Method for improving the positional accuracy of a mobile carrier

Abstract

The invention relates to a method for improving the position accuracy of a moving carrier, comprising the steps of: S1) predicting a position of the moving carrier (1), and determining a set of predicted distances between the moving carrier (1) and a set of obstacles (2, 3, ) in the environment of the moving carrier (1); S2) acquiring a channel impulse response (CIR) between a first node (T1) and a second node (T0), at least one of the two nodes (T1, T0) being mounted on the moving carrier (1); S3) determining a set of multipath components from the channel impulse response (CIR); S4) associating at least one predicted distance with at least one corresponding multipath component so as to obtain an updated set of distances between the moving carrier (1) and the set of obstacles (2, 3, ). Figure for the abstract: Fig. 1

Description

Title of the invention: Method for improving the positional accuracy of a moving carrier. Technical field

[0001] The invention relates to the field of telecommunications, and in particular to the field of mobile carrier localization using Ultra Wide Band (UWB). The invention finds its application for localization in areas that do not offer a sufficiently accurate satellite positioning service, or in areas not covered by a satellite positioning service.

[0002] In the absence of a reliable GNSS (satellite positioning) service, UWB technology offers very good accuracy, generally around 10 cm, due to its very wide bandwidth, generally exceeding 500 MHz.

[0003] The UWB radio is characterized by occupying a very wide frequency spectrum, that is to say a band of at least 20% of the center operating frequency or greater than 250MHz (EU) or 500MHz (US).

[0004] The use of a very wide spectrum (conventional radios generally use bandwidths ranging from a few kilohertz up to about ten megahertz) gives this technology the ability to measure very precisely the time of arrival of the radio signal (or TOA for Time of Arrival), or the differences in time of arrival (TDOA for "Time Difference of Arrival"), which subsequently allows the calculation of a flight time and therefore a very precise distance.

[0005] Indeed, the large bandwidth of the signal in the spectral domain translates into the ability to emit very short pulses in the time domain, each with a duration on the order of a few nanoseconds. Since the accuracy of the arrival time measurement is related to the pulse duration, distances can be measured with accuracies on the order of tens of centimeters.

[0006] In such cases, one or more UWB receivers (also called "tags"), placed on the mobile carrier to be located (for example a handheld terminal, a drone, or any type of vehicle) measure the time of arrival or the difference in time of arrival with UWB anchors of known positions, in order to calculate the distances between the anchors and the UWB receivers, and thus determine the position of the carrier.

[0007] Thus, in the present application, UWB anchors are stationary telecommunications nodes installed at a known fixed location in a location system, which emit (and can also receive) UWB signals. UWB tags (also called beacons in this technical field) are telecommunications nodes carried by the objects being tracked. They receive signals emitted by UWB anchors at different times depending on their distance from the anchors (and can also emit signals themselves). By measuring the time of flight of these signals, the UWB beacons can determine their distance from the anchors. By combining these calculated distances, the exact position of the mobile device can be determined through trilateration.

[0008] Ideally, the anchors are deployed around the target area (i.e., the area in which the carrier to be located is located) to ensure good geometric dilution of accuracy. However, for applications such as locating a drone in a warehouse, the anchors are mounted at ground level, while the drone may fly at high altitude, or the anchor deployment area is limited to a small surface. Thus, the geometric dilution of accuracy can be significant, which can negatively impact the accuracy of the carrier's location.

[0009] Some applications using UWB technology include UWB radar systems that are well suited for use on board a drone to detect obstacles and estimate the drone's height, as described in [Barra's] article. With these systems, the UWB radar and localization are based on two different devices.

[0010] Some authors (notably [Gentner], [Bocus], [Ledergerber], [Li], and [Santoro]) have demonstrated the feasibility of bistatic radar measurements using UWB locating devices of the DW 1000 or DW3000 type from Qorvo. These devices make it possible to measure the channel impulse response (CIR) collected during a telemetry transmission between two devices in order to detect echoes and use the direct path for synchronization purposes.

