Autonomous navigation system of an underwater vehicle, navigation system and navigation method implemented by such a system

Abstract

Autonomous navigation system for an underwater vehicle, navigation system, and navigation method implemented by such a system. The invention relates to a navigation system (100) for an underwater vehicle comprising: - a data acquisition module (200) for collecting data from sensors of said vehicle and a surface unit; - a fusion module (300) for retrieving a portion of said collected data, including a processing algorithm for discriminating said retrieved data and for determining a position of said vehicle; - a movement module (400) for calculating the difference between said determined position and a setpoint position and for sending a movement command to said vehicle when the difference between said determined position and said setpoint position exceeds a threshold value; - a monitoring module (500) for ensuring that the movement command sent is consistent. Figure 2

Description

Title of the invention: Autonomous navigation system for an underwater vehicle, navigation assembly and navigation method implemented by such a system. Technical field

[0001] The invention relates to the field of underwater inspection.

[0002] More specifically, the invention relates to a navigation system for an underwater vehicle.

[0003] The invention also relates to a navigation system comprising such a system and a navigation method implemented by such a system. Previous technique

[0004] Underwater inspection may be necessary for example for the purpose of inspecting underwater infrastructure such as offshore wind turbines, offshore platforms, oil platforms, pipelines, for defense, etc., as part of the inspection of installations such as bridges, dikes, dams, anchorages etc. or more generally any submerged element.

[0005] It may also be necessary to resort to underwater inspection, for example, for observation purposes, or for dimensional measurements or monitoring of submerged structures. Depth measurements or the creation of relief maps, and the detection of unknown objects, can also be carried out in the context of the oil, energy, scientific, or military industries.

[0006] As is known, such underwater inspections can be carried out using remotely operated underwater vehicles from the surface, which can be designated by the acronym "ROV" for "Remotely Operated Vehicle" in English.

[0007] ROVs are typically used for inspecting underwater structures, for maintenance and repair operations of offshore installations, for search and rescue missions, and for scientific studies.

[0008] The ROV is a remotely controlled underwater vehicle, piloted from the surface by a human operator, via an umbilical cable which provides the electrical power and which allows the transmission of data between the ROV and the surface.

[0009] Thus, the fact that data transmission between the ROV and the surface is carried out via an umbilical cable allows for real-time and high-resolution transmission of acquired images and videos.

[0010] Also, the fact that the ROV is controlled in real time by an operator allows for direct interventions and adjustments immediate, thus giving the ROV a good ability to perform complex tasks with direct human intervention.

[0011] However, the length of the umbilical cable limits the range of action of the ROV.

[0012] Also, the vessel from which the ROV is piloted remains within the perimeter of the area to be inspected for the entire duration of the ROV's navigation.

[0013] However, the cost incurred to keep the ship within the perimeter of the area to be inspected for the entire duration of the ROV's navigation, on which a team of operators is embarked, is extremely high.

[0014] Moreover, the cost of maintaining an ROV is very high.

[0015] As is known, underwater inspections can also be carried out using autonomous underwater vehicles with a pre-programmed route, which can be designated by the acronym "AUV" for "Autonomous Underwater Vehicle" in English.

[0016] AUVs define a second type of underwater vehicle which, like ROVs, can be used for various missions in aquatic environments.

[0017] AUVs are typically used to establish underwater mapping, to carry out oceanographic surveys, for defense, in the context of environmental monitoring operations or exploration of inaccessible areas.

[0018] The AUV is programmed before the mission to follow a specific route autonomously and perform predefined tasks without the need for piloting by an operator.

[0019] Unlike the ROV, no umbilical cord is used to control the AUV, as the AUV is powered by batteries. Therefore, the use of the AUV allows it to cover greater distances than an ROV and to operate in challenging environments.

[0020] Also, since the operation of the AUV does not require piloting by an operator, the operating costs of an AUV are generally lower than those of an ROV.

[0021] However, the fact that the AUV is powered by batteries limits mission duration to the battery life. Also, the AUV's ability to react to unforeseen situations is limited. Similarly, the AUV is relatively inflexible for performing complex tasks.

[0022] Furthermore, among the autonomous navigation methods used for underwater positioning of AUVs, inertial navigation is commonly used through the use of an inertial measurement unit (IMU) integrated into the AUV, comprising accelerometers and gyroscopes, including fiber optic gyroscopes. The inertial measurement unit allows for measure the accelerations on all axes of the AUV and determine a motion and velocity of the AUV.

[0023] However, obtaining a high-precision measurement with this type of method can be very expensive.

[0024] Thanks to advances in sonar technology, modern AUVs can use a Doppler Velocimeter (DVL) acoustic sensor instead of the accelerometers in the inertial measurement unit. The DVL acoustic sensor allows for the direct measurement of the AUV's 3D position and velocity, without integration.

[0025] Despite this, the combination of a DVL sensor with a fiber optic gyroscope still generates accumulated errors, limiting navigation accuracy to 0.2% of the distance traveled.

[0026] To correct these accumulated errors, several methods are currently employed.

[0027] Accumulated error correction can be based on the use of GPS data. To do this, the AUV periodically surfaces to receive GPS signals and correct its position, as GPS signal reception is impossible underwater. However, implementing this method increases the AUV's energy consumption and thus limits its operating time and range.

[0028] Acoustically assisted navigation can also correct accumulated errors. To this end, the AUV can include a navigation system incorporating a so-called "long baseline" (LBL) acoustic positioning system or a so-called "short baseline" (SBL) positioning system. However, the use of long or short bases within the AUV's working area limits the AUV's effective range.

