Road object positioning method

Abstract

The invention relates to a method and a device for localizing road objects based on a plurality of trajectories transmitted by at least one vehicle traveling on a road network, the trajectories transmitted by the vehicle comprising a plurality of successive vehicle localizations acquired during a drive and at least one specific road object detected by a sensor of the vehicle during this drive, said object being associated with a vehicle localization at the time it was detected. The method is characterized in that it comprises a step of selecting trajectories comprising at least one calibration road object included in a list of calibration road objects, in which a specific road object is associated with a localization, this list being initialized with at least one road object whose actual localization is known; and in that, for each selected trajectory: at least one correction parameter is calculated, which represents the deviation between the localization associated with an object in this trajectory and the localization associated with said object in the calibration list; the calculated correction parameter is applied to the objects included in this trajectory to obtain a calibration trajectory; and the calibration object list is updated with the localizations of the road objects included in the corrected trajectory; the steps of calculating, applying the parameter and updating are repeated as long as at least one calculated correction parameter is higher than a certain threshold.

Description

TECHNICAL FIELD

[0001] The present invention relates to the field of high resolution mapping of road infrastructure, and in particular to a method and device enabling to accurately determine the positioning of road objects identified by a vehicle during its travel on a road network. BACKGROUND

[0002] To perform high definition mapping, the industry generally equips vehicles with various sensors adapted to collect data during the travel of the vehicle.

[0003] Due to the inaccuracy of the sensors used and / or the high speed of motion of the vehicle, it is often observed that the positioning of infrastructure elements determined by the vehicle is offset from their actual position. Therefore, before being able to exploit these data, they need to be processed to correct the observed offset. To this end, reference points whose absolute positioning is known are generally used: when detected by the vehicle, these reference points enable to estimate the error introduced by the sensors and to determine correction parameters enabling to correct the collected data.

[0004] However, this technique has limitations. In particular, only the data transmitted by the vehicles that pass by and detect a reference point can be corrected. However, since reference points are artificially created at a high cost, they only cover a small part of the road network. Moreover, the lifetime of reference points can be limited: for example, a signboard whose position is known can be moved or abolished.

[0005] Therefore, there is a need for a method that enables to correct the data transmitted by a vehicle during a drive even when no reference point is detected by the vehicle during this drive. SUMMARY

[0006] To this end, a method is proposed for positioning road objects based on a plurality of traces transmitted by at least one vehicle traveling on a road network, the traces transmitted by the vehicle comprising:

[0007] - a plurality of successive vehicle positions acquired during a drive,

[0008] - at least one specific road object detected by a sensor of the vehicle during the drive, said object being associated with a vehicle position at which it was detected,

[0009] The method comprises the steps of

[0010] - initializing a calibration road object list in which specific road objects are associated with positions, wherein the actual position of at least one road object is known,

[0011] - selecting a trace comprising at least one road object that appears in the calibration road object list, and

[0012] for each selected trajectory,

[0013] - calculating at least one correction parameter representative of a deviation between the position associated with an object in the trajectory and the position associated with said object in the list of positions and calibrations,

[0014] - applying the calculated correction parameter to the objects included in the trajectory to obtain a calibrated trajectory, and

[0015] - updating the list of calibration objects with the positions of the road objects included in the corrected trajectory,

[0016] the steps of selection, calculation, application and updating are repeated as long as at least one calculated correction parameter is higher than a certain threshold.

[0017] Thus, when a reference road object, i.e. a road object whose actual position is known, is detected by the vehicle during its travel on the road network, the difference between the actual positioning of this road object and the positioning of this object determined by the vehicle is used to correct the positions of the other road objects detected by the vehicle during its travel. The road objects whose positioning is then corrected in this way can be added to the list of calibration objects, the objects in this list being used as a reference to adjust the positions of the road objects detected by another vehicle which has detected one of these road objects.

