CN114762011A - Vehicle-mounted driving assistance method and device for drawing lane crossing contour - Google Patents

Abstract

The invention relates to an on-board driving assistance device and method for a Vehicle (VEH), comprising: -a step of detecting a lane contour located in front of a reference point of the vehicle, -a step of generating a representation (C) of the detected lane contour, -a step of driving along a portion of the detected lane contour via a movement of the vehicle, -a step of projecting a Vehicle (VEH) reference point (O) onto the representation (C), -a step of mapping an over-lane contour according to the projected point (M').

Description

Vehicle-mounted driving assistance method and device for drawing lane crossing contour

The present invention relates to an on-vehicle driving assistance method and apparatus that generates a drawing line crossing a lane contour.

The lane profile on the ground is usually detected by sensors of the vehicle to assist its driving; however, these lane contours are not stored for later use.

In document WO 201468094, some historical information, such as curve records, is used to generate a view of an advanced driving assistance system.

In document US 10160281, a map of driving records is constructed from measurements of the roughness of the road along the route that has been taken, and this characteristic, which is representative of the road quality, is then used to anticipate the selection of suspension attitude to be applied to the suspension system.

These recordings of driving behaviour or road quality therefore do not allow locating the rear vehicle relative to the lane and predicting the behaviour of the rear vehicle, nor do they allow providing the driver with a representation of the rear environment without environmental perception means (such as a camera) behind the vehicle.

It is an object of the present invention to overcome at least some of the disadvantages of the prior art by providing an on-board driving assistance method for mapping an outline of an over-the-lane. Such a feature in fact allows to delimit lanes delimited by drawn lines crossing the lane outline and to associate detected objects with said lanes and to apply driving rules to these objects and/or to display said lanes and said objects on said lanes without a rear camera. In practice, for example, two rear-side radars provide information, but this information is not sufficiently reliable to qualitatively represent the environment behind the driver and to form a decision aid for them, given that false information is instead a source of danger. Further, radar can detect objects and lane edges, but cannot detect ground marker lines.

To this end, the present invention proposes an on-vehicle driving assistance method for a vehicle, the method including:

-a step of detecting a lane contour located in front of a reference point of the vehicle;

-a step of generating a representation of the detected contour;

-a step of defining a movement;

-a step of projecting a reference point of the vehicle onto the representation;

-a step of mapping said crossing profile according to the projected points;

this allows reconstructing the crossing profile based on available geometric and kinematic information without the need for additional sensors or connections to other vehicles.

By virtue of the invention, which is also applicable to autonomous vehicles, the detected lane contour allows a drawn line to be formed from the ground markings that have been passed over, without the need for a rear camera or any other rear sensor to detect the lane contour.

According to a particular embodiment, the lane profile is a ground mark and/or a lane edge.

According to an advantageous feature, the steps are carried out simultaneously on a plurality of lane profiles, which makes the method suitable for any type of traffic lane, even complex traffic lanes.

According to another advantageous feature, the representation of the contour is a clothoid, or a polynomial of order 3, or a series of polynomials of order 3 or clothoid splines.

According to another advantageous feature, the projection step comprises at least one iterative sub-step of searching for the projected point of the reference point of the vehicle on the representation, in particular by using a newton-raphson optimization method that optimizes the projection time.

According to another advantageous feature, the projection step comprises a sub-step of initializing the at least one iterative search sub-step starting from a first identification point of the representation of said contour.

Advantageously, the representation of the contour is a clothoid of length L, with a curvature of origin c₀And a rate of change of curvature of c₁And the first point is the center of the clothoid.

According to an advantageous feature, the projections are orthogonal, within a tolerance of about 5 °, which is easily identified by trigonometry.

According to another advantageous feature, the step of plotting comprises a sub-step of calculating intermediate points whose discretized distances are a function of the speed of the vehicle, which makes it possible to provide an optimized plotted line without low-value additional points.

According to another advantageous feature, the output of the step of drawing the profile of the crossing lane is a point cloud, which then allows the storage and management of a single memory.

According to another advantageous feature, the step of generating the representation of the detected contour comprises the sub-step of recalculating the length of the representation of the contour.