[0029] Topography-assisted navigation, using accurate underwater topographic charts for positioning and navigation, can also correct accumulated errors. However, this requires detailed charts that are only available for certain coastal regions.

[0030] Furthermore, with the continued development of autonomous underwater navigation technology, it is essential to reduce reliance on external navigation tools to enable more extensive and less costly autonomous navigation.

[0031] The simultaneous localization and mapping method, known as "SLAM", an English acronym for "Simultaneous Localization And Mapping", is considered a key method for enabling underwater vehicles to navigate autonomously in unknown environments.

[0032] The SLAM method allows the underwater vehicle to create a map of the underwater environment and to position itself using information from onboard vision cameras.

[0033] Sonar technologies, particularly scanning sonars, provide good imaging capabilities and can also work with inertial sensors on the underwater vehicle to achieve autonomous navigation. However, scanning sonars are not suitable for precise autonomous navigation in complex local maritime areas, and computer vision synchronized with all sensors can complement this aspect.

[0034] Also, carrying out inspections autonomously, i.e. without human intervention, requires the use of precise real-time positioning solutions to enable the underwater vehicle's navigation system to evolve in the underwater environment.

[0035] However, the underwater positioning solutions available on the market use technologies that allow the underwater vehicle to position itself autonomously, at best between approximately 5m and approximately 3m away from an object.

[0036] Indeed, their limited precision prevents them from maintaining a stable position at short distances, i.e. at a distance between 0.5m and 1.5m.

[0037] Thus, the intervention of a human operator is always necessary for the majority of detailed inspections, in particular to inspect defects, welds, damage, etc. Description of the invention

[0038] The present invention aims to overcome the aforementioned drawbacks, and to this end relates to a navigation system for an underwater vehicle, remarkable in that it comprises: - a data acquisition module, designed to collect data from a plurality of sensors arranged on an underwater vehicle and on a surface unit capable of communicating with said underwater vehicle, - a fusion module, communicating at least with said data acquisition module and designed to retrieve at least a portion of said data collected by said data acquisition module, said fusion module comprising a processing algorithm adapted to discriminate said retrieved data according to its relevance and to determine, on the basis of said discriminated data, a position of said underwater vehicle in a predetermined reference frame, - a movement module, communicating at least with said fusion module, designed to calculate the difference between said position determined by said fusion module and a pre-programmed target position and to send a movement command said underwater vehicle when the difference between said position determined by said fusion module and said pre-programmed setpoint position exceeds a predetermined threshold value, - a monitoring module, communicating at least with said movement module, designed to ensure that the movement order sent by said movement module is consistent.

[0039] Thus, by providing a navigation system comprising a fusion module including a processing algorithm discriminating the data retrieved by the acquisition module according to their relevance and determining a position of said underwater vehicle in a predetermined reference frame, it is possible to take into account the different data from the different sensors and thus determine which sensor(s) is / are reliable and which sensor(s) has / have drifted according to the situations of use of the underwater vehicle.

[0040] Thanks to the aggregation of all the sensors of the navigation system and thanks to the fusion and discrimination of the data carried out by the fusion module, it is possible to overcome the lack of positioning accuracy which can occur in particular when the underwater vehicle is at a short distance from the subject to be inspected.

[0041] In this way, the position of the underwater vehicle is known precisely and at all times, which makes it possible to reprogram, automatically thanks to the movement module and the monitoring module, the trajectory of the underwater vehicle without human intervention, in real time and as the underwater vehicle navigates, thus making the navigation of the underwater vehicle autonomous and thus allowing an autonomous inspection of underwater structures.

[0042] According to optional features of the navigation system according to the invention: - said processing algorithm adapted to discriminate the data retrieved according to their relevance includes a statistical calculation and risk determination algorithm; - said fusion module includes, as input to said fusion module, a filtering system for said retrieved data; - said filtering system includes at least one Kalman filter and / or a machine learning and artificial intelligence solution and / or a simultaneous localization and mapping algorithm; - said fusion module includes a weighting module designed to assign to each of said sensors a weight representative of the measurement accuracy of said sensors according to the conditions of use of said underwater vehicle; - said weighting module is designed to take into account measurement uncertainties of said sensors at a given time.

[0043] The invention also relates to a navigation system comprising an underwater vehicle and a surface unit capable of communicating with said underwater vehicle, comprising a plurality of sensors arranged on said underwater vehicle and on said surface unit, said underwater vehicle being piloted by a navigation system, said navigation system being notable in that said navigation system is according to the invention.

[0044] According to optional features of the navigation system according to the invention: - said underwater vehicle includes at least the following sensors: an acoustic sonar, adapted to detect an object located near said underwater vehicle, a stereoscopic optical system, designed to capture at least two synchronized images, a Doppler velocimeter, designed to measure the ground displacement of said underwater vehicle, an acoustic positioning beacon, adapted to position said underwater vehicle relative to said surface unit, an inertial measurement unit, designed to measure the linear and angular accelerations of said underwater vehicle, at least one vision camera, adapted to acquire images of the environment around said underwater vehicle, a pressure sensor, designed to measure the pressure of a column of water in which said underwater vehicle is moving; - said monitoring module includes an extrapolation module integrating a suitable computer solution to generate, from a single image acquired by said at least one vision camera, a depth map; - said monitoring module includes a proximity module communicating with said extrapolation module and with said acoustic sonar, said proximity module being adapted to determine whether the data from the acoustic sonar and / or said at least one vision camera are representative of a risk of collision of the underwater vehicle; - the data from said Doppler velocimeter, said acoustic positioning beacon and said inertial measurement unit are filtered by said at least one Kalman filter of said filtering system and the data from said stereoscopic optical system and said acoustic sonar are filtered by said simultaneous localization and mapping algorithm; - said sensors are synchronized with each other and in that said data collected by said data acquisition module are synchronized; - the synchronization of the readings from each of the said sensors is achieved via an artificial metronome.