[0018] In this way, the accuracy of the positioning of the road objects detected by a vehicle which has not passed by a reference road object, i.e. a road object whose actual position is known, can be improved.

[0019] “Road object” is understood to mean an element of road infrastructure, such as a sign, a traffic light, a roundabout, a bridge, a tunnel, or even an identifiable element of the surrounding environment, such as a building, a city utility or even a tree.

[0020] In the context of the application, a reference road object is a road object whose actual positioning is known. This positioning can be determined manually by an operator on the basis of particularly accurate positioning means.

[0021] “Calibration road object” is understood to mean a road object for which a reliable positioning can be obtained, either because the actual positioning of this object is known, or because the positioning of this object has been corrected by applying a correction parameter according to the steps of selection, calculation, application and updating. The calibration road objects make it possible to estimate the deviation between the position of an object detected by a vehicle and the reliable position of this object.

[0022] According to the application, the corrected trajectory is a trajectory in which the positioning of the detected objects has been corrected by applying a correction parameter, or in which the successive vehicle position samples have been modified by adding or subtracting a correction value. For example, the processing time of the images taken by the on-board camera for detecting the road objects, or the reaction time of the vehicle sensors, can introduce a shift between the vehicle position given by the GNSS receiver and the actual detected position of the object: at the time when the object is considered to be detected, the vehicle can have already passed the object, in particular if the vehicle is travelling at high speed. By estimating this shift and applying it to the data of the trajectory, a corrected trajectory is obtained in which the positioning better corresponds to the actual positioning of the objects.

[0023] According to one particular embodiment, the step of calculating the correction parameter comprises:

[0024] - calculating a first deviation between the positioning associated with a first object in the trajectory and the positioning associated with said first object in the list of calibrations, and

[0025] - calculating at least a second deviation between the positioning associated with a second object in the trajectory and the positioning associated with said second object in the list of calibrations.

[0026] Thus, when the vehicle passes at least two road objects present in the list of calibrations for which reliable positioning can be obtained, either because the actual positioning of the object is known, or because the positioning of the object has been corrected by applying a correction parameter according to the steps of selection, calculation, application and updating of the positioning method, it is proposed to calculate a correction parameter taking into account the positioning deviations calculated for these two objects. For example, the correction parameter can be calculated on the basis of the minimum deviation calculated or the average of the deviations.

[0027] According to one particular embodiment, the first and second deviations calculated are weighted with a confidence index associated respectively with the positioning of the first and second calibration objects, the confidence index being inversely proportional to the number of iterations of the steps of selection, calculation, application and updating performed before updating for the corresponding calibration object in the list.

[0028] It is thus proposed to associate a confidence index with a calibration road object when it is added to the list of calibrations. For example, a road object for which the actual position is known is associated with a high confidence index, while a calibration object for which the corrected position is obtained after one or more iterations of the steps of the method is associated with a lower confidence index. Indeed, a calibration object obtained by a corrected trajectory is less reliable than a calibration object obtained on the basis of the actual position of the object. The positioning reliability thus decreases with the number of iterations of the steps of the method.

[0029] According to one particular embodiment, the step of computing a correction parameter comprises computing an average of a first deviation between a position associated with a first object in the trajectory and a position associated with said first object in the list of positions and calibrations, and a second deviation between a position associated with a second object in the trajectory and a position associated with said second object in the list of positions.

[0030] In this way, the trajectory is corrected based on the observed average deviation when the vehicle passes a plurality of road objects for which a reliable positioning is available. This measure enables to obtain the positioning of the road objects with a higher accuracy by taking into account the variations of the deviation that can occur during the travel.

[0031] In one preferred embodiment, the step of computing a correction parameter comprises computing an average of a first deviation between a position associated with a first object in the trajectory and a position associated with said first object in the list of positions and calibrations, and a second deviation between a position associated with a second object in the trajectory and a position associated with said second object in the list of positions, the computed average being weighted with a confidence index associated with the first and second calibration objects respectively.

