Swarm of drones with collaborative positioning by ultra-wideband communication and associated positioning method

Abstract

SWARM OF DRONES WITH COLLABORATIVE POSITIONING BY ULTRA-WIDEBAND COMMUNICATION AND ASSOCIATED POSITIONING METHOD Swarm of drones (100) comprising a plurality of drones (10), in particular aerial ones, capable of performing a coordinated flight, in which each drone (10) comprises an inertial measurement unit (31), called IMU, making it possible to estimate the attitude of the drone, an ultra-wideband communication module (32), called UWB module, configured to measure a relative position when it communicates with another UWB module, and a central computer (33) configured to perform a fusion of the data from the IMU and the UWB module. No drone (10) uses a satellite geolocation module, such as a GPS receiver; and each drone (10) is configured to communicate, via its UWB module, with one or more neighboring drones (10), so as to ensure its relative positioning in the swarm. Figure for abstract: Figure 2

Description

Title of the invention: Swarm of drones with collaborative positioning by ultra-wideband communication and associated positioning method Technical field

[0001] The present invention belongs to the field of swarm robotics, in particular drone swarms. It relates more particularly to a swarm of drones in which the drones position themselves collaboratively by ultra-wideband communication, better known by its English acronym UWB (Ultra Wide Band), as well as a positioning method implementing such a swarm of drones.

[0002] The present invention finds a direct, but not exclusive, application in drone air shows and in coordinated drone flights for technical inspection or defense mission. State of the art

[0003] Swarm drone flight is a growing discipline that has opened up many technical challenges, particularly regarding the precise positioning of drones in the swarm.

[0004] The precise positioning of drones in a swarm is particularly crucial in the field of drone shows due to the high precision and synchronization requirements necessary for the successful execution of aerial figures.

[0005] Indeed, during a drone show in the form of a sequence of artistic figures, each drone in the swarm must follow, with great precision, a predefined trajectory from the start to the end of the show, so that the overall rendering corresponds as closely as possible to the sequences and transition movements previously simulated and validated.

[0006] Following a given trajectory requires that the drone can, in real time, locate itself and position itself correctly accordingly.

[0007] Technically, the positioning of drones in a swarm is achieved by estimating the state of each drone. This estimation can be carried out individually in each drone using its on-board computer.

[0008] Estimating the state of a drone can be efficiently achieved by conventional fusion of data from different sensors of the drone, including an inertial measurement unit (IMU) and an absolute position sensor, such as a GPS sensor, and / or a relative position (distance) sensor, such as a UWB sensor.

[0009] Patent FR3100895, in the name of the applicant company, relates to a method for automatically positioning a plurality of drones moving in a swarm, each drone comprising an IMU, a GPS module, a UWB module and a central computer implementing a Kalman filter, said method comprising for each drone: a step of measuring the position and attitude of the drone by the IMU and by the GPS module; a step of measuring distances by means of the UWB module and fixed UWB antennas (anchors), allowing a determination of the position of the drone; a step of merging the data from all or part of the sensors of the drone with the distance and position measurements obtained in the previous step, in the central computer by means of the Kalman filter; and a step of estimating the state of said drone.

[0010] More particularly, in this solution, positioning by UWB technology is based on locating the drones of the swarm by measuring distance (ranging), using location algorithms based on the TWR (Two Way Ranging) technique for example, or preferably, on its improved version SDS-TWR (Symmetrical Double Sided Two Way Ranging).

[0011] Localization can also consist of a direct calculation of the propagation time of the UWB signal between the target and the anchors without synchronization of their antennas, this time being known as Time of Arrival (ToA).

[0012] This latter method makes it possible to reduce the pulse traffic and therefore to increase the number of targets that can be located simultaneously as well as the location frequency, and then presents an effective solution for the location of drones in a swarm.

[0013] In the same type, document KR102459019 describes a UWB-based positioning method for a swarm of drones. More specifically, appropriate positioning for the flight of the swarm is ensured in real time by calculating relative positions using the TWR technique according to a master-slave model.

