Method for detecting lane boundaries

Abstract

A method (P1) for detecting boundaries (L1, L2, L3) of lanes (VC1, VC2) of a motor vehicle (1) is disclosed, characterized in that it comprises: a step (E1) of detecting said boundaries by vehicle environment detection means (3), said boundaries being defined by a function, in particular a polynomial function; a step (E4) of determining a plurality of first vectors (MXi) characterizing the boundaries, based on map data and on data relating to the current position and orientation of the vehicle; a step (E6) of determining a plurality of second vectors (Fj) characterizing the boundaries, each second vector being determined by an orthogonal projection of a first vector onto said function; and a step (E8) of calculating the Mahalanobis distance between each first vector and each second vector.

Description

[Technical field]

[0001] The invention relates to the field of methods for detecting lane boundaries of a motor vehicle. The invention also relates to the field of methods for determining the position of a vehicle on a road. The invention finally relates to a motor vehicle equipped with means for implementing such a method. [Background technology]

[0002] Lane boundary detection plays a key role in the development of autonomous vehicles. Correct detection is required to accurately represent the vehicle's environment and make appropriate decisions regarding the control of the vehicle. In particular, incorrect detection of lane boundaries, commonly referred to as "false positives", can have harmful consequences, for example, incorrect vehicle heading or untimely braking.

[0003] Lane boundaries can take many forms, such as lines painted on the ground, road surface boundaries, barriers, sidewalks, etc. These boundaries are therefore particularly complex to detect. To detect these boundaries, environmental detection means such as cameras, radars, or lidars are traditionally used. Although these detection means are becoming increasingly sophisticated, the data they provide may contain uncertainties and errors. The complexity of detection means, which are generally based on neural networks, makes false positives particularly difficult to understand or analyze. Such deficiencies limit the deployment of autonomous vehicles.

[0004] To improve lane boundary detection, methods are known which consist of combining the data received by the environment detection means with map data. Patent document 1 discloses an example of such a method. However, the known methods are still not efficient enough. In particular, false positive detections cannot always be avoided. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] U.S. Patent Publication No. 20210004017A1 Summary of the Invention [Problem to be solved by the invention]

[0006] A first subject of the invention is a method for detecting lane boundaries of a motor vehicle, making it possible to avoid false positive detections.

[0007] A second subject of the invention is a detection method making it possible to improve the determination of the position of a vehicle on a given road. [Means for solving the problem]

[0008] The present invention relates to a method for detecting lane boundaries in a motor vehicle, the method comprising the steps of: - a step of detecting said boundaries, performed by detection means for detecting the environment of the vehicle, said detection means being integrated into the vehicle, and the boundaries detected by the detection means being defined by a function, in particular a polynomial function; - determining a plurality of first vectors characterizing said boundary, the first vectors being determined based on map data and on data relating to a current position and orientation of the vehicle; - determining data relating to the uncertainty of each of the first vectors; - determining a plurality of second vectors characterizing the boundary, each second vector being determined by an orthogonal projection of a first vector onto the function; - determining uncertainty data for each of the second vectors; - calculating a Mahalanobis distance between each first vector and each second vector obtained from an orthogonal projection of the first vector under consideration, each Mahalanobis distance being calculated on the basis of data relating to the uncertainty of each first vector and each second vector under consideration; - classifying the boundaries detected by the detection means as positive or negative based on the previously calculated Mahalanobis distance.

[0009] The first vector and the second vector may be formulated in the same reference frame associated with the vehicle.

[0010] Each first vector is - a first component equal to the distance from the origin of the first vector under consideration to the vehicle along the longitudinal axis of the vehicle; - a second component equal to the distance from the origin of the first vector under consideration to the vehicle along the lateral axis of the vehicle; a third component characterizing an orientation of a tangent to the boundary at the origin of the first vector, Each second vector is - a first component equal to the distance from the origin of the second vector under consideration to the vehicle along the longitudinal axis of the vehicle; - a second component equal to the distance from the origin of the second vector under consideration to the vehicle along the lateral axis of the vehicle; a third component characterizing the orientation of a tangent to said boundary at the origin of the second vector.

[0011] Each first vector may be determined such that its origin is within the range of the detection means.

[0012] Uncertainty data for each of the first vectors may be calculated based on uncertainty values ​​for the current position and heading of the vehicle and based on uncertainty values ​​for the map data.

[0013] Data relating to the uncertainty of each of the second vectors may be calculated by the detection means.

[0014] The detection method may include a step of calculating a maximum Mahalanobis distance among the Mahalanobis distances calculated between each first vector and each second vector obtained from an orthogonal projection of the first vector under consideration, and a step of classifying the second lane boundary as a positive detection or a false detection is performed based on the calculated maximum Mahalanobis distance.