[0032] In this way, the deviations observed on the most reliable calibration road objects contribute more to the correction of the trajectory. The accuracy of the positioning of the road targets is improved.

[0033] According to one particular embodiment, the method enables, when the trajectory comprises at least a first and a second calibration object for which a first and a second correction parameter have been respectively computed, estimating a third correction parameter for a road object in the trajectory located in time between the first and the second calibration objects, this estimation being performed by regression based on the correction parameters computed for the first and second calibration objects and the respective detection instants of the first, second and third objects.

[0034] This measure enables to apply a non-uniform correction to the trajectory, but on the contrary to finely adapt the correction to the various road objects detected by the vehicle, in particular when the deviation between the position of the objects detected by the vehicle and their actual position varies over time, for example when the deviation is related to the travel speed of the vehicle.

[0035] In one particular embodiment, the regression is weighted with the confidence indices associated with the first and second calibration objects.

[0036] In this way, when a road object is associated with a high confidence index, the determined deviation between the positioning of this object determined by the vehicle and its actual position is more taken into account in the estimation of the intermediate deviation. The estimation of the positioning of the intermediate object is thus improved.

[0037] According to another aspect, the application relates to a positioning device for positioning road objects based on a plurality of trajectories transmitted by at least one vehicle travelling on a road network, the trajectories transmitted by the vehicle comprising:

[0038] - a plurality of successive vehicle positions acquired during a drive,

[0039] - at least one particular road object detected by a sensor of the vehicle during the drive, said object being associated with the vehicle position at which it was detected,

[0040] The device comprises a processor and a memory, the memory having recorded therein computer program instructions adapted to configure the processor to implement the following steps:

[0041] - initializing a calibration road object list in which the particular road objects are associated with positions, at least one of the road objects being known in its actual position,

[0042] - selecting a trajectory comprising at least one road object appearing in the calibration road object list, and

[0043] for each selected trajectory,

[0044] - calculating at least one correction parameter representative of the deviation between the position associated with the object in the trajectory and the position associated with said object in the calibration list,

[0045] - applying the calculated correction parameter to the object comprised in the trajectory to obtain a calibrated trajectory, and

[0046] - updating the calibration object list with the positions of the road objects comprised in the calibrated trajectory,

[0047] the steps of selection, calculation, application and updating being repeated as long as at least one of the calculated correction parameters is higher than a certain threshold.

[0048] The application also relates to a server comprising a positioning device as described above.

[0049] Finally, the application relates to a processor-readable information carrier having recorded thereon a computer program comprising instructions for executing the steps of the road object positioning method as described above.

[0050] The information carrier can be a non-transitory information carrier, such as a hard disk, a flash memory or an optical disk.

[0051] The information carrier can be any entity or device capable of storing instructions. For example, the carrier can include a memory device, such as a ROM, a RAM, a PROM, an EPROM, a CD ROM, or a magnetic recording means, such as a hard disk.

[0052] On the other hand, the information carrier can be a transmissible carrier that can be routed via cable or optical fiber, by radio or by other means, such as electrical signals or optical signals.

[0053] Alternatively, the information carrier may be an integrated circuit incorporating a program, which is adapted to execute or can be used to execute the methods discussed.

[0054] The various embodiments and features described above can be added independently or in combination to the steps of the road object localization method.

[0055] The devices, servers, and information carriers possess at least the advantages conferred by the methods associated with them. Attached Figure Description

[0056] Other features, details, and advantages of the invention will become apparent from the following detailed description and analysis of the accompanying drawings, wherein:

[0057] Figure 1a A road network comprising multiple road objects is shown, with the travel distances of vehicles marked on these road objects.

[0058] Figure 1b The road network after calibrating the first trajectory is shown.

[0059] Figure 1c The road network after calibrating the second trajectory is shown.