[0045] The invention also relates to a method for navigating an underwater vehicle implemented by a navigation system according to the invention, remarkable in that it comprises the following steps aimed at: - to collect data from a plurality of sensors arranged on an underwater vehicle and on a surface unit capable of communicating with said underwater vehicle, - to discriminate said retrieved data according to its relevance, - to determine, on the basis of said discriminated data, a position of said underwater vehicle in a predetermined reference frame, - calculate the difference between said determined position and a pre-programmed setpoint position and determine a movement command to be sent to said underwater vehicle when the difference between said determined position and said pre-programmed setpoint position exceeds a predetermined threshold value, - ensure that the said movement order to be sent is consistent, - send the said movement order to the said underwater vehicle if the movement order to be sent is consistent. Brief description of the drawings

[0046] Other features, purposes and advantages of the invention will become apparent from the following detailed description, for the understanding of which reference should be made to the accompanying drawings in which:

[0047] [Fig-1] shows a navigation assembly according to the invention.

[0048] [Fig.2] shows the overall architecture of the navigation system according to the invention.

[0049] [Fig.3] illustrates the navigation system according to an example of an embodiment of the invention.

[0050] [Fig.4] represents measurement uncertainty curves as a function of time for some of the sensors of the navigation assembly according to the invention.

[0051] [Fig.5] schematically shows the steps of the navigation process according to the invention. Description of the implementation methods

[0052] In the following description, elements having an identical structure or analogous functions are designated by the same reference.

[0053] Reference is made to [Fig.1] showing a navigation assembly 1 according to the invention.

[0054] The navigation system 1 comprises an underwater vehicle 3 and a surface unit 5. In one embodiment, the surface unit 5 may be a vessel. Alternatively, the surface unit 5 may be a buoy, a float, an offshore platform, etc. More generally, the surface unit is adapted to be able to carry a GPS (Global Positioning System) satellite geopositioning beacon and a sensor acoustic positioning for example of the type "ultra-short base" or "USBL", English acronym for "Ultra Short Base Line".

[0055] The surface unit 5 is capable of communicating with a central control system of an on-board computer 9 present on board the underwater vehicle 3.

[0056] The underwater vehicle 3 may consist of an ROV, that is to say an underwater vehicle remotely controlled and piloted from the surface by a human operator via an umbilical cable connecting the surface unit 5 to the on-board computer 9 of the underwater vehicle 3.

[0057] Alternatively, the underwater vehicle 3 may consist of an AUV, that is to say an autonomous underwater vehicle.

[0058] The navigation assembly 1 comprises a plurality of sensors arranged on the underwater vehicle 3 and on the surface unit 5.

[0059] For this purpose, the navigation assembly 1 includes a GPS (Global Positioning System) satellite geo-positioning system, comprising a GPS antenna 11 arranged on the surface, on the surface unit 5.

[0060] The GPS antenna 11 allows the surface unit 5 to be located on a nautical chart.

[0061] The navigation assembly 1 further comprises a positioning system acoustics adapted to position the underwater vehicle 3 relative to the surface unit 5.

[0062] The acoustic positioning system can for example be of the "ultra-short base" or "USBL" type, an English acronym for "Ultra Short Base Line".

[0063] In one embodiment, the acoustic positioning system includes an acoustic positioning beacon 201a, for example a USBL sensor, mounted on the underwater vehicle 3, capable of communicating with an acoustic positioning beacon 201b, for example a USBL sensor, arranged on the surface unit 5.

[0064] The GPS antenna 11 retrieves the satellite position and transmits it to the acoustic positioning beacon 201b arranged on the surface unit 5. The position of the acoustic positioning beacon 201b is thus given by the received GPS signal.

[0065] The acoustic positioning beacon 201b arranged on the surface unit 5 emits an acoustic signal and retrieves at defined time intervals the position of the acoustic positioning beacon 201a on board the underwater vehicle 3. The retrieved data is transmitted to the surface unit 5 to cross-reference it with the GPS position of the surface unit 5, in order to define a relative position of the underwater vehicle 3 with respect to the surface unit 5.

[0066] The position of the underwater vehicle 3 relative to the surface unit 5 is given with an accuracy of approximately 3m to approximately 5m.

[0067] The acoustic positioning beacon 201a can for example be arranged on the upper part of the underwater vehicle 3.

[0068] As will be seen later in the description, the presence of the DVL sensor is of particular interest in the event of a loss of visibility of the underwater vehicle. Indeed, the integration of the DVL sensor with the other sensors ensures continuity of navigation in the event of signal loss.

[0069] The navigation assembly 1 further includes an acoustic sensor of the Doppler velocimeter type 203 (DVL sensor).

[0070] The Doppler velocimeter 203 is mounted on the underwater vehicle 3 and allows, via the emission of sound waves, the measurement of the displacements of the underwater vehicle 3 on the ground and the measurement of the altitude of the underwater vehicle 3 relative to the bottom of the water mass in which it is submerged, for a distance ranging from 5cm to 50m of water.

[0071] The Doppler velocimeter 203 allows for relative positioning with respect to the first detected ground signal. The accuracy of the Doppler velocimeter 203 is between approximately 3m and approximately 5m.

[0072] In one embodiment, the Doppler velocimeter 203 can for example be located under the underwater vehicle 3. It is precisely oriented vertically relative to the underwater vehicle 3.

