METHOD FOR DETECTING A CENTER LINE OF A TRAFFIC ROAD

Description

Technical field of the invention

[0001] The present invention relates generally to driving aids for motor vehicles.

[0002] It applies particularly to cars and other motorized vehicles traveling on roads, but also applies to other fields such as robotics.

[0003] The invention relates to a method for detecting a center line of a lane belonging to a road which is being used by a motor vehicle.

[0004] It also concerns a motor vehicle adapted to implement such a detection process. State of the art

[0005] In an effort to improve the safety of motor vehicles, they are currently being equipped with driver assistance systems and even highly automated driving systems. An example of prior art is the document CHENHAO WANG ET AL: "Precise curvature estimation by cooperating with digital road map", INTELLIGENT VEHICLES SYMPOSIUM, 2008 IEEE, IEEE, PISCATAWAY, NJ, USA, June 4, 2008 (2008-06-04), pages 859-864, XP031318945, ISBN: 978-1-4244-2568-6.

[0006] These are typically lane keeping assist systems (better known by the English acronym LKA for "Lane Keeping Assist"), adaptive cruise control systems...

[0007] Many of these systems need, in order to function, to know the position of the center line of the traffic lane used by the vehicle, or even the positions of the center lines of other traffic lanes (for example, the one towards which the vehicle is heading when it changes lanes).

[0008] Currently, it is known to use a sensor, such as a camera, which incorporates image processing means to determine the position of this center line of track.

[0009] It is also known to use several sensors, such as a camera coupled with a RADAR remote detector, and then process the data from these different sensors together, for example by data fusion, in order to deduce the position of the center line of the track.

[0010] These different solutions allow us to approximate the position of the center line.

[0011] Unfortunately, they do not provide complete satisfaction since sometimes the central line thus identified does not allow for a comfortable driving experience that offers the desired safety guarantees. Presentation of the invention

[0012] In order to remedy the aforementioned drawback of the prior art, the present invention proposes to construct several central lines that can potentially be used, and then to select the one that is most suitable for the traffic conditions of the motor vehicle.

[0013] The invention is defined by the claims.

[0014] More specifically, a detection method is proposed which includes the following steps: acquisition of information characterizing two track edge lines (those which delimit the traffic lane whose center line we wish to know), acquisition of data relating to the posture and / or movement of the motor vehicle on the road, and in particular on the traffic lane, creation of at least two candidate center lines located between the two track edge lines, during which an electronic and / or computer unit calculates coefficients characterizing said at least two candidate center lines according to the information acquired, selection of one of the candidate center lines according to the data acquired, and deduction of the center line of the lane according to the selected candidate center line.

[0015] Thus, thanks to the invention, several candidate center lines are calculated, and the selected one depends on the instantaneous behavior of the motor vehicle (for example, its position on the road and / or its velocity vector and / or its acceleration vector and / or its heading angle and / or its yaw rate and / or its yaw acceleration). In this way, the method provides center line detection that adapts to the behavior of the motor vehicle, ensuring good driving comfort and safe driving.

[0016] Indeed, the result of the invention, when used to steer a motor vehicle automatically, allows the vehicle to follow more consistent trajectories given its position and speed. In other words, the vehicle's trajectory is smoother and more reliable.

[0017] It should be noted that two identical vehicles (therefore equipped with the same detection system) successively using the same traffic lane will not necessarily detect the same center line if their positions and / or dynamic characteristics are not the same.

[0018] One advantage of the invention is that it does not generate any additional cost on the vehicle in which it is integrated, since the sensors or equipment used are already conventionally embedded in mid-range vehicles.

[0019] It should also be noted that this solution does not require any connectivity with other vehicles or infrastructure, making it inexpensive and robust.