[0060] Figure 1d The road network after calibrating the third trajectory is shown.

[0061] Figure 2 This is a flowchart illustrating the main steps of a road object localization method according to a specific embodiment.

[0062] Figure 3 The trajectories of two detected calibration objects are shown.

[0063] Figure 4a It is a graphical representation of the specific linear change of the correction parameter over time.

[0064] Figure 4b It is a graphical representation of the specific nonlinear changes of the correction parameter over time, and

[0065] Figure 4c It is a graphical representation of another specific linear change in the correction parameter over time. Detailed Implementation

[0066] Figure 1 schematically illustrates a road network 100 comprising multiple road objects, such as signs 101 to 105 and traffic lights 106.

[0067] The itineraries 107, 108, and 109 followed by the vehicle during separate driving periods on road network 100 are also shown.

[0068] Trips 107, 108, and 109 are obtained from a collection vehicle equipped with a positioning device (e.g., a GNSS receiver). Therefore, the collection vehicle periodically queries the GNSS receiver to obtain a trajectory containing its continuous positioning, each location associated with a timestamp. The collection vehicle is also equipped with one or more sensors, such as cameras, lidar, and / or radar, enabling it to detect specific objects in its surrounding environment during travel. By analyzing images captured by the cameras, the collection vehicle can thus detect road objects, such as road infrastructure elements (bridges, tunnels, etc.) or signs. For this purpose, the collection vehicle includes a processing unit, such as an ECU (electronic control unit) implementing appropriate identification algorithms. These algorithms, in particular, enable the detection of the presence of road objects and the definition of their types. Thus, during the trip, the collection vehicle generates a trajectory including a series of geolocations it occupies, each location associated with the time it was acquired and, where appropriate, with a road object detected at that location. The trajectory is transmitted to server 118 via communication network 119 in order to generate, for example, a high-definition map of the road network on which the positions of road objects detected by each collecting vehicle are accurately indicated.

[0069] The locations of road objects 101 to 106 detected by the collection vehicle as it travels on network 100 are shown by black dots on the path. For example, in path 107, sign 102 is detected at location 110, and sign 103 is detected at location 111. For ease of reading, the figures indicate the times when these signs should theoretically be detected during the path by dashed lines perpendicular to the road and starting from the respective signs. As can be seen, in these examples, the time when an object is detected does not correspond to the time when the vehicle is closest to the object, causing signs 102 and 103 to be incorrectly located in path 107. Such positioning errors are, for example, due to the asynchrony of the vehicle's various sensors, each with its own clock and operating frequency (e.g., the GNSS receiver obtains positioning at a lower frequency than the camera captures images). Therefore, there may be an offset between the time when the camera captures an image and the time when GNSS positioning is obtained. This offset leads to an error that increases with the vehicle's travel speed.

[0070] Returning to Figure 1, we can see that trajectories 108 and 109 also include offsets: in trajectory 108, objects 102, 101, and 106 are positioned at positions 112, 113, and 114, respectively, thus having a delay relative to where they should theoretically be positioned; while in trajectory 109, objects 104, 106, and 105 are positioned at positions 115, 116, and 117, respectively, i.e., before their actual positions.

[0071] It should be noted that, in this description, the location of a road object determined by the vehicle is understood as the position of the vehicle when it is closest to that road object.

[0072] The server 118 in Figure 1 is, for example, a computer server connected to a communication network 119 and adapted to process data transmitted by the collection vehicle via the communication network 119. For this purpose, the server 118 includes a processing unit (e.g., one or more processors) and memory. The server is connected to a database 120, which stores the locations of road objects known as reference road objects whose actual locations are known. Records in database 120 include, for example, the signature of a particular reference road object and its geographic coordinates.