[0011] However, in the prior art, the "alternative" use of the CIR of a UWB circuit dedicated to range measurement is described only between two static radios relative to their environment. The main reason is that it is very complex to identify a target from a CIR because it does not offer high resolution (typically, a CIR comprises 1000 measurements sampled at 1 ns, but three-quarters of these measurements contain only the noise preceding the arrival of the signals) and the background noise can be significant (especially when the transmissions / receptions – radar measurements – are performed from an object that is itself moving). Indeed, the primary ranging function requires antennas that are rather omnidirectional, that is to say, that they potentially receive echoes from all directions in space, which considerably increases this background noise, while the radar functionality protects itself from this by using directional antennas.

[0012] Consequently, the processing of this background noise using known techniques is carried out by recording a "basic response" of it, and the detection of a moving target is performed by analyzing the variations of the measurement relative to this "basic response". This approach therefore only works if the background noise – and thus the radar antennas – are static.

[0013] There is therefore a need to provide a method for improving the accuracy of the position of a mobile carrier, using CIR measurements with mobile antennas relative to their environment. Summary of the invention

[0014] An object of the invention is therefore a method for improving the accuracy of the position of a moving carrier, comprising the steps of:

[0015] SI) prediction of a position of the mobile carrier, and determination of a set of predicted distances between the mobile carrier and a set of obstacles in the environment of the mobile carrier;

[0016] S2) acquisition of a channel impulse response between a first node and a second node, at least one of the two nodes being carried on the mobile carrier;

[0017] S3) determination of a set of multipath components from the response channel impulse;

[0018] S4) association of at least one predicted distance to at least one multi-component corresponding path so as to obtain a set of updated distances between the mobile carrier and the set of obstacles.

[0019] Advantageously, the method includes, after step S4), a step S5) of updating the position of the mobile carrier according to the updated set of distances.

[0020] Advantageously, determining a set of multipath components includes the substeps of:

[0021] - determination of a first reference time corresponding to the direct path between the first knot and the second knot;

[0022] - determination of n additional times corresponding to the different indirect routes between the first node and the second node;

[0023] - calculation of n distances measured as a function of the n additional times, from the first reference time and the distance between the first node and the second node.

[0024] Advantageously, the association of a predicted distance with a multipath component includes, for each obstacle in the set of obstacles, the comparison against a threshold of the difference between the predicted distance and each of the measured distances.

[0025] Advantageously, the threshold is determined dynamically as a function of the dispersion of the predicted distances.

[0026] Advantageously, the association of a predicted distance with a multipath component includes, for all possible associations between the set of multipath components and the set of obstacles in the environment of the mobile carrier, the following substeps:

[0027] - calculation of a position of the mobile carrier from the predicted distances and the distances measured, using said association;

[0028] - from this position, calculation of the theoretical distances between the mobile carrier and a plurality of fixed anchors and obstacles;

[0029] - calculation of residuals defined as the difference between the theoretical distances and the measured distances;

[0030] - calculation of a likelihood score based on residuals;

[0031] - determination of the association for which the likelihood score is the most pupil.

[0032] Advantageously, the determination of a first reference time includes, in a first search window:

[0033] - a determination of a first sample whose amplitude is greater than one first threshold;

[0034] - a determination of a maximum sample amplitude subsequent to the first sample ;

[0035] - a determination of a second threshold which is lower than the maximum amplitude;

[0036] - a determination of a second sample whose amplitude precedes immediately the second threshold, and a third sample whose amplitude immediately follows the second threshold;

[0037] - the first reference time being determined by a linear interpolation carried out between the second sample and the third sample.

[0038] Advantageously, the first threshold is a multiple of the ambient noise measured in the first search window.

[0039] Advantageously, the second threshold is equal to half of the maximum amplitude.

[0040] Advantageously, the determination of n additional times includes, for each additional time between 1 and n:

[0041] - a determination of the sample having the maximum amplitude of the response channel impulse;

[0042] - an extraction of the arrival time of the multipath component in a window of research focused on the sample having the maximum amplitude of the channel impulse response;

[0043] - a zero forcing of all samples in an elimination window centered on the sample having the maximum amplitude of the channel impulse response.

[0044] Advantageously, the prediction of a position of the moving carrier is implemented by an extended Kalman filter.

[0045] Alternatively, the prediction of a position of the mobile carrier is implemented by particle filtering.