[0015] The present invention also relates to a method for detecting multiple lane boundaries of a motor vehicle, the method comprising the steps of: - detecting a first set of lane boundaries, the detection being performed by detection means for detecting an environment of the vehicle, the detection means being integrated into the vehicle; - detecting a second set of lane boundaries based on the map data and based on data related to the current position and orientation of the vehicle; - calculating a matrix of Mahalanobis distances between each boundary of the first set and each boundary of the second set, each element of the matrix being equal to the maximum Mahalanobis distance calculated by implementing the method for detecting boundaries defined above, with a boundary of the boundaries of the first set and with a boundary of the boundaries of the second set; - classifying the boundaries of the first set as positive or negative by the matrix of Mahalanobis distances by implementing a nearest neighbor determination algorithm on the matrix of Mahalanobis distances.

[0016] A method for detecting multiple boundaries may include a step of comparing the type of each boundary of a first set with the type of each boundary of a second set, and in the step of calculating a matrix of Mahalanobis distances, only elements of the matrix that satisfy the type comparison may be calculated.

[0017] The invention also relates to a method for determining the position of a vehicle on a road, the method comprising the steps of: - determining at least two positioning hypotheses for the vehicle; Then, for each vehicle positioning hypothesis, - implementing the method for detecting multiple boundaries as defined above, - calculating an accuracy index based on the determined number of positive and negative positives; and then - selecting a positioning hypothesis by comparing the accuracy figures calculated for the various hypotheses.

[0018] The invention also relates to a computer program product comprising program code instructions recorded on a computer readable medium for carrying out the steps of the method defined above.

[0019] The invention also relates to a computer readable data storage medium having recorded thereon a computer program comprising program code instructions for implementing the method defined above.

[0020] The invention also relates to a motor vehicle comprising detection means for detecting the environment of the vehicle, geolocation means for determining the geolocation of the vehicle, determination means for determining the orientation of the vehicle, a memory in which map data are stored or means for accessing the memory, and a calculation unit equipped with a data storage medium as defined above. [Brief description of the drawings]

[0021] [Figure 1] 1 is a schematic diagram of a motor vehicle according to one embodiment of the present invention. [Diagram 2] FIG. 2 is a first schematic plan view of a vehicle on a road including two lanes, showing the lane boundaries detected by detection means installed in the vehicle; [Diagram 3] FIG. 1 is a block diagram of a method for detecting lane boundaries according to one embodiment of the present invention. [Figure 4] FIG. 2 is a second schematic plan view of the vehicle showing lane boundaries determined by map data. [Diagram 5] FIG. 11 is a third schematic plan view of a vehicle showing, by vector modelling, lane boundaries detected by detection means installed in the vehicle; [Figure 6] 1 is a block diagram of a method for detecting multiple lane boundaries according to one embodiment of the present invention. [Figure 7] FIG. 1 is a schematic plan view of a vehicle on a road including five lanes. [Figure 8] 1 is a block diagram of a method for determining a position of a vehicle on a road according to one embodiment of the present invention. [Figure 9] 4 is a graph showing the evolution over time of the accuracy index calculated for three positioning hypotheses of a vehicle. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0022] 1 shows a schematic representation of a motor vehicle 1 according to one embodiment of the invention. The vehicle 1 can be of any type. In particular, the vehicle 1 can be, for example, a private car, a utility vehicle, a truck or a bus. The vehicle 1 comprises a computation unit 2 to which are connected detection means 3 for detecting the environment of the vehicle, geolocation means 4, determination means 5 for determining the orientation of the vehicle and a memory 6 in which map data are stored.

[0023] The detection means 3 can be, for example, a camera, a radar or a lidar. The detection means 3 can also be formed by the cooperation of one or more cameras, radars or lidars. It is for example capable of detecting objects present in the environment of the vehicle, such as lines painted on the ground, barriers, sidewalks or even any other form of boundary defining the edge of a lane. Advantageously, the detection means 3 are intelligent detection means, i.e. capable of interpreting the signals it senses and transmitting data resulting from an initial processing of these signals to a computation unit.

[0024] The geolocation means 4 may provide information on the geographical position of the vehicle 1, such as GPS coordinates. The geolocation means 4 may, for example, comprise a GPS sensor. The determination means 5 for determining the vehicle's orientation may provide data on the orientation of the vehicle around an axis perpendicular to the plane in which the vehicle is stationary. For simplicity of explanation, the vehicle is considered to be stationary on a horizontal ground. The determination means 5 for determining the vehicle's orientation may therefore provide data on the orientation of the vehicle around a vertical axis. The data on the vehicle's orientation may be equal to or established on the basis of the heading or azimuth angle followed by the vehicle. This data may, for example, be provided by a GPS sensor, a gyroscope or a compass built into the vehicle.

