Method and apparatus for calibrating an environmental depth sensor

Abstract

The present invention relates to an apparatus and method for calibrating the yaw angle of an environmental depth sensor (11, 12, 13) mounted by a vehicle (1), the method comprising the steps of: storing data relating to planes identified by the environmental depth sensor (11, 12, 13), the data including at least information regarding the orientation of the planes; statistically analyzing the stored data in a database capable of identifying a most prevalent plane orientation, the orientation being parallel to a vertical plane passing through the longitudinal axis of the vehicle (1) or perpendicular to the longitudinal axis of the vehicle (1); and calculating a calibration angle formed between the environmental depth sensor (11, 12, 13) and the most prevalent plane orientation, the orientation being parallel to a vertical plane passing through the longitudinal axis of the vehicle (1) or perpendicular to the longitudinal axis of the vehicle (1).

Description

Technical Field

[0001] The technical field of the present invention is to calibrate an environmental depth sensor mounted on a vehicle.

Background Art

[0002] Vehicles generally include an environmental depth sensor that enables detection of obstacles located around the vehicle, and more generally all obstacles located around the vehicle. Depending on the vehicle, these sensors can be used for parking assistance, to generate a warning when there is another vehicle in the blind spot of the rearview mirror, to apply emergency brakes when there is an obstacle in front of the vehicle, and even to enable fully autonomous driving of the vehicle.

[0003] The environmental depth sensor is most often a radar. The environmental depth sensor can also be a lidar, in which depth measurement is obtained by measuring the return time of light reflected by the environment after emitting light. Therefore, it is known to use a flash lidar on a vehicle, and the flash lidar emits flashes of multi-directional light in the near-infrared and analyzes the return time of the reflected light in multiple directions. Such a flash lidar is described, for example, in the document of US Patent Application Publication No. 2019146067A1.

[0004] Other types of environmental depth sensors are also known. For example, a scanning lidar in which a laser beam scans an area of the environment is known. Radars, sonars, or stereo cameras that enable calculation of the depth of an object located in the environment of a vehicle are also known.

[0005] The present invention can be applied to any type of environmental depth sensor mounted on a vehicle, particularly but not exclusively to the types of sensors described above.

[0006] To facilitate calculation, the position of environmental depth sensors relative to the vehicle on which they are mounted must be precisely known in relation to the vehicle the position of objects detected in the environment by those sensors. The term “position of the sensor relative to the vehicle” in this application refers to both the position of the sensor on the vehicle on which it is mounted and the angular position of the sensor relative to the vehicle. However, often, manufacturing tolerances of the vehicle or sensor can cause slight variations in the position of the sensor relative to the vehicle. In other cases, maintenance work or collisions involving the vehicle can cause the sensor to move.

[0007] Consequently, the precise location of the environmental depth sensor relative to the vehicle cannot be known in real time. Such uncertainty is particularly undesirable with respect to the sensor's yaw angle, i.e., the angle the sensor makes relative to the vehicle in the horizontal plane. In particular, even a small offset between the sensor's actual yaw angle and the theoretical yaw angle used by the computational means that process the data from the sensor to infer the position of objects around the vehicle can cause a significant shift in the calculated position of an object relative to its actual position.

[0008] Therefore, it is necessary to know precisely the position of the environmental depth sensor relative to the vehicle it will be mounted on, and especially the yaw angle that the sensor creates relative to that vehicle.

[0009] This position is generally measured during the calibration step, during which the environmental depth sensor detects the position of a test pattern with feature points, the position of which is precisely known relative to the vehicle. The data from this detection makes it possible to precisely determine the yaw angle of the sensor.

[0010] Such calibration is typically performed at the end of vehicle manufacturing (in other words, "end-of-line" calibration). In addition to adding an extra step to vehicle manufacturing, this calibration does not enable the detection of subsequent changes in the sensor's position relative to the vehicle due to, for example, shock, vibration, or maintenance work. Changes in the sensor's position may occur, and these changes can only be detected when the vehicle is later inspected.

