EP3300525A1

EP3300525A1 - Method for estimating geometric parameters representing the shape of a road, system for estimating such parameters and motor vehicle equipped with such a system

Info

Publication number: EP3300525A1
Application number: EP16714787.5A
Authority: EP
Inventors: Brouk WEDAJO; Amaury BREHERET; Julien Rebut
Original assignee: Valeo Schalter und Sensoren GmbH
Current assignee: Valeo Schalter und Sensoren GmbH
Priority date: 2015-03-18
Filing date: 2016-03-18
Publication date: 2018-04-04
Also published as: FR3033912B1; WO2016146823A1; FR3033912A1

Abstract

The invention relates to a method for estimating geometric parameters (R1M) representing the shape of a road, using at least one sensor and one analysis module equipping a motor vehicle (100) situated on said road, comprising in particular the following steps: - estimating, as a function of the position of detected points marking an edge (LM; LL) of at least one traffic lane (LA) of said road, parameters (R1M) of a first geometric model, representing the shape of at least a portion of said road, - determining a limit position with respect to the motor vehicle, up to which the first previously configured geometric model describes, with a given precision, the shape of said road, and - estimating the parameters of a second geometric model representing the shape of said road beyond said limit position. Also described are a system for estimating said parameters, and a motor vehicle (100) equipped with such a system. Figure 5

Description

Method for estimating geometric parameters representative of the shape of a road, system for estimating such parameters and motor vehicle equipped with such a system TECHNICAL FIELD TO WHICH THE INVENTION RELATES

The present invention generally relates to the field of processes and systems for assisting the driving of a motor vehicle.

It relates more particularly to a method of estimating geometric parameters representative of the shape of a road, by means of a sensor and an analysis module equipping a motor vehicle situated on said road, the sensor, for example a image sensor, delivering data representative of the environment facing the motor vehicle.

The invention also relates to a system for estimating such parameters, and a motor vehicle equipped with such a system and a driver assistance module.

The invention is particularly applicable in highway exit warning systems, in brake assist systems, or in power steering systems.

BACKGROUND

It is useful, in driving assistance systems, to determine geometrical parameters representing in a synthetic manner the shape of a road on which a motor vehicle circulates, in order, for example, to control a power steering device enabling the motor vehicle to remain substantially in the middle of the lane of traffic that it borrows, without intervention of its driver, or with a reduced intervention of the latter.

Such geometrical parameters can be estimated from a survey of the positions of points situated along an edge of this traffic lane, and on the basis of a geometrical road model.

Various geometric models are known for making such an estimate, in particular:

a circular road model, in which the portion of road facing the motor vehicle is described by an arc of a circle,

- a model representing this portion of road by a clothoid arc, or

- A model representing this portion of road, in an orthonormal frame located in the plane that it defines, by a polynomial piecewise function commonly called "spline".

But the precise description of a given section of road, by a single model, strongly depends on the context and can in certain situations greatly limit the performance of a driving assistance system.

When driving on a motorway lane for example, whose turns are particularly long and slightly curved, a section of road facing a motor vehicle can be accurately described by an arc. But in contrast, when driving on a winding country road, a section of road facing a motor vehicle can frequently have a rectilinear portion followed by a bend. An arc of a circle or clothoid then imprecisely describes the whole of such a section.

It is known from EP 2 1 72 826 to determine the distance up to which a portion of road extending towards a motor vehicle is likely to be accurately described by an arc of a circle, or by a clothoid arc. , before estimating the geometric parameters of this arc. This distance is determined on the basis of map data of this road and a geolocation of the motor vehicle.

OBJECT OF THE INVENTION

In this context, the invention proposes a method for estimating geometric parameters representative of the shape of a road, by means of a sensor and an analysis module equipping a motor vehicle situated on said road, including the following steps :

acquisition by the sensor of data representative of the environment facing said motor vehicle,

detection, by the analysis module, according to said data, of points marking an edge at least of a traffic lane of said road, and determination of the positions of said detected points,

estimating, as a function of said positions of the detected points, parameters of a first geometric model, representative of the shape of at least part of said road,

- determination (based on the positions of detected points) of a limit position, with respect to the motor vehicle, up to which the first previously defined geometric model describes with a given precision the shape of said road, and

estimating, according to said positions of the detected points, located beyond said limit position, parameters of a second geometric model representative of the shape of said road beyond said limit position.

As explained above, on a country road, for example, a section of road facing a motor vehicle frequently has two successive parts of different geometries, for example a rectilinear part followed by a bend, a bend followed by a rectilinear part, or two successive turns of different directions.

The method described above is particularly well suited to the estimation of geometric parameters representative of the shape of such a road section, since it relies on two geometric models respectively describing two successive road portions.

The determination (on the basis of the positions of the detected points) of said limit position makes it possible to determine the transition point between two such successive road portions of different geometries. This then makes it possible, in particular by determining the parameters of the second geometric model from the positions of the detected points situated beyond said limit position, to ensure an optimum description of the overall accuracy of such a road section.