[0073] The navigation assembly 1 further comprises an inertial navigation unit 205, directly integrated into the underwater vehicle 3.

[0074] The inertial measurement unit 205 is designed to measure linear and angular accelerations on all axes of the underwater vehicle 3, and thus deduce a movement and a speed of the underwater vehicle 3.

[0075] As will be seen in the rest of the description, the inertial unit 205 can for example be used to check the positions provided by the other sensors in real time and to detect an aberrant event.

[0076] The set of sensors including the GPS antenna 11, the USBL acoustic positioning system comprising the acoustic positioning beacon 201a, the Doppler velocimeter type acoustic sensor 203 and the inertial unit 205 can be considered as so-called "direct" position sensors in the sense that they all three return directly positions of the underwater vehicle 3.

[0077] The navigation assembly 1 further includes a photogrammetric sensor defined by a stereoscopic optical system 207 mounted on the underwater vehicle 3, designed to capture at least two synchronized images.

[0078] The stereoscopic optical system 207 comprises at least two cameras designed, by means of synchronized photo pairs, to reconstruct to scale the detailed environment of the area to be inspected.

[0079] In one embodiment, the cameras of the stereoscopic optical system 207 are located at the front of the underwater vehicle 3.

[0080] The cameras of the stereoscopic optical system 207 are inclined relative to a longitudinal axis 13 of the underwater vehicle 3, which allows for optimal observation of the environment. In this way, the observable area by the cameras is maximized at a distance of between approximately 60 cm and approximately 3 m from the underwater vehicle 3.

[0081] Providing a stereoscopic optical system allows a reference dimension to be assigned to the viewed object, which makes it possible to determine the distance traveled between two frames and thus to determine the dimensions of a calculated object, rendering the navigation system 1 autonomous. By comparison, when the optical system is monocular, as is the case in some prior art devices, the optical system does not allow the dimensions of a target to be determined during image capture. Indeed, when the optical system is monocular, the determination of the size of a target is carried out using calibration objects that are positioned near the target, thus rendering navigation tools equipped with a monocular optical system non-autonomous.

[0082] The navigation system 1 also includes an acoustic sonar 209 mounted on the underwater vehicle 3 adapted to detect an object located near the underwater vehicle 3.

[0083] For example, the acoustic sonar 209 makes it possible to detect any object located in a defined angular sector upstream of the underwater vehicle 3, even without visibility.

[0084] The acoustic sonar 209 allows the sonar echo of the environment in front of it to be displayed in real time. It makes it possible to locate shapes and objects with good accuracy and to maintain a position in low visibility conditions, thus allowing it to take over when conditions are too difficult for the stereoscopic optical system 207, for example.

[0085] Thus, the acoustic sonar 209 makes it possible, by determining which objects are positioned in the direction of movement of the underwater vehicle 3, to avoid a collision.

[0086] The navigation assembly 1 also includes one or more vision cameras 501 mounted on the underwater vehicle 3, each adapted to acquire images of the environment around the underwater vehicle 3.

[0087] The 501 vision cameras used are, for example, low-cost cameras. In the embodiment illustrated in the figures, the underwater vehicle 3 has five 501 vision cameras positioned at the cardinal points of the underwater vehicle 3, i.e., at the rear, left, right, top, and bottom of the underwater vehicle 3, thus providing a 360° view of the obstacles during the inspection. In the embodiment illustrated in the figures, only the 501 vision cameras positioned at the rear, top, and bottom of the underwater vehicle 3 are visible. The view at the front of the underwater vehicle 3 can be managed by the system's cameras. stereoscopic optics 207. In an alternative embodiment, the vision cameras 501 can be replaced by acoustic sensors.

[0088] The presence of the vision cameras 501 allows, for example, the underwater vehicle 3 to be stopped in an emergency when an object is in its path, thus limiting the risk of collision of the underwater vehicle 3.

[0089] The 501 vision cameras thus act as security cameras for the underwater vehicle 3.

[0090] The navigation assembly 1 further includes a pressure sensor (not shown) designed to measure the pressure of a water column in which the underwater vehicle 3 is moving in order to determine the depth of movement of the underwater vehicle 3.

[0091] The sensors listed above, mounted on the underwater vehicle 3, are positioned rigidly and fixed relative to each other, which makes it possible to know precisely the offsets between each of them and to fuse them with respect to each other.

[0092] The sensors on board the underwater vehicle 3 all operate at the same time and communicate the data collected to the on-board computer 9, then the on-board computer 9 aggregates and synchronizes the data collected and provides a synchronized data stream to the surface unit 5.

[0093] As will be seen in the rest of the description, the synchronization of all the sensors makes it possible to obtain a coherent estimate of the position of the underwater vehicle.

[0094] In one embodiment of the invention, the synchronization of readings from each sensor is achieved by means of an artificial metronome. The artificial metronome operates as follows: A master board triggers the sensors simultaneously via a physical "trigger," which allows the position / displacement information from each sensor to be retrieved in a synchronized manner. The trigger can be external or originate from another sensor. The trigger is unique for all sensors. The artificial metronome can be independent or dependent on one of the sensors present on board the underwater vehicle 3. In one particular embodiment, the artificial metronome is generated by the stereoscopic optical system 207.Thus, when the artificial metronome is generated by the stereoscopic optical system 207, the stereoscopic optical system 207 synchronizes the readings of all the sensors of the navigation assembly 1.

[0095] The underwater vehicle 3 is piloted by a navigation system 100, the operation of which is described in [Fig.2] to which reference is now made.

[0096] The navigation system 100 includes hardware and software means capable of implementing the navigation process detailed in the rest of the description.