[0046] The invention also relates to a system for improving the positional accuracy of a mobile carrier, the system being configured to:

[0047] SI) predict a position of the mobile carrier, and determination of a set of predicted distances between the mobile carrier and a set of obstacles in the environment of the mobile carrier;

[0048] S2) acquire a channel impulse response between a first node and a second node, at least one of the two nodes being carried on the mobile carrier;

[0049] S3) determine a set of multipath components from the response channel impulse;

[0050] S4) associate a predicted distance with a corresponding multipath component so as to obtain a set of updated distances including at least one updated distance;

[0051] Advantageously, the two nodes are carried on the mobile carrier, the first node being an Ultra Wideband tag, the second node being an Ultra Wideband receiver.

[0052] Advantageously, the mobile carrier is a drone. Description of the figures

[0053] Other features, details and advantages of the invention will become apparent from the description made with reference to the accompanying drawings given by way of example.

[0054] Fig. 1 illustrates an overview of the system according to the invention.

[0055] Figure 2 illustrates an architecture for implementing the process according to the invention.

[0056] Fig. 3, Fig. 4, Fig. 5 and Fig. 6 illustrate an example of determining the first reference time on graphs showing the IRC standard as a function of time.

[0057] Figure 7 illustrates the steps for determining the additional times corresponding to indirect routes.

[0058] Figure 8 illustrates an example of an association between a predicted distance and a multipath component. Detailed description

[0059] Figure 1 shows an example of a system configuration according to the invention. It comprises an array of anchors (4, 5) whose positions are known, and which enable UWB localization through time-of-flight measurements between the anchors (4, 5) and the mobile carrier 1. At least two UWB tags (T0, T1) are integrated into the mobile carrier 1. They enable the positioning of the mobile carrier 1 by cooperating with the anchors (4, 5). In Figure 1, only two UWB tags are shown, but it may be possible to equip the mobile carrier 1 with more than two UWB tags, which increases the number of measured ranges and provides information on the attitude of the mobile carrier 1, particularly in the case where the mobile carrier 1 is a drone.

[0060] Emissions according to Ultra Broadband technology follow a time division multiple access scheme, or TDMA for "Time Division Multiple Access Scheme", and are organized into frames / , where each node (i.e. the beacons and anchors) has a dedicated location for its transmission.

[0061] One of the nodes (for example, node T0) can act as a coordinator to indicate the start of a new frame / and synchronize all the nodes in the network. Each of the other nodes i then sends a packet in its allocated time slot, allowing the other nodes j to measure the arrival time yL.

[0062] The packet also contains the start time pF1 and the arrival time measured by the node » during the previous frame f - 1, corresponding to the transmissions of the nodes j T^- depending on the size of the packet, a limited number of measurements can fit in the packet.

[0063] With reference to Figure 1, it is assumed that the packets transmitted by the anchors (4, 5) contain the time of arrival measured by node 1 during the transmissions of the coordinating node jd'1. Thus, a set of measurements [W'1 1 is obtained, which allows the distance to be measured by two-way ranging relative to the T0 beacon, eliminating the problems of synchronization between the clocks of the anchors and the tags.

[0064] The method according to the invention is described in more detail in Figure 2. A position determination block 6 enables the implementation of a first step 1S1) which consists, at a given instant, of predicting the position Σ of the mobile carrier 1, and determining a set of predicted distances {1} between the mobile carrier 1 and a \dkj J / =w set of obstacles (2, 3, Nobs) in the environment of the mobile carrier 1.

[0065] The position determination block 6 can be a Kalman filter, an extended Kalman filter, a Bayesian estimator, an estimator using the method of least squares or maximum likelihood, or any estimator capable of determining a current position from a prediction and an update.

[0066] In applications such as navigation and trajectory tracking, where measurement models are often non-linear (as is the case with UWB measurements), the extended Kalman filter is capable of providing more accurate position and velocity estimates.

[0067] Alternatively, the position determination block 6 can be a particle filter. The particle filter estimates a complete probability distribution (and not a value, as the Kalman filter would) of the quantity sought (for example, a position). The particle filter allows the estimation of probability distributions not limited to Gaussian distributions, and can therefore, in theory, handle more complex problems. In practice, the probability distribution is modeled by a multitude (hundreds or thousands) of particles, each representing a possible value associated with a weight. The weights of the particles are updated from observations (measurements).

[0068] The arrival time measurements [t1 t ] are transmitted to the block of determination of position 6, and in particular to a state update block 7. A state prediction block 11 determines a set of predicted distances / between the mobile carrier 1 and the surrounding set of obstacles.