A precise description of a section of road facing a motor vehicle can thus be obtained in a wide variety of situations (both on a motorway and on a winding road in rural areas or in the mountains) and up to a significant distance of the motor vehicle since this distance is not limited by the extent of a first part of road which would have a given simple geometry.

Such a description by two different geometrical models, each associated with a part of a road, makes it possible, while preserving a high description accuracy, to use simple geometrical models, for example circular road models, thanks to which the determination of such representative geometric parameters is advantageously fast. This determination can for example be made in real time, which is particularly interesting when the parameters thus determined are used by a module driving assistance.

The invention also provides, in a preferred embodiment, that the parameters of the first geometric model are estimated as a function of weighting coefficients respectively associated with the detected and decreasing points as a function of the distance from the point associated with the motor vehicle, so as to to make the parameters of the first geometric model depend all the more strongly on the position of one of the points detected that this point is close to the motor vehicle.

This arrangement is particularly interesting because it makes it possible to obtain a first set of geometric parameters describing the shape of a road in a proximal part thereof relative to the motor vehicle, and without prior knowledge of the extension of the proximal part of this route can be described by a given simple geometry.

Preferably, it is also provided that said determination of the limit position comprises a step of determining said precision of description of the shape of a given section of said road by the first geometric model previously parameterized, according to:

parameters of said first geometric model, and

positions of the detected points belonging to the given section, each of said positions contributing in an equivalent manner and without weighting to said determination.

More particularly, it is provided that said accuracy is determined by evaluating a magnitude representative of a distance between the first previously parameterized geometric model and the set of detected points belonging to the given section, said quantity being evaluated by:

calculating, in an equivalent manner for each detected point, an intermediate quantity representing a distance between said detected point and the first geometric model previously parameterized, and

summing said intermediate quantities.

Other non-limiting and advantageous features of the estimation method described above are as follows:

said limit position is determined so that the accuracy of description, by the first geometric model previously parameterized, of a road section extending from a proximal zone of the motor vehicle to beyond said limit position, is less than said given accuracy,

the first geometrical model and / or the second geometrical model correspond to an arc of circle, the parameters of said geometrical model including the radius of said arc of circle and the position of its center with respect to the motor vehicle, and

the first geometric model and / or the second geometric model corresponds to a clothoid arc.

It is also proposed that the step of estimating the parameters of the first geometric model or the second geometric model comprises an iterative process comprising steps of:

- random selection of only a part of the detected points,

determination of a temporary set of parameters of said geometrical model, as a function of the positions of said randomly selected points,

estimating a precision of description of the shape of said road by the geometric model parameterized by said temporary set of parameters, said estimate taking into account all the points detected,

selecting from the set of temporary parameter sets the set of parameters corresponding to the best description accuracy.

This estimation method advantageously reduces the influence of measurement noise on the parameters thus determined.

It can also be provided that the estimation method described above comprises a step of determining a second limit position, with respect to the motor vehicle, up to which said second previously parameterized geometric model describes with a given precision the shape of the said road.

The invention also provides a system for estimating geometric parameters representative of the shape of a road, for a motor vehicle, comprising a sensor adapted to acquire data representative of the environment facing the motor vehicle, and a module for analysis adapted to:

detecting, on the basis of said data, points marking an edge of at least one traffic lane of said road, and determining positions of said detected points, estimating, as a function of said positions of the detected points, the parameters of a first geometrical model, representative of the shape of at least a part of said road,

determining a limit position, with respect to the motor vehicle, up to which the first previously defined geometric model describes with a given precision the shape of said road, and

The invention also provides a motor vehicle comprising an estimation system as described above, and a driver assistance module, the estimation system being adapted to transmit to the driver assistance module the parameters of the first and second geometric model and said limit position previously estimated, the driver assistance module being adapted to control reaction means to trigger a controllable functionality according to said parameters and said limit position received from the system of estimate.

DETAILED DESCRIPTION OF AN EXEMPLARY EMBODIMENT The following description with reference to the accompanying drawings, given by way of non-limiting examples, will make it clear what the invention consists of and how it can be implemented.

In the accompanying drawings:

FIG. 1 schematically represents a motor vehicle equipped in particular with a system for estimating geometric parameters representative of the shape of a road according to the invention,

FIG. 2 represents the main steps of a method implemented by the estimation system of FIG. 1, in accordance with the teachings of the invention,

FIG. 3 diagrammatically represents the detail of a step of the method of FIG. 2,

FIG. 4 diagrammatically represents the detail of another step of the method of FIG. 2,

FIG. 5 schematically represents a portion of road on which the motor vehicle of FIG. 1 is located, and two models of said road portion, the parameters of which are estimated by means of the method of FIG.