[0097] The navigation system 100 comprises a data acquisition module 200, a fusion module 300, a displacement module 400 and a monitoring module 500. The data acquisition module 200, fusion module 300, displacement module 400 and monitoring module 500 are distributed between the surface unit 5 and the underwater vehicle 3.

[0098] The data acquisition module 200 is designed to collect data from all sensors of the navigation assembly 1.

[0099] The data collected by the data acquisition module 200 are pre-synchronized, for example via the artificial metronome.

[0100] The navigation system 100 is designed to transmit the data acquired by the data acquisition module 200 to the fusion module 300, as well as some of the data acquired by the data acquisition module 200 to the monitoring module 500, as will be seen in the rest of the description.

[0101] The fusion module 300 retrieves at least part of the data collected by the acquisition module 200, for processing.

[0102] According to the invention, the fusion module 300 includes a processing algorithm adapted to discriminate the data retrieved by the acquisition module 200 according to their relevance. In this way, it is possible to take into account the different data from different sensors of the navigation system 1, and thus to determine which sensor(s) is / are reliable and which sensor(s) has / have drifted depending on the operating conditions of the underwater vehicle 3.

[0103] The fusion module 300 then determines, based on the discriminated data, a position of the underwater vehicle 3 within a predetermined reference frame. Thus, the fusion module 300 makes it possible to know precisely, in real time and at any moment, the position of the underwater vehicle 3.

[0104] In one embodiment, the fusion module can be integrated into the on-board computer 9 of the underwater vehicle 3.

[0105] The displacement module 400 communicates with the fusion module 300. The 400 displacement module is designed to calculate the difference between the position determined by the 300 fusion module and a pre-programmed target position, and to send a displacement command to the underwater vehicle 3 when the difference between the position determined by the 300 fusion module and the pre-programmed target position exceeds a predetermined threshold value. For example, a displacement command can be sent to the underwater vehicle 3 as soon as the position determined by the 300 fusion module differs from the pre-programmed target position.

[0106] Thus, the displacement module 400 uses the position of the underwater vehicle 3, calculated by the fusion module 300, to send the underwater vehicle 3 a real-time repositioning command, if necessary. In this way, the combined use of the fusion and displacement modules allows the underwater vehicle 3 to be guided in real time, making its navigation autonomous.

[0107] The monitoring module 500 communicates with the movement module 400 to ensure that the movement command sent by the movement module 400 is consistent. The monitoring module 500 also communicates with the fusion module 300 to verify the consistency of the returned data.

[0108] Also, in the embodiment illustrated in the figures, the monitoring module 500 also communicates with the acquisition module 200. In this way, the data from the different sensors recorded by the acquisition module 200 can be sent for verification to the monitoring module 500 to ensure that there is no measurement aberration.

[0109] Reference is made to [Fig.3] showing an example of an embodiment of the navigation system 100 of the invention.

[0110] The data acquisition module 200 collects data from the acoustic positioning beacon 201a (for example a USBL sensor), the acoustic Doppler velocimeter type sensor 203 (DVL sensor), the inertial measurement unit 205, the stereoscopic optical system 207 and the acoustic sonar 209.

[0111] The fusion module 300 includes an input filtering system 301 for the retrieved data. The presence of the filtering system 301 helps to limit errors in determining the position of the underwater vehicle 3.

[0112] In the embodiment illustrated in the figures, the filtering system 301 includes one or more Kalman filters 303 and one or more SLAM algorithms 305, a simultaneous localization and mapping algorithm developed to enable the navigation system 1 to generate maps of its environment while tracking its own position, by combining data from various sensors.

[0113] Data from the acoustic positioning beacon 201a, the Doppler velocimeter type acoustic sensor 203 and the inertial measurement unit 205 are post-processed by the Kalman filters 303.

[0114] Kalman 303 filters can be, for example, extended Kalman filters, frequently referred to by the acronym "EKF" for "Extended Kalman Filter" in English.

[0115] The use of Kalman 303 filters allows for real-time merging and filtering of data from the acoustic positioning beacon 201a, from the acoustic sensor type Doppler velocimeter 203 and inertial measurement unit 205, thus improving the accuracy of navigation and localization of the underwater vehicle 3.

[0116] Alternatively, the Kalman filters 303 and / or the SLAM algorithms 305 can be replaced by a machine learning and artificial intelligence solution (of the "deep learning" type). Deep learning techniques are used to analyze and interpret sensor data, enabling more accurate detection and classification of underwater objects and features.

[0117] The SLAM 305 algorithms used include a visual SLAM 305v, developed to perform post-processing of data from the stereoscopic optical system 207, and an acoustic SLAM 305a developed to perform post-processing of data from the acoustic sonar 209 and thus to position the underwater vehicle 3.

[0118] The successive data from the cameras of the stereoscopic optical system 207 are processed in such a way as to allow a relative positioning of the underwater vehicle 3 with respect to the first processed image.

[0119] In this way, it is possible to determine and track precisely the underwater vehicle 3 in relation to a visible environment using the visual SLAM algorithm 305v, which makes it possible to generate in real time the trajectory and position of the stereoscopic optical system 207 around the object to be inspected.

[0120] The visual SLAM algorithm 305v can use a Deep Learning type artificial intelligence approach to reduce the risk of loss of tracking of the object to be inspected which could occur in the event of poor visibility conditions or reduce the risk of loss of positioning of the underwater vehicle 1. Indeed, the use of neural networks trained on complex synthetic scenarios allows the definition of geometric primitives more robust than those defined manually by image processing algorithms for example of type ORB (English acronym for "Oriented fast Rotated Brief") or SIFT (English acronym for "Scale-Invariant Feature Transform").

