METHOD FOR ESTIMATING THE MOTION OF AN OBJECT MOVING IN AN ENVIRONMENT AND IN A MAGNETIC FIELD

Description

GENERAL TECHNICAL FIELD

[0001] The present invention relates to the field of GPS-free navigation. More specifically, it relates to a method for estimating the movement of an object using magneto-visio-inertial techniques. STATE OF THE ART

[0002] The techniques known as visio-inertial navigation, or VINS (vision-aided inertial navigation system), make it possible to precisely estimate the speed (and hence the position by integration) of a solid in an environment considered static.

[0003] For example, document US 8,761,439 B1 describes a method for determining the pose of a solid in real time, in particular for augmented reality applications, by combining inertial data, magnetic data and visual data.

[0004] Specifically, in addition to the equations already used in classical inertial techniques, we compare successively acquired images of the environment to estimate the (linear) velocity vector V, from the angular velocity vector Ω (usually measured by gyrometers of the inertial unit). Images of the solid can also be acquired from the static environment for the same result.

[0005] Thanks to this information on the speed of the solid, the error in the position estimation will be in the worst case proportional to time, unlike classic inertial techniques where it is quadratic in time.

[0006] These visio-inertial techniques thus provide satisfaction, and require inertial sensors of lower performance than classic inertial methods (based on the integration through a Kalman filter or another observer of information coming from gyrometers and accelerometers embedded on the solid whose position and speed we wish to estimate: typically, the gyrometers "maintain" a reference frame, in which a double temporal integration of the accelerometer measurements makes it possible to estimate the movement) and therefore less bulky.

[0007] Thus, at least one inertial unit is coupled to at least one camera attached to the inertial unit. The data from these sensors are transmitted to an electronic acquisition and calculation card which provides the user with position and speed information by solving the system of equations.

[0008] However, we now see that WINES suffer from two defects: they have high power consumption. Indeed, while the consumption of standard MEMS inertial sensors is negligible, it has been shown that that of a camera and especially that of the processors needed to process the images is almost two orders of magnitude higher (see the document "Visual-Inertial Odometry on Chip: An Algorithm-and-Hardware Co-design Approach". Zhengdong Zhang, Amr Suleiman, Luca Carlone, Vivienne Sze, Sertac Karaman, 2017). It is thus recognized that approximately 10 Watts of power are required to operate the system on a CPU architecture, which is problematic if the system is battery powered.they need images that are easily differentiated from the environment, which is why they are often used in drones that are generally equipped with a camera as standard and operate in an outdoor environment (called "outdoor", typically natural or urban) which presents lit and varied landscapes, but they can become almost inoperative in an indoor environment (called "indoor"), in which we encounter situations such as a dark tunnel or a room with walls with little texture. In addition, the reaction of the system faced with such a situation is to increase the gain of the camera, which increases the noise in the image. Alternatively, we can provide a better quality camera (more sensitive to light), which in addition to being significantly more bulky (sensor and lens size), consumes more than a low-end camera.Similarly, in the case of systems where the camera is fixed in the environment, there may be an obstacle that hides the solid target.

[0009] Furthermore, these systems rely, to varying degrees, on the assumption of a field of view intersecting mostly a static part of the world, which may be temporarily false if the scene is dynamic. While purely vision-based methods can be found to segment static areas from dynamic areas in the image, these are computationally intensive and also tend towards increased consumption.

[0010] To solve the consumption problem, it is possible to change the hardware support, in particular by carrying out heavy calculations on an FPGA or by founding an ASIC or even a specialized processor (see for example the "HPU" from Microsoft), which remains very expensive.

[0011] Alternatively, it has been proposed to find a compromise between performance and consumption, for example by reducing the resolution of the acquired image, its quality (using a monochrome camera) or the acquisition frequency. Such a solution further degrades performance in the problematic cases mentioned above.

[0012] It would be desirable to have a new method of visio-inertial navigation that significantly reduces power consumption, while improving performance, particularly in difficult environments such as indoors. PRESENTATION OF THE INVENTION

[0013] The present invention thus relates, according to a first variant of a first aspect, to a method for estimating the movement of an object evolving in an environment and an ambient magnetic field according to the appended claim 1.

[0014] The present invention relates, according to a second variant of the first aspect, to a method for estimating the movement of an object evolving in an environment and an ambient magnetic field according to the appended claim 2.