[0014] However, the use of GPS and / or anchors remains necessary in these solutions which apply to outdoor and / or indoor flights.

[0015] Such a swarm cannot therefore evolve precisely in an environment without anchors and without access to GPS. Such an environment could, for example, be the interior of an abandoned building that requires inspection or exploration.

[0016] It is known that UWB technology makes it possible to measure a relative position with respect to another object: a mobile drone or a fixed anchor. It thus allows the positioning of drones during an indoor flight unlike GPS which only works properly outdoors due to the nature of its signal.

[0017] Furthermore, the use of fixed anchors, or ground stations, can induce a lack of robustness due to the quality of the connection between the drones and the ground stations.

[0018] There are many methods for estimating the states of a swarm of independent robots more generally, in which recourse is mainly made to Kalman filters, as mentioned in the solution above, and in particular the extended Kalman filter (EKF).

[0019] Data fusion methods for estimating the state of a drone in a swarm, although varied, systematically include GPS measurements in the fused data in order to guarantee a certain precision in the estimated positions.

[0020] For example, in "RUBIN ZOU. Cooperative positioning or U AV swarms by fusing IMU / UWB / GPS with Federal Kalman Filter. INTERNATIONAL JOURNAL OF COMPUTER ENGINEERING AND TECHNOLOGY, 10, 09 2019", a fusion method, called Federal Kalman Filter (FKF), is described, which uses a double Kalman filter to fuse the data from the three measurements IMU, UWB and GPS, in order to estimate the state of each drone in the swarm. Each drone carries its own FKF. A first Kalman filter is used to fuse the IMU and GPS measurements and then a second Kalman filter is used to fuse the output of the first filter with the UWB measurements.

[0021] Finally, it should be noted that the operation of GPS is significantly less reliable than that of UWB, that the use of IMUs alone leads to a significant divergence of the covariance, which is absolutely not compatible with swarm flight, and that the addition of GPS to the IMU significantly improves the navigation of drones by eliminating the divergence of the covariance, but GPS measurements are not always available (jamming, multi-path, indoor). Summary of the invention

[0022] The present invention aims to overcome all or part of the drawbacks of the prior art set out above by proposing a positioning solution for a swarm of drones, making it possible to do without GPS, or the like, and anchors, and therefore to adapt without technical constraints to any type of environment.

[0023] More fundamentally, by exploiting the fact that a swarm of drones offers advantages over individual drones, such as cooperation, redundancy and the ability to perform complex tasks in a distributed and collaborative manner, the present invention provides a collaborative positioning solution using a distributed UWB system.

[0024] An advantage of the invention is therefore to avoid the use of GPS or to limit it significantly (a single GPS in a drone of the swarm for example). This latter case remains possible, but is not of great interest in the context of the invention.

[0025] Another advantage of the invention is to limit the drift of the swarm and to ensure a good compromise between the drift and the geometric distortion of the aerial figures.

[0026] To this end, the present invention relates to a swarm of drones comprising a plurality of drones, in particular aerial ones, capable of carrying out a coordinated flight, in which each drone comprises an inertial measurement unit, called IMU, for estimating the attitude of the drone, an ultra-wideband communication module, called UWB module, configured to measure a relative position when it communicates with another UWB module, and an on-board central computer, configured to perform a fusion of the data from the IMU and the UWB module. This swarm of drones is remarkable in that no drone uses a satellite geolocation module, such as a GPS receiver, and in that each drone is configured to communicate, via its UWB module, with one or more neighboring drones, so as to ensure its relative positioning in the swarm.

[0027] Advantageously, the estimation is decentralized in the swarm of drones: each drone in the swarm includes its own Kalman filter.

[0028] UWB allows to reduce geometric distortion thanks to an adequate adjustment of a covariance parameter specific to the UWB module.

[0029] According to one embodiment, each measurement carried out by the UWB module of a drone of the swarm has a standard deviation determined as a function of covariance information from a neighboring drone having been used to establish said measurement, so as to disseminate the information throughout the swarm.