FIG. 1 diagrammatically shows a motor vehicle 100 situated on a taxiway LA of a road R here comprising two traffic lanes LA and LA '.

The motor vehicle 100 is equipped in particular with a system 109 for estimating geometric parameters representative of the shape of a road, according to the invention.

The estimation system 109 comprises a sensor 101, such as an image sensor, a radar, or a lidar, enabling the acquisition of data representative of the environment facing the motor vehicle 1 00, in particular data representative of object positions of this environment with respect to the motor vehicle 100. In the embodiment described below, the sensor 101 is an image sensor, in this case a video camera.

The estimation system 109 also comprises an analysis module 108 comprising a processor 102 performing logical operations, for example a microprocessor, and a storage module 103 made for example in the form of a hard disk or a memory. vivid.

The data acquired by the sensor 101 is transmitted to the analysis module 108, where it is analyzed in real time to determine positions of points marking at least one edge of the taxiway LA on which the motor vehicle 100 is located.

Said point positions are defined in this example by their Cartesian coordinates in an orthonormal coordinate system (0, x, y) located in the plane defined by the section of road on which the motor vehicle 100 is located. The reference (0, x , y) is related to the motor vehicle 100.

The axis (Ο, χ) of this marker extends for example from an origin O located substantially in the center of the front face FA of the motor vehicle 1 00, in a direction perpendicular to this front face FA, and to the zone located at the front of the motor vehicle 100.

In the example shown in FIG. 1, a first edge of the taxiway LA, which separates it from the taxiway LA ', is marked on the ground by a dotted marking line LM, for example white or yellow. The second edge of the taxiway LA, which separates it from the outside of the road R, is materialized on the ground by a continuous marking line LL, for example white or yellow.

The LL and LM marking lines, which are designed to be easily visualized, are clearly distinguishable visually from their environment. The analysis module 108 can therefore accurately locate these lines in an image of the environment facing the motor vehicle 100, acquired by the image sensor 101 and thus extract the positions, within this image, of points located along the LM marking line and along the LL marking line. For each of these points, the position within said image can then be converted by the analysis module 108 so as to obtain the position of this point in the physical space facing the motor vehicle 100, in this case in the mark (0, x, y).

The step of converting the position of a given point, within such an image, to the position corresponding to this point in the (0, x, y) frame, can for example be carried out using a table of correspondence that connects them. This correspondence table can for example be produced during preliminary calibration steps in which source points are placed at known positions in the (0, x, y) frame. In this situation, it is at the same time the position of a source point within an image, and its position in the coordinate system (0, x, y) which are known, which makes it possible to determine the link existing between them , which link can be stored here in the storage module 103 in the form of said correspondence table.

The positions in the mark (0, x, y) of points situated along the marking line LM, and along the marking line LL can thus be determined by the analysis module 1 08, on the basis of an image of the environment facing the motor vehicle 100 acquired by the image sensor 101.

The field of view of the image sensor 101 corresponds approximately, in the plane (0, x, y), to a truncated angular sector, delimited by two straight lines D and D ', and whose truncated peak is located opposite the motor vehicle 1 00. This field of vision makes it possible to visualize (then to locate) points of the LM marking line located beyond a point A of this line, the point A being situated at a distance d from the front face of the motor vehicle 100 for example between 0.1 meters and 5 meters.

The distance D up to which points can be located along the marking line LM, by acquisition of an image by the image sensor 101, is even greater than the resolution of the latter is high. In practice, the value of this limit distance D may be equal to or greater than 100 meters.

The points of the LM marking line that can be identified in this way are thus between the point A mentioned above and a point C (visible in FIG. 5) located substantially at a distance D from the front face FA of the motor vehicle 100.

In a comparable manner, the portion of the marking line LL along which points can thus be located is between a point E situated substantially at a distance d from the front face of the motor vehicle 100, and a point G (visible FIG. 5) substantially at a distance D therefrom.

The positions of points situated along the marking lines LM and LL, determined by the analysis module 109, allow the estimation by the latter, according to a method described below, of geometric parameters representative of the shape of each of the lines LM and LL marking, said parameters including in particular data identifying the position of such a marking line with respect to the motor vehicle 1 00.

In other embodiments, the shape of the road portion facing the motor vehicle 100 may be determined on the basis of object positions, such as guardrails, located along said driving lane, such as objects that can be detected by an image sensor, or by a radar, for example.

The geometric parameters thus estimated can be transmitted by the analysis module 109 to a driver assistance module 1 07.

The driver assistance module 107 can control reaction means in order to trigger a controllable functionality, in particular according to the result of said estimation of geometric parameters representative of the shape of the road R.

For example, the driver assistance module 107 may control a signaling and / or alarm device, for example a loudspeaker so as to trigger an audible alarm signaling to a driver of the vehicle automobile 100 that the distance separating, in the direction y, the motor vehicle 100 of the marking line LL is less than a given determined limit.