[0011] The paper “Automatic data selection and boresight adjustment of LiDAR systems”, Rabine Keyetieu and Nicolas Seube, 07 / 05 / 2019, MDPI, describes a method for automatically calibrating the boresight angle of LiDAR and inertial measuring devices based on the construction of a boresight error observability criterion, particularly applicable to drones, and the automatic selection of the point most sensitive to boresight error to enable boresight angle adjustment based on statistical analysis of the adjustment results.

[0012] This method provides so-called "online" automatic calibration, or in other words, calibration that is not at the "end of the chain," but it still requires an inertial measurement device consisting of three accelerometers and three gyrometers, optionally hybridized with a GNSS system, and therefore the implementation of this calibration method is costly and complex.

[0013] The objective of this invention is to overcome these drawbacks of the prior art. [Overview of the Initiative] [Problems that the invention aims to solve]

[0014] Specifically, the objective of the present invention is to enable calibration of the position of the environmental depth sensor relative to the vehicle on which it is mounted, and to allow for the determination of the precise angular position of the sensor relative to the vehicle.

[0015] Another object of the present invention is to enable such calibration throughout the use of the vehicle in order to quickly detect any correction in the position of the sensor relative to the vehicle.

[0016] Another object of the present invention is to enable such calibration to be performed using sensors with limited accuracy and resolution, and using only relatively low computing power.

[0017] A particular objective of the present invention is to enable such calibration to be performed by a relatively moderate-capacity computer that can be mounted in a vehicle at a reasonable cost. [Means for solving the problem]

[0018] These objectives, as well as other objectives more clearly shown below, are achieved in accordance with the present invention by a method for calibrating the yaw angle of an environmental depth sensor mounted on a vehicle, and this method is implemented simultaneously and continuously in accordance with the present invention. - A step of storing data relating to a plane identified by an environmental depth sensor in a database, wherein the data includes at least information relating to the orientation of the plane relative to the environmental depth sensor. - A step of statistically analyzing the data stored in the database, wherein the most dominant planar direction is identified among the data, - A step of calculating the calibration angle created between the most dominant plane direction and the ambient depth sensor. Includes.

[0019] Therefore, calibration is continuously performed during vehicle movement based on the plane within the environment in which the vehicle is moving. Such calibration is advantageous over conventional calibration, which had to be performed occasionally using a test pattern.

[0020] Preferably, this method includes a step of preliminaryly selecting data sent by an environmental depth sensor before the step of storing the data, in which case related data to be stored in a database is selected according to predetermined criteria.

[0021] Although this step precedes the step of storing data, it should be noted that, in the logical order of the steps of the method, it is implemented simultaneously with and in succession to the other steps of the method.

[0022] Preferably, this method also includes the step of associating at least one weighting factor with the data for each of the planes stored in the database before the step of statistically analyzing the data.

[0023] Note that this step is located before the step of statistically analyzing the data, but in the logical order of the method steps, it is implemented simultaneously and continuously with other steps of the method.

[0024] Advantageously, the step of associating at least one weighting factor includes - an environmental depth sensor, - a vehicle, - an analysis of data regarding other planes stored in the database, or - the most dominant plane direction identified in the past the step of determining at least one weighting factor according to the data sent by.

[0025] Preferably, during the step of statistically analyzing, the most dominant plane direction is identified as the direction that obtains the maximum score, and this score is determined according to the number of planes having this angular direction or a similar direction in which the data is stored and weighted by at least one weighting factor related to those data.

[0026] According to an advantageous embodiment, during the step of statistically analyzing, the most dominant plane direction is identified as the average direction of the planes to which the data is associated with the most favorable weighting factor.

[0027] The present invention relates to - means for storing data sent by an environmental depth sensor, the data including at least information regarding the direction of the environmental depth sensor with respect to the plane identified by the sensor, for storing means, and - means for performing a statistical analysis capable of identifying the most dominant plane direction among the data stored in the storage means, -Means for calculating the calibration angle created between the most dominant plane direction and the ambient depth sensor. This also relates to equipment for calibrating the yaw angle of environmental depth sensors, including [specific equipment / devices].