[0030] Advantageously, the covariance information is transmitted from drone to drone by UWB communication using the UWB modules.

[0031] According to one aspect of the invention, each drone comprises a Kalman filter implemented in its central computer.

[0032] According to one embodiment, the swarm of drones further comprises at least one stationary drone serving as an anchor for the other drones.

[0033] The present invention also relates to a method for collaborative positioning of drones in a swarm, implementing a swarm of drones as presented, comprising: • a preliminary digital simulation step allowing the choice of a covariance to adjust a drone positioning model; and • a step of executing the positioning model in each drone.

[0034] More particularly, the simulation step makes it possible to estimate an aerial figure of the drones as a function of a set aerial figure and to choose the covariance as a function of an estimation precision and performance indicators among a distortion, a translation drift and a rotation drift of the estimated figure with respect to the set figure.

[0035] According to one embodiment, the swarm of drones uses at least one anchor provided with a UWB transmitter.

[0036] The present invention also relates to: a computer program product downloadable from a communications network and / or stored on a readable medium by microprocessor and / or executable by a microprocessor, characterized in that it comprises program code instructions for the execution of such a collaborative positioning method, as well as a storage medium readable by a terminal and non-transitory, storing a computer program comprising a set of instructions executable by a computer or a processor to implement said method.

[0037] The fundamental concepts of the invention having just been set out above in their most elementary form, other details and characteristics will emerge more clearly on reading the description which follows and with reference to the appended drawings, giving by way of non-limiting example an embodiment of a swarm of drones with collaborative positioning and its associated positioning method, in accordance with the principles of the invention. Presentation of the drawings

[0038] The figures are given purely for illustrative purposes for a better understanding of the invention without limiting its scope. The various elements may be represented schematically and are not necessarily to scale. Throughout the figures, identical or equivalent elements bear the same numerical reference.

[0039] It is thus illustrated in: [Fig.l]: a schematic view of a swarm of drones according to the invention; [Fig.2]: A diagram of two drones in the swarm exchanging UWB signals, showing the main on-board systems of a drone; [Fig.3]: an aerial figure of the swarm called “square” on a 2D plane; [Fig.4]: another aerial figure of the swarm called “ellipses”; [Fig.5]: another aerial figure of the swarm called “flower”; [Fig.6]: a first result of a simulation of the “square” figure, showing a significant drift by rotation of the trajectories; [Fig.7]: a graph of the position error along the x and y axes, corresponding to the simulation of [Fig.6]; [Fig.8]: the evolution of performance indicators as a function of UWB covariance; [Fig.9]: a second result of a simulation of the “square” figure with a tuned UWB covariance, showing a reduction in drift compared to the first result of [Fig.6]; [Fig. 10]: a graph of the position error along the x and y axes, corresponding to the simulation of [Fig.9]; [Fig.l 1]: the evolution of performance indicators according to the number of anchors (a) and according to the number of fixed drones (b); [Fig. 12]: the evolution of performance indicators as a function of UWB covariance in the case of the “ellipses” (a) and “flower” (b) figures. Detailed description of embodiments

[0040] It should be noted that certain technical elements well known to those skilled in the art are recalled here to avoid any insufficiency or ambiguity in the understanding of the present invention.

[0041] In the embodiment described below, reference is made to a swarm of drones, intended primarily for carrying out drone air shows and for carrying out coordinated flights in the context of military missions. This non-limiting example is given for a better understanding of the invention and does not exclude the use of the principles of the invention in other coordinated sets of robots (terrestrial or marine) with a view to carrying out other types of missions.

[0042] In the present description, the terms “drone” and “swarm” designate by extension, respectively, an aerial drone, or unmanned aerial vehicle (UAV), and a group of independent drones capable of flying simultaneously and in a coordinated manner.

[0043] [Fig.l] schematically represents a swarm of drones 100 comprising n independent drones 10 (n being greater than 2 and typically of the order of several hundred or a few thousand), each evolving along a well-defined trajectory, so that the swarm 100 performs an aerial figure within the framework of a drone show or any coordinated flight (technical inspection, maintenance, surveillance, combat, etc.).