[0073] For example, a signature for a specific road object can be calculated based on images, or features extracted from signals such as those from sensors, by analyzing camera images, radar echoes, or lidar echoes. The signature can also include the geographic region where the object is located, such as a geographic hash. Generally, such a signature can be calculated by a moving vehicle based on images captured by a camera, and is calculated such that for the same road object, the signature calculated by each vehicle is the same.

[0074] Among the road objects in Figure 1, sign 103 is a reference road object whose actual location is known and stored in database 120.

[0075] Now we will combine Figure 2 To describe the steps of a positioning method according to a specific embodiment.

[0076] In the first step 200, server 118 receives multiple trajectories transmitted by at least one collecting vehicle. The trajectories are transmitted, for example, via a 2G, 3G, 4G, 5G cellular access network, Wi-Fi, or WiMAX to which the vehicle is connected, or via a removable storage medium in JSON or XML format, or in any other suitable format.

[0077] The tracks received by server 118 are stored in a database, such as database 120, or in a file system, awaiting processing by server 118. In the example of Figure 1, tracks 107, 108, and 109 are received by the server and stored in a database, such as database 120.

[0078] In step 201, a list of calibrated road objects is initialized. This list is stored, for example, in database 120, enabling server 118 to obtain the reliable location of a specific road object based on its signature. To do this, the server, for example, sends an SQL request to database 120, adapted to select a record corresponding to the signature of a specific road object, so that if the signature corresponds to the record, the location of the road object is obtained as a response.

[0079] Therefore, the calibration object list is initialized with reference road objects whose geographic locations are accurately known. Thus, taking Figure 1 as an example, the calibration database 120 is initialized with a unique road object whose actual location is known (i.e., sign 103).

[0080] In step 202, server 118 selects from the received trajectories all trajectories that include signatures of at least one road object included in the calibration database. To do this, the server examines the signature of the road object detected in each received trajectory and requests the database 120 for each signature to determine whether the vehicle transmitting the trajectory detected a road object present in the calibration object list during driving. As a variation, the server selects trajectories based on geographic criteria, such that only trajectories included in a specific geographic region are selected; the request then includes a geographic region identifier, such as a geographic hash, based on which the trajectories are selected.

[0081] Therefore, referring to Figure 1, server 118 makes a request to database 120 using the signatures of road objects detected in tracks 107, 108, and 109. In this example, for track 107, the server searches using the signatures of object 102 detected at location 110 and object 103 detected at location 111. Since the database is initialized with reference object 103, only the signature of object 103 detected at location 111 will receive a response. Thus, track 107 is selected. Since the other tracks 108 and 109 do not include detected road objects whose signatures appear in database 120, they are not selected at this stage.

[0082] In step 203, server 118 calculates at least one correction parameter representing the deviation between the location (i.e., position 111) associated with object 103 in the selected trajectory 107 and the location associated with object 103 in calibration database 120. To this end, server 118 determines a position 121 corresponding to the vehicle's position when it is closest to the actual position of sign 103 obtained from database 120, and calculates a correction parameter P representing the deviation between the vehicle's position 111 for locating object 103 and the position 121 where the vehicle should theoretically locate sign 103. r1 The value. This deviation is, for example, the distance or time interval between positions 121 and 111.

[0083] Then, in calibration step 204, the correction parameters thus calculated are applied to other objects detected in trajectory 107. For this purpose, server 118 assigns parameter P... r1 Applied to position 110 to correct positioning errors.

[0084] Since the position of the road object 102 is now more reliable after this correction, the object is added to the calibration database 120 in step 205. Figure 3 The road network of Figure 1 is shown, on which trajectory 107 is calibrated: the detection positions 110 and 111 of the corresponding road objects 102 and 103 are updated and marked by white dots, indicating that they are now calibrated objects.