The driver assistance module 107 can also, depending on the geometric characteristics of the portion of road facing the motor vehicle 1 00, control an actuator, such as a brake assist system, or a braking system emergency, or a power steering system.

The motor vehicle 100 may also include:

a speed sensor 104 delivering data representative of the speed of the motor vehicle 100 with respect to the road R,

an accelerometer and / or a gyrometer 105, and

a sensor 106 enabling the motor vehicle 100 to be geolocated, for example a GPS signal sensor (English acronym for "Global Positioning System") equipped with an analysis module, adapted to locate the motor vehicle 100.

The data delivered by the sensors 104, 105 and 106 are transmitted to the analysis module 108 which determines, in particular on the basis of these data, for each previously detected marking line, if it is a a marking line separating two traffic lanes in opposite directions, or a marking line separating two lanes of traffic in the same direction, or a marking line marking an edge of Route R, or another type of marking line.

The data delivered by the sensors 104, 105 and 106 can also be advantageously transmitted to the driver assistance module 1 07, to be combined with the geometric parameters representative of the shape of the road portion facing the motor vehicle 100 , in order to determine the operating conditions of the motor vehicle 1 00 with respect to the road R, and in particular to anticipate the future positions of the motor vehicle 100 on the road R. The above-mentioned reaction means can then be controlled by the module driving assistance 107 according to the operating conditions of the motor vehicle 100, especially when these operating conditions do not seem appropriate to the environment in which the motor vehicle 100 circulates. FIG. 2 represents the main steps of an exemplary method implemented by the estimation system 1 09 of FIG. 1.

Such a method starts in step 201 by the acquisition, by the sensor 101, data representative of the environment facing the motor vehicle 100, here an image of this environment.

The data thus acquired are then analyzed by the analysis module 108, during the step 202, to determine positions of points marking at least one edge of the taxiway LA on which the motor vehicle 100 is located.

These positions correspond, in the example described here, to the positions in the reference (0, x, y) of points situated:

- along the LM marking line, between points A and C, and

- along the LL marking line, between points E and G.

The points located along the LM marking line, and whose positions are thus determined, are noted here M1, M2, ..., Mn. They are divided, in increasing order of index, between the point A and the point C. The point M1 is thus that which is located closest to the motor vehicle 1 00 (it is about point A), and the point Mn is the one that is farthest away (this is point C).

The following step 203 is an estimation step, by the analysis module 108, as a function of said position of points, of parameters of a first geometric model, representative of the shape of at least a portion of said road R. .

This estimation step 203 is described here in the nonlimiting case of a circular road model, that is to say here a model in which a portion of each of the marking lines LM and LL describes an arc circle.

As mentioned in the introduction, the part of the road R facing the motor vehicle 100, considered in its entirety, may have a complex shape that a single arc may not be sufficient to describe.

On the other hand, an arc of circle can provide a very precise description of the geometry of the road R, when this description is limited to a portion of it.

The parameters of this first road model are determined so as to optimize the description accuracy, by this model, of the portion of the road R located immediately opposite the motor vehicle 100. In the context of the circular road model used here, the geometric parameters characterizing the shape of the portion of the LM marking line facing the motor vehicle 100 include:

the coordinates (XC, YC), in the coordinate system (0, x, y), of the center Ci ^M of an arc ARC1 M describing this marking line, and

- The radius Ri ^M of this arc.

Similarly, the arc ARC1 L which describes the portion of the marking line LL facing the motor vehicle 1 00 is characterized by its radius and the position of its center in the reference (0, x, y).

The method used in this exemplary embodiment for estimating the geometrical parameters of each of the arcs of circle ARC1 M and ARC1 L is shown schematically in FIG. 3. This estimation method, similar for arcs of circle ARC1 M and ARC1 L is described below for arcuate arc ARC1 M.

Step 203 here comprises the iterative execution of a substep sequence 301, 302, 303, 304, possibly 305, and 306. This sequence is for example executed a number k of times. During execution number i of this sequence of substeps, an ARCI arc Mi is determined. At the end of step 203, the arc which most precisely describes the geometry of the marking line LM, among the arcs of circle ARC1 M1, ARC1 M2,..., ARC1 Mk thus determined, is preserved for use in later stages of the process.

In the sub-step 301, three points P1, P2 and P3 are randomly selected from the points M1, M2, ..., Mn located along the LM marking line. The points P1, P2 and P3 can therefore be located over the entire portion of the marking line LM between points A and C, that is to say over the entire portion of the LM marking line for which positions points could be determined in step 202.

The geometrical parameters of an arc ARC1 mi passing through these three points Pi, P ₂ and P ₃ are then determined during the sub-step 302. These parameters include the coordinates (XCi, YCi) of the center Ci ^M i of this arc of circle, and its radius Ri ^M i.