[0028] Preferably, the device includes means for performing preliminary selection of data transmitted by an environmental depth sensor, which can select relevant data stored in a storage means according to predetermined criteria.

[0029] Finally, the present invention relates to a processor-readable data medium on which a computer program containing instructions for performing the steps of the calibration method described above is stored.

[0030] Other features, advantages, and details of the present invention will become clearer when the remainder of the following description is read in relation to the drawings. [Brief explanation of the drawing]

[0031] [Figure 1] This is a schematic top view of a vehicle equipped with an environmental depth sensor and objects within that environment. [Figure 2] This is a schematic representation of the data measured by one of the vehicle's environmental depth sensors in Figure 1. [Figure 3] This is a flowchart showing the main steps of a calibration method according to one embodiment of the present invention. [Modes for carrying out the invention]

[0032] Environmental depth sensor Figure 1 is a schematic top view of vehicle 1 equipped with three environmental depth sensors. The left sensor 11, located near the left side mirror of vehicle 1, is oriented laterally to the left. The right sensor 12, located near the right side mirror of vehicle 1, is oriented laterally to the right. The rear sensor 13, located near the rear bumper of vehicle 1, is oriented rearward.

[0033] In the illustrated embodiment, these three environmental depth sensors are flash-lidar sensors that emit a flash of light in the near-infrared region and collect the light reflected by surrounding objects at a given field of view. Measuring the return time of the reflected light in each direction makes it possible to determine the distance of the object that reflected the light.

[0034] In other embodiments, sensors 11, 12, or 13 may be located at other positions on the vehicle, or they may be different types of sensors, such as scanning lidars, stereoscopic cameras, sonars, radars, or any other type of environmental depth sensor known to those skilled in the art.

[0035] In the illustrated embodiment, the three sensors 11, 12, and 13 each have angle detection fields 111, 121, and 131, respectively, which are 120° (angle α in Figure 1) in the horizontal plane and 30° in the vertical plane. Each of these angle detection fields is centered on the main sensor axes 110, 120, and 130, respectively.

[0036] A method for detecting a plane schematically represented by rectangle 31 in Figure 3 is implemented by an environmental depth sensor. In this application, the environmental depth sensor is considered to be a device that includes computing means suitable for collecting data relating to the separation of planes and for processing that data to generate information relating to the depth of objects in the environment. The expression “computing means” as used herein means a processor, calculator, automated device, computer or a set of computers, or any other equivalent means that enables the processing of data.

[0037] In the situation shown in Figure 1, vehicle 1 is near three objects: a parallelepiped object 21 whose one face 211 is parallel to the vehicle's longitudinal axis 100, a second parallelepiped object 22 whose face is not parallel to the vehicle's longitudinal axis 100, and a third object 23 which has no plane. These three objects are located within the angle detection field 121 of the environmental depth sensor 12.

[0038] Figure 2 schematically illustrates how the environmental depth sensor 12 detects these three objects. The sensor 12 measures the positions of multiple points on a surface that reflects the flash of light it emits. Therefore, Figure 2 shows in the horizontal plane the point 210 corresponding to the detection by the sensor 12 on the surface 211 of object 21, the point 220 corresponding to the detection by the sensor 12 on the surface 221 of object 22, and the point 230 corresponding to the detection by the sensor 12 on the surface of object 23. Of course, in reality, the sensor 12 measures the positions of these points not only in two dimensions but also in three dimensions, as schematically shown in Figure 2.

[0039] This step of measuring the positions of points 210, 220, and 230 on a surface in the environment relative to sensor 12 is performed by sensor 12 at a frequency of multiple measurements per second throughout the use of sensor 12. This step, represented by rectangle 311 in Figure 3, forms part of method 31 for detecting a plane, and the step is preferably implemented simultaneously and continuously.

[0040] Since the positions of these points are known, several mathematical processing steps are performed by the environmental depth sensor 12.