An advantage of including the step of storing at least some of the drawn lines of the crossing lane profile is that it enables the display of a route history to be provided.

According to an advantageous feature, the projection step comprises a sub-step of verifying said reference point of the vehicle projected onto the representation, which improves the reliability of the projection.

According to another advantageous feature, the rendering step comprises a sub-step aimed at geometrically ordering said intermediate points, which allows to store in memory only the most relevant points.

According to another advantageous feature, the method comprises a sub-step aimed at evaluating the error associated with the projected point by calculating the variance and the error propagation, which improves the reliability of the method.

Advantageously, the reference point of the vehicle corresponds to the centre of the line segment formed by the rear axle of the vehicle, which allows a better representation of the vehicle (in particular during lane changes) in order to avoid any false detection of a lane change.

Advantageously, the method comprises the step of displaying said drawn line crossing the contour on a screen.

According to another advantageous feature, the method comprises the step of associating the object with a lane delimited at least in part by said drawn line crossing the lane contour, thereby allowing the position of the object detected behind the vehicle to be displayed in relation to the displayed drawn lane.

The present invention also relates to an in-vehicle driving assistance apparatus including:

-means for detecting a lane contour located in front of a reference point of the vehicle, in particular a sensor, and more particularly a front camera;

-means for generating a representation of the contour;

-means for defining a movement;

-means for projecting a reference point of the vehicle onto the representation;

-means for mapping said crossing lane profile according to the projected points, such a device having similar advantages to those of the method.

The invention also relates to a motor vehicle comprising such a device, wherein the vehicle is in particular free of contour detection means located behind the reference point.

The invention also relates to a computer program comprising instructions for implementing the steps of such a method when said program is executed on a computer, the advantages of which are the same as those of the method.

Description of the drawings:

FIG. 1a illustrates the principle of the present invention at vehicle start-up; and

FIG. 1b illustrates the principle of the invention at the next instant after the move;

FIG. 2 shows a functional logic diagram of the method according to the invention;

FIG. 3 shows a schematic view of the method according to the invention;

FIG. 4 shows a logic diagram of the projection step of the method according to the invention;

FIG. 5 shows a diagram of the rendering steps of the method according to the invention;

fig. 6a shows a diagram of the sub-steps of the drawing step of the method according to the invention for a first example;

FIG. 6b shows a diagram of a drawing step of the method according to the invention for sub-steps of a second use case;

FIG. 7 shows a schematic view of a display using the method according to the invention;

fig. 8 shows a schematic view of the use of the driving assistance using the method according to the invention.

Throughout the text, directions and orientations are expressed with reference to the direct XYZ orthogonal coordinate system conventionally used in automotive design, where X denotes the longitudinal direction of the vehicle, pointing forward, Y is the direction transverse to the vehicle, pointing to the left, and Z is the vertical direction, pointing upwards. The angle θ is the yaw angle of the vehicle, which is conventionally positive when the vehicle is turning left and negative when the vehicle is turning right. The concepts of "front" and "rear" are provided with reference to the normal forward direction of travel of the vehicle. Throughout the description, the term "substantially" refers to minor deviations that may be allowed with respect to a determined nominal position or orientation, e.g., "substantially perpendicular" is intended to mean within the scope of the invention deviations of about 5 ° from a strictly perpendicular orientation are allowed. For purposes of clarity, the same reference numbers will be used throughout the drawings to identify the same or similar elements.

Vehicles equipped with driving assistance services, or even autonomous vehicles, receive data from a plurality of sensors. The present invention aims to optimize the use of data from vision sensors, i.e. sensors such as cameras, lidar or radar. More specifically, the invention derives therefrom a drawn line crossing the lane contour on the basis of information relating to the lane contour originating from these sensors. The lane profile may be a ground mark detected by a camera, in particular, or may even be a lane edge detected by a radar. Using data from a single sensor or using a perception algorithm that may combine or correlate data from different sensors, the vehicle obtains a representation of the contour of the lane in front of it. It is an object of the invention to derive from such information a representation of the contour of a lane that has been passed over and that falls behind the own vehicle, i.e. in particular a road marking or a lane edge, in order to improve the driving assistance service.