The center Ci ^M i of an arc passing through three points Pi, P ₂ and P ₃ is located at the intersection:

the mediator Ai of the segment connecting the points P ₂ and P ₃ , the mediator Δ ₂ of the segment connecting the points Pi and P ₃ , and

the mediator Δ ₃ of the segment connecting the points Pi and P ₂ .

The coordinates (XCi, YCi) in the coordinate system (0, x, y) of the center Ci ^M i can therefore be obtained by:

determining the Cartesian equations E1, E2 and E3, respectively describing, in the frame (0, x, y), each of the mediators Δ ₂ and Δ ₃ , and

determining the pair of coordinates (XCi, YCi) solution of the system of equation {E1, E2, E3}.

Two of the three equations E1, E2 and E3 are theoretically sufficient to determine the pair of coordinates (XCi, YCi). Taking into account the three equations E1, E2 and E3 may, however, be advantageous for improving in practice the accuracy of determining the coordinates (XCi, YCi).

The value of the radius R- | ^M i of the ARCI arc Mi through the three points Pi, P ₂ and P ₃ can then be determined for example by calculating the distance between the center Ci i of this arc and one of the points Pi, P ₂ or P ₃ .

A quantity ECi, representative of the precision with which the arcuate arc ARC1 Mi describes the geometry of the portion of the marking line LM situated immediately opposite the motor vehicle 100, is then calculated during the substep 303.

The size ECi is function:

geometric parameters (XCi, YCi, Ri ^M i) characterizing the ARCI arc Mi,

coordinates of the points M1, M2,..., Mn, and

weighting coefficients c1, c2, ..., cn respectively associated with each of the points M1, M2, ..., Mn.

The weighting coefficients c1, c2, ..., cn are chosen to give, in the calculation of the magnitude ECi, a greater weight at the points of the LM marking line located near the motor vehicle 100, than to those located away from this one.

The quantity ECi can for example be calculated according to the formula F1: ECi = cj. dist (ARClMi, Mj) / cj (Fl)

or :

dist (ARC1 Mi, Mj) represents the distance between the arc ARC1 Mi and the point Mj (in the plane (0, x, y)), and where

the value of a weighting coefficient cj increases as the point Mj with which it is associated is close to the motor vehicle 100. The value of a coefficient cj may, for example, decrease from a value close to 1 for the point M1 to a value close to 0.25 for the point Mn.

The magnitude ECi is, in this embodiment, even smaller than the description of the portion of the marking line LM between the points A and C, by the arc ARC1 arc is accurate. The quantity ECi thus represents a distance between the arc ARC1 Mi and the set of points detected Mi.

In the sub-step 304, the magnitude ECi is compared with a magnitude EC (defined below), in order to determine whether the ARCI arc Mi is the one that most accurately describes the portion of the LM marking line. between the points A and C, among arcs of circle ARC1 M1 to ARCI Mi previously determined, during executions number 1 to i of the sequence of substeps 301 to 306.

If this is the case, the parameters of arc ARC1 Mi are assigned to arc ARC1 M, during substep 305:

XC = XCi, YC = YCi, = R ^ i.

This process is for example implemented by performing the substep 304 as follows.

For i = 1, that is to say if it is the first execution of the sequence of substeps 301 to 306, step 203 continues with substep 305, and the magnitude EC is initialized by assigning it the value of the quantity EC1:

EC = EC1.

For i> 1, and if ECi <EC, which translates here that the ARCI arc Mi is the one that most accurately describes the portion of the LM marking line between the points A and C, among the arcs of circle ARC1 M1 to ARCI Mi previously determined, then: step 203 continues with substep 305, and

the value of the quantity ECi is attributed to the quantity EC: EC = ECi. In other cases, step 203 continues directly by substep 306.

The quantity EC corresponds to the smallest of the quantities EC1 to ECi obtained during the executions numbers 1 to i of the sequence of substeps 301 to 306. It is even smaller than the arc ARC1 M describes with precision the portion of the LM marking line between points A and C.

In the sub-step 306, the analysis module 108 tests whether the sequence of substeps 301 to 306 is to be executed again or if step 203 can be completed, the method then continuing through the step 204. The test carried out in sub-step 306 may relate both to:

the number i of realizations of the sequence of substeps 301 to 306 already carried out during step 203 (this number being for example limited here to k realizations), and / or on

the precision with which the arc ARC1 M describes the portion of the marking line lying between the points A and C, the precision translated here by the value of the quantity EC; the step 203 may for example be completed, at the end of the substep 306, as soon as the magnitude EC is less than a given determined threshold EClim.

The value of the threshold EClim can for example be recorded in the storage means 103 of the analysis module 108 during commissioning of the latter.

FIG. 5 shows an example of arc ARC1 whose parameters (XC, YC, Ri ^M ) have been determined during step 203.

In this example, the portion of the road R facing the motor vehicle 100 first describes a turn, in this case on the left, and then a straight line.