[0020] Other advantageous and non-limiting features of the detection method according to the invention, taken individually or in all technically possible combinations, are as follows: The coefficients characterizing at least one of the candidate centerlines are calculated based solely on the information characterizing one of the two trackside lines and the track width; the coefficients characterizing at least one of the candidate centerlines are calculated based on the information characterizing both trackside lines; exactly two candidate centerlines are created (to reduce the computational load); exactly three candidate centerlines are created (which offers a greater variety in the choice, without overloading the calculations); more than three candidate centerlines are created (to guarantee better accuracy);It is planned to have a step of estimating data relating to the posture and / or movement that the motor vehicle will exhibit at a future time, then another step of selecting one of the candidate centerlines based on the estimated data, the centerline of the lane then being deduced based on the two selected candidate centerlines (the one initially selected in the selection step and then the one selected during said other selection step); if the same candidate centerline has been selected twice, in the deduction step, it is deduced that the centerline of the lane is the selected candidate centerline; if the two selected candidate centerlines are different, in the deduction step, it is planned to construct the centerline of the lane based on the coefficients characterizing the two selected candidate centerlines;At each selection stage, it is planned to calculate a distance between the motor vehicle and each candidate centerline, and to select the candidate centerline for which the calculated distance is the smallest; the calculated distance between the motor vehicle and each candidate centerline is a statistical distance, for example a Mahalanobis distance, which depends on a covariance matrix associated with said candidate centerline.

[0021] The invention also proposes a motor vehicle comprising means for acquiring information characterizing two lane edge lines of a road traveled by the motor vehicle and data relating to the posture and / or movement of the motor vehicle on the road, and a computer adapted to implement a detection method as described above.

[0022] Of course, the different features, variants and embodiments of the invention can be combined with each other in various ways as long as they are not incompatible or mutually exclusive. Detailed description of the invention

[0023] The description that follows, with regard to the attached drawings, given by way of non-limiting examples, will make it clear what the invention consists of and how it can be carried out.

[0024] Regarding the attached drawings: [ Fig.1 ] is a schematic view of a motor vehicle adapted to implement a method according to the present invention; [ Fig. 2 ] is a schematic top view of the motor vehicle of the [ Fig.1 ], on which central and lane edge lines appear; [ Fig.3 ] is a block diagram illustrating the different stages of a process according to the present invention.

[0025] On the [ Fig.1 ], a vehicle 10 adapted to implement the invention was represented.

[0026] This refers to a car. Alternatively, it could be another type of vehicle (truck, motorcycle, etc.).

[0027] Here, this vehicle 10 classically includes a passenger compartment in which there is, among other things, a seat for the driver 20 of the vehicle, a dashboard with a display screen, and a steering wheel 12.

[0028] This vehicle 10 includes a powertrain, a braking system, and a steering system for turning the vehicle (not shown in the figure). Typically, the steering system includes an electronically controlled power steering actuator, the powertrain includes an electronically controlled engine control actuator, and the braking system includes an electronically controlled brake actuator.

[0029] The vehicle 10 also includes an electronic and / or computer processing unit (hereinafter referred to as computer 11) comprising at least one microprocessor or microcontroller, at least one memory and input and output interfaces.

[0030] Thanks to its input interfaces, the calculator 11 is adapted to receive various input data, which come from third-party sensors or calculators.

[0031] Among these sensors, the following are planned, for example: a device such as a front camera, enabling the position of the motor vehicle 10 to be located in relation to its lane of travel, and a device such as a RADAR remote detector and / or a LIDAR remote detector, enabling the detection of the edges of the road.

[0032] Thanks to its output interfaces, the calculator 11 is suitable for controlling the display screen, the power steering actuator, the engine control actuator, and the brake actuator.

[0033] Thus, the computer is adapted to implement driver assistance functions and / or automated driving functions (in which the vehicle can move in traffic autonomously, without intervention from the driver).

[0034] Thanks to its memory, the calculator 11 stores a computer application, consisting of computer programs including instructions whose execution by the calculator allows the implementation of the process described below.

[0035] On the [ Fig. 2 [ ] We have represented the motor vehicle 10 seen from above, while it is taking a traffic lane 1 of a road.

[0036] In the example considered here for illustrative purposes, this road is of the "motorway" type. Only one of the traffic lanes 1 of this motorway is represented here, namely the one used by the motor vehicle 10.

[0037] In this figure, we observe that traffic lane 1 is delimited by two lane edge lines, namely left line 1G and right line 1D. In this particular example, traffic lane 1 widens since the vehicle is at a motorway exit.

[0038] Of course, the procedure described below will apply to any other traffic situation.