[0044] In the swarm 100, each drone 10 is able to communicate wirelessly, by UWB, with the other drones of the swarm to know its relative position. Indeed, each drone 10 of the swarm 100 is configured to receive the positions of the other drones and send its own position to the other drones via a UWB module comprising a UWB signal transmitter-receiver.

[0045] More particularly, each drone 10 is capable of communicating by UWB at least with one or more drones, called neighbors, located in its vicinity delimited by the range of the exchanged UWB signals.

[0046] Thus, a near-near UWB communication takes place even if the swarm 100 has an extent (largest dimension) greater than the range of the UWB signal used.

[0047] For the purpose of estimating its state and its positioning in the swarm 100, each drone 10 comprises a UWB module, an inertial measurement unit, called IMU, and an on-board central computer, together making it possible to measure the movement of the drone and to obtain relative position information from neighboring drones.

[0048] Optionally, at least one drone 10 of the swarm 100 may comprise a GPS-type geolocation module, the use of which in the context of the invention is not necessary, but remains compatible if necessary.

[0049] [Fig.2] schematically represents two drones 10 of the swarm 100, communicating via UWB.

[0050] With reference to this figure, each drone 10 comprises a main body 20 in which the IMU 31, the UWB module 32 and the central computer 33 are embedded.

[0051] The drones 10 of course include motors 40, in particular propeller motors in quadrotor configuration, for their propulsion as well as all the services necessary for their autonomous flight.

[0052] Thus constituted, the drones 10 of the swarm 100 are capable of carrying out precise collaborative positioning thanks to step-by-step UWB communication and a state estimation model which will be presented below. Simulation results making it possible to validate the operation of the invention will be presented and commented on with reference to simple 2D (two-dimensional) aerial figures so as not to burden the description. The validated estimation model naturally applies to any 3D (three-dimensional) aerial figure.

[0053] Figures 3 and 4 represent examples of simple aerial figures (a square and concentric ellipses), on a gridded simulation plane. The small crosses represent the drones on their trajectories (here ten drones); and the thick dots at the corners of each graph represent anchors equipped with UWB transmitters, which can be used in the context of the invention as explained later.

[0054] [Fig.5] represents a more complex aerial figure, called a “flower”, which is representative of the different movements carried out by the drones whatever their trajectory (rectilinear and curvilinear translations, and rotations), and which therefore makes it possible to test the estimation model on several aspects of performance.

[0055] An initial estimation model is described below.

[0056] Each drone in the swarm is located by its x and y coordinates in the plane. The state vector of each drone is thus defined by:

[0057] The discretization according to a time step A^ then makes it possible to express the behavior of the drone by the following system of equations: = with (vt-N(0,e) There yw ï HN(o. / o * ~ wk)

[0058] R depends on the measurements considered, namely the IMU measurements and the UWB measurements.

[0059]

[0060]

[0061]

[0062]

[0063]

[0064]

[0065]

[0066]

[0067]

[0068]

[0069] First, we consider a "stone" prediction model, in which the predicted state is equal to the state at the previous instant. This translates to: / (¾ = ^ + ^ vk We therefore obtain: OiEt: 1 / In each drone, the UWB module, operating as a receiver, gives the distance (relative position) in relation to another UWB module, operating as a transmitter, with a standard deviation auwB. For a drone i having received a relative position measurement with respect to a drone j, we have: ?uwb = hUWB (W y J wuwb î = 11 W _ yj 11 + wsEt: - N ( 0, ) ; with RÜ™ = It should be noted that for a drone j of the swarm, depends directly on the associated covariance which appears as an uncertainty on the position of the drone j which is not fixed, unlike a UWB anchor. The observation matrix is ​​expressed: pUWB / • • \ _dh , yn _ i ^k {i, J) - 1 The prediction and correction equations of the extended Kalman filter (EKF) are used in the following. The prediction is made according to the following two equations: = / (¾ 0)pt+||t = FP^l + GkQG{ The correction is made based on the available measurements. For example, for two UWB measurements (with the drones ji and j2), we have: UWBI k~ , / 1 °ï k \0 1 / The correction, or update, equations are therefore as follows. Innovation, which corresponds to the gap between measurement and prediction: - S<+i