[0069] A radar processing block 8 is configured to perform a step S2) of receiving a channel impulse response (CIR for "Channel Impulse Response") between the first node T1 and the second node T0, at least one of the two nodes being carried on the mobile carrier 1, and the distance between the first node and the second node being known (i.e. fixed or measured).

[0070] According to an advantageous embodiment, the first node T1 and the second node T0 are UWB beacons mounted on the mobile carrier 1, which may be a drone. In this case, the spacing between the beacons is known and fixed. Alternatively, one of the two nodes is mounted on the mobile carrier 1, and the other node is positioned at a fixed location, for example on the ground or on a building. This node may also be co-located with one of the anchors (in this case, the distance between the first and second nodes is also known by a two-way ranging measurement).

[0071] The radar processing block 8 receives the CIR measurements. It is recalled that the impulse response of a channel is the output obtained when a channel is excited by a pulse (a Dirac signal). Since it can be difficult to generate a true Dirac pulse, various methods, known to those skilled in the art, can be implemented to measure the CIR, for example by using a pseudo-random sequence that has spectral properties similar to a Dirac pulse.

[0072]

[0073]

[0074]

[0075]

[0076]

[0077] The SI and S2 steps can be executed concurrently, although this is not essential. An MPC 9 extraction block ('MultiPath Component' or multipath component) is configured to extract a predefined number of NMPC multipath components from the channel impulse response. The multipath components are consisting of a set of distances [l IW ii=kN reflecting a direct route and PC Indirect paths between the first node T1 and the second node T0, at time k. An MPC 10 selection block is configured to associate the detected echoes. by the MPC extraction block 9 to the known obstacles determined by the position determination block 6. For this purpose, the MPC selection block 10 establishes correspondences between the predicted distances fl and the multi-W components route 1 \akJ in order to obtain a set of updated .MPC distances between the mobile carrier 1 and the set of obstacles (2, 3). i=^riu! It is assumed that obstacles are limited to flat surfaces and polygonal, with obstacles usually modeled as sets of small flat surfaces. The set of obstacles includes, in particular, the ground. The outputs of radar processing block 8 consist of a maximum of Nrad radar distances associated with the corresponding obstacles, with jV j < min((V Thus, the method according to the invention improves the accuracy of distance measurements between the mobile platform and obstacles in its immediate or surrounding environment. Indeed, after the association step, the distance used is the distance derived from the CIR (Controlled Internal Reference), which is inherently more accurate than the predicted distance. In an application such as navigating a drone within the racking of a logistics warehouse, the invention allows for precise and accurate control of the distance between the drone and the building walls or racking, particularly in the horizontal plane. Existing solutions have shown a degradation in horizontal positioning accuracy, with the associated risks (degradation caused by the drone's altitude relative to the anchors). This improved positioning is made possible by the radar function implemented between the first and second nodes. Ranging and radar measurements (CIR measurement) are performed using the same UWB equipment, without adding any specific equipment to the mobile platform.

[0078] Thus, the determination of a set of multipath components (radar measurements) is obtained by "diverting" the UWB radio from its primary functionality, which is achieved by exploiting internal data (the CIR) available in the UWB chips.

[0079] The method may also optionally include a step for updating the position of the mobile carrier, based on the updated distances. The calculation of the updated position is performed by triangulation, in a manner known to those skilled in the art.

[0080] The position . is updated according to the known equations of the extended Kalman filter.

[0081] In order to describe in detail these different sub-steps of the determination of a set of multipath components, it is necessary to define, beforehand, the signal ^t) received during a transmission between the first node T1 and the second node T0, which can be modeled by the following relation:

[0082] , , . zx \ J ( O = Li=1 ^-(0^(^-^-) +

[0083] mpc corresponds to the number of multipath components, aî and ri represent the complex amplitude and propagation delay of the zth multipath component, s ( t) is a received unit signal taking into account the transmitted signal and the antenna transfer function, and $(O is a random Gaussian white noise.

[0084] The reception of the signal ^t) Can be implemented by the second node T0; alternatively, the two nodes Tl and T0 can also be connected to a common processing unit which receives all the measurements made by the nodes Tl and T0.

[0085] A first substep SS1 of determining a set of multipath components consists of determining a first reference time ro corresponding to the propagation delay on the direct path between the first node and the second node. The first reference time can be provided by the UWB transmitter / receiver integrated into the UWB nodes.