[0015] Further advantageous features are provided in appended claims 3 to 8.

[0016] According to a second aspect, the invention relates to equipment for estimating the movement of an object evolving in an environment and an ambient magnetic field according to the appended claim 9.

[0017] According to other advantageous and non-limiting characteristics: The equipment is a box comprising the magnetic measuring means and / or the inertial measuring means and / or the optical acquisition means; The equipment is a mobile terminal or a server, adapted to communicate with the magnetic measuring means, the inertial measuring means and the optical acquisition means.

[0018] According to a third aspect, the invention relates to a system comprising the equipment according to the second aspect of the invention and the magnetic measurement means, the inertial measurement means and the optical acquisition means in connection.

[0019] According to a fourth and a fifth aspect, the invention relates to a computer program product comprising code instructions for executing a method for estimating the movement of an object moving in an ambient magnetic field according to the first aspect of the invention; and a storage means readable by computer equipment on which a computer program product comprises code instructions for executing a method for estimating the movement of an object moving in an environment and an ambient magnetic field according to the first aspect of the invention. PRESENTATION OF FIGURES

[0020] Other characteristics and advantages of the present invention will appear on reading the following description of a preferred embodiment. This description will be given with reference to the appended drawings in which: THE figures 1 is an equipment diagram for implementing a variant of the method according to the invention; The figure 2 represents in more detail an example of a box for implementing the method according to the invention according to another variant. figure 3 represents in more detail an example of a box for implementing the method according to the invention according to the first variant. DETAILED DESCRIPTION Principle

[0021] The present invention proposes to add magnetometers to the VINS so as to implement additional magneto-inertial navigation techniques. More precisely, a "visio-magneto-inertial" system called V-MINS is used, which can be seen as both a "visio-inertial" and "magneto-inertial" system. It is recalled that magneto-inertial navigation is well known, see document FR2914739.

[0022] It may seem counterintuitive to try to reduce the system's power consumption not by removing but by adding sensors, but it turns out that the vision and magnetic aspects are perfectly complementary and that using them together is more economical than using the vision aspect alone.

[0023] More specifically, the vision aspect and the magnetic aspect are effective in opposite environments: while indoor vision is more likely to be deficient, magnetometry is perfectly adapted. On the contrary, in an outdoor environment, while the absence of a magnetic gradient can hinder purely magneto-inertial navigation, vision is perfectly adapted.

[0024] Various ways of making the most of this data will be described later, but it will be understood that above all, the addition of magnetometers in a VINS system provides information allowing "dead-reckoning" (or dead reckoning) to be carried out autonomously, which is not possible with the inertial unit alone.

[0025] Indeed, in classic VINS systems, vision data is used to correct inertial data, i.e. to perform recalibration. We therefore understand that if the camera is switched off (for example because we are in a dark area and there is no longer any point in acquiring images and therefore it is better to switch off the camera to save energy) there is no more recalibration.

[0026] In the current V-MINS system, it is still possible to recalibrate even when the camera is switched off, thanks to magnetometric data.

[0027] This therefore makes it possible to reduce the necessary frequency of camera-based recalibration for the same level of performance, or even to do without the camera and the associated heavy processing over a certain time interval (in the operating areas of magneto-inertial navigation). Architecture

[0028] In reference to the figure 1 ,the present method allows the estimation of the movement of an object 1 evolving in an environment Σ and an ambient magnetic field (typically the Earth's magnetic field, possibly slightly altered by metallic objects or nearby electric currents), noted B The magnetic field is a vector field in this case three-dimensional, that is to say associating a three-dimensional vector with each three-dimensional point in which the object is mobile.

[0029] This object 1 can be any mobile object whose position knowledge is desired, for example a wheeled vehicle, a drone, etc., but also a person or a part of this person's body (their hands, their head, etc.).

[0030] The object 1 is equipped with magnetic measuring means 20, which are magnetometers and / or gradiometers. More precisely, instead of measuring the value of the components of the magnetic field Bthe latter directly measure the value of the components of the gradient of the magnetic field B , i.e. the value of the spatial derivatives. Such magnetic gradiometers 20 are known to those skilled in the art. According to certain embodiments of the invention, it is possible to use gradiometers directly measuring the value of the second derivatives (gradient of order two), and generally of the i-th derivatives (gradient of order i).

[0031] In the remainder of the description we will take the example of magnetometers but those skilled in the art will be able to transpose this to gradiometers.