[0041] During the first processing step, usually called the clustering step, various points are grouped into groups or clusters corresponding to objects or surfaces. In this step, points corresponding to road or ground detection are usually removed, and these points are identified using information about the height of the points and the substantially horizontal orientation of the surface containing them. Then, points that do not belong to road or ground are clustered using mathematical 3D nearest neighbor search. In the embodiment shown in Figure 2, this step may allow, for example, point 210 to be clustered into one cluster, point 220 into another cluster, and point 230 into yet another cluster.

[0042] This clustering step, represented by rectangle 312 in Figure 3, also forms part of method 31 for detecting planes.

[0043] During the second processing step, a plane search is performed for each of the obtained clusters of points. This search can be performed using the RANSAC algorithm (RANSAC stands for "Random Sample Consensus"), which searches within the cluster of points for the largest point corresponding to a plane extracted from subsamples of those points.

[0044] The face containing the most points found in this way is refined using a numerical optimization method that takes those points into consideration. This optimization can be implemented, for example, using the SVD method (SVD stands for "Singular Value Decomposition"). It should be noted that the identified faces are not necessarily perpendicular faces.

[0045] During the application of the plane-searching step shown in the example in Figure 2, a pair of points 210 may enable the identification of a face corresponding to plane 211 of object 21. A pair of points 220 may enable the identification of another face corresponding to plane 221 of object 22. In contrast, a pair of points 230 does not enable the identification of a face.

[0046] The step of searching for this plane, represented by rectangle 313 in Figure 3, also forms part of method 31 for detecting the plane.

[0047] The steps of measuring the location of points, clustering them, and searching for planes are generally performed by an environmental depth sensor. In other embodiments, the environmental depth sensor may also perform these steps in conjunction with external computing means, or implement steps different from those described above to obtain identification of planes of objects in the environment.

[0048] In order for vehicle 1 to be able to accurately determine the position of objects in its environment by measuring the environmental depth sensors 11, 12, and 13, it is necessary to know the precise position of each of these sensors relative to vehicle 1. More precisely, it is especially important to know the yaw angle of each of these sensors, that is, the angle created in the horizontal plane between the vehicle's longitudinal axis 100 and the principal axes 110, 120, and 130 of each sensor.

[0049] To this end, vehicle 1 is advantageously equipped with an environmental depth sensor calibration device capable of implementing an environmental depth sensor calibration method, which is schematically represented by rectangle 32 in Figure 3, and its steps are preferably implemented simultaneously and continuously.

[0050] The inventors have revealed that vehicles very often move parallel to a plane such as a building wall, or perpendicular to a plane such as a road sign and especially a sign located on the road, or to the rear of a truck traveling in front of the vehicle. The calibration method and calibration device provided take advantage of this to continuously calibrate the angular position of the environmental depth sensor relative to the vehicle 1 while the vehicle is moving by referring to these planes that are parallel or perpendicular to the vehicle.

[0051] This method for calibrating an environmental depth sensor can be implemented independently of the method used to obtain identified plane information transmitted by the environmental depth sensor. This method is preferably implemented by a calibration device which may consist of software operating on the sensor's computing means, a dedicated computing means, or even on any other computing means such as a vehicle's onboard computer.

[0052] The calibration method 32 according to the present invention preferably includes at least the following steps, which are implemented simultaneously and continuously: - Step 322 to store data, - Step 324: Statistically analyze the data. - Step 325: Calculate the calibration angle.

[0053] During the data storage step, represented by rectangle 322 in Figure 3, the calibration device stores data about the planes identified by the environmental depth sensor, particularly information about the orientation of those planes relative to the sensor, in a database. This data storage step can advantageously continue for the entire operating time of the sensor being calibrated.

[0054] The moment a sufficient number of plane-related data points are stored in the database, the calibration device implements a statistical analysis step, represented by rectangle 324 in Figure 3. This step aims to identify the most dominant plane direction from the stored data.

[0055] During this step, the device can sort the planes on which the data is stored according to the angular direction of those planes relative to the sensor and search for the most dominant plane direction among the planes on which the data is stored.