[0086] A second substep SS2 consists of determining n additional times Ti corresponding to the propagation delays on the different indirect paths between the first node and the second node.

[0087] A third substep SS3 consists of calculating n measured distances as a function of the n additional times Tc of the first reference time To and the distance between the first node and the second node.

[0088] To do this, each of the n distances d, is calculated using the following formula:

[0090] d0 corresponds to the distance between the first node T1 and the second node T0, etc. corresponds to the speed of light.

[0091] The UWB transceiver integrated into one of the nodes digitizes the received CIR into memory. A peak detection algorithm is executed to calculate the first reference time r0. It is particularly advantageous to extract the multipath components (i.e., the reference time T0 and the additional times U) with better accuracy than the sampling period. Having subsampling also improves the accuracy of the mobile carrier localization 1.

[0092] Several techniques can be considered for peak detection, including the MUSIC (MUltiple Signal Classification) algorithm described in the article by [Jiokeng], or the highest peak detection algorithm described in the article by [Froehle]).

[0093] Figures 3 to 6 illustrate the sub-steps enabling the implementation of a particularly advantageous embodiment for determining a first reference time To.

[0094] A preliminary substep SS4 consists of setting a value for the first threshold Thl, and a search window ([Twl Tw2]). The preliminary substep SS4 therefore consists of a parameterization step, which can be executed before the mission, for example as a setting made by the user.

[0095] Advantageously, the first threshold Thl is a multiple of the ambient noise measured in the first search window [Twl Tw2]. This allows the threshold level to be calibrated relative to the noise level, which can vary. By setting the threshold in this way (for example, at three times the noise level), the lowest possible threshold is obtained (which improves the probability of detection) while minimizing the probability of a false alarm (when the noise exceeds the threshold).

[0096] Figures 3 to 6 illustrate the C1 norm ci of the CIR as a function of time, represented by a time index K. Indeed, if we denote C1 = || l.dtc) ||, with dtc the time resolution of CIR acquisition per I / Q channel, the value on the x-axis corresponds to the time index

[0097] In figures 3 to 6, "sample" means the time index1 of an acquisition point of the CIR, and "amplitude" means the corresponding CIR value.

[0098] A first sub-step SS5 consists of determining a first sample lSUpThl whose amplitude csupThl is greater than the first threshold Thl.

[0099] A second substep SS6 consists of determining a maximum amplitude c / m“ of a sample subsequent to the first sample (see [Fig. 4]). The determination of the maximum amplitude can be carried out by traversing the signal until it becomes decreasing (i.e., the next value is less than the current value).

[0100] A third substep SS7 consists of calculating a second threshold Th2 which is lower than the maximum amplitude cl^x (see [Fig. 5]). Advantageously, one can fix:

[0101]

[0102]

[0103]

[0104]

[0105] Thl = Cf^H Choosing a second threshold Th2 at half the maximum amplitude ct"ax allows detection in the steepest part of the peak, which therefore offers the best precision. It is possible to determine the second threshold Th2 differently, provided that it is less than the maximum amplitude cï"“x. A fourth step, SS8, involves determining a second sample, lpreTh2, which is the last sample before the second threshold, Th2, i.e., CipreTh2 < Thl. In other words, the amplitude of the sample following IpreThl ( / a- is greater or equal to Th2. The fourth step SS8 also includes the determination of a third sample IpostThl (i.e., IpreTh2 + 1), which is the first sample after the second threshold Th2, i.e., Ci Ti(Z > Th2 (see Figure 5). In other words, the amplitude of the previous sample IpostThl is less than or equal to IpostThl_ Th2.

[0106] When considered in terms of time scale (and no longer in terms of the number of samples), the following relationship can be written:

[0107] ti T.2 + dtc LJ lpostTh2 lpreTh2 c

[0108] [Fig.6] illustrates a detailed view of samples 742 to 752 of [Fig.5].

[0109] In a final substep SS9, the first reference time To is determined by a linear interpolation performed between the second sample IpreThl ct 'c third sample lp(Ktrm : [OHO] _ Thl-c^ ,. . . 1 0 “ + lpreTh2atc

[0111] Thus, the algorithm described above makes it possible to determine the first reference time corresponding to the direct path between the first node T1 and the second node T0, with a temporal resolution lower than that of UWB measurements. Furthermore, the algorithm described above does not require CIR accumulation or the use of an anti-aliasing filter.