[0032] The magnetometer(s) 20 are “axial”, i.e. capable of measuring a component of said magnetic field, i.e. the projection of said magnetic field vector. B along their axis (or in the case of a gradiometer a component of a spatial derivative of order i of said magnetic field, i.e. the variation of a component of ∇ i- 1< Bat the level of said gradiometer along their axis).

[0033] Advantageously, the magnetometers 20 are at least 3n+3 in number, advantageously organized in groups of three in “triaxes”, i.e. a triplet of two-by-two orthogonal magnetometers 20 associated with the same spatial position and measuring the magnetic field along the three axes.

[0034] Preferably, the orthonormal reference frame associated with the object is chosen by convention (and for ease for the remainder of this description) such that the triaxes are advantageously oriented in accordance with said orthonormal reference frame, so as to further facilitate the calculations.

[0035] But the person skilled in the art will in any case be able to transpose this to any spatial arrangement of gradiometers / magnetometers.

[0036] The object 1 is also equipped with inertial measurement means 24. Such means are widely known to those skilled in the art and typically constitute an inertial unit comprising at least one accelerometer and / or one gyrometer, preferably three accelerometers and three gyrometers, also arranged in a triaxial arrangement. The gyrometers measure the instantaneous angular velocity of the inertial unit relative to the terrestrial reference frame, denoted Ω. The accelerometers are sensitive to external forces other than gravitational forces applied to the sensor, and make it possible to measure an acceleration denoted γ.

[0037] The inertial and magnetic measuring means 20, 24 are integral with the object 1, i.e. they have a substantially identical movement in the terrestrial frame of reference. Preferably, the frame of reference of the object 1 is provided with an orthonormal Cartesian reference in which the coordinates are noted ( x 1 , x 2 , x3), the magnetometers 20 thus have a predetermined position in this reference frame.

[0038] The system for implementing the method comprises at least optical acquisition means 26. In a main embodiment of the invention shown in the figure 2(which will be taken as an example in the remainder of this description), the optical acquisition means 26 are also integral with the object 1 (in other words they have the same movement) and observe the environment Σ. To facilitate understanding of the present method, the assembly formed by the object 1 and the optical acquisition means 26 will be reduced to a point, located at the optical center of the optical acquisition means 26. These optical acquisition means 26 can be of varied nature but preferably consist of one or more cameras and advantageously a single camera to limit electrical consumption. The camera(s) can be depth cameras, but preferably a conventional camera will be used which restores a visible image of the environment and not a depth image, and in particular a simple CCD or CMOS camera (color or black and white) which is sufficient, less bulky and less energy consuming.

[0039] In the alternative embodiment (which will be described more specifically later) and corresponds to the figure 1 , the optical acquisition means 26 are fixed relative to the environment Σ and observe the object 1. It will be understood that the first embodiment (on-board optical acquisition means 26) is preferred, because it allows the continuous estimation of the movement of the object 1, including if it moves over long distances. In the second mode, one is limited to the field of vision of the optical acquisition means 2, but this is perfectly suitable if one follows the movement of parts of a user's body for example.

[0040] In reference to the figure 2, the inertial and magnetic measurement means 20, 24, and where appropriate the optical acquisition means 26, are preferably those of a housing 2 having means 23 for attachment to the object 1. These attachment means 23 consist, for example, if the object 1 is a person, of a bracelet, for example with a self-gripping band which encloses the limb (for example the wrist, the hand or the head) and allows the integral connection, see the example of the figure 3 . It will be understood that the invention is not limited to the estimation of the movement of a person, but it is particularly advantageous in such use because it allows a very reduced size, which is necessary for the box to be portable by a human in an ergonomic manner.

[0041] The housing 2 may comprise processing means 21 (typically a processor) for the direct implementation in real time of the processing of the present method, or the measurements may be transmitted via communication means 25 to an external device such as a mobile terminal (smartphone) 3, or even a remote server 4, or the measurements may be recorded in local data storage means 22 (a memory, for example of the flash type) local memory for subsequent processing, for example on the server 4.

[0042] The communication means 25 may implement short-range wireless communication, for example Bluetooth or Wifi (in particular in an embodiment with a mobile terminal 3) or even be means of connection to a mobile network (typically UMTS / LTE) for long-distance communication. It should be noted that the communication means 25 may be, for example, a wired connection (typically USB) for transferring data from the local data storage means 22 to those of a mobile terminal 3 or a server 4.