[0056] This most dominant direction may be the one that corresponds to a significantly larger number of planes than other directions, for example, along with a determined tolerance.

[0057] According to one embodiment, this statistical analysis may consist of identifying a pair of planes that have approximately the same orientation with respect to the sensor and are likely to be parallel to the vehicle's longitudinal axis, and calculating the average of the orientations of the identified planes. This average can be considered to correspond to the most dominant plane orientation.

[0058] According to the inventors' considerations, this most dominant planar direction corresponds in most cases to a direction parallel to the vertical plane passing through the vehicle's longitudinal axis, and in certain cases to a direction perpendicular to the vehicle's longitudinal axis.

[0059] Therefore, step 324 involves statistically analyzing the aforementioned data stored in the aforementioned database, during which the most dominant planar direction among the aforementioned data is identified. During step 324 of the statistical analysis, this direction may be parallel to the vertical plane passing through the longitudinal axis of the vehicle 1, or perpendicular to the longitudinal axis of the vehicle 1.

[0060] Therefore, the most dominant identified planar direction, which is either parallel to the vertical plane passing through the longitudinal axis of vehicle 1 or perpendicular to the longitudinal axis of vehicle 1, forms the reference direction for the remaining calibration of the yaw angle of the environmental depth sensor mounted on vehicle 1.

[0061] One of the reference directions, which is either parallel to the vertical plane passing through the longitudinal axis of vehicle 1 or perpendicular to the longitudinal axis of vehicle 1, is selected according to the nominal position of the environmental depth sensor on the vehicle.

[0062] The calibration device can then implement the step of calculating the calibration angle represented by rectangle 325 in Figure 3. This step consists of calculating the calibration angle, which is determined to be the angular direction of the sensor relative to the vehicle, based on the most dominant detected planar direction that is parallel to the vertical plane passing through the longitudinal axis of the vehicle 1 or perpendicular to the longitudinal axis of the vehicle 1. In order to implement this step, the most dominant planar direction is estimated to be perpendicular to the longitudinal axis of the vehicle, in particular cases where the step of the method is specifically adapted to identify the most dominant planar direction that is parallel to the vertical plane passing through the longitudinal axis of the vehicle or perpendicular to the vehicle.

[0063] According to a preferred embodiment, the calibration method may include an additional step to take into account data other than information regarding the orientation of the identified plane. Thus, during the statistical analysis step, the most dominant plane orientation that is estimated to be parallel to or perpendicular to the vertical plane passing through the vehicle's longitudinal axis can be identified with greater reliability and efficiency.

[0064] Therefore, the calibration method 32 according to the preferred embodiment shown in Figure 3 preferably includes the following steps, which are implemented simultaneously and continuously: - Step 321 to select data as a preliminary step, - Step 322 to store data, - Step 323 to associate weighting coefficients, - Step 324: Statistically analyze the data. - Step 325: Calculate the calibration angle.

[0065] Accordingly, the calibration device may include means for performing a preliminary selection, which implements step 321 of preliminaryly selecting data to be stored in the database. During this step, the device selects data so as to store in the database only data relating to the plane that has the highest probability of having a desired direction parallel or perpendicular to the longitudinal axle of the vehicle.

[0066] Therefore, in the illustrated embodiment, this device introduces data corresponding to a plane into the database only when that plane is detected by the environmental depth sensor while the vehicle is traveling in a straight line at a speed exceeding a predetermined threshold speed. In particular, planes detected under these conditions have a high probability of being parallel to the vehicle's longitudinal axis, unlike planes detected when the vehicle is turning or moving at low speed.

[0067] The information necessary to make this selection may be provided, for example, by the vehicle's onboard computer, by a geolocation instrument, or by a gyroscope or inertial instrument. The step of obtaining this information is represented by rectangle 33 in Figure 3.

[0068] In other embodiments, other criteria may be used to preliminarily select the data to be introduced into the database.