The portion of the LM marking line situated between the points A and C thus comprises a first section, substantially in the shape of an arc of a circle, extending from the point A to a point B, followed by a second substantially rectilinear section. , between points B and C.

As can be seen in FIG. 5, the arc of circle ARC1 M describes the first section AB with precision, but departs clearly from the marking line LM at the second section BC. The accuracy of description of the first section AB by the arc ARC1 M is an advantageous result of the predominant weight given to the points of the marking line LM which are close to the motor vehicle 100, by means of the weighting coefficients c1, c2, ..., cn, when calculating the magnitude ECi.

This is particularly interesting because, as described below, arc ARC1 M forms a first model intended to approach the shape of the road (here, precisely, the LM marking line) in a proximal portion of this one (compared to the motor vehicle).

Such a weighting, in the calculation of the magnitude ECi, also makes it possible to take into account all the points situated along the portion AC of the marking line LM (while privileging, as described above, the points the closest to the motor vehicle 100). This characteristic is interesting since the extent of a first section AB of the marking line LM, which can be accurately described by a single arc, is not known in advance of the estimation device 109.

Otherwise formulated, the limit position, corresponding to point B marking the transition between:

a first section of road AB that can be precisely described by a first road model, here circular, and

a second section of road BC, which can be described precisely by a second road model, here also circular and distinct from the first,

is not known a priori of the estimation device 109. The position of this transition point B in the reference (0, x, y) also evolves progressively as the motor vehicle 100 progresses along. from taxiway LA.

This limit position is determined in the next step 204.

Note also that a section of straight road is a special case of a circular section of road, very large radius, and can therefore be accurately described by the circular road model used during the stage 203.

It will also be noted that the estimation method implemented during step 203 is similar to a "RANSAC" type estimate (in the English acronym of "RANdom SAmple Consensus"), during which:

only a part of the points of the set of points M1, M2,... Mn, chosen at random (the points Pi, P ₂ and P ₃ ) serves to determine the parameters (XCi, YCi, Ri ^M i) of the arc ARC1 Mi,

the precision of description of the line of marking LM by the arc ARCI Mi is then evaluated by taking into account all the points M1, M2, ... Mn,

- ARC1 Mi arch with the best description accuracy is finally retained.

Such a method advantageously makes it possible to reduce the influence of the measurement noise on the parameters (XC, YC, Ri ^M ) of the arc ARC1 thus determined, while minimizing the influence of possible aberrant measurement points, by virtue of this process of random selection of subsets of points (P1, P2, P3).

The parameters of the arc of circle ARC1 L, describing the marking line LL, are determined during the step 203, in a manner comparable to what is presented above for the circular arc ARC1 M.

The position of the point B, up to which the first road model precisely describes the geometry of the marking line LM, is determined by the analysis module 108 in the step 204 shown in more detail in FIG. 4.

During this step, the description accuracy of a section AMI of the marking line LM, by the arc ARC1 M, is evaluated iteratively, for Ml points for example closer and closer to the motor vehicle 1 00, until said description accuracy is better than a given accuracy. The AMI section thus determined then corresponds to section AB.

The accuracy of description of a section AMI of the marking line LM, by the arc ARC1 M, is evaluated by giving the same weight all the points M1, M2, ... , MI of this stretch. This accuracy of description is for example evaluated by calculating a quantity EC '(I) according to the formula F2:

EC '(l) = - ^ dist (ARClM, Mj) (F2)

i = i

which amounts to assigning the same weighting coefficient to all the points M 1, M2, ... ,MID.

The quantity EC '(I) therefore represents an average distance between the arc ARC1 M and the set of points M 1 to Ml.

In this embodiment, the point B is then defined as a point, among the detected points Mi, such as the distance EC '(I) between the arc ARC1 M and the set of points M1,. ,., ΜΙ (located between the point A and the point B), is lower than a given determined distance, denoted EC'lim. On the other hand, the distance EC '(I) between the arc ARC1 M and a set of points extending beyond the point B, is a priori greater than the given determined distance EC'lim.

The value of the distance EC'lim, which determines the precision intended for the description of a first section of the marking line LM, is recorded, for example, in the storage means 103 of the analysis module 1 08 when setting in service of the latter. Its value can for example be chosen between 1 and 50 centimeters.

Step 204 begins with a substep 401, during which a counter I is initialized with the value n: 1 = n. This makes it possible to start by calculating the quantity EC '(I) for all the section AC of the marking line LM.

The following substeps 402, 403, and optionally 404, are then executed iteratively.

In sub-step 402, the quantity EC '(I) is calculated. Then, during step 403, the previously calculated quantity EC '(I) is compared with a quantity EC'lim.

If EC '(l)> EC'lim, which translates that the AMI segment is described by the ARC1 M arc with a precision estimated as insufficient, the process continues with the sub-step 404. During the Sub-step 404, the counter I is decremented by one unit, and the process then resumes in step 402.