[0043] If it is a mobile terminal 3 (respectively a server 4) which hosts the “intelligence”, it comprises processing means 31 (respectively 41) such as a processor for implementing the processing of the present method which will be described. When the processing means used are those 21 of the box 2, the latter can also include communication means 25 for transmitting the estimated position. For example, the position of the wearer can be sent to the mobile terminal 3 to display the position in an interface of navigation software.

[0044] In the remainder of this description, it will be seen that the data processing means 21, 31, 41 respectively of the box 2, of a smartphone 3 and of a remote server 4 can indifferently and depending on the applications carry out all or part of the steps of the method. Process

[0045] In a first step (a), the method comprises the acquisition (in particular in a repository of the object 1) of up to three types of data: by the inertial measurement means 24 integral with said object 1, of at least one component: o of the acceleration and / or o of the angular velocity of the object 1 (preferably three components of the angular velocity, and three components of the acceleration). by the magnetic measurement means 20 (the gradiometer(s) / magnetometer(s)) integral with said object 1, of a component: o of the magnetic field and / or o of at least one i-th derivative of said magnetic field B , with one or more values i> 1 (as explained, various gradiometers / magnetometers can be used to measure various orders of derivatives or directly the values ​​of the magnetic field). by the optical acquisition means 26, either: o If they are attached to the object 1, consecutive images of the environment Σ; o If they are attached to the environment Σ of the object 1, consecutive images of the object 1.

[0046] For convenience, in the remainder of this description, the term “inertial data” will be used to refer to the component(s) of the acceleration and / or angular velocity of the object 1 acquired by the inertial measurement means 24; “magnetic data” will be used to refer to the component(s) of the magnetic field and / or of at least one i-th derivative of said magnetic field. B acquired by the magnetic measuring means 20; and “vision data” the images acquired by the optical acquisition means 26.

[0047] These data are advantageously acquired with dt sampling (i.e. every “dt” seconds) with dt very small compared to the characteristic time of the movements of object 1, typically 40 ms.

[0048] As will be seen later, each of these quantities is not necessarily measured at each time step. More precisely, at each instant the magnetic measuring means 20 and / or the optical acquisition means 26 can be switched off so as to permanently acquire only two types of data out of three, or even only one: inertial data and, preferably, magnetic data or vision data. It is noted that it is already known in VINS systems to acquire images at a lower frequency than the inertial data to limit consumption, but this frequency remains relatively high under penalty of having a significant drift.

[0049] In a step (b), the data processing means 21, 31, 41 estimate at least one component of the movement of the object 1, in particular all the components of the linear speed V. One or more components of the angular speed vector Ω can also be estimated or at least verified. This estimation is made using the inertial data, as well as the magnetic data and / or the vision data. In other words, the data set used can vary dynamically: either inertial + magnetic, or inertial + vision, or inertial + magnetic + vision.

[0050] Different ways to estimate the object's motion from this data will be described later. Data selection

[0051] Note that step (a) always includes the continuous acquisition of inertial data, but possibly only intermittent acquisition of magnetic data or vision data. Step (a) can, however, alternatively include the continuous acquisition of the three types of data, but not necessarily their consideration all the time: remember that the majority of the energy consumption is due to the associated heavy processing.

[0052] In this respect, step (a) advantageously includes the selection of the magnetic data and / or the vision data. As explained, it is generally not necessary to have two data simultaneously since one of the two is always valid (and most often both): outdoors, at least the vision data will be reliable, and indoors at least the magnetic data will be reliable.

[0053] In this respect, the selection of the magnetic data and / or the vision data is advantageously the consequence of a test of the validity of the magnetic data and / or the vision data.

[0054] For example, it may be decided that one of these data types is the default type (the other being designated the recalibration data), and it is this type whose validity is checked. It is indeed cleverly assumed that if the default data is not valid, then the other data is. It can be provided that when the default data is not reliable, either the recalibration data or both data are selected.

[0055] For example, according to a first example (case of the figure 2) object 1 is a drone operating outdoors. It is known that processing vision data will always give good results in recalibrating inertial data, but that such processing of vision data will be energy-intensive. Saving energy is crucial for a drone, which runs on batteries and has only a limited autonomy. We can then define that the default data type is the magnetic type, that is to say that in normal situations the magnetic data is selected and used, but not the vision data (since the energy consumption of the magnetometers is much lower than that of the camera). If the validity test is not conclusive for the magnetic data (typically in the case of magnetic disturbance), then the vision data is selected instead to perform a recalibration.