[0069] Therefore, to limit the data introduced to the data relating to the plane with the highest probability of having the desired orientation, for example, data relating to planes detected within the precise angular zone of the detector can be selected.

[0070] For example, it might even be decided not to include data in the database for planes identified within geographical areas that have been previously identified as not having sufficient planes parallel to the vehicle's route.

[0071] In cases where sensor calibration must be based on recognizing a plane perpendicular to the vehicle's longitudinal axis, the introduction of plane data into the database can be conditional based on specific criteria, such as the approximate position of the plane relative to the vehicle, the approximate position of the plane at a height for identifying road signs in front of the vehicle and overhanging the road, and the orientation of the plane that is close to the expected vertical direction.

[0072] This preliminary selection step can be implemented within the calibration method independently of the other additional steps in the embodiment shown in Figure 3.

[0073] In the preferred embodiment shown in Figure 3, step 323 is implemented in which the calibration device associates weighting coefficients with stored data. During step 323, the calibration device associates data for each plane in the database with one or more weighting coefficients.

[0074] These weighting coefficients preferably represent the probability that the plane has a desired orientation.

[0075] Such weighting coefficients can, for example, be used as a quality score for the detected plane, allowing evaluation of whether the plane can form a highly reliable standard or whether it has a high or low probability of being parallel to the vehicle's longitudinal axle. This score is determined by the environmental depth sensor and can be sent to a calibration device simultaneously with the data for each plane.

[0076] Therefore, for each identified plane, the sensor: - Primarily horizontal planes, and to a lesser extent vertical planes, -The planarity of a plane obtained by calculating the positional deviation of each point on the plane from a theoretical plane. - The proportion of points in a cluster of points belonging to a plane. The quality score can be determined by taking these criteria into consideration.

[0077] Therefore, in the situations shown in Figures 1 and 2, the plane corresponding to surface 211 may be assigned a higher quality score than the plane corresponding to surface 221, due to the greater degree of this issue.

[0078] The step of determining the quality score of the plane represented by rectangle 314 in Figure 3 forms part of the method 31 for detecting the plane.

[0079] In other embodiments, the calibration device itself can calculate a score corresponding to the surface quality, particularly based on data generated from sensors.

[0080] Another weighting factor may relate to the recency of the surface identification by the sensor. Therefore, data corresponding to older surfaces may be weighted less than more recently identified surfaces. In particular, if an event alters the sensor's position relative to the vehicle, taking the sensor's actual position into account, newly identified surfaces are more likely to have a direction parallel to the vehicle than previously identified surfaces.

[0081] Such coefficients related to newness can possibly be set to zero for the oldest planes, preventing those data from being considered in the statistical analysis steps. In one modified form, the data for the oldest planes can also be removed from the database.

[0082] This weighting allows for quicker identification of changes in the sensor's orientation relative to the vehicle.

[0083] Another weighting factor may relate to the vehicle's condition when identifying the surface.

[0084] This coefficient can be high in situations where the vehicle is likely traveling parallel to a plane parallel to its own path. Therefore, this coefficient is preferably high when the vehicle is moving at high speed on a straight path and low when the vehicle is moving slowly or on a curved path.

[0085] In a particular embodiment, this coefficient can also take into account the geographical area in which the vehicle is located. Thus, this coefficient may be higher in areas where planes are known to be parallel to the driving lanes, and lower in areas where fewer planes are known to be parallel to the driving lanes.

[0086] The information that enables the generation of these weighting coefficients can come from the vehicle's onboard computer or from geolocation information or a navigation system. The step of obtaining this data is represented by rectangle 33 in Figure 3.

[0087] Another weighting factor may relate to the consistency of the data with respect to the plane relative to the previously recorded data. Therefore, a plane whose orientation relative to the sensor differs from the orientation of the previously recorded plane may be assigned a lower factor than a plane whose orientation is the same as many of the previously recorded planes relative to the sensor. Similarly, a plane with a large angular offset relative to the most prevailing plane orientation, such as those previously identified, may be assigned a lower factor.