In fact, the information related to the passing lane contour will be used for the driving assistance method, in particular by associating an object located behind the own vehicle with a lane, which allows to apply driving rules, not only kinematic rules, to said object. These objects are detected, for example, by means of lateral and/or rear radars and/or 360 ° lidar located on the roof of the vehicle. Thus, for functions such as collision warning or trajectory prediction, for example, a coherent association of objects with lanes will allow for the prediction of the potential behavior of these objects (lane change, staying in the lane, etc.), which will then allow for a more efficient prediction of the likely trajectory and collision of the own vehicle with various objects around it.

In addition, the history of the lane profile is also useful when the vehicle is reversing, especially when the vehicle does not have a rear camera to provide information about lines located behind. Similarly, in the case of a vehicle equipped with such a sensor, but during the time when the sensor is temporarily unable to detect a line, the history of the lane profile may allow interpolation of a predetermined maximum distance.

Furthermore, in order to assist the driver's driving, it is advantageous to display a representation of the history of the lane profile on a human-machine interface.

Fig. 1a illustrates a starting point preferably identified by the starting time of the own vehicle VEH, from which the detection of the lane contour in front of the own vehicle VEH starts. More specifically, this involves detecting the lane contour located in front of a reference point O of the own vehicle, to which reference point the geometric coordinates (0,0) are assigned in the coordinate system of the own vehicle VEH. The reference point O associated with the own vehicle is preferably chosen as the center of the line segment formed by the rear axle of the own vehicle VEH and is shown in fig. 1a by a clear cross. This reference point O thus selected is more stable than the calculation of the rotation and vibration of the vehicle and avoids any false detections associated with lane changes in particular with respect to the reference point to be selected in front of the vehicle.

The sensors, preferably cameras located in front of and oriented forward of the own vehicle VEH, detect the lane contours via ground markings (and in this example, in particular lines located to the right of the own vehicle VEH). The raw data from the sensors may be in the form of points or cubic polynomials, and when several sensors are used, when their data correspond to the same element, these data are combined, and then the lane contour data are transformed to generate a clothoid C describing the shape of the detected line. A clothoid curve (also known as a crux or euler spiral) is a curve of a two-dimensional affine space. Clothoids are well known for use in the road field, as the geometric design of roads themselves uses these planar profiles. Thus, the road layout is usually designed and defined by clothoids. Clothoid curves also allow straight lines, right turns, left turns to be represented and for them to be easily placed one after the other.

Clothoid curves are defined by three elements:

-the arc length L of the clothoid curve;

curvature c at origin of clothoid₀(ii) a And

rate of change of curvature c of clothoid₁；

Wherein the clothoid curvature c (l) is proportional to its length I:

[Math.1]

c(l)＝c₀+c₁l

clothoid curve C is described as:

[Math.2]

C＝(M₀，Ψ₀，c₀，c₁，L)，

wherein:

-M₀is the origin of the clothoid curve C, having the geometric coordinates (x) in the coordinate system of the vehicle₀，y₀) (ii) a And is

-Ψ₀Is the tangent angle to the origin of the clothoid curve.

Along the clothoid, the orientation angle varies, and at length I it is equal to:

[Math.3]

and the coordinates of the points (x (l), y (l)) belonging to the clothoid curve are:

[Math.4]

and

the evaluation of the points (x (l), y (l)) on the clothoid curve, which will be done in the various steps of the method, can advantageously use fresnel integration as seen below.

In FIG. 1a, a clothoid C, shown as length L, has the projection of the reference point O on the clothoid C as the origin M₀In FIG. 1a, the origin is represented by the coordinates (x)₀,y₀) Shown as a solid cross.