[0056] According to a second example (case of the figure 1) the object 1 is a virtual reality headset, with the optical acquisition means 26 observing the headset and not integral with the latter. Since the means 26 are not mounted on the object 1, their power consumption is less problematic, which is why we can then define that the default data type is the vision type, that is to say that in a normal situation the vision data is selected and used, but not the magnetic data. If the validity test is not conclusive for the vision data (typically if the user's hands hide the object 1), then the magnetic data is selected instead to perform a recalibration.

[0057] Various validity tests can be implemented, and possibly combined. It will be possible to check indiscriminately (i.e. to test the validity of both the magnetic data and the vision data) the quality of the trajectory by estimating, for example, a drift uncertainty, particularly in the form of a cone. If the width of this cone is beyond a threshold, it means that the default data is not reliable.

[0058] More specifically for vision data, we could for example: Check that the images have at least a predetermined minimum number of points of interest; Check that the light intensity is above a threshold; Check that the contrast in the image is sufficient; etc.

[0059] More specifically for magnetic data, we can for example: Evaluate the stationarity of the magnetic field (see request FR1756958); Check the non-uniformity of the gradient. Etc.

[0060] It should also be noted that there is not necessarily an assessment of the validity for the selection, and that at least at certain times both the magnetic data and the vision data can be selected by default independently of the validity of the data, for example at the initialization of the process, or when the quality of the mapping declines (see below), or quite simply after a certain duration.

[0061] Thus, an inter-image temporal distance criterion can, for example, be taken into account to at least check the drift of navigation based on inertial and magnetic data, using image information, and correct only if necessary.

[0062] If accuracy is the primary criterion of the application, it is also possible to use both default measurements, and periodically check the consistency of the two measurements with each other. Validity tests are then only performed in the event of inconsistency, to temporarily disable a sensor. Movement estimation

[0063] As explained, in a step (b), the data processing means 21, 31, 41 estimate at least one component of the movement of the object 1, in particular all the components of the linear speed V and / or angular speed Ω, using the inertial data and the magnetic data.

[0064] Many techniques are known for estimating one or more components of motion either using inertial data and magnetic data or using inertial data and vision data, including techniques using inertial data continuously and vision data intermittently.

[0065] Thus, step (b) can simply implement a magneto-inertial navigation algorithm and a visio-inertial navigation algorithm, and use one or the other depending on the selected data.

[0066] Alternatively, it is simple and effective to use known visio-inertial navigation techniques and add consideration of magnetic data.

[0067] For example, visio-inertial filtering methods are known, in particular of the "inverse square root" type. In these filtering methods, the state is composed of the attitude of the object as well as the attitude of a set of "key images", and a filter evolution model based on inertial equations is used. It is sufficient to add at least one magnetic equation to this filter, such as Ḃ = -Ω ∧ B + ∇ B. V, to make the filter magneto-visual-inertial.

[0068] In the absence of the magnetic data (if only the vision data is selected), the contribution of the magnetic equation(s) can be suppressed. And in the absence of the vision data (if only the magnetic data is selected), the filter will continue to work without problems.

[0069] Generally, step (b) preferably comprises a first sub-step (b1) of constructing mapping elements of the environment using the inertial data as well as the magnetic data and / or the vision data, and a second sub-step (b2) of estimating at least one component of the movement of the object 1 from said mapping elements (by locating the object 1).

[0070] The mapping elements are either the complete map or just landmarks called "landmarks", fixed and identifiable in the environment (like lighthouses in maritime navigation).

[0071] Note that these two steps are not necessarily separate and can be performed simultaneously. In the case where the environment is mapped, this technique is known to those skilled in the art as SLAM ("simultaneous localization and mapping").

[0072] In the case where there is no complete intermediate mapping but only the determination of the position of landmarks in step (b1), the technique is known to those skilled in the art as "visual odometry". Note that the position of the landmarks may be an implicit data of the odometry method. For example, in a filtering case, the position of these landmarks is not always included in the state. Instead, they can be triangulated and instantly marginalized when the landmarks become more visible.