[0085] Steps 202 through 205 are repeated when at least one new calibration object is added to the calibration list, or when the correction parameter calculated for one of the trajectories in step 203 is higher than a predetermined threshold. For example, steps 202 through 205 are repeated when at least one selected trajectory is calibrated with a correction parameter that means the object has been repositioned at least 30 meters or at least 1 second. In a particular embodiment, a maximum number of iterations is configured to guarantee the stopping of the algorithm. The maximum allowed number of iterations may be predetermined or proportional to the number of trajectories selected in step 202. Thus, the method includes step 206, where a stopping condition is tested, which may be that the correction parameter has stabilized between two iterations or that the maximum number of iterations has been reached.

[0086] In this example, since a new object has been added to the calibration list, server 118 repeats steps 202 to 205 to select a trajectory using the new calibration objects 102 and 106 added to database 120.

[0087] During this second iteration, trajectories 107 and 108 were selected in step 202 because they included signatures of road objects 102 and 103 that appeared in the calibration database. In step 203, the server calculated the correction parameter P for trajectory 108 based on the deviation between the location 112 of object 102 in the trajectory and the associated corrected location 110 of the same object 102 added to the calibration list in the previous iteration. r2 The correction parameter P is calculated in this way. r2 This enables the correction of the positioning of objects 101 and 106 detected at positions 113 and 114 in trajectory 108, so that a calibrated trajectory can be obtained in step 204. Then, the correction parameter P is applied... r2 Objects 101 and 106, whose positioning has been corrected, have been added to the list of calibrated road objects.

[0088] Since two new objects have been added to the calibration list, the server executes steps 202 to 205 of the method again. In this third iteration, trajectory 109 is selected because it includes a reference to object 106, which is now present in the calibration road object database 120. The correction parameter P is calculated based on the deviation between the position 11 of object 106 detected in trajectory 109 and the corrected position of that object that has been added to the calibration database 120. r3 This is then applied to trajectory 109 to correct objects 104 and 105. Objects 104 and 105 are then added to the calibration database in association with their corrected positions, which are determined by applying correction parameter P. R3 .

[0089] In this way, the location of road objects detected by the vehicle during its journey can be made more reliable, even if the vehicle does not pass any reference objects. Figure 3 Road network 100 is shown, on which trajectories 107, 108, and 109, corrected by implementing the three iterations described above, are illustrated. This improves the localization of detected road objects. These objects can be used to generate high-resolution road network maps.

[0090] Figure 3The trajectory 400 transmitted by the vehicle during its driving is shown. During this driving period, the vehicle detected road objects 401, 402, 403, and 404 at times 405, 406, 407, and 408, respectively. Road objects 401 and 404 are calibration road objects that have been added to the calibration object list, either because their actual location is known or because they have been corrected based on data transmitted by other vehicles according to the steps described above. The location data indicator sign 401 associated with road object 401 in the calibration list should be detected at time 409, and the location data indicator sign 404 associated with road object 404 in the calibration list should be detected at time 410. Therefore, these two objects 401 and 404 can be used to calibrate trajectory 400.

[0091] According to one particular embodiment, in order to calculate the correction parameters in step 203, server 118 calculates a first deviation Δa between the position of object 401 indicated in trajectory 400 and, for example, the position of the same object 401 indicated in a calibration list.

[0092] Server 118 also calculates a second deviation Δb between the position of object 404 indicated in trajectory 400 and the position of the same object 404 in the calibration list.

[0093] According to a particular embodiment, the correction parameter is an average deviation calculated by averaging the first deviation Δa and the second deviation Δb. This average deviation is then applied to trajectory 400 in step 204 to correct the corresponding positions 406 and 407 of road objects 402 and 403, thereby obtaining a calibrated trajectory.