[0025] The memory 6 is a data carrier for storing high-definition map data. These map data comprise position and orientation information for the boundaries of the lanes in a given area. The boundaries of the lanes can in particular be stored in the form of a number of vectors, the configuration of which is explained in detail below. The map data can in particular be presented in the form of the ADASISv3 standard. As a variant, the memory 6 can be external to the vehicle 1, in which case the vehicle is equipped with means for accessing this memory, for example 4G or 5G communication means. The memory 6 can also be integrated in the computing unit 2.

[0026] The calculation unit 2 comprises in particular a memory 7 and a microprocessor 8. The memory 7 is a data carrier on which a computer program is recorded comprising program code instructions for implementing a boundary detection method according to an embodiment of the invention. The microprocessor 8 is capable of executing said computer program.

[0027] The detection means 3, the localization means 4 and the determination means 5 for determining the orientation of the vehicle are sensors integrated in the vehicle. They are able to provide information with a given uncertainty, in other words with a given resolution. Similarly, the map data recorded in the memory 6 also have a given uncertainty. The uncertainty therefore characterizes the accuracy of the information transmitted. These uncertainties are quantifiable. They may be presented in the form of a covariance matrix specific to each item of information provided by the various sensors or specific to each map data. Advantageously, as will be explained below, these uncertainty values ​​are exploited in a detection method to achieve a better detection of said boundaries. In the following, it is assumed that the uncertainties associated with each item of information are unbiased and follow the Gaussian law.

[0028] Figure 2 shows a vehicle 1 on a road comprising two lanes VC1, VC2. The first lane VC1 is bounded on both sides by two boundaries L1, L2. In this case, the two boundaries L1, L2 are respectively a solid line and a dashed line drawn on the ground. As a variant, these two boundaries can also be of different types, for example a barrier, a footpath or a covered edge. The second lane VC2 is adjacent to the first lane VC1 and is bounded by boundaries L2, L3.

[0029] A first reference frame, called the global reference frame, is defined, fixed with respect to the lane and independent of the vehicle 1. This first reference frame is formed by the axes X1 and Y1. For example, the axis X1 may be oriented along the north-south axis and the axis Y1 along the east-west axis. A second reference frame, called the vehicle reference frame, is also defined, tied to the vehicle 1 and formed by the axes X2 and Y2. The axis X2 corresponds to the longitudinal axis of the vehicle (i.e. the axis along which the vehicle moves linearly). The axis Y2 corresponds to the transverse axis of the vehicle and is perpendicular to the longitudinal axis X2.

[0030] The detection means 3 has a certain range, which is represented diagrammatically by a dashed line ZP in front of the vehicle. The range of the detection means is limited in particular by a minimum longitudinal range Xmin and a maximum longitudinal range Xmax. In some cases, the minimum longitudinal range may be considered equal to zero. The maximum range Xmax may, for example, be of the order of several tens or hundreds of meters.

[0031] 3 is a block diagram illustrating various steps of a method P1 for detecting lane boundaries according to one embodiment of the present invention. As will be understood below, the method may be implemented to detect any number of lane boundaries. For ease of explanation, a description of the implementation of the detection method for detecting a single boundary, for example boundary L2 shown in FIG. 2, is given first.

[0032] This detection method aims to ascertain whether the detection of a boundary performed by the detection means 3 corresponds to a correct detection (i.e. the detected boundary actually exists) or to a false detection (i.e. the detected boundary results from an incorrect interpretation of the signal received by the detection means 3 and does not correspond to a real boundary). The different steps of one particular embodiment of the method are now described.

[0033] In a first step E1, the detection means 3 detect the portion of the boundary within the range. The detection means 3 provide the calculation unit 2 with digital data characterizing the detected boundary at a preset frequency. The boundary is characterized by a function of the y=f(x) type defined by the detection means 3. In particular, this function is a polynomial function, in particular a third order polynomial function. As a variant, this function may be a polynomial function of different degrees, or any other mathematical function.