[0073] In both cases, step (b1) includes the construction of a mathematical representation of the environment possessing the information necessary for the relocation of object 1 evolving in the volume of the mapped environment, or at least provided with landmarks.

[0074] This representation may contain information from the optical acquisition means 26 (description of visual landmarks) or from the magnetic measurement means 20 (map of the magnetic field and its gradient), in particular depending on what the default data is.

[0075] Note that it is possible to add information on the magnetic field to the map in addition to visual landmarks.

[0076] A simple example might be to associate for each landmark the angle between the direction of the measured field when the landmarks are detected and the line of fire toward the landmark. Even if the field does not give precise north, this can make relocation easier on a large map, since we only have to associate the landmarks we currently see with the landmarks on the map that were seen at a time when the field was roughly aligned with the currently measured field.

[0077] A preferred embodiment of the process of constructing mapping elements of step (b1) can be described in three sub-steps which can optionally be executed simultaneously: (i) Reception of inertial, magnetic and / or vision data; (ii) Rough but efficient construction of mapping elements, preferably implemented only with magnetic and inertial data, so as to reduce the number of images to be recorded (images are mainly used for reconstruction and are less important for rough localization); (iii) Refinement of mapping elements in order to correct possible drifts from step ii (potentially carried out a posteriori), in particular with vision data.

[0078] We can therefore see, for example, that as long as the magnetic data is reliable and selected, we can remain at step ii, and implement step iii if this data is evaluated as unreliable and unselected.

[0079] The mapping elements necessarily contain 3D positioning information of the visual landmarks: this information can come from an optical sonar type method (see application WO 2011147947) or from traditional “Structure From Motion” (SfM) or “Bundle Adjustment” (BA) methods, iterative non-linear optimization on all the data finding the 3D position of the points best explaining all the measurements. These SfM methods exist in the state of the art in versions incorporating or not inertial data. Note that all or part of the mapping elements can just as well be known a priori (in particular to initialize the process).

[0080] In practice, the accuracy and integrity of these mapping elements is paramount for the localization phase. And SfM methods are generally necessary to achieve the final level of accuracy and integrity required for the application. However, these methods are cumbersome, and their computational time increases much more than linearly with the size of the mapping.

[0081] The basis of these SfM methods being a non-linear optimization, the quality of the initialization of the position variables of the landmarks and the cameras having observed them during the mapping phase is essential for convergence towards the good local minimum and rapid convergence.

[0082] We see that the present process cleverly combining the use of inertial, magnetic and vision data then allows: to significantly reduce the computing power required to construct the position variables of the initialized landmarks compared to the method used today, including the VINS filtering methods mentioned above. These lower computing powers may result, for example, from the fact that it is possible to decouple the localization and triangulation part or from the fact that the localization apriori based on magnetic and inertial data makes it possible to reduce the search for points of interest on a part of the image. To help the initialization of the variables (step ii) as close as possible to the final solution in cases where the vision is ambiguous (flat wall, dark place), This allows faster convergence of the SfM algorithms. To make the initialization of the variables of the SfM problem more robust with respect to the type of movement carried out during data collection.This allows convergence in the good local minimum of the SfM problem even if the movements made by the user, a priori untrained, are not optimal for the reconstruction of the mapping. (essential in practice, the state-of-the-art devices for 3D reconstruction advise avoiding pure rotation movements, which poorly condition the optimization problem, the addition of the magnetic data makes it possible to avoid these conditioning problems polluting the trajectory when the movement made by the user is degenerate). Calibration

[0083] Preferably, step (a) comprises the prior calibration of one of the inertial measurement means 24, the magnetic measurement means 20 and the optical acquisition means 26, with the data acquired by the other two. Preferably, either the optical acquisition means 26 are calibrated with the inertial data and the magnetic data, or the magnetic measurement means 20 with the inertial data and the vision data.

[0084] Indeed, to the extent that there is potentially redundancy in the information carried by the inertial data, the magnetic data and the vision data, when all three are considered it is possible to correct parameters linked to the sensors (bias or scale factors of the magnetic measurement means 20, distance between the optical axis of the optical acquisition means 26 and the measurement means 20, 24, etc.) and therefore to improve the estimation of the movement. in fine. Equipment and system

[0085] According to a second aspect, the invention relates in particular to the equipment 2, 3, 4 for implementing one or other of the embodiments of the method.