If, on the contrary, EC '(I) <EC'lim, which indicates that the segment AMI is described by the arc ARC1 M with an accuracy considered as sufficient, the process continues, after the sub-step 403, by the sub-step 405 during which the coordinates of the point Ml are assigned to the point B.

In another embodiment, the position of the point B can be determined, also iteratively, by evaluating the description accuracy of a section AMI of the marking line LM, by the arc ARC1 M, but for Ml points increasingly distant from the motor vehicle 100, and until said accuracy of description becomes lower than a given specific accuracy (which is reflected here by the fact that the distance EC '(I) becomes greater at the given determined distance EC'lim), which then marks the limit of the first section of the LM marking line correctly described by the ARC1 M. arc

The position of a point F, up to which the arc of circle ARC1 L precisely describes the geometry of the marking line LL, is determined during the step 204, in a manner comparable to that presented above for point B, in step 204.

Note that, because the arc ARC1 M has been chosen as the optimum model by favoring the proximal part of the road (here the part AB) thanks to the use of the weights cj, it is not necessary to reiterate a search for the optimal model for each value of I in the process above.

The parameters of a second road model describing the second section BC of the marking line LM, here by an arc ARC2M arc, are determined during step 205.

The ARC2M arc parameters are determined as a function of the positions of the points of the marking line LM lying between the points B and C. This determination is made in a manner comparable to the determination of the parameters of the ARC1 arc. M, performed during step 203 described above, as a function of the positions of the points of the marking line LM between points A and C.

Likewise, a second road model describing the second section FG of the marking line LL, here by an arc ARC2L, is determined during step 205.

Optionally, the position of a point B2, corresponding to a second limit position up to which the arc ARC2M accurately describes the geometry of the marking line LM, can be determined in step 205.

The position of point B2, located on the LM marking line between points B and C, can in this case be determined by a process comparable to that of step 204 described above.

Similarly, the position of a point F2, up to which the arc of a circle

ARC2M accurately describes the geometry of the LM marking line, can be determined in step 205.

As can be seen in FIG. 5, the use of a first and second distinct circular road model, the parameters of which are determined according to the estimation method described above, makes it possible to accurately describe a portion of a road comprising a turn and then a straight section, even though the limit position corresponding to the transition between these two sections is not known. beforehand.

In general, this estimation method makes it possible to obtain a precise geometrical characterization of a portion of road comprising two successive sections of distinct shapes, which is frequent for a winding road, in rural areas for example.

These two sections may for example correspond to a rectilinear section followed by a turn, or, as in the case of Figure 5, to a turn followed by a straight section, or to two successive turns of different rays or directions .

The estimation method described above is also adapted to a portion of a road that can be described by a single circular road model, for example a straight road portion from point A to point C. In this case, the position of point B, determined in step 204, is simply the point

C. In this case, it is possible to go directly from step 204 to step

206 without executing step 205.

Furthermore, in the embodiment presented above, the first, as the second section of the road R is described by an arc. In other embodiments, one and / or the other of these two sections may be described by another geometric shape, for example a clothoid arc.

Finally, during step 206, the parameters of the first and second geometric models of the road R, previously determined by the analysis module 1 08, are transmitted to the driver assistance module 107, as well as the coordinates points B and F, and optionally, the coordinates of points B2 and F2. Here, these parameters include in particular:

the value of the radius Ri ^M and the coordinates (XC, YC) of the center Ci ^M of the arc ARC1 M, and

- the radius value and the coordinates of the center of the arc

ARC1 L.

This data is then used by the driver assistance module 107 to control reaction means, as explained above in the description of FIG. In the embodiment described above, geometric parameters representative of the shape of each of the two marking lines LM and LL are estimated.

In another embodiment, such geometric parameters can be estimated for only one of the two marking lines, LM or LL. The description of the part of the road R facing the motor vehicle is less complete in this case than in the previous one, but it is also faster to obtain, and can thus be well adapted for certain driving assistance applications requiring a particularly short reaction time.

Claims

1. Method for estimating geometric parameters (XC, YC, Ri) representative of the shape of a road (R) by means of a sensor (101) and an analysis module (108) fitted to a motor vehicle (1 00) located on said road (R), comprising the following steps:

acquisition, by the sensor (101), of data representative of the environment facing said motor vehicle (100),

- Detecting, by the analysis module (108), according to said data, points (M1, M2, ..., Mn) locating an edge (LM; LL) of at least one traffic lane (LA) said road (R), and determining the positions of said detected points (M1, M2, ..., Mn),

estimating, as a function of said positions of the detected points (M1, M2, ..., Mn), parameters (XC, YC, R ^) of a first geometric model, representative of the shape of at least a portion of said road (R),

characterized in that it further comprises the following steps:

determining a limit position, with respect to the motor vehicle, up to which the first previously parameterized geometric model describes with a given precision the shape of said road (R), and

- estimation, according to said positions of the detected points

(M1, M2, ..., Mn), located beyond said limit position, parameters of a second geometric model representative of the shape of said road beyond said limit position.