[0034] For each boundary Mi detected by the detection means 3, these digital data may be transmitted in the following form: M i =[c 0 ,c 1 ,c 2 ,c 3 ,xmin ,x max , M Σ p ,M type ] Where: - c0, c1, c2, c3 are the coefficients of the cubic polynomial function P(x), Xmin and Xmax are the minimum and maximum longitudinal extents of the detection means 3, MΣP is the covariance matrix expressing the uncertainty regarding the detection of the boundary L2 carried out by the detection means 3, - Mtype is data describing the nature or type of the boundary detected (which may include, for example, a line painted on the ground, a barrier, a walkway, etc.).

[0035] Therefore, each boundary is P(x)=c 0 +c 1 x+c 2 x 2 +c 3 x 3 The vector y is defined in the vehicle reference frame by the curve y=P(x) where

[0036] A polynomial function P(x) characterizes the shape of the boundary under consideration. This function is finite for any value of x between Xmin and Xmax. Thus, at a given instant, the boundary detected by the detection means 3 is characterized by four coefficients of a polynomial function. Such a way of representing the boundary makes it possible to reduce the amount of data exchanged between the detection means 3 and the calculation unit 2, compared to a model in which all points belonging to the boundary are transmitted to the calculation unit 2.

[0037] The covariance matrix MΣP may be expressed in the following form:

[0038]

number

[0039] FIG. 2 represents a representation of each of the boundaries L1, L2, L3 calculated by the detection means 3 using three curves M1, M2, M3. For the line M2, a zone Z1 is used to roughly represent the uncertainty associated with the detection of the boundary L2. The zone Z1 indicates the zone in which the boundary L2 is considered to be located. If the boundary L2 is well recognized by the detection means 3 (e.g. good contrast of the line drawn on the ground and good visibility), the zone Z1 may be particularly limited around the line M2. Conversely, if the recognition of the boundary L2 by the detection means 3 is poor (e.g. poor contrast of the line drawn on the ground and / or poor visibility), the zone Z1 around the line M2 may be particularly wide. Of course, such zones can also be represented for the lines M1, M3.

[0040] In a second step E2, the geolocation means 4 and the determination means 5 for determining the vehicle's orientation determine, respectively, the current position and the orientation of the vehicle 1 in the global reference frame. The geolocation means 4 and the determination means 5 provide digital data, at a pre-set frequency, to the calculation unit 2. These digital data may be expressed in the following form:

[0041]

number

[0042] In a third step E3, the measurement uncertainty of the geolocation means 4 and of the determination means 5 for determining the orientation of the vehicle is determined. This uncertainty may be quantified by the geolocation means 4 and of the determination means 5 for determining the orientation of the vehicle and transmitted to the calculation unit 2. This uncertainty may for example depend on the quality of the GPS signals received by the geolocation means 4 or on any other factors that may affect the operation of the geolocation means 4 and / or of the determination means 5 for determining the orientation of the vehicle. As a variant, this uncertainty may also be calculated by the calculation unit 2 or set to a predetermined value. This uncertainty may be expressed in the form of a 3×3 dimensional covariance matrix OΣM, the form of which is as follows:

[0043]

number

[0044] In a fourth step E4, a plurality of first vectors MXi characteristic of the boundary are determined on the basis of the map data and on the basis of data relating to the current position and orientation of the vehicle. Advantageously, the set of first vectors MXi determined circumscribes a predetermined boundary line around the vehicle. This boundary line, also called e-horizon or electronic horizon, may correspond to the range of the detection means 3. It will thus be understood that the first vectors are obtained from a map database and are circumscribed in a zone defined by the current position and orientation of the vehicle.

[0045] In a fifth step E5, data relating to the uncertainty of each of the first vectors MXi are determined.

[0046] More precisely, the memory 6 provides the calculation unit 2 with digital data relating to the various boundaries present within a given boundary line around the vehicle. The memory 6 containing the map data can be considered as a sensor integrated in the vehicle and providing information with a given accuracy. Each boundary is defined by a set of first vectors MXi. The map data therefore provides discrete information for defining each of the boundaries. Each boundary MLI present within the boundary line around the vehicle can be defined in the following form:

[0047]

number

[0048] Each of the first vectors MXi can be expressed in the vehicle reference frame in the following form:

[0049]

number

[0050] Each covariance matrix Var(OXi) of the set of covariance matrices Var(OXi=1...Ni) is a 3x3 dimensional matrix and can be expressed in the following form:

[0051]

number

[0052] These matrices represent the uncertainty associated with the map data of each vector MXi. This uncertainty may be due to the means used for the development of the map data. Data characterizing the uncertainty of the map data are also stored in the memory 6.