[0086] As explained previously, according to a first embodiment the equipment is an autonomous box 2 comprising the magnetic measurement means 20 and / or the inertial measurement means 24 (preferably the magnetic measurement means 20 and the inertial measurement means 24) and the data processing means 21 configured for the implementation of the steps of the method.

[0087] The box 2 further comprises means 23 for attaching the box 2, and where appropriate, data storage means 22 (for storing, for example, estimated movements) and / or communication means 25 for exporting the results.

[0088] Either the box 2 includes the optical acquisition means 26 (if they observe the environment), or these are distinct and arranged in a fixed manner and communicate with the box 2. In this mode where the box 2 contains the intelligence, we are preferentially in the case where the box 2 also contains the optical acquisition means 26 (case where the object 1 is for example a drone)

[0089] According to a second embodiment, the equipment is a mobile terminal 3 or a server 4, adapted to communicate with the magnetic measurement means 20 and the inertial measurement means 24, and the optical acquisition means 26.

[0090] Typically, the magnetic measuring means 20 and the inertial measuring means 24 are arranged in a housing 2, and the optical acquisition means 26 are arranged in a fixed manner in the environment, but it is possible to have the three means 20, 24, 26 in the same housing 2, or two or even three separate housings.

[0091] In other words, the terminal 3 or the server 4 comprises the processing means 31 or 41 configured for the implementation of the steps of the method. Each box 2 can still comprise data processing means 21 for the control of the means 20, 24, 26 and the transmission (via communication means 25) of the measured data to the data processing means 31, 41.

[0092] It should be noted that the means 21, 31, 41 may, where appropriate, share steps of the process.

[0093] The invention in this case also concerns the system comprising the equipment 3, 4 according to this embodiment and the “satellite” box(es) 2 in connection

[0094] In all cases, the data processing means 21, 31, 41 of the “main” equipment 2, 3, 4 are configured to implement: A module for receiving: ∘ at least one component of an acceleration and / or an angular velocity of the object 1, called inertial data, acquired by inertial measurement means 24 integral with said object 1; ∘ at least one component of the magnetic field and / or an i-th derivative of the magnetic field, called magnetic data, by magnetic measurement means 20 integral with said object 1; ∘ consecutive images of the environment Σ or of the object 1, called vision data, by optical acquisition means 26 respectively integral with said object 1 or the environment Σ; A module for estimating at least one component of the movement of said object 1 using the inertial data, as well as the magnetic data and / or the vision data. Computer program product

[0095] According to a third and a fourth aspect, the invention relates to a computer program product comprising code instructions for the execution (on the processing means 21, 31, 41) of a method for estimating the movement of an object 1 evolving in an environment Σ and an ambient magnetic field according to the first aspect of the invention, as well as storage means readable by computer equipment (for example data storage means 22) on which this computer program product is found.

Claims

1. Method for estimating the movement of an object (1) moving in an environment (Σ) and an ambient magnetic field, the method comprising the steps of: (a) Acquisition: ∘ by inertial measurement means (24) fixed with reference to said object (1), of at least one component of an acceleration and / or an angular velocity of the object (1), designated inertial datum; ∘ by magnetic measurement means (20) fixed with reference to said object (1), of at least one component of the magnetic field and / or an i-th derivative of the magnetic field, at the magnetic measurement means (20), designated magnetic datum; ∘ by optical acquisition means (26) fixed with reference to said object (1), of consecutive images of the environment (Σ), designated vision datum; evaluation of the validity of the magnetic datum and / or the vision datum, and selection of the magnetic datum and / or the vision datum as a function of the result of said validity evaluation; one of the magnetic datum and vision datum being a default datum and the other a resetting datum, the evaluation of the validity being implemented for the default datum, and the selected datum being: o the default datum if the default datum is evaluated valid, o the resetting datum if the default datum is not evaluated valid; (b) Estimation by the data processing means (21, 31, 41) of at least one component of the movement of said object (1) using the inertial datum, as well as the datum or data selected from among the magnetic datum and the vision datum.