[0070] The covariance of innovation: ^+1 ^À+lA+l^In + Rk

[0071] The optimal Kalman gain: Kk+1 - ^k+^k+\^k+l

[0072] Status update: = ^k+ik + Kk+lvk+l

[0073] The covariance update: Pk+M+1 “ " %k+ i^k+^k+ik

[0074] The standard deviation avwB of a UWB measurement is determined based on the covariance information of the neighboring drone used to establish the measurement. The neighboring drone can be an “interlocutor” drone or an anchor. This ensures the dissemination of the information, in particular the precision, throughout the swarm. The transmission of the covariance information is carried out by UWB, the UWB being in this case used as a means of communication.

[0075] According to an exemplary embodiment, the standard deviation j) for a measurement between two drones i and j is calculated by summing the standard deviation of the UWB sensor auwB with the semi-major axis of the covariance ellipse of the reference drone and the semi-major axis of the covariance ellipse of the drone which takes the measurement.

[0076] The semi-major axis is calculated by taking the square root of the maximum eigenvalue of the covariance matrix of the drone considered. If the drone is an anchor, this maximum eigenvalue is 0. Otherwise, its expression at time k is:

[0077] Where eig(A) represents the set of eigenvalues ​​of A.

[0078] The consideration of the drone measurements in the initial model above is described below.

[0079] It should be noted that in the absence of an IMU, the drones in the swarm are not able to follow a given trajectory and simply maintain the same relative distances that they have with each other.

[0080] The IMU provides two acceleration measurements, along the x and y axes. The IMU measurement vector can therefore be noted:

[0081] The IMU providing speed measurements in addition to the positions (geometric coordinates) of the drone, the state vector of each drone becomes: X iy vx \Vyl

[0082] The drone evolution model with the IMU becomes: jVx\ ay I \vyl -me)

[0083] Furthermore, we have:

[0084] The discretization of the evolution model above gives: Either : Xk+^f(Xk,Uk,vk) = yk + Vy>k-Xt VX* axjc ' vxJï ', + VyJc ' I X^F^ + Giu. + v^'- / 1 0 A / 0 \ Et 0 1 0 At 0 0 1 0 \0 0 0 1 / / 0 0 Gt=f(W 0 0 A / 0 \0 Garlic

[0085] The model then applies the prediction equations with the new function f and the new matrices F and G.

[0086] Regarding the correction step, the addition of the IMU only involves a modification of the size of certain matrices. More precisely, the observation matrix corresponding to the UWB measurement becomes: ^17)=^(440) = (^ $4 oo)Aw: K \ J / <)X ' KK \ d(i,j) dyij) /

[0087] The matrix remains unchanged.

[0088] The correction equations are finally applied with the new h functions and the new observation matrices.

[0089] Thus, a collaborative positioning algorithm based on the model obtained is implemented in the central computer of each drone of the swarm, and makes it possible to execute a positioning method implementing a swarm of drones according to the invention. Results

[0090] The simulations and tests enabling validation of the positioning model described were carried out with ten drones, on a 2D plan of 20m x 20m and over a duration of 20 seconds, for obvious reasons of simplification.

[0091] The variable parameters are the covariance, the number of anchors, the predefined trajectory of the drones (the aerial figure of the swarm).

[0092] The analysis of the results uses three performance indicators, which are geometric and quantifiable and which make it possible to evaluate the consistency of the estimated aerial figure of the swarm with respect to its actual figure. These indicators are: • Distortion D: value in meters measuring the conformity of the estimated (obtained) figure with respect to the real figure. For example, if the real figure is a square of a certain size, an estimate with a low distortion will give an estimated figure representing a square of the same dimension. The estimated figure can nevertheless be shifted or oriented without increasing the distortion. • Translation T: value in meters measuring the distance difference between the center of gravity of the estimated figure and the center of gravity of the real figure. • Rotation R: value in degrees measuring the rotation of the estimated figure relative to the actual figure.