[0039] On this [ Fig. 2 We have also represented a coordinate system (X, Y) attached to the motor vehicle 10, which will be used for the calculations described below. This coordinate system has an abscissa extending forward along the longitudinal axis of the vehicle, and a ordinate extending to the left of the vehicle. Its origin is located, for example, at the center of gravity of the motor vehicle 10.

[0040] The objective is to determine the position and shape of the centerline of traffic lane 1.

[0041] On the [ Fig.3 ], we illustrated the process that calculator 11 is adapted to implement in order to characterize this central line.

[0042] It should be noted that this central line can be defined as a line which is substantially positioned in the center of the traffic lane 1, between the right line 1D and left line 1G, and which will be taken into account to drive the motor vehicle in an automated manner.

[0043] As will become clear after reading this explanation, this central line will not necessarily be central in the geometric sense of the term. Rather, it will be a line close to this geometric central line.

[0044] Here, the traffic lane considered will be the lane in which vehicle 10 is traveling. Of course, it could be another lane, in particular an adjacent lane towards which the vehicle wishes to move (for example during a lane change).

[0045] The centerline detection process involves several successive steps, all of which are repeated in a loop at regular intervals so as to regularly update the shape and position of this centerline as the vehicle moves along the road.

[0046] In summary, the method according to the invention comprises steps consisting of acquiring information on the geometry of the traffic lane 1 and on the attitude of the motor vehicle 10, then creating different "candidate center lines" according to the information on the lane, and finally selecting one of these candidate center lines according to the attitude of the motor vehicle 10.

[0047] Put another way, as will be described in detail in steps E1 to E5 below, the idea is to create candidate center lines based on the characteristics of the traffic lane 1 used, and then to select the center line best suited to the current behavior (at time t 0) of the motor vehicle.

[0048] It could be envisaged that the candidate centerline thus selected forms the centerline of traffic lane 1 to be considered for piloting the motor vehicle in an automated manner.

[0049] However, preferably, as will be described in detail in steps E6 to E10, an additional approach will be to predict the behavior of the motor vehicle in the short term (at a time t' > t0), and then select the candidate center line best suited to the future behavior (at time t') of the motor vehicle. From this, it will be possible to deduce the position of the center line of lane 1 based on the candidate center lines selected at times t0 and t' (which lines may be the same or different).

[0050] The detailed steps of this process are as follows.

[0051] During the first stage E1, the computer 11 receives information characterizing the two track edge lines (the left line 1G and the right line 1D).

[0052] These road edge lines are typically formed by the edge of the road shoulder and / or by a road marking line.

[0053] This information can be provided directly by a sensor or be pre-processed.

[0054] Preferably, only information measured by the vehicle's sensors is used to detect these lane edge lines. No information received from road infrastructure or third-party vehicles is used, for example.

[0055] Here, each left line 1G and right line 1D is modeled in analytical form.

[0056] As an example, one might expect these lines to be modeled in polynomial form. However, here they are modeled as a clothoid, this form being preferred because it provides better results (for the same number of coefficients taken into account).

[0057] Such a clothoid is defined as having a curvature c that depends on its length, which can be written: c l = c 0 + c 1 . l

[0058] In this equation: c is therefore the equation of the clothoid, 1 is the curvilinear abscissa of the arc of the clothoid, which varies between 0 inclusive and L inclusive, L is the length of the arc of the clothoid, c 0 is the curvature at the origin of the arc, and c 1 is the rate of change of curvature of the arc.

[0059] In other words, a clothoid can be defined by six coefficients, which are: x 0 , the abscissa of the origin point of the clothoid in the (X, Y) coordinate system, y 0 the ordinate of this origin point, Ψ 0 which is the angle of heading at the origin of the arc, L, c 0 , and c 1 .

[0060] In the following, each clothoid will be defined by the set C of these six coefficients, which can be noted in the following way: C (x 0 , y 0 , Ψ 0 , c 0 , c 1 , L).

[0061] The clothoid characterizing the shape of the left line 1G will then be defined by the set C g (x 0g , y 0g , Ψ 0g , c 0g , c 1g , L g ) while the clothoid characterizing the shape of the right line 1D will be defined by the set Cd (x 0d , y 0d , Ψ 0d , c 0d , c 1d , L d ).