[0112] It is recalled that the accumulation of the CIR consists of aggregating several CIRs measured at the same point, so that the sampling times are not exactly the same between the CIRs. This amounts to increasing the number of CIR samples (assuming that the latter remains constant).

[0113] The process according to the invention advantageously determines n additional times '& corresponding to the different indirect paths between the first node T1 and the second node T0 (cf. [Fig.7] which illustrates the substeps).

[0114] For this purpose, the time window comprising all the samples used in determining the first reference time To, that is to say the samples between the second sample Iprejhlct and the sample having the maximum amplitude Z* .max , • kz , is reset to zero.

[0115] Alternatively, a shorter or longer time window can be set to zero and used to determine additional times

[0116] A preliminary substep SS 10 consists of fixing the number n of MPC components to be extracted, a search window width ([Twl Tw2]), and an elimination window width ([Tw3 Tw4]). The number n of MPC components to be extracted can be fixed, for example experimentally, particularly according to the topological context in which the mobile carrier 1 will operate (logistics warehouse, urban canyon, etc.). The search window width ([Twl Tw2]) and the elimination window width ([Tw3 Tw4]) can be parameterized.

[0117] For each of the MPC components (i=l.. .NMpc) to be extracted, the following substeps are executed.

[0118] A first sub-step SSII consists of determining the sample 4'0 having the maximum amplitude Clm over the entire CIR.

[0119] In a second substep SS12, the arrival time of the ith multipath component is determined on the search window [Z / o • dtc-Twl, liQ. dtc + Tw2], in a manner identical to the determination of the first reference time described previously.

[0120] Thus, a first sample lXUpThi is determined whose amplitude csupThl is greater than the first threshold Thl. Then, a second threshold Th2 is calculated which is less than the maximum amplitude cko, preferably Th2 “Ci !2. A second sample lpreTh2, which is the last sample before the second threshold Th2, and a third sample lpostTh2, which is the first sample after the second threshold Th2, are determined. Finally, the linear interpolation described above is performed between the second sample lpreThl and the third sample lpostTh2.

[0121] Thus, the arrival time of the i-th multipath component is determined by the following formula:

[0122] Th2-ct dtc +1 preTh2^^c

[0123] A final substep SS13 consists of zeroing all samples in the elimination window [dtc - Tw3, li0. dtc + TwA ].

[0124] The process is repeated until the multipath NMpc components have been treated.

[0125] A set of multipath components, materialized by a set of NMPCs radar distances f | , is thus obtained from the arrival times according the following formula:

[0126]

[0127] c is the speed of light, d0 the distance between the first node Tl and the second node T0, assuming that d^ << CT^.

[0128] The algorithm for determining multipath components is lightweight in terms of computational complexity and the volume of data processed, particularly due to the reset performed before each iteration of the process for each new multipath component. Thus, the process is easily integrated and suitable for use on a mobile platform such as a drone, which has limited payload capacity. Furthermore, the accuracy is improved compared to known solutions.

[0129] As previously described, the MCP 10 selection block establishes correspondences between the predicted distances f "^d 1 and the multi-\dk.j J • =w J obs components route 1 fi™? | in order to obtain a set of updated distances IO^] between the mobile carrier 1 and the set of obstacles (2, 3). I ki ' J ; [. y * ' rad

[0130] According to an embodiment illustrated by Figure 8, the association of a predicted distance ["^ / 1] with a multipath component J l comprises, for For each obstacle in the set of obstacles, the comparison is made against a threshold of the difference between the predicted distance and each of the measured distances:

[0131] Thus, for each known obstacle qÎ (knowledge by prediction), all measured distances corresponding to a multipath component are considered to be associated with the obstacle qJ if the following relationship is satisfied: I jrad jrud I vrad \ak / ~akj |

[0132] Krad being a constant.

[0133] In Figure 8, the round marks correspond to the predicted distances [ "^1 1 , and the dot marks correspond to the distances 1 W ​multipath components [ | . For a frame k, if a known obstacle '= ^MPC q) is found in the frame, the measured distance of an ith multipath component is compared to each of the predicted distances ^rad- H, so this is a frame analysis ak,j by frame.