[0053] Advantageously, the first vectors MXi are formulated in the vehicle reference frame and the covariance matrix Var(OXi) characterizing the uncertainties of the first vectors MXi is formulated in the global reference frame. Since the map data initially available in the memory 6 are formulated in the global reference frame, the fourth step E4 advantageously comprises a sub-step E41 of calculating first vectors in the vehicle reference frame on the basis of the first vectors expressed in the global reference frame. In this sub-step, each first vector MXi is calculated on the basis of data relating to the position and orientation of the vehicle determined in step E2. In particular, the following formula may be used:

[0054]

number

[0055] The rotation matrix MRo can be expressed as follows:

[0056]

number

[0057] Similarly, the fifth step E5 includes a substep E51 of calculating a set of covariance matrices Var(OXi) in the vehicle reference frame. This calculation can be carried out using the following formula:

[0058]

number

[0059] The Jacobian matrix δMXi / δOX6 may be defined by the following equation:

[0060]

number

[0061] Finally, at this stage of the detection method, on the one hand, a vector model of the boundary based on map data is available. The boundary is characterized by a set of first vectors MXi expressed in the vehicle reference frame. Data characterizing the uncertainty of this set of first vectors is also available. This uncertainty is also expressed in the vehicle reference frame. Figure 4 shows diagrammatically the boundaries ML1, ML2, ML3 obtained from the map data and defined by carrying out a fourth step E4 for each of the boundaries L1, L2, L3. The origin of each vector MXi is identified by a point belonging to the boundaries ML1, ML2, ML3. In particular, for each vector MXi characterizing the boundary L2, a zone Z2 is used to express the uncertainty regarding the location of the origin of the vector MXi. The modeling of the boundary using map data is a form of discrete modeling, i.e. the boundary is defined by a finite set of vectors. On the other hand, a continuous model of the boundary generated by the detection means 3 is available. In fact, the boundary is expressed in the form of a function of the y=f(x) type, so that the boundary is finite at any point within the range of the detection means 3. To compare these two models, the continuous model of the boundary from the detection means 3 is discretized.

[0062] In a sixth step E6, a number of second vectors Fj characteristic of the boundaries identified by the detection means 3 are determined. Each second vector Fj is determined by orthogonally projecting a vector MXi from the set of first vectors onto the function y=P(x) defined above. In other words, each second vector Fj is determined such that its origin belongs to the function y=P(x), a straight line passing through the origin of the vector Fj and the origin of the vector MXi is orthogonal to the function y=P(x), and the orientation of the vector Fj is equal to the orientation of the tangent of the function y=P(x) at the origin of the vector Fj. It should be noted that the function y=P(x) generally has a radius of curvature large enough that there is a single orthogonal projection of the vector MXi onto the function y=P(x). In the very rare hypothesis where there may be several possible orthogonal projections of the vector MXi onto the function y=P(x), one of these projections, for example the first one found, may be used arbitrarily, i.e. the search for orthogonal projections is stopped as soon as an orthogonal projection is found.

[0063] In this way, a model of the boundaries detected by the detection means 3 is discretized. Discretizing this model involves representing each boundary by a finite set of vectors Fj, rather than a function (finite at any point between Xmin and Xmax). Each vector Fj locally characterizes the boundary by the coordinates of a point of the boundary in the vehicle reference frame and a component that characterizes the orientation of the boundary under consideration at this point. Thus, for a particular boundary, the vector Fj can be expressed in the vehicle reference frame as follows:

[0064]

number

[0065] Figure 5 shows an example of the discretization of lines M1, M2 and M3. Each line M1, M2 or M3 is represented by a set of vectors Fj, whose origin is the point (x(j),P(x(j)) of coordinates in the vehicle reference frame and whose orientation is equal to arctan(P'(xj)).

[0066] In a seventh step E7, the uncertainty associated with each of the second vectors Fj is determined. In fact, the uncertainty of the detection means 3 for the detection of each of the boundaries may also be discretized. For each vector F(j), the measurement uncertainty may be calculated using the following formula:

[0067]

number

[0068] In Fig. 5, for each vector Fj characterizing the boundary L2, a zone Z11 is used to represent the uncertainty regarding the location of the origin of the vector Fj, and a zone Z12 is used to represent the uncertainty regarding the orientation of the vector Fj. Due to the uncertainty regarding the orientation of the vehicle, the uncertainty is larger for the vectors Fj that are farthest from the vehicle. Thus, the zones Z11, Z12 are larger the further away from the vehicle.

[0069] In an eighth step E8, the Mahalanobis distance is calculated between each first vector MXi and each second vector Fj resulting from the orthogonal projection of the first vector MXi under consideration. The calculation of the Mahalanobis distance is based not only on the components of the vectors MXi and Fj but also on data on the uncertainty of these two vectors. The Mahalanobis distance between vectors Fj and MXi can be calculated using the following formula:

[0070]

number

[0071] The Mahalanobis distance calculated in this manner is a value expressing the degree of similarity between the vector Fj and the vector MXi. This Mahalanobis distance can be calculated for each vector MXi whose origin is within the range of the detection means 3.