2. Method for estimating the movement of an object (1) moving in an environment (Σ) and an ambient magnetic field, the method comprising the steps of: (a) Acquisition: ∘ by inertial measurement means (24) fixed with reference to said object (1), of at least one component of an acceleration and / or an angular velocity of the object (1), designated inertial datum; ∘ by magnetic measurement means (20) fixed with reference to said object (1), of at least one component of the magnetic field and / or an i-th derivative of the magnetic field, at the magnetic measurement means (20), designated magnetic datum; o by optical acquisition means (26) fixed with reference to the environment (Σ), of consecutive images of said object (1), designated vision datum; evaluation of the validity of the magnetic datum and / or the vision datum, and selection of the magnetic datum and / or the vision datum as a function of the result of said validity evaluation; one of the magnetic datum and vision datum being a default datum and the other a resetting datum, the evaluation of the validity being implemented for the default datum, and the selected datum being: o the default datum if the default datum is evaluated valid, o the resetting datum if the default datum is not evaluated valid; (b) Estimation by the data processing means (21, 31, 41) of at least one component of the movement of said object (1) using the inertial datum, as well as the datum or the data selected from among the magnetic datum and the vision datum.

3. Method according to one of claims 1 and 2, in which the evaluation of the validity of the magnetic datum and / or the vision datum includes the evaluation of an uncertainty on the estimation of the movement.

4. Method according to one of claims 1 to 3, in which the evaluation of the validity of the magnetic datum includes the evaluation of the stationarity of the ambient magnetic field, or the evaluation of the non-uniformity of the gradient.

5. Method according to one of claims 1 to 4, in which the evaluation of the validity of the vision datum includes the evaluation of the number of points of interest, the evaluation of the light intensity, the evaluation of an indicator of the image contrast, or the evaluation of the success of the detection of the object in the image in the case where the optical acquisition means (26) are fixed with reference to the environment (Σ).

6. Method according to one of claims 1 to 5, in which step (b) includes in a preferred manner a first sub-step (b1) of constructing elements for mapping the environment using the inertial datum as well as the magnetic datum and / or the vision datum, and a second sub-step (b2) of estimating the at least one component of the movement of the object (1) from said mapping elements.

7. Method according to claim 6, in which the first sub-step (b1) of constructing elements for mapping the environment using the inertial datum as well as the magnetic datum and / or the vision datum includes a first rough construction of elements for mapping the environment using the inertial datum as well as the magnetic datum, then refining the elements for mapping the environment using the vision datum.

8. Method according to one of claims 1 to 7, in which step (a) includes the prior calibration of one among the inertial measurement means 24, the magnetic measurement means 20 and the optical acquisition means 26, with the data acquired by the two others.

9. Equipment (2, 3, 4) for estimating the movement of an object (1) moving in an environment (Σ) and an ambient magnetic field, comprising data processing means (21, 31, 41) configured to implement: - A module for receiving: o at least one component of an acceleration and / or an angular velocity of the object (1), designated inertial datum, acquired by inertial measurement means (24) fixed with respect to said object (1); o at least one component of the magnetic field and / or an i-th derivative of the magnetic field, designated magnetic datum, by magnetic measurement means (20) fixed with respect to said object (1); o consecutive images of the environment (Σ) or the object (1), designated vision datum, by optical acquisition means (26) respectively fixed with respect to said object (1) or the environment (Σ); of evaluation of the validity of the magnetic datum and / or the vision datum, and selection of the magnetic datum and / or vision datum as a function of the result of said validity evaluation; one of the magnetic datum and vision datum being a default datum and the other a resetting datum, the evaluation of the validity being implemented for the default datum, and the selected datum being: o the default datum if the default datum is evaluated valid, o the resetting datum if the default datum is not evaluated valid; - A module for estimating at least one component of the movement of said object (1) using the inertial datum, as well as the datum or data selected from among the magnetic datum and the vision datum.

10. Equipment according to claim 9, being a case (2) including the magnetic measurement means (20) and / or the inertial measurement means (24) and / or the optical acquisition means (26).

11. Equipment according to claim 10, being a mobile terminal (3) or a server (4), suited for communicating with the magnetic measurement means (20), the inertial measurement means (24) and the optical acquisition means (26).

12. System including the equipment (3, 4) according to claim 11 and the magnetic measurement means (20), the inertial measurement means (24) and the optical acquisition means (26) in connection.

13. Computer programme product including code instructions for the execution of a method for estimating the movement of an object (1) moving in an environment (Σ) and an ambient magnetic field according to one of claims 1 to 8, when said programme is executed on a computer.

14. Storage means readable by a computer equipment on which a computer programme product includes code instructions for the execution of a method for estimating the movement of an object (1) moving in an environment (Σ) and an ambient magnetic field according to one of claims 1 to 8.