[0062] For your information, the angle of heading along the clothoid can be written as follows: Ψ l = Ψ 0 + ∫ 0 l c u . du = Ψ 0 + c 0 . l + c 1 2 . l 2

[0063] The Cartesian coordinates (x, y) of a point located along the clothoid can be written as follows: x l = x 0 + ∫ 0 l cos Ψ u . du y l = y 0 + ∫ 0 l sin Ψ u . du

[0064] During this first step E1, it is possible to consider that the computer 11 receives the two sets C g , C d from one of the sensors on board the vehicle (for example from the camera if the latter is equipped with a processor adapted to carry out the operations necessary to determine the coefficients sought).

[0065] Alternatively, the calculator 11 could receive data collected by several of the vehicle's sensors, merge this data, and deduce the desired coefficients.

[0066] Alternatively, the calculator could receive coefficients defining the two lines, right 1D and left 1G, in polynomial form (for example, as coefficients of third-order polynomials). The calculator would then need to transform these coefficients to find, by calculation, the values ​​of the six required coefficients. Such a calculation is well known to those skilled in the art and will therefore not be provided here.

[0067] During this step E1, in addition to the aforementioned coefficients, the calculator 11 receives the two variance-covariance matrices (hereafter referred to as covariance matrices) associated respectively with the coefficients of the two right lines 1D and left lines 1G. Each covariance matrix takes into account the uncertainty in the detection of each track edge line by the sensor(s).

[0068] These matrices are, for example, provided by the manufacturer of the sensors used. Alternatively, they can be determined during a sensor testing campaign and stored in the computer's memory. The coefficients of these matrices are preferably constant.

[0069] It should be noted here that if several sensors are used to determine the coefficients characterizing a track edge line, their uncertainties will be associated together in order to obtain the covariance matrix of a detected line.

[0070] Finally, during step E1, the computer 11 acquires data relating to the posture and / or kinematics of the motor vehicle 10 in its lane of travel 1. This is kinematic data which is preferentially obtained using only the sensors on board the motor vehicle 10.

[0071] Here, this data includes values ​​for: speed V of the motor vehicle relative to the ground, expressed in m / s, longitudinal acceleration ax of the motor vehicle along the X axis, expressed in m / s², lateral acceleration ay of the motor vehicle along the Y axis, expressed in m / s², yaw rate of the motor vehicle, expressed in rad / s, yaw rate of the motor vehicle, expressed in rad / s², front wheel angle of the motor vehicle, expressed in rad, rear wheel angle of the motor vehicle, expressed in rad, and instantaneous curvature c₀,ego of the trajectory of the motor vehicle (this curvature being defined here by the inverse of the radius of curvature, expressed in m⁻¹).

[0072] The second step E2 consists, for the computer 11, of constructing by calculation several candidate central lines.

[0073] The idea here is to construct at least two candidate center lines, so that the best one can then be selected taking into account the conditions under which the motor vehicle 10 travels on its lane.

[0074] This construction is entirely performed by computer 11. However, as an alternative, some of the candidate center lines could be created by the camera's processor. However, this solution is not preferred as it is unnecessarily expensive.

[0075] Here, and preferentially, exactly three candidate central lines are constructed (see [ Fig. 2 Alternatively, this number could be more restricted (equal to two) or larger.

[0076] Each of these lines is modeled by a clothoid.

[0077] One of these candidate center lines (called candidate left line 2G) is created based solely on the left edge line 1G and the track width. In other words, this candidate left line 2G is defined by a set of six coefficients whose values ​​are derived from the coefficient values ​​of the clothoid representing the left line 1G and the track width.

[0078] One of these candidate center lines (called the 2D straight candidate line) is created on the basis of only the 1D straight track edge line and the track width.

[0079] Finally, the third of these central candidate lines (called intermediate candidate line 2DG) is created on the basis of both the straight line 1D and the left line 1G.

[0080] Alternatively, these three lines could be modeled differently. Typically, all three lines could be calculated based on the shapes of the two trackside edge lines.

[0081] Alternatively, at least one of the lines could not depend on either of the two trackside lines, typically when one of the sensors (e.g., the camera) directly provides the coefficients of a candidate centerline.