[0072] In a ninth step E9, the maximum Mahalanobis distance of the Mahalanobis distances calculated between each first vector of an identical boundary and each second vector associated with the first vector under consideration is calculated. This maximum value is an indication of the similarity between the boundary detected by the detection means 3 and the boundary identified in the map data. The lower the maximum value, the closer the boundary detected by the detection means 3 is considered to be to the boundary identified in the map data. The Mahalanobis distance between the boundary detected by the detection means 3 and the boundary identified in the map data may be defined as equal to this maximum value.

[0073] In a tenth step E10, the lane boundaries detected by the detection means 3 are classified as positive or negative based on the maximum distance calculated in step E9. For this purpose, it is possible, for example, to compare the previously calculated maximum value with a predefined threshold value determined in a calibration phase of the method.

[0074] Alternatively, the classification of lane boundaries detected by the detection means 3 as positive or negative can be based on other indicators, which themselves are based on a set of previously calculated Mahalanobis distances. For example, the average or minimum of the set of Mahalanobis distances calculated between vectors MXi and Fj can be compared with a threshold. However, a comparison of maximum values ​​is simple to implement and allows a reliable classification as positive or negative.

[0075] A description of one embodiment of a method P2 for detecting a plurality of lane boundaries is now given with reference to figure 6. Method P2 comprises a step E01 of detecting a first set of lane boundaries on the basis of data provided by the detection means of the vehicle, and a step E02 of detecting a second set of lane boundaries on the basis of map data and on the basis of data relating to the current position and orientation of the vehicle. Steps E01 and E02 are carried out prior to the implementation of method P1, as explained above.

[0076] According to a first variant embodiment of the invention, the method P1 defined by steps E1 to E10 may be performed for each possible pair formed by a boundary of the first set and by a boundary of the second set. If the number of boundaries detected by the detection means 3 is N1 and the number of boundaries identified in the map data is M1, the method P1 is repeated a number of times equal to N1×M1. Such a variant may require particularly large amounts of computational resources.

[0077] According to a second, more advantageous variant embodiment of the invention, the detection method may comprise a step E03 of comparing the type of boundary detected by the detection means 3 with the type of boundary identified in the map data. The method is then only implemented to calculate the Mahalanobis distance between boundaries of the same type, and possibly between a boundary of any type and a boundary of unknown type. For example, the detection means may detect three boundaries M1, M2, M3 of type "line drawn on the ground", of unknown type and of barrier type, respectively. At the same time, the map data may identify three boundaries ML1, ML2, ML3 of type "line drawn on the ground", of unknown type and of "barrier", respectively. In this case, the method will be used to calculate the Mahalanobis distance between pairs of boundaries: - M1, ML1 - M1, ML2 - M2, ML1 - M2, ML2 - M2, ML3 - M3, ML2 - M3, ML3

[0078] This method is not used to calculate the Mahalanobis distance between pairs of boundaries: - M1, ML3 - M3, ML1

[0079] Thus, by performing this comparison step, the number of calculation operations performed by the calculation unit 2 can be reduced.

[0080] Then, in an eleventh step E11, the method P1 defined by steps E1 to E10 is carried out for each pair of identified boundaries, and the maximum Mahalanobis distance for each of these pairs is determined. Then, in a twelfth step E12, a matrix of Mahalanobis distances between each boundary of the first set and each boundary of the second set is calculated, each element of the matrix being equal to the maximum Mahalanobis distance calculated by carrying out the method P1 with a boundary of the first set of boundaries and with a boundary of the second set of boundaries. Of course, assuming that the step E03 of comparing the types of boundaries has been carried out beforehand, only the elements of the matrix that satisfy the type comparison described above are calculated.

[0081] Then, in a thirteenth step E13, the Mahalanobis distance matrix classifies the boundaries of the second set as positive or false positives by implementing a nearest neighbor determination algorithm on the matrix of Mahalanobis distances. Such algorithms are also commonly referred to as "global nearest neighbor (GNN)" algorithms. This makes it possible to obtain the best possible association between the boundaries of the second set and the boundaries of the first set. If it is not possible to associate the boundary of the first set with the boundary of the second set, it is possible to deduce therefrom that the boundary of the first set corresponds to a false positive, i.e. is the result of a false positive carried out by the detection means 3.