2. An estimation method according to claim 1, wherein the parameters (XC, YC, Ri ^M ) of the first geometric model are furthermore estimated according to weighting coefficients (c1, c2, ..., cn) respectively associated at the detected points (M1, M2, ..., Mn) and decreasing as a function of the distance from the associated point (Mj) to the motor vehicle (100), so as to make the parameters (XC, YC, Ri ^M ) dependent on the first geometric model all the more strongly of the position of one of the detected points (Mj) that this point is close to the motor vehicle (100).

3. Estimation method according to one of claims 1 or 2, wherein the determination of the limit position comprises a step of determining said accuracy of description of the shape of a given section (A-MI) of said route (R) by the first geometric model previously parameterized, according to:

parameters (XC, YC, Ri ^M ) of said first geometric model, and

positions of the detected points belonging to the given section (M1, M2, ..., MI), each of said positions contributing in an equivalent manner and without weighting to said determination.

4. An estimation method according to claim 3, wherein said accuracy is determined by evaluating a quantity (EC '(I)) representative of a distance between the first geometric model previously parameterized and the set of detected points belonging to the section. given (M1, M2, ..., MI), said quantity (EC '(I)) being evaluated in:

calculating, in an equivalent manner for each detected point (Mj), an intermediate quantity (dist (ARClM, Mj)) representing a distance between said detected point (Mj) and the first geometric model previously parameterized, and

summing said intermediate quantities (dist (ARClM, Mj)).

5. estimation method according to one of claims 1 to 4, wherein said limit position is determined so that the accuracy of description, by the first geometric model previously set, a road section (R) s' extending from a proximal zone of the motor vehicle to beyond said limit position, is less than said given accuracy.

6. Estimation method according to one of claims 1 to 5, wherein the first geometric model or the second geometric model corresponds to an arc (ARC1 M ARC1 L ARC2M ARC2L), the parameters of said geometric model comprising the radius (Ri ^M ) of said arc and the position (XC, YC) of its center relative to the motor vehicle (100).

7. Estimation method according to one of claims 1 to 6, wherein the first geometric model or the second geometric model corresponds to a clothoid arc.

8. Estimation method according to one of claims 1 to 7, wherein the step of estimating the parameters of the first geometric model (XC, YC, Ri ^M ) or the second geometric model comprises an iterative process comprising steps from:

- random selection of only a part (P1, P2, P3) of the points detected (M1, M2 Mn),

determination of a temporary set of parameters (XCi, YCi, Ri ^M i) of said geometric model, as a function of the positions of said randomly selected points (P1, P2, P3),

estimating a precision of description of the shape of said road by the geometric model parameterized by said temporary set of parameters (XCi, YCi, Ri i), said estimation taking into account all the detected points (M1, M2, ..., Mn),

selecting, from the set of temporary parameter sets (XCi, YCi, Ri ^M i), the set of parameters (XC, YC, Ri ^M ) corresponding to the best description accuracy.

9. An estimation method according to one of claims 1 to 8, further comprising a step of determining a second limit position, relative to the motor vehicle (1 00), to which said second geometric model previously parameterized describes with a given precision the shape of said road (R).

1 0. An estimation system (1 09) of geometric parameters (XC, YC, Ri ^M ) representative of the shape of a road (R), for a motor vehicle (1 00), comprising a sensor (101) adapted to acquiring data representative of the environment facing the motor vehicle (100), and an analysis module (108) adapted to:

detecting, on the basis of said data, points (M1, M2,..., Mn) identifying an edge (LM; LL) of at least one traffic lane (LA) of said road (R), and determining positions of said detected points (M1, M2, ..., Mn),

- estimate, according to said positions of the detected points

(M1, M2, ..., Mn), parameters (XC, YC, R- | ^M ) of a first geometric model, representative of the shape of at least part of said road (R),

characterized in that the analysis module (108) is further adapted to:

determining a limit position, with respect to the motor vehicle (100), up to which the first previously parameterized geometric model describes with a given precision the shape of said road (R), and

estimating, according to said positions of the detected points (M1, M2,..., Mn), located beyond said limit position, parameters of a second geometric model representative of the shape of said road (R) at -of the of said limit position.

1 1. A motor vehicle (100) comprising an estimation system (109) according to claim 10, and a driver assistance module (107), the estimating system (109) being adapted to transmit to the assistance module to the conduct (1 07) the parameters of the first geometric model (XC, YC, R ^) and the second geometric model and said limit position previously estimated, the driver assistance module (1 07) being adapted to control means in order to trigger a controllable functionality based on said parameters (XC, YC, Ri ^M ) and said limit position received from the estimation system (109).