[0082] In the embodiment considered here, the left candidate line 2G is the result of a translation operation of the clothoid defining the left line 1G, across half the width of the track and towards its center. Similarly, the right candidate line 2D is the result of a translation operation of the clothoid defining the right line 1D, across half the width of the track and towards its center.

[0083] Thus, we can write that: The left candidate line 2G is defined by the set of coefficients C gg (x 0g , (y 0g +y 0d ) / 2, Ψ 0g , c 0g , c 1g , L g ). The right candidate line 2D is defined by the set of coefficients C dd (x 0d , (y 0g +y 0d ) / 2, Ψ 0d , c 0d , c 1d , L d ).

[0084] The intermediate candidate line 2DG could be obtained in various ways. Here, a barycentric type operation is used. Thus, we can consider that this line is defined by the set of coefficients C dg ((x 0g +x 0d ) / 2, (y 0g +y 0d ) / 2, (Ψ 0g + Ψ 0d ) / 2, (c 0g +c 0d ) / 2, (c 1g + c 1d ) / 2, (L g +L d ) / 2).

[0085] The third step E3 consists of the calculator 11 calculating the covariance matrix of each candidate row.

[0086] This calculation is performed based on how each of these candidate lines was developed.

[0087] More specifically, the covariance matrix of each candidate row is calculated based on the coefficients of the clothoid associated with that candidate row and based on the covariance matrix(s) associated with the right row 1D and left row 1G (depending on whether one or both rows were considered to define the candidate row in question).

[0088] Several uncertainty transfer algorithms could be implemented to perform this calculation. Since one such uncertainty transfer algorithm is well-known, it will not be described in detail here. We will only specify that the calculation of the covariance matrix can be based on the following formulas: Var AX = A * Var X * A T Var X + Y = Var X + cov X , Y + cov Y , X + Var Y

[0089] In both equations, "Var" denotes the covariance matrix and "cov" denotes the cross-covariance matrix. X and Y are the vectors of the variables in the detected rows, then of size 6x1. A is a constant matrix of size 6x6 in this case.

[0090] It should be noted that assuming that the lines are detected and estimated independently, the cross covariance matrices cov(X,Y) and cov(Y,X) are zero.

[0091] In the example shown on the [ Fig. 2 where the motor vehicle 10 is at a motorway exit, the left candidate line 2G seems most appropriate if the vehicle remains on the motorway, while the right candidate line 2D seems most appropriate if the vehicle exits the motorway.

[0092] At this stage, the present invention therefore proposes to choose the candidate line most appropriate to the situation.

[0093] To this end, a preliminary filtering operation could possibly be implemented in order to set aside the absurd candidate line(s) (for example a candidate line which would get too close to one of the track edge lines).

[0094] In any case, during a fourth step E4, the computer calculates a parameter which will allow the selection of the best candidate line, taking into account the data relating to the motor vehicle 10.

[0095] This parameter is preferably a statistical distance. More precisely, here it is a Mahalanobis distance dM.

[0096] The method used then consists, for the calculator 11, of calculating the Mahalanobis distance d M between each candidate line and the motor vehicle 10 (at the current time t 0, at the level of the motor vehicle 10).

[0097] To do this, the calculator performs the following calculation: d M = u − ν T . Su + Sv . u − ν

[0098] In this equation, u is a first vector relating to the candidate line under consideration, and Su is its covariance matrix. Similarly, v is a second vector relating to the motor vehicle 10, and Sv is its covariance matrix.

[0099] Preferably, the vectors have the same number of coefficients, between two and six.

[0100] Thus, one could predict that the first vector u has six coefficients corresponding to the six coefficients of the clothoid.

[0101] However, a smaller number of coefficients will be used so that the calculation does not require excessive computing power. Here, only two coefficients will be used since it has been observed that the results remain very reliable with only two coefficients.

[0102] One of the coefficients of each vector is a heading angle (such an angle indeed giving a good indication of the driver's intention), while the other coefficient is a curvature (the heading angle not being sufficient to indicate this intention if the road and the vehicle's trajectory are curved).

[0103] Here, the first vector u is therefore defined by: u = Ψ 0 , c 0 T

[0104] The second vector v is defined by: v = θ , c 0 , ego T .

[0105] θ is the heading angle of the motor vehicle 10 and is equal to either the steering angle of the front wheels, or the difference between the steering angle of the front wheels and the steering angle of the rear wheels if the latter is not zero.