[0121] In one embodiment of the invention, the visual SLAM algorithm 305v can be coupled with other types of artificial intelligence processing designed to recognize characteristic elements (for example anchor chains, pipelines or any other object of interest) and to track them over time to recognize areas already inspected or to ensure the complete reconstruction of a target object with respect to a previously known and constructed three-dimensional model.

[0122] Similarly, in one embodiment of the invention, the acoustic sonar 209, and the acoustic SLAM 305a developed to perform data post-processing data from the 209 acoustic sonar can be coupled with other types of artificial intelligence processing designed to recognize characteristic features (e.g. anchor chains, pipelines or any other object of interest).

[0123] The filtered data from the Kalman filters 303 and the SLAM algorithms 305 are positions of the underwater vehicle 3 in a predetermined frame of reference. The filtered data are then sent to a weighting module 311 integrated into the fusion module 300 and mounted at the output of the filtering system 301.

[0124] The weighting module 311 is designed to assign a weight to the various sensors of the navigation assembly 1 according to the operating conditions of the underwater vehicle 3. The weight assigned by the weighting module 311 is representative of the accuracy of the sensor measurement. For example, in very poor visibility, the SLAM algorithm 305 may give a position that is offset from the actual position. In this case, the weighting module 311 of the fusion module 300 will assign a lower weight than the SLAM algorithm 305 compared to the weight of other sensors in the navigation assembly 1.

[0125] Once the weighting module 311a has applied a weight to all the sensors in the navigation set, the fusion module 300 aggregates the sensor data and merges them by applying the corresponding weighting.

[0126] In one embodiment of the invention, the processing algorithm of the fusion module 300, adapted to discriminate the retrieved data according to their relevance, comprises a statistical calculation and risk determination algorithm whereby the fusion module 300 estimates a confidence index for any position of the underwater vehicle 3, the position being calculated at the output of the filtering system 301. The confidence index can, for example, be expressed as a percentage. Based on the estimated confidence index, the fusion module 300 deduces a percentage of uncertainty. On this basis, the fusion module 300 determines which sensor(s) is / are correct and which sensor(s) is / are incorrect.

[0127] The position retained at the end of the processing by the fusion module 300 is sent to the displacement module 400.

[0128] The displacement module 400 includes a displacement position determination module 401 designed to calculate the difference between the position determined by the fusion module 300 and a pre-programmed setpoint position.

[0129] The displacement module 400 further includes a displacement control module 411 designed to send a displacement command to the underwater vehicle 3 when the difference between the position determined by the displacement position determination module 401 and the pre-programmed target position exceeds a predetermined threshold value.

[0130] For example, a movement order can be sent to the underwater vehicle 3 as long as the difference between the position calculated by the movement position determination module 401 and the pre-programmed target position is not zero.

[0131] In order to avoid any risk of collision which could be induced by the movement command of the underwater vehicle 3 requested by the movement module 400, the monitoring module 500 of the navigation system 100 integrates the vision camera(s) 501 which acquire images of the environment around the underwater vehicle 3.

[0132] The monitoring module 500 further includes an extrapolation module 511 incorporating a computer solution, based for example on neural network technology, adapted to generate, from a single image acquired by the vision cameras 501, a depth map representing the underwater environment near the image capture point. This makes it possible to reconstruct the immediate environment of the underwater vehicle 3 at a frequency, for example, higher than 1 Hz, which allows the distance between the underwater vehicle 3 and the object seen by the vision camera 501 to be determined, thus further preventing the risk of collision.

[0133] The monitoring module 500 also includes a proximity module 521, communicating with the extrapolation module 511 and with the acoustic sonar 209 and being adapted to determine whether the data from the acoustic sonar 209 and / or the vision cameras 501 are representative of a risk of collision of the underwater vehicle 3. Thus, the acoustic sonar 209 and the vision cameras 501 define anti-collision safety sensors for the underwater vehicle 3.

[0134] The monitoring module 500 further includes a movement authorization module 531, communicating with the proximity module 521 and communicating with the movement control module 411 of the movement module 400.

[0135] When neither the data from the acoustic sonar 209 nor those from the vision cameras 501 are representative of a risk of collision of the underwater vehicle 3, the movement authorization module 531 of the monitoring module 500 sends a movement authorization signal to the movement control module 411 of the movement module 400.

[0136] Reference is made to [Fig.4] representing measurement uncertainty curves of sensors of navigation assembly 1 as a function of time.

[0137] The acoustic positioning beacon 201a (for example a USBL sensor), the acoustic Doppler velocimeter type sensor 203 (DVL sensor), the inertial measurement unit 205 and the SLAM algorithm 305 have measurement uncertainties expressed in meters, which vary according to time and / or according to measurement scenarios and / or according to their intrinsic characteristics.

[0138] Curve 15 represents the measurement uncertainty of the acoustic positioning beacon 201a (USBL sensor) combined with the GPS antenna 11, which acquires the satellite position and transmits it to the acoustic positioning beacon 201b arranged on the surface unit 5. The uncertainty curve 15 is average and constant over time. In other words, the USBL sensor is inaccurate at all times, but the error is constant over time. Furthermore, the uncertainty value is not very high. For example, the measurement uncertainty of the USBL sensor is on the order of at least approximately 1 m and is generally between approximately 1 m and approximately 10 m. However, this uncertainty increases with the distance between the submerged acoustic positioning beacon 201a and the acoustic positioning beacon 201b arranged on the surface unit 5.