[0088] Naturally, in order to evaluate the probability that a plane has a desired orientation parallel to or perpendicular to the vertical plane passing through the vehicle's longitudinal axle, other weighting coefficients can be used in addition to or instead of the weighting coefficients described above. These coefficients can be associated with data on planes identified by sensors at any point prior to the step of statistically analyzing these data. Thus, this association can be made before or after the step of storing the data. After the data has been subjected to statistical analysis, it is also possible to modify the weighting coefficients associated with that data before a new statistical analysis step.

[0089] If data regarding the orientation of planes is associated with weighting coefficients that represent the probability that those planes have a desired orientation, the step of statistically analyzing that data is carried out taking those coefficients into account.

[0090] Therefore, the step of statistical analysis may include searching for the angular direction that yields the highest score, and the score for each angular direction is determined by the number of planes having this angular direction or a similar direction, weighted by one or more weighting coefficients associated with each of those planes.

[0091] According to one preferred embodiment, this statistical analysis may be based on the calculation of the mean and variance of angular direction measurements of a plane weighted by various weighting coefficients. Thus, in this statistical analysis, greater weight is given to data of planes associated with higher coefficients, and therefore with a higher probability of having the desired direction.

[0092] According to one possible embodiment, the calibration device sorts the data during this statistical analysis step for the purpose of incorporating the measurements into the calculation of the most dominant planar direction according to various criteria, or conversely, for the purpose of excluding such measurements from this calculation.

[0093] Therefore, data corresponding to the plane direction can be incorporated or removed depending on the values ​​of the associated weighting coefficients.

[0094] These data can be added or removed depending on their deviation from the mean of the pre-combined data. The acceptable magnitude of the deviation can be set according to the variance of the pre-combined data. Therefore, if the pre-combined data is very consistent with each other, it becomes more difficult to incorporate new data that deviates significantly from its mean into the calculation.

[0095] When deciding to include or exclude new data from the average, the number of data points that have already been included or excluded can also be taken into consideration. Therefore, if a lot of data has been recently excluded, the criteria for including new measurements in the average calculation can be made less selective in order to facilitate the estimation of new values ​​when the sensor moves.

[0096] Those skilled in the art can implement other known statistical methods that enable the identification of the most dominant planar direction from among the data stored in the database.

[0097] As the inventors have observed, a vehicle traveling in a straight line is often seen traveling parallel to a plane that is parallel to its own path. Therefore, a statistical analysis of the plane orientation relative to the sensor must reveal a number of planes that share the same orientation relative to the sensor. Thus, the most dominant plane orientation resulting from this analysis can be estimated to be parallel to the vertical plane passing through the vehicle's longitudinal axis.

[0098] The basis for this estimation may be stronger if, when searching for the most dominant plane direction, the data selection criteria or weighting coefficients are chosen to support planes that have a higher probability of having a direction parallel to the vertical plane passing through the vehicle's longitudinal axis.

[0099] In contrast, when searching for the most dominant plane direction, if the data selection criteria or weighting coefficients are chosen to support planes that have a higher probability of being perpendicular to the vehicle's longitudinal axis, then the most dominant plane direction resulting from the statistical analysis can be estimated to be perpendicular to the vehicle's longitudinal axis.

[0100] Once this most dominant planar direction is identified with sufficient probability, the calibration device can calculate the angle the sensor makes with respect to this direction, which is estimated to be horizontal or perpendicular to the vertical plane passing through the vehicle's longitudinal axle, and thus called the calibration angle the sensor makes with respect to the vertical plane passing through the vehicle's longitudinal axle or to a plane perpendicular to the vehicle's longitudinal axle.

[0101] Therefore, the orientation of the sensor relative to the vehicle can be determined without the need to place the test pattern close to the vehicle. This orientation measurement allows for the adaptation of calculation parameters as needed, enabling the identification of the position of objects in the environment relative to the vehicle.

[0102] Advantageously, the calibration device continues this calibration throughout the vehicle's lifespan. Therefore, as soon as the vehicle travels in a straight line, data corresponding to the new plane is added to its database. Thus, the position of the sensors relative to the vehicle can be constantly monitored, and changes in this position can be quickly taken into account.