Fig. 1b shows the time step after the first movement of the vehicle VEH after the plotting step (T). The Vehicle (VEH) thus follows an initial clothoid C, a portion of which has been crossed, this portion being defined by the original origin M of the initial clothoid C₀The dashed line between the orthogonal projections on the initial clothoid C from the reference point O of the own vehicle is shown. This projected point M ' (shown by a solid cross in FIG. 1b) has coordinates (x ' 0, y ' 0) and divides the clothoid into a length L ' in the initial clothoid C that has not been crossed by the reference point O and a line segment [ M0M ']Corresponds to the crossing length L-L', which corresponds to the crossing portion of the initial clothoid C. Thus, the updated clothoid C 'has a length L' that has not yet been crossed, and the crossed portion of the initial clothoid C is shown by a dashed line. In this case, the not yet mentioned yaw angle remains zero while the straight course is shown.

Fig. 2 shows the steps of the method according to the invention. As mentioned above, the method starts with a step D of detecting a lane contour (ground mark, such as a line, or a lane edge), wherein the detected data of the lane contour correspond to, for example, raw data, such as cubic polynomials, line segments and/or points associated with the ground mark and originating from a sensor, such as a camera. This step D of detection by the sensors, once the vehicle is started, is carried out continuously with an execution frequency specific to each sensor. Each polynomial associated with the contour of the lane may be associated with a characteristic field via a data structure that describes the polynomial by associating with the polynomial, for example, the quality of the received signal, the nature (continuous, discontinuous) of the detected ground markings (if the lane contour corresponds to ground markings).

The following steps are preferably performed by a dedicated computer, such as a vehicle computer supporting the driving assistance method. The step R of receiving the raw lane contour data from the sensors is performed in determined time steps, in particular greater than or equal to the response time of the slowest sensor, i.e. the duration (for example, a few milliseconds) required for the slowest time sensor to perform and deliver its detection result. This receiving step R preferably comprises a sub-step (not shown) of sorting in time sequence the data received from the various sensors for each lane profile and a sub-step of combining these data and associating an identifier with each detected profile. The format of the data output from this receiving step R remains the same as the format used for the input raw data.

The next step Δ of defining the movement involves determining whether the movement of the vehicle VEH between data reception and the current time is known (i.e. defined), which is to be understood as that even if the movement (along the x, y, z coordinates) is zero, it remains defined, and if this movement is defined, the method continues using this data, otherwise this data is eliminated in the step S of removing the lane contour data, because they are not located.

Then, a step G of generating a representation of the input data is performed, more particularly a step G of generating or updating a representation of the detected ground mark, in particular by transforming the raw data received in the preceding reception step R. Thus, each detected lane contour (preferably identified by its identifier) is thus preferably transmitted in the form of a polynomial and then converted in step G into a geometric shape representative of the detected ground mark, preferably into a clothoid C or a clothoid spline, or for example into a polynomial of order 3 or into a series of polynomials of order 3. Thus, each lane contour is represented by a single clothoid. However, the lane contour may also be modeled by clothoid splines (i.e., curves comprising a plurality of clothoids connected in series). The modeling method enables a better representation of the elongated lane contour. For example, when the invention is applied to a clothoid spline, when a clothoid segment that has just been traversed is identified, the entire clothoid may correspond to this segment. The present invention is described below with reference to the case of a single clothoid, which is a more basic case, because a clothoid spline is constituted only by a clothoid, and searching for a history from the clothoid spline means searching for a history of a lane contour represented by a single clothoid. The invention can also be applied to a lane contour modeled by a polynomial by replacing the equation precisely associated with the clothoid model with the reciprocal equation of the polynomial model. Preferably, each clothoid C associated with an identified lane contour polynomial retains an identifier of the detected contour and its characteristic fields.

This clothoid generation step G also comprises, when it concerns the updating of the clothoid C, a sub-step (not shown) of compensating the movement performed by the vehicle VEH and determined in the step Δ of defining the movement, so as to allow the clothoid C to be updated. Therefore, the characteristic M of the clothoid C is updated by taking into account the prediction of the movement of the own vehicle during the following steps₀(x₀，y₀)Ψ₀L, L: calculating dx along the x-axis, dy along the y-axis, and d θ for yaw angle by applying trigonometric rules, where x is₀Become (x)₀-dx).cos(θ)+(y₀-dy), sin (θ) and y₀Becomes- (x)₀-dx).sin(θ)+(y₀-dy), cos (θ) and Ψ₀Becomes Ψ₀-theta, and c₀，c₁