[0139] Curve 17 represents the measurement uncertainty of the Doppler velocimeter 203 acoustic sensor (DVL sensor). Uncertainty curve 17 shows that the Doppler velocimeter 203 drifts constantly over time. Thus, each time a new position is acquired by the Doppler velocimeter 203, a measurement error accumulates, linearly increasing the measurement uncertainty. This error is intrinsic to the Doppler velocimeter 203.

[0140] Curve 19 represents the measurement uncertainty of the inertial measurement unit 205. Uncertainty curve 19 shows that the inertial measurement unit 205 is very accurate over a relatively short period of time, for example on the order of a few tens of seconds, but that the measurement uncertainty increases exponentially. The measurement uncertainty depends largely on the movements and scenarios.

[0141] Curve 21 represents the measurement uncertainty resulting from the SLAM 305 algorithm. The accuracy and uncertainty of the SLAM 305 algorithm as a function of time and the scenarios encountered are highly variable depending on the artifacts and measurement conditions. Indeed, the sensors associated with SLAM 305, namely the stereoscopic optical system 207 and the acoustic sonar 209, inevitably make random errors due to measurement noise. One of the objectives of SLAM 305 is to weight these errors over time to reduce the drift of the estimated trajectory relative to the actual trajectory of the underwater vehicle 3. To correct this drift, loops can be included in the path of the underwater vehicle 3, allowing the recognition of an area already visited in order to estimate the positioning difference between two readings and to determine the magnitude of the drift.

[0142] In one embodiment of the invention, the measurement uncertainties of the sensors of the navigation assembly 1 are taken into account by the weighting module 311 of the fusion module 300 of the navigation system 100. For example, a specific weighting, representative of the measurement uncertainty of the sensor in question at a given time, can be applied to each sensor in order to take into account the measurement drifts of each sensor.

[0143] Reference is made to [Fig.5] showing the steps of the navigation process implemented by the navigation system 100 according to the invention.

[0144] The navigation method of the underwater vehicle 3 includes a first step El implemented by the acquisition module 200, aimed at collecting data from all the sensors arranged on the underwater vehicle 3 and on the surface unit.

[0145] During a step E2, the fusion module 300 discriminates, according to their relevance, at least a part of the data collected during the step El and then determines, during a step E3, on the basis of the discriminated data, a position of the underwater vehicle 3 in a predetermined reference frame.

[0146] The displacement module 400 then calculates during a step E4 the difference between the position determined by the fusion module 300 and a pre-programmed setpoint position and determines a displacement order to be sent to the underwater vehicle 3 when the difference between the determined position and the pre-programmed setpoint position exceeds a predetermined threshold value.

[0147] During an E5 step, the monitoring module 500 ensures that the movement order to be sent to the underwater vehicle 3 is consistent, in particular according to predetermined assessment criteria.

[0148] If the movement order is consistent, the movement module 400 sends (step E6) a movement order to the underwater vehicle 3.

[0149] One mode of operation of the invention is described below by way of illustrative example representative of a navigation situation.

[0150] The fusion of data from acoustic positioning beacons 201a, 201b with the other sensors of the navigation assembly 1 makes it possible to reposition the trajectory of the underwater vehicle 3 with respect to the GPS coordinates of the surface unit 5, thus allowing georeferencing of the underwater vehicle 3 and enabling its guidance over long distances in open water.

[0151] When the underwater vehicle 3 is at a distance from the target to be inspected less than 50m and more than 3m, the acoustic positioning beacon 201a (USBL sensor) is master of navigation.

[0152] The acoustic sonar 209 ensures that the underwater vehicle 3 does not encounter any unexpected objects.

[0153] The pressure sensor ensures that the programmed depth is maintained, which allows for redundancy in the estimation of the position of the underwater vehicle 3 and for more reliable arbitration on the trajectory estimated by all the sensors.

[0154] The inertial measurement unit 205 and the Doppler velocimeter 203 (DVL sensor), which are sensors which drift little in a very short time period as shown in [Fig.4], ensure as far as possible that the data provided by the acoustic positioning beacon 201a are consistent.

[0155] When the underwater vehicle 3 is at a distance of less than 3m from the target to be inspected, the stereoscopic optical system 207 and the visual SLAM 305v provide a master position for navigation.

[0156] The GPS position is determined using the GPS antenna 11 and the acoustic positioning beacon 201a. Artificial intelligence makes it possible to recognize the characteristic element to be observed and to ensure that it remains in the center of the images.

[0157] The inertial measurement unit 205 and the Doppler velocimeter 203 ensure, as far as possible, that the trajectory determined by the visual SLAM 305v is consistent.

[0158] In the event of reduced visibility, the acoustic sonar 209 allows the target to be inspected to remain in close proximity. Artificial intelligence also allows the target to be kept in the center of the observed area until visibility is sufficient for the use of the visual SLAM 305v.

[0159] In the event that the visibility of the submarine 3 is suddenly lost, for example when a fish raises a cloud of particles, the visual SLAM 305v is also lost, which means that the location of the underwater vehicle 3 is lost.

[0160] Given that the acquisition module 200 knows the position of the underwater vehicle 3 at the time when visibility was lost, the fusion module 300 discriminates the data and determines the position of the underwater vehicle 3 using here both the inertial measurement unit 205 and the Doppler velocimeter 203 (DVL sensor).

[0161] A movement command aimed at moving away from the target to be inspected can be commanded and controlled by the inertial unit 205 and the Doppler velocimeter 203 (DVL sensor) to avoid a collision until visibility is regained.

[0162] When the lifted particles fall back down, the fusion module 300 again uses the data collected by the visual SLAM 305v.

[0163] Thus, the present invention makes it possible to deal with certain problems regularly observed in the context of underwater navigation, for example that of the reduction or sudden loss of visibility.