[0094] According to a particular embodiment, road objects in the calibration list are associated with confidence indices. The highest confidence value is associated with a reference road object whose actual location is known and which is available in the calibration list. Road objects added to the calibration list after correction are associated with lower confidence indices. For example, referring to Figure 4, road object 13 is associated with the highest confidence index in the calibration list, such as an index of value 0. Calibration object 102, added to the calibration list after correction of trajectory 107, is associated with a lower confidence index, such as an index of value 1. Sign 101 is added to the calibration list during the second iteration, during which trajectory 108 is calibrated based on trajectory 107. Therefore, a confidence index of value 2 can be associated with this road object in the calibration database. Thus, the confidence associated with a calibrated road object is inversely proportional to the number of iterations of the method steps performed before adding the object to the calibration list.

[0095] In this example, the confidence index was chosen to vary inversely with the accuracy of the associated object's location; that is, the index value increases as accuracy decreases. However, it is entirely conceivable to reduce the index value proportionally to the number of iterations without altering the invention.

[0096] In one particular embodiment, the confidence indices associated with road object 401 and road object 404 in FIG4 are used to calculate correction parameters, which correspond to the average of a first deviation Δa and a second deviation Δb weighted by the respective confidence indices of objects 401 and 404.

[0097] According to another specific embodiment, when there are multiple calibration objects in the calibration list that can be used to calibrate a specific trajectory, only the calibration road object associated with the highest confidence index is considered to calculate the correction parameters.

[0098] According to a particular embodiment, when trajectory 400 includes at least a first calibration object 401 and a second calibration object 404 for which a first deviation Δa and a second deviation Δb have been calculated respectively, individual correction parameters are estimated for each road object 402 and 403 in the trajectory that is temporally located between the first calibration object 401 and the second calibration object 404. This estimation is performed by regression based on the deviations Δa and Δb calculated for the first calibration object 401 and the second calibration object 404 respectively, and the respective detection times of the first, second, and third objects. Therefore, for example, if calibration objects 401 and 404 are detected at times T0 and T3 respectively, the correction parameters estimated for object 402 detected at time T1 are given by an affine function.

[0099] Therefore, the correction parameter at detection time T1 of object 402 is estimated by the following formula:

[0100] f(T1) = a.T1 + b

[0101] And for object 403:

[0102] f(T2) = a.T2 + b

[0103] in:

[0104]

[0105] and:

[0106]

[0107] Figure 4aThis is a graphical representation of the deviation (vertical axis) between the vehicle-determined object position and its actual position as a function of time (horizontal axis). Therefore, given the known deviation between the actual position of the calibrated object 401 and its position detected by the vehicle at time T0, and the known deviation between the actual position of the object 404 and its position detected by the vehicle at detection time T3, specific correction parameters to be applied to objects 402 and 403 during the vehicle's detection period are determined using a linear function.

[0108] According to one particular embodiment, the deviations Δa and Δb are weighted with confidence indices associated with road objects 401 and 404, respectively, to estimate correction parameters to be applied to objects 402 and 403 in the trajectory.

[0109] f(Tn) = a.(d.Tn+e) b +c

[0110] in:

[0111] a: The difference between the deviation determined at time T0 and the deviation determined at time T3.

[0112] b: The ratio of the confidence index associated with object 401 to the confidence index associated with object 404.

[0113] c: The observed deviation between the actual position of object 401 and the position detected by the vehicle at time T0, and

[0114] d and e: Factors that make the value [T0-T3] fall into the interval [0-1].

[0115] Figure 4b It is a graphical representation of the deviation (vertical axis) between the vehicle's determined position and its actual position as a function of time (horizontal axis). Figure 4b In the example, the confidence index associated with calibration road object 404 is higher than that associated with calibration road object 401. Therefore, the bias does not evolve linearly over time. Figure 4c This is another graphical representation of the deviation (vertical axis) between the position of an object determined by the vehicle and its actual position as a function of time (horizontal axis), where the confidence index associated with a calibrated road object detected at time T0 is higher than the confidence index associated with a calibrated road object detected at time T3.

[0116] When the stopping condition is verified in step 206, in step 207, the geospatial database, which serves as a high-definition map, is updated using the object localization corrected by trajectory correction in one or more iterations from steps 202 to 205.