[0134]

[0135]

[0136]

[0137]

[0138]

[0139]

[0140]

[0141]

[0142] According to one embodiment, the threshold is adapted dynamically, that is to say, according to the dispersion of the predicted distances: LraJ ^7 / 1 ^rad rai] | ™ki " ®k,j | < '&kj corresponds to the standard deviation of the uncertainty of the predicted distances [ 'That's 1 for a k frame. J 1.4 T obs Alternatively, the following steps are repeated for all possible distance / obstacle associations between the set of multipath components and the set of obstacles in the environment of the mobile carrier: - calculation of a position of the mobile carrier 1 from the predicted distances and the measured distances, using said association; - from this position, calculation of the theoretical distances between the mobile carrier 1 and a plurality of fixed anchors and obstacles; - calculation of residuals defined as the difference between theoretical distances and measured distances; - calculate a likelihood score based on the residuals, which can advantageously be the inverse of the sum of the squared residuals; - determination of the association for which the likelihood score is the highest.

[0143] With this variant, it is no longer the predicted position that is used, but the set of UWB measurements and radar measurements.

[0144] It may be noted that the ground is one of the obstacles. The method according to the invention thus allows UWB beacons to act as altimeters with high accuracy, which makes it possible to do without a radar altimeter or a barometer, thereby allowing a saving in volume and mass.

[0145] Documents cited

[0146] [Barra]: J. Barra, T. Creuzet, S. Lesecq, G. Scorletti, E. Blanco, and M. Zarudniev, “Micro-drone ego-velocity and height estimation in gps-denied environments using an fmcw mimo radar” IEEE Sensors Journal, vol. 23, no. 3, pp. 2684-2692, 2023.

[0147] [Gentner] : C. Gentner and M. Schmidharnmer, « Ranging and multipathenhanced device-free localisation with densely-meshed ultrawideband deviees », 1ET Microwaves, Antennas & Propagation, vol. 17, no. 8, pp. 667—676, 2023. [Online]. Available: https: / / ietresearch.onlinelibrary. wiley.com / doi / abs / 1 0.1 049 / mia2.12373

[0148] ][Bocus] : M. J. Bocus and R. J. Piechocki, « Passive unsupervised localization and tracking using a multi-static uwb radar network », in 2021 IEEE Global Communications Conférence (GLOBECOM), 2021, pp. 01-06.

[0149] [Ledergerber] : A. Ledergerber and R. D'Andrea, « A multi-static radar network with ultra-wideband radio-equipped deviees », Sensors (Basel, Switzerland), vol. 20, 2020. [Online]. Available: https: / / api.semanticscholar.Org / CorpusiD:212750232

[0150] [Li] : C. Li, E. Tanghe, J. Fontaine, L. Martens, J. Romme, G. Singh, E. De Poorter, and W. Joseph, « Multistatic uwb radar-based passive human tracking using cots deviees », IEEE Antennas and Wireless Propagation Letters, vol. 21, no. 4, pp. 695— 699, 2022.

[0151] [Santoro] : L. Santoro, M. Nardello, D. Fontanelli, and D. Brunelli, « Uwb bistatic radar sensor: Across channels évaluation », IEEE Sensors Letters, vol. 7, no. 10, pp. 1-4, 2023.

[0152] [Jiokeng] : K. Jiokeng, G. Jakllari, A. Tchana, and A.-L. Beylot, « When fun discovered music: Accurate wifi-based ranging in the presence of multipath », in IEEE INFOCOM 2020 - IEEE Conférence on Computer Communications, 2020, pp. 1857-1866.

[0153] [Froehle] : M. Froehle, P. Meissner, and K. Witrisal, « Tracking of uwb multipath components using probability hypothesis density filters », in 2012 IEEE International Conférence on Ultra-Wideband, 2012, pp. 306-310.

Claims

Demands

1. A method for improving the position accuracy of a moving carrier, comprising steps of: S1) predicting a position of the moving carrier (1), and determining a set of predicted distances / between the moving carrier (1) and a set of obstacles (2, 3, q!) in the environment of the moving carrier (1); S2) acquiring a channel impulse response (CIR) between a first node (T1) and a second node (TO), at least one of the two nodes (T1,TO) being carried on the moving carrier (1); S3) determining a set of multipath components / from the channel impulse response (CIR); S4) associating at least one predicted distance / w 1 to at least one multipath component f 1 l \ \ if PC / corresponding so as to obtain a set of updated distances O1) between the mobile carrier (1) and the set of obstacles (2, 3, qJ).