[0082] FIG. 7 shows the result of implementing the method P2 on a road with five parallel lanes. The vehicle 1 is located on the central lane. The detection means 3 detects four boundaries M1, M2, M3, M4, represented by solid lines in FIG. 7. Three boundaries M1, M2, M3 are straight, while the fourth boundary M4 is curved. For each boundary, a circle or an ellipse is used to represent a zone Z11, which represents the uncertainty regarding the position of the origin of the vector Fj. Furthermore, the map data is used to identify eight boundaries ML1 to ML8, represented by dashed lines in FIG. 7. Then, the method P2 for detecting multiple lane boundaries is implemented. After the execution of the nearest neighbor determination algorithm, the boundaries M1, M2, M3 are associated with the boundaries ML3, ML4, ML5, respectively. None of the boundaries defined by the map data can be associated with the boundary M4. The latter is then correctly classified as a false positive.

[0083] The method for detecting multiple lane boundaries may also be used in the method P3 for determining the position of a vehicle on a road. One embodiment of such a method is shown in FIG.

[0084] In a first step E21, at least two positioning hypotheses for the vehicle are determined. For example, assuming that the vehicle 1 is on a road, such as a highway, which includes three parallel lanes, a first hypothesis H1 consists in assuming that the vehicle 1 is located in the rightmost lane. A second hypothesis H2 consists in assuming that the vehicle 1 is located in the middle lane. A third hypothesis H3 consists in assuming that the vehicle 1 is located in the leftmost lane.

[0085] Then, in a second step E22, the above-described method P2 for detecting a number of lane boundaries is performed for each positioning hypothesis H1, H2, H3 of the vehicle. In successive executions of the method P2, the current position of the vehicle determined by the geolocation means 4 is corrected to position the vehicle according to each of the positioning hypotheses. In this case, the determined current position of the vehicle is corrected to position the vehicle successively in the center of the right lane, then in the center of the middle lane, then in the center of the left lane. This correction can be performed by applying to the components Myi an offset corresponding to the coordinate along the axis Y2 of the origin of each vector MXi.

[0086] Then, in a third step E23, an accuracy index is calculated based on the number of positive and negative positives determined for each hypothesis H1, H2, H3. For example, the accuracy index may be calculated using the following formula:

[0087]

number

[0088] FIG. 9 shows the progress of the accuracy index for each of the hypotheses H1, H2, H3. It is observed that the accuracy index associated with the hypothesis H1 is generally higher than the accuracy index associated with the hypothesis H2, which itself is generally higher than the accuracy index associated with the hypothesis H3. It is therefore possible to distinguish between the various hypotheses. In a fourth step E24, a positioning hypothesis is therefore selected by comparing the accuracy indexes calculated for the various hypotheses. For this purpose, it is possible, for example, to retain the hypothesis that has the highest average value of the accuracy index over a given time window, i.e. the hypothesis H1 according to the example of FIG. 9. This makes it possible to determine the lane in which the vehicle is actually traveling. This information can be utilized in a vehicle autonomous control system or in a navigation system.

[0089] The present invention provides a method for detecting one or more boundaries of a lane, which makes it possible to identify false positive detections by a detection means for detecting the environment installed in a vehicle, which detection method can be advantageously implemented in a method for determining the position of a vehicle on a road in order to determine the position of the vehicle with greater reliability.

[0090] It should be noted that the enumerated terms, such as first, second, etc., are intended merely to distinguish between the various steps of the method. These terms do not characterize an order relationship between the various steps, which may be performed in any order, provided that the necessary data is available at the input. The block diagrams shown in Figures 3, 6, and 8 use arrows to indicate one possible order between the various steps. However, one skilled in the art would be able to envision a different order to achieve the same result. The described method may be repeated indefinitely with a predefined repetition frequency. [Explanation of symbols]

[0091] 1 vehicle 2 Computational Unit 3. Detection Methods 4 Geolocation means 5. Decision-making procedure 6. Memory 7 Memory and data recording media 8 Microprocessors

Claims

1. 1. A method for detecting lane boundaries in a motor vehicle, comprising: The method comprises: - detecting said boundary, which is performed by detecting the environment of said vehicle from on board said vehicle, wherein said detected boundary is defined by a function; determining a plurality of first vectors characterizing said boundary, said first vectors being determined on the basis of map data and on data relating to the current position and current heading of said vehicle; determining data relating to the uncertainty of each of said first vectors; determining a plurality of second vectors characterizing said boundary, each second vector being determined by an orthogonal projection of one of the respective first vectors onto said function; determining data relating to the uncertainty of each of said second vectors; calculating a Mahalanobis distance between each first vector and a respective second vector obtained from said orthogonal projection of each first vector, wherein each Mahalanobis distance is calculated based on data relating to the uncertainty of each first vector and each second vector; and classifying the detected boundaries as positive or negative based on the calculated Mahalanobis distance.