[0106] Note that the first vector u will have different values ​​from one candidate line to another.

[0107] It should be noted here that the covariance matrix Su will be deduced from the covariance matrix attached to the candidate row considered.

[0108] The covariance matrix Sv will then be read (its coefficients are stored in the computer's memory and depend on the sensor type and the method used to determine the uncertainty of the sensor(s) used – if several sensors are used in combination). This covariance matrix is ​​therefore determined in a manner analogous to those associated with the two trackside lines. Thus, here again, if several sensors are used in combination, it is possible to associate the uncertainties related to each sensor.

[0109] During this fourth step E4, the Mahalanobis distance is then calculated for each of the candidate lines, at time t 0.

[0110] The fifth step E5 then consists of selecting the candidate line for which the Mahalanobis distance thus calculated is the smallest.

[0111] In the example shown on the [ Fig. 2 ], this is the candidate straight line 2D, which appears preferable as the vehicle is traveling towards the motorway exit.

[0112] As previously stated, the process could end here.

[0113] However, this process continues here in order to obtain even more reliable results, especially when the vehicle driver suddenly changes behavior (for example, ultimately staying on the highway and not getting off it).

[0114] To summarize, the plan is to estimate where the vehicle will be at a future time t' and what its dynamic characteristics will be, then to repeat part of the aforementioned steps in order to select one of the candidate lines (another one or the same one as the one selected in step E5).

[0115] The instant t' is separated from the instant t 0 by a duration between 0.5 and 2 seconds, for example equal to 1 or 1.3 seconds.

[0116] We can describe these different subsequent steps in more detail.

[0117] During a sixth step E6, the computer 11 uses a predictive model of the vehicle's movement in order to determine the kinematic data that the vehicle will present at time t'.

[0118] This model simplifies the calculations. Its purpose is to calculate the coefficients of the second vector v at time t' as well as those of the covariance matrix Sv.

[0119] The chosen model can be more or less simple and precise.

[0120] As an example, it is possible to use a constant curvature and zero acceleration evolution model, according to which the predicted trajectory of the vehicle is deduced from the velocity vector of this vehicle, considering that the magnitude of this velocity vector will not evolve in the next few seconds and that the direction of this velocity vector will continue to evolve at a constant yaw rate in the next few seconds.

[0121] Alternatively, it is possible to use a constant curvature and acceleration evolution model, according to which the predicted trajectory of the vehicle is deduced from the velocity vector and the acceleration vector of that vehicle.

[0122] Other variations would of course be conceivable.

[0123] Thanks to this modeling, the curvilinear abscissa 1 indicating the position of the motor vehicle at time t' is calculated for each candidate line.

[0124] The first vector u can thus be calculated at time t'. Its covariance matrix Su at time t' can also be calculated using a well-known algorithm for transferring uncertainty along a line.

[0125] We can only specify regarding this algorithm that at curvilinear abscissa 1, the covariance matrix can be deduced from the following equations: Σ l = B * Σ 0 * B T

[0126] In this equation, B is a matrix defined as follows: B = J 0 0 1

[0127] In this matrix B, J is the Jacobian of the function fl, which can be defined as follows: s l = x l y l psi l c l c 1 T = fl x 0 y 0 psi 0 c 0 c 1 T c 1 T

[0128] Note that the first vector u can also be recalculated taking into account the equation of the clothoid of the candidate line considered, as well as its covariance matrix Su.

[0129] During a sixth step E7, knowing the vectors u and v at time t' and the covariance matrices Su and Sv at that same time t', the Mahalanobis distances at that time t' are calculated and the candidate line with the minimum distance is selected and then stored in memory (as in steps E4 and E5).

[0130] At step E8, the candidate line selected at time t 0 (step E5) is compared by the computer 11 with the candidate line selected at time t' (step E7).

[0131] If it is the same candidate line, then this candidate line is definitively chosen as the central line to be considered (step E9).

[0132] Otherwise (for example because the driver starts to turn his steering wheel to the right and ultimately does not leave the motorway), a hybrid line is constructed and then definitively chosen as the central line to be considered (step E10).

[0133] This hybrid line can be constructed in various ways.

