeferenced point cloud (300), the map representing at least a part of said set of at least one construction element; the flight plan (400) comprising a flight trajectory in which are defined a set of successive positions that the drone (100) must adopt, the flight trajectory being such that the drone (100) passes within the processing distance of the surface to be treated, whose coordinates of the points are known; - Define the position of the drone (100) during the flight and selectively activate the surface treatment device according to the position of the drone (100).

2. A method according to claim 1, wherein the topographic mapping is provided by a preliminary reconnaissance step comprising steps consisting of: - To fly the drone (100) at a processing distance of said assembly of at least one construction element, so that the assembly of at least one distance sensor of the drone (100) can detect the distance between the drone (100) and the surface to be treated of said construction element; - Record the three-dimensional coordinates of the drone (100) for a set of its positions during the flight; - For at least one of the said positions, record the value of the distance between the drone (100) and the surface of said construction element, and - deduce the position of the points of said construction element.

3. A method according to claim 2, wherein the step of recording the three-dimensional coordinates of the drone (100) in flight is implemented by a simultaneous localization and mapping (SLAM) algorithm embedded in the drone (100).

4. A method according to any one of the preceding claims, wherein the surface treatment device is configured to implement, when activated: • An optical, visible or IR recognition step of the surface to be treated.

5. Method according to claim 4, further comprising a comparison step consisting of: • Comparing the images from the optical recognition step, visible or IR, of the surface to be treated to reference images for the same three-dimensional coordinates of the drone (100).

6. A method according to any one of the preceding claims, wherein the surface treatment device is configured to implement, when activated: • A spraying of a treatment fluid onto the surface to be treated.

7. A method according to any one of the preceding claims, comprising at least one of the following steps: • A step of calculating the offset / drift of the drone (100) in flight; • a step of compensating for drift during flight, so as to maintain the treatment distance substantially constant, which allows following any curvature of the surface to be treated.

8. A method according to any one of the preceding claims, wherein the selective activation step of the surface treatment device is implemented depending on the position three-dimensional of the drone (100), and its yaw angle relative to the normal to the surface to be treated.

9. Drone (100) comprising a computer and a memory in which is stored a computer program comprising program code instructions for implementing the method according to any one of the preceding claims when said program is executed on said computer.

10. Drone (100) according to claim 9, comprising a controller including a PID loop configured to control the position of the drone (100) to that defined in the flight plan (400).

Images