[0093] The choice of the covariance aüWB must guarantee a good compromise between the estimation precision, which must be high, and these three indicators which must remain low. This choice can be made manually or automatically during a step of adjustment of the positioning process called "noise tuning".

[0094] The noise tuning step can also take into account the number of anchors and the nature of the trajectories of the drones in the swarm.

[0095] It is recalled that the model only uses UWB modules and IMUs.

[0096] [Fig.6] represents the result obtained for the “square” figure. The continuous and broken lines correspond respectively to the exact and estimated trajectories of the drones. The ellipses represent the UWB covariances of the drones which are all equal according to the illustrated example.

[0097] The "square" figure drifts slowly in translation and rotation. This is explained by the absence of an absolute position sensor (GPS) which would allow drones to correct their absolute position according to the predefined trajectory.

[0098] [Fig.7] represents error graphs on the x and y positions of an arbitrarily chosen drone in the swarm.

[0099] The error on the x position remains within the confidence interval at 3cr, while the error on the y position tends to diverge and leave the confidence interval after a certain time (16s). This corresponds to the drift observed in [Fig.6] which is in fact greater along the y axis (downward drift of the trajectories).

[0100] This first result was obtained with an initial covariance auwB = 1.5.

[0101] This result can be improved by noise tuning, in order to reduce the drift of the aerial figure.

[0102] Figure 8 represents the performance indicators of the “square” figure as a function of auws.

[0103] The distortion increases by oscillating around the value 0.2. Thus, it is advisable to choose a low auwe value to limit the distortion. However, a low guwb value implies a greater rotation of the aerial figure.

[0104] A compromise must therefore be made between these two parameters. Depending on the nature and shape of the aerial figure, rotation or distortion may prove to be more or less troublesome. For example, for a spherical figure, rotation will have no impact on the overall position of the swarm.

[0105] Considering the above analysis, taking auwB = 2, we obtain the result of [Fig.9] and the error graphs of [Fig.10].

[0106] [Fig.9] clearly shows that the drift has been reduced overall, notably due to a reduction in the rotation of the swarm. [Fig.10] clearly confirms that the error in y is significantly reduced compared to the previous result and remains largely within the confidence interval.

[0107] Furthermore, [Fig. 10] simply shows a slow translation of the swarm along x, of the same order as that obtained with a GPS (according to tests carried out by the inventors).

[0108] [Fig. 11] shows the influence of the number of anchors or stationary drones on performance.

[0109] It turns out that the use of a single anchor significantly reduces translational drift while the use of two anchors makes it possible to overcome rotational drift.

[0110] Thus, by combining these results with the previous results, very good results can be obtained by choosing a value of auwB sufficiently large to limit rotational drift and an anchor to limit translational drift.

[0111] In the case of the absence of anchors, the correction provided by a stationary drone is less significant.

[0112] Figure 12 shows that these conclusions are also valid for the other trajectories (ellipses and flower). In particular, it shows that the translational drift remains the same. However, the rotational drift is significantly less significant for the “ellipses” figure in (a) than for the “flower” figure in (b). Consequently, a lower value of ^uwb for “ellipses” than for “flower” can be suitably chosen.

[0113] Thus, getting rid of GPS while retaining 1TMU and UWB is very interesting in the case where the swarm interacts with an anchor (or more) since the translational drift is eliminated with the anchor, and the rotational drift is eliminated with a good choice of avwB. However, in the case of an absence of anchors (unknown environment for example), the translational drift of the swarm is inevitable. Nevertheless, the distortion and rotation of the swarm can be limited by choosing the value of auwB carefully according to the simulation parameters.

[0114] Finally, adding stationary drones can slightly improve distortion and rotational drift but the effects remain slight.