[0164] As will be understood, the present invention is not limited to the only embodiments of this navigation system, assembly and method described above solely by way of illustrative examples, but on the contrary it encompasses all variants involving the technical equivalents of the means described as well as their combinations if these fall within the scope of the invention.

Claims

Demands

1. Navigation system (100) of an underwater vehicle (3), characterized in that it comprises: - a data acquisition module (200), designed to collect data from a plurality of sensors arranged on an underwater vehicle (3) and on a surface unit (5) capable of communicating with said underwater vehicle (3), - a fusion module (300), communicating at least with said data acquisition module (200) and designed to recover at least a portion of said data collected by said data acquisition module (200), said fusion module (300) comprising a processing algorithm adapted to discriminate said recovered data according to its relevance and to determine, on the basis of said discriminated data, a position of said underwater vehicle (3) in a predetermined frame of reference, - a displacement module (400), communicating at least with said fusion module (300),designed to calculate the difference between said position determined by said fusion module (300) and a pre-programmed setpoint position and to send a movement command to said underwater vehicle (3) when the difference between said position determined by said fusion module (300) and said pre-programmed setpoint position exceeds a predetermined threshold value, - a monitoring module (500), communicating at least with said movement module (400), designed to ensure that the movement command sent by said movement module (400) is consistent.

2. Navigation system (100) according to claim 1, characterized in that said processing algorithm adapted to discriminate the retrieved data according to their relevance comprises a statistical calculation and risk determination algorithm.

3. Navigation system (100) according to any one of claims 1 or 2, characterized in that said fusion module (300) comprises, at the input of said fusion module (300), a filtering system (301) for said retrieved data.

4. Navigation system (100) according to claim 3, characterized in that said filtering system (301) comprises at least one Kalman filter (303) and / or a machine learning and artificial intelligence solution and / or a simultaneous localization and mapping algorithm (305).

5. Navigation system (100) according to any one of claims 1 to 4, characterized in that said fusion module (300) comprises a weighting module (311) designed to assign to each of said sensors a weight representative of the measurement accuracy of said sensors according to the conditions of use of said underwater vehicle (3).

6. Navigation system (100) according to claim 5, characterized in that said weighting module (311) is designed to take into account measurement uncertainties of said sensors at a determined time.

7. Navigation assembly (1) comprising an underwater vehicle (3) and a surface unit (5) capable of communicating with said underwater vehicle (3), comprising a plurality of sensors arranged on said underwater vehicle (3) and on said surface unit (5), said underwater vehicle (3) being piloted by a navigation system (100), said navigation assembly (1) being characterized in that said navigation system (100) is according to any one of claims 1 to 6.

8. Navigation assembly (1) according to claim 7, characterized in that said underwater vehicle (3) comprises at least the following sensors: - an acoustic sonar (209), adapted to detect an object located in the vicinity of said underwater vehicle (3), - a stereoscopic optical system (207), designed to acquire at least two synchronized images, - a Doppler velocimeter (203), designed to measure the ground displacement of said underwater vehicle (3), - an acoustic positioning beacon (201a), adapted to position said underwater vehicle (3) relative to said surface unit (5), - an inertial measurement unit (205), designed to measure the linear and angular accelerations of said underwater vehicle (3), - at least one vision camera (501), adapted to acquire images of the environment around said underwater vehicle (3), - a pressure sensor, designed to measure the pressure of a column of water in which said underwater vehicle is moving (3).

9. Navigation assembly (1) according to claim 8, characterized in that said monitoring module (500) comprises an extrapolation module (511) integrating a computer solution adapted to generate, from a single image acquired by said at least one vision camera (501), a depth map.

10. Navigation assembly (1) according to claim 9, characterized in that said monitoring module (500) comprises a proximity module (521) communicating with said extrapolation module (511) and with said acoustic sonar (209), said proximity module (521) being adapted to determine whether the data from the acoustic sonar (209) and / or from said at least one vision camera (501) are representative of a risk of collision of the underwater vehicle (3).

11. Navigation assembly (1) according to any one of claims 8 to 10, wherein said navigation system (100) is according to claim 4 or according to any one of claims 5 or 6 combined with claim 4, characterized in that: - the data of said Doppler velocimeter (203), of said acoustic positioning beacon (201a) and of said inertial measurement unit (205) are filtered by said at least one Kalman filter (303) of said filtering system (301), - the data of said stereoscopic optical system (207) and of said acoustic sonar (209) are filtered by said simultaneous localization and mapping algorithm (305).

12. Navigation assembly (1) according to any one of claims 8 to 11, characterized in that said sensors are synchronized with each other and in that said data collected by said data acquisition module (200) are synchronized.

13. Navigation assembly (1) according to claim 12, characterized in that the synchronization of the readings of each of said sensors is achieved by means of an artificial metronome.

14. A method for navigating an underwater vehicle (3) implemented by a navigation system (100) according to any one of claims 1 to 6, characterized in that it comprises the following steps aimed at: - collect (step El) data from a plurality of sensors arranged on an underwater vehicle (3) and on a surface unit (5) capable of communicating with said underwater vehicle (3), - discriminate (step E2) the retrieved data according to its relevance, - determine (step E3), on the basis of said discriminated data, a position of said underwater vehicle (3) in a predetermined frame of reference, - calculate (step E4) the difference between said determined position and a pre-programmed setpoint position and determine a movement command to be sent to said underwater vehicle (3) when the difference between said determined position and said pre-programmed setpoint position exceeds a predetermined threshold value, - ensure (step E5) that the said movement order to be sent is consistent, - send (step E6) said movement order to said underwater vehicle (3) if the movement order to be sent is consistent.