[0134] As an example, the results of the modeling used in step E6 make it possible to determine the position that the motor vehicle should probably occupy at time t' and the yaw angle that it should exhibit.

[0135] Knowing the current position of this vehicle and its yaw angle, it is possible to construct a continuous and continuously differentiable line passing through these two positions. This line can, for example, be modeled as a clothoid or in another form (typically polynomial).

[0136] It is then definitively chosen as the central line to be considered.

[0137] The center line from step E9 or E10 can then be used to control the vehicle's actuators so that it remains centered in its lane without driver intervention. It can also be used to implement other driver assistance functions, such as emergency braking, obstacle avoidance, overtaking, adaptive cruise control, and collision warning.

[0138] In addition, the center line can be displayed on the vehicle's display screen, superimposed on an already displayed image of the vehicle and its surroundings.

[0139] The present invention is in no way limited to the embodiment described and represented, but a person skilled in the art will be able to make any variation in accordance with the invention.

[0140] As an example, the method could be used to determine the shape of the centerline of a traffic lane adjacent to the one used by the motor vehicle.

[0141] The invention also applies to the field of robotics and to any other field in which an object must follow a particular trajectory on a road (this road being understood here in its most general sense, as a traffic area for the robot).

Claims

1. Method for detecting a central line of a lane (1) of a road taken by a motor vehicle (10), comprising steps of: - acquisition, by a sensor such as a camera, or a RADAR remote sensor, or a LIDAR remote sensor, of information characterizing two lane edge lines (1D, 1G), - acquisition of data relating to the position and / or the movement of the motor vehicle (10) on the road, - creation of at least two candidate central lines (2D, 2G, 2DG) located between the two lane edge lines (1D, 1G), during which an electronic and / or computer unit (11) calculates coefficients characterizing said at least two candidate central lines (2D, 2G, 2DG) on the basis of the acquired information, - selection of one of the candidate central lines (2D, 2G, 2DG) on the basis of a statistical distance calculated between the at least two candidate lines (2D, 2G, 2DG) and the motor vehicle (10), and - deduction of the central line of the lane (1) on the basis of the selected candidate central line (2D, 2G, 2DG).

2. Detection method according to the preceding claim, wherein the coefficients characterizing at least one of the candidate central lines (2D, 2G) are calculated solely on the basis of the information characterizing one of the two lane edge lines (1D, 1G) and the width of the lane (1).

3. Detection method according to one of the preceding claims, wherein the coefficients characterizing at least one of the candidate central lines (2DG) are calculated on the basis of the information characterizing the two lane edge lines (1D, 1G).

4. Detection method according to one of the preceding claims, wherein exactly three candidate central lines (2D, 2G, 2DG) are created.

5. Detection method according to one of the preceding claims, wherein: - a step of estimating data relating to the position and / or the movement that the motor vehicle (10) will have at a future time is provided, - a further step of selecting one of the candidate central lines (2D, 2G, 2DG) on the basis of the estimated data is provided, and - the central line of the lane (1) is deduced also on the basis of the candidate central line (2D, 2G, 2DG) selected in said further selection step.

6. Detection method according to the preceding claim, wherein, if the same candidate central line (2D, 2G, 2DG) has been selected twice in said selection step then in said further selection step, it is deduced, in the deduction step, that the central line of the lane (1) is the selected candidate central line (2D, 2G, 2DG).

7. Detection method according to one of the two preceding claims, wherein, if the two candidate central lines (2D, 2G, 2DG) selected in said selection step then in said further selection step are different, provision is made, in the deduction step, to construct the central line of the lane (1) on the basis of the coefficients characterizing the two selected candidate central lines (2D, 2G, 2DG).

8. Detection method according to one of the preceding claims, wherein the calculated distance between the motor vehicle (10) and each candidate central line (2D, 2G, 2DG) is a Mahalanobis distance, which depends on a covariance matrix associated with said candidate central line (2D, 2G, 2DG).

9. Motor vehicle (10) comprising means for acquiring information characterizing two lane edge lines (1D, 1G) of a road taken by the motor vehicle (10) and data relating to the position and / or the movement of the motor vehicle (10) on the road, characterized in that it comprises a computer (11) designed to implement a detection method according to one of the preceding claims.