[0115] In summary, UWB is a technology that uses signals capable of providing precise information on the distance and flight time of radio waves, which makes it possible to determine the relative position between different drones with high accuracy.

[0116] In the context of the present invention, UWB technology is used to enable near-field communication and precise relative positioning between the drones in the swarm. Each drone in the swarm is equipped with a UWB module, generally in the form of an electronic chip or a module integrated into the computer. The same applies to 1TMU.

[0117] When drones are in flight, they communicate with each other using UWB signals. Each drone measures the time of flight of UWB signals from other drones in the swarm. By knowing the propagation speed of UWB signals in the environment, each drone can calculate the relative distance to the other drones. By combining this distance information with the 1TMU measurements, the drones can determine their relative position in the swarm with high accuracy.

[0118] UWB technology also enables real-time updates of the relative positions of drones, which facilitates coordination and control of the swarm. For example, drones can adjust their position and trajectory based on location information provided by UWB technology, which helps maintain accurate formation and avoid collisions.

[0119] Besides its application in the field of light drone shows, the present invention offers promising opportunities in other fields such as technical inspection and defense. In the context of light drone shows, the UWB relative positioning system allows for breathtaking aerial choreographies, where each drone is in perfect harmony with the others, thus creating a striking and immersive visual spectacle. The artistic and creative possibilities are endless, offering spectators a unique and memorable experience.

[0120] With regard to technical inspection activities, the drone swarm can be used to inspect complex infrastructures, such as buildings or industrial facilities, providing a global and detailed view of the environment, and quickly identifying areas requiring intervention or repair.

[0121] It is apparent from the present description that certain non-essential elements may be modified, replaced or deleted without departing from the scope of the invention defined by the claims below.

Claims

Claims

1. A swarm of drones (100) comprising a plurality of drones (10), in particular aerial drones, capable of performing a coordinated flight, in which each drone (10) comprises an inertial measurement unit (31), called an IMU, for estimating the attitude of the drone, an ultra-wideband communication module (32), called a UWB module, configured to measure a relative position when it communicates with another UWB module, and a central computer (33) configured to perform a fusion of the data from the IMU and the UWB module, said swarm being characterized in that no drone (10) uses a satellite geolocation module, such as a GPS receiver; and in that each drone (10) is configured to communicate, via its UWB module, with one or more neighboring drones (10), so as to ensure its relative positioning in the swarm.

2. Swarm of drones according to claim 1, in which each measurement carried out by the UWB module of a drone (10) has a standard deviation determined as a function of covariance information from a neighboring drone used to establish said measurement, so as to disseminate the covariance information throughout the swarm.

3. A drone swarm according to claim 2, wherein the covariance information is transmitted from drone to drone by ultra-wideband communication using the UWB modules.

4. A swarm of drones according to any preceding claim, wherein each drone (10) comprises a Kalman filter implemented in its central computer (33).

5. A drone swarm according to any preceding claim, further comprising at least one stationary drone serving as an ultra-wideband communication anchor for the other drones.

6. Method for collaborative positioning of drones, implementing a swarm of drones (100) according to any one of claims 1 to 5, comprising: • a prior step of digital simulation making it possible to choose a covariance to adjust a positioning model of the drones; and • a step of executing the positioning model in each drone (10).

7. Positioning method according to claim 6, in which the simulation step makes it possible to estimate an aerial figure of the drones as a function of a set aerial figure and to choose the covariance as a function of an estimation precision and performance indicators among a distortion, a translational drift and a rotational drift of the estimated figure relative to the set figure.

8. A positioning method according to claim 6 or 7, wherein the swarm of drones (100) uses at least one anchor provided with an ultra-wideband transmitter.

9. Computer program product downloadable from a communication network and / or stored on a medium readable by a microprocessor and / or executable by a microprocessor, characterized in that it comprises program code instructions for the execution of a collaborative positioning method according to one of claims 6 to 8.

10. A terminal-readable and non-transitory storage medium storing a computer program comprising a set of instructions executable by a computer or a processor to implement a collaborative positioning method according to one of claims 6 to 8.

Images