2. A method according to claim 1, comprising, after step S4), a step S5) of updating the position of the mobile carrier as a function of the updated set of distances

3. 3. A method according to any one of the preceding claims, wherein the determination of a set of multipath components (d^) comprises the substeps of: - determining a first reference time (T1) corresponding to the direct path between the first node (T1) and the second node (T0); - determining n additional times (T1) corresponding to the different indirect paths between the first node and the second node; - calculating n distances measured as a function of the n additional times (T1), the first reference time, and the distance between the first knot and second knot.

4. 4. A method according to any one of the preceding claims, wherein the association of a predicted distance with a multipath component includes, for each obstacle in the obstacle set, the comparison against a threshold of the difference between the predicted distance and each of the measured distances.

5. 5. Method according to claim 4, wherein the threshold is determined dynamically as a function of the dispersion of the predicted distances.

6. 6. A method according to any one of claims 1 to 3, wherein the association of a predicted distance to a multipath component comprises, for all possible associations between the set of multipath components and the set of obstacles in the environment of the mobile carrier, the substeps of: - calculating a position of the mobile carrier (1) from the predicted distances and the measured distances, using said association; - from this position, calculating the theoretical distances between the mobile carrier 1 and a plurality of fixed anchors and obstacles; - calculating residuals defined as the difference between the theoretical distances and the measured distances; - calculating a likelihood score as a function of the residuals; - determining the association for which the likelihood score is the highest.

7. 7. A method according to any one of claims 3 to 6, wherein the determination of a first reference time (Tk$) comprises, in a first search window ([Twl Tw2]): - a determination of a first sample (lsupThl) whose amplitude (CsupThï) is greater than a first threshold (Thl); - a determination of a maximum amplitude of a sample subsequent to the first sample (IsupThÙ); - a determination of a second threshold (Th2) which is less than the maximum amplitude; - a determination of a second sample (lpreThï) whose amplitude immediately precedes the second threshold (Th2), and of a third sample (lpostTh2) whose amplitude immediately follows the second threshold (Th2); - the first reference time (Tt°) being determined by a linear interpolation performed between the second sample (LpreHû) and the third sample (IpostThl)-

8. Method according to claim 7, wherein the first threshold (Thl) is a multiple of the ambient noise measured in the first search window ([Twl Tw2]).

9. A method according to any one of claims 7 or 8, wherein the second threshold (Th2) is equal to half the maximum amplitude (

10. 'i )■ 10. A method according to any one of claims 3 to 9, wherein the determination of n additional times (' ^)) comprises, for each additional time between 1 and n: - a determination of the sample (Zro) having the maximum amplitude of the channel impulse response (CIR); - an extraction of the arrival time of the multipath component in a search window ([Z® dtc-Twl, Z,-o. dtc + Tw2]) centered on the sample (Zjq) having the maximum amplitude (¾) of the channel impulse response (CIR); - a zero-forcing of all samples in an elimination window ( [Z,-o- dtc- Tw3. li(y dtc+Tw^] ) centered on the sample (Z / o) having the maximum amplitude of the channel impulse response (CIR).

11. 11. A method according to any one of the preceding claims, wherein the prediction of a position of the moving carrier (1) is implemented by an extended Kalman filter.

12. 12. Method according to any one of claims 1 to 10, in the prediction of a position of the moving carrier (1) is implemented by particle filtering.

13. 3System for improving the accuracy of the position of a mobile carrier, the system being configured to: S1) predict a position of the mobile carrier, and determine a set of predicted distances between the mobile carrier and a set of obstacles in the environment of the mobile carrier; S2) acquire a channel impulse response (CIR) between a first node and a second node, at least one of the two nodes being onboard on the mobile carrier; S3) determine a set of multipath components from the channel impulse response (CIR); S4) associate a predicted distance with a corresponding multipath component so as to obtain a set of updated distances including at least one updated distance;

14. 14. Method according to claim 13, wherein the two nodes are carried on the mobile carrier, the first node (Tl) being an Ultra Wideband tag, the second node (TO) being an Ultra Wideband receiver.

15. 15. System according to any one of claims 13 or 14, wherein the mobile carrier (1) is a drone.

Images