[0103] Finally, the present invention relates to a processor-readable data medium on which a computer program containing instructions for performing the steps of the calibration method described above is stored.

[0104] The data medium may be a non-volatile data medium such as a hard disk, flash memory, or optical disc.

[0105] The data medium can be any entity or device capable of storing instructions. For example, the medium may include storage means such as ROM, RAM, PROM, EPROM, CD-ROM, and even magnetic recording means, such as a hard disk.

[0106] Furthermore, the data medium may be a transmittable medium such as an electrical signal or optical signal that can be routed via an electrical cable or optical cable, wirelessly, or by other means.

[0107] Alternatively, the data medium may be an integrated circuit into which the program is incorporated, and the circuit may be suitable for use in performing or in performing the method.

Claims

1. A method for calibrating the yaw angle of an environmental depth sensor mounted on a vehicle (1), comprising the following steps, which are simultaneously and continuously implemented: - A step (322) of storing data relating to a plurality of planes identified by environmental depth sensors (11, 12, 13) in a database, wherein the data includes at least information relating to the orientation of the plurality of planes with respect to the environmental depth sensors (11, 12, 13), - A step (324) of statistically analyzing the data stored in the database, wherein during the statistical analysis step (324), the most dominant planar direction is identified among the data, and the most dominant planar direction is parallel to the vertical plane passing through the longitudinal axis of the vehicle (1) or perpendicular to the longitudinal axis of the vehicle (1), - A step (325) of calculating the calibration angle formed between the most dominant planar direction, which is parallel to or perpendicular to the longitudinal axis of the vehicle (1) passing through the longitudinal axis of the vehicle (1), and the environmental depth sensors (11, 12, 13) and A method characterized by including

2. Before the step of storing the aforementioned data (322), - A step of preliminaryly selecting data sent by environmental depth sensors (11, 12, 13), wherein related data stored in the database during the preliminary selection step is selected according to a predetermined criterion. The method according to claim 1, characterized by including

3. Before the step of statistically analyzing the aforementioned data (324), - A step of associating at least one weighting coefficient with the data relating to each of the plurality of planes stored in the database. The method according to claim 1 or 2, characterized by including the following:

4. The step of associating the at least one weighting coefficient is: - The aforementioned environmental depth sensors (11, 12, 13), - The aforementioned vehicle (1), - Analysis of the data relating to multiple other planes stored in the database, or - The most dominant planar direction previously identified The method according to claim 3, characterized by comprising the step of determining at least one weighting coefficient in accordance with the data sent by.

5. The method according to claim 3 or 4, characterized in that during the step (324) of statistically analyzing the data, the most dominant plane direction is identified as the direction that yields the highest score, and the score is determined according to the number of planes having this angular direction or similar direction, the data of which is stored and weighted by at least one weighting coefficient associated with the data.

6. The method according to any one of claims 1 to 5, characterized in that during the step (324) of statistically analyzing the data, the most dominant plane direction is identified as the average direction of the plane to which the data is associated with the most favorable weighting coefficient.

7. - A means for storing data transmitted by environmental depth sensors (11, 12, 13), wherein the data includes at least information relating to the orientation of a plurality of planes relative to the environmental depth sensors (11, 12, 13) as identified by the environmental depth sensors (11, 12, 13), - Means for performing a statistical analysis capable of identifying the most dominant planar direction among the data stored in the means for storage, - Means for calculating the calibration angle created between the most dominant planar direction and the environmental depth sensors (11, 12, 13) and A device for calibrating the yaw angle of an environmental depth sensor, including [specific component / tool].

8. The apparatus according to claim 7, characterized in that it includes means for performing a preliminary selection of data transmitted by environmental depth sensors (11, 12, 13), which can select relevant data stored in the means for storage according to predetermined criteria.

9. A processor-readable data medium on which a computer program comprising instructions for performing a step of the method according to any one of claims 1 to 6 is stored.

Images