[0117] In a preferred embodiment, steps 200 to 207 of the method are implemented by computer program instructions recorded in memory, which are adapted to the processor of the configuration device such that the method is implemented when the processor executes the instructions. For example, the instructions are loaded into the memory of server 118 during initialization and executed by the processor of server 120.

Claims

1. A method for locating road objects based on multiple trajectories transmitted by at least one vehicle traveling on a road network, wherein the trajectories transmitted by the vehicle include: - Multiple consecutive vehicle locations collected during driving. - At least one specific road object detected by the vehicle's sensors during the driving period, the object being associated with the vehicle's position at the time of detection. The method includes the following steps - Initialize a list of calibrated road objects, in which specific road objects are associated with locations, and the actual location of at least one road object is known. - Select the trajectory that includes at least one road object that appears in the list of calibrated road objects, and For each selected trajectory, - Calculate at least one correction parameter, which represents the deviation between the positioning associated with the object in the trajectory and the positioning associated with the object in the calibration list. - The calculated correction parameters are applied to objects included in the trajectory to obtain a calibrated trajectory, and - Update the list of calibration objects using the locations of road objects included in the calibration trajectory. If at least one calculated correction parameter is higher than a certain threshold, repeat the steps of selection, calculation, application, and updating.

2. The method according to claim 1, wherein, The steps for calculating the correction parameters include: - Calculate the first deviation between the localization and calibration list associated with the first object in the trajectory and the localization associated with the first object, and - Calculate at least a second deviation between the positioning and calibration list associated with the second object in the trajectory and the positioning associated with the second object.

3. The method according to claim 2, wherein, The first and second deviations are calculated by weighting confidence indices associated with the positioning of the first and second calibration objects, respectively. The confidence indices are inversely proportional to the number of iterations of the selection, calculation, application, and updating steps performed on the corresponding calibration objects in the list before the update.

4. The method according to any one of claims 2 to 3, wherein, The steps for calculating the correction parameters include calculating the average of the first and second deviations.

5. The method according to claim 4, wherein, The calculated average is the average weighted by the confidence indices associated with the first and second calibration objects, respectively.

6. The method according to any one of claims 2 to 3, wherein, When the trajectory includes at least first and second calibration objects for which first and second correction parameters have been calculated respectively, a third correction parameter is estimated for road objects in the trajectory that are located between the first and second calibration objects in time. This estimation is performed by regression based on the correction parameters calculated for the first and second calibration objects and the detection times of the first, second and third objects respectively.

7. The method according to claim 6, wherein, The regression was weighted using confidence indices associated with the first and second calibration objects.

8. A device for locating a road object based on multiple trajectories transmitted by at least one vehicle traveling on a road network, the trajectories transmitted by the vehicle including: - Multiple consecutive vehicle locations collected during driving. - At least one specific road object detected by the vehicle's sensors during the driving period, the object being associated with the vehicle's position at the time of detection. The device is characterized in that it includes a processor and a memory, the memory containing computer program instructions adapted to configure the processor to perform the following steps: - Initialize a list of calibrated road objects, in which specific road objects are associated with locations, and the actual location of at least one road object is known. - Select the trajectory that includes at least one road object that appears in the list of calibrated road objects, and For each selected trajectory, - Calculate at least one correction parameter, which represents the deviation between the positioning associated with the object in the trajectory and the positioning associated with the object in the calibration list. - The calculated correction parameters are applied to objects included in the trajectory to obtain a calibrated trajectory, and - Update the list of calibration objects using the locations of road objects included in the calibration trajectory. If at least one calculated correction parameter is higher than a certain threshold, repeat the steps of selection, calculation, application, and updating.

9. A server comprising the apparatus of claim 8.

10. A processor-readable information carrier having a computer program recorded thereon, the computer program including instructions for performing the steps of the positioning method according to any one of claims 1 to 7.

Images