2. The detection method of claim 1 , wherein the first vector and the second vector are formulated in the same reference frame associated with the vehicle.

3. Each first vector is a first component equal to the distance from the origin of each of said first vectors to said vehicle along the longitudinal axis of said vehicle; a second component equal to the distance from the origin of each of the first vectors to the vehicle along the lateral axis of the vehicle; and a third component characterizing the orientation of a tangent to the boundary at the origin of each of the first vectors, Each second vector is a first component equal to the distance from the origin of each of said second vectors to said vehicle along the longitudinal axis of said vehicle; a second component equal to the distance of each of the second vectors from the origin to the vehicle along the lateral axis of the vehicle; and a third component characterizing the orientation of a tangent to the boundary at the origin of each of the second vectors.

4. The detection method according to claim 1 , wherein each of the first vectors is determined such that the origin of each of the first vectors is within a detection range from on-board the vehicle.

5. 2. The detection method of claim 1, wherein the data regarding the uncertainty of each of the first vectors is calculated based on uncertainty values regarding the current position and current heading of the vehicle and based on uncertainty values regarding the map data.

6. the method comprises calculating the maximum Mahalanobis distance of the Mahalanobis distances calculated between each first vector and a respective second vector obtained from the orthogonal projection of the respective first vector; and 2. The detection method according to claim 1, wherein classifying the second lane boundary as a positive or a negative detection is performed based on the calculated maximum Mahalanobis distance.

7. 1. A method for detecting multiple lane boundaries in a motor vehicle, comprising: The method comprises: - detecting a first set of lane boundaries, which is performed by detecting the environment of the vehicle from on-board the vehicle; - detecting a second set of lane boundaries based on map data and on data relating to the current position and current heading of said vehicle; - calculating a matrix of Mahalanobis distances between each boundary of said first set and each boundary of said second set; wherein each element of the matrix is equal to the maximum Mahalanobis distance calculated by implementing the detection method of claim 6 using the boundaries of the first set and the boundaries of the second set; and classifying the boundaries of said first set as positive or negative according to said matrix of Mahalanobis distances by implementing a nearest neighbor determination algorithm on said matrix of Mahalanobis distances.

8. - comparing each boundary type of said first set with each boundary type of said second set; and A detection method according to claim 7, wherein when calculating the matrix of Mahalanobis distances, only elements of the matrix that satisfy a type comparison are calculated.

9. 1. A method for determining the position of a vehicle on a roadway, comprising: The method comprises: determining at least two positioning hypotheses for said vehicle; Then, for each positioning hypothesis of the vehicle: - implementing the detection method according to claim 7, calculating an accuracy index based on the determined number of true positives and false positives; after that, - selecting a positioning hypothesis by comparing said accuracy indices calculated for different hypotheses.

10. A non-transitory computer readable medium comprising program code instructions for performing the method of claim 1 when the program is run on a computer.

11. 11. A computer program product comprising the non-transitory computer-readable medium of claim 10.

12. A motor vehicle, detection means for detecting the environment of the vehicle; geolocation means for determining the geolocation of the vehicle; determining means for determining a heading of the vehicle; a memory in which map data is stored or means for accessing the memory; and A computing unit comprising a computer program product according to claim 11. A motor vehicle comprising:

13. The detection method of claim 1, wherein the function is a polynomial function.

14. A detection method as described in claim 1, wherein the boundary detection is performed by a detection means built into the vehicle.

15. The detection method of claim 14, wherein each first vector is determined such that the origin of each of the first vectors is within the range of the detection means.

16. The detection method of claim 14, wherein the data relating to the uncertainty of each of the second vectors is calculated by the detection means.

17. The detection method of claim 7, wherein the detection of the boundary of the first set is performed by a detection means built into the vehicle.

18. A method for determining the position of a vehicle on a roadway, comprising: The method comprises: determining at least two positioning hypotheses for said vehicle; Then, for each positioning hypothesis of the vehicle: - implementing the detection method according to claim 8, calculating an accuracy index based on the determined number of true positives and false positives; after that, - selecting a positioning hypothesis by comparing said accuracy indices calculated for different hypotheses.

19. A non-transitory computer-readable medium comprising program code instructions for performing the method of claim 7 when the program is run on a computer.

20. A non-transitory computer-readable medium comprising program code instructions for performing the method of claim 9 when the program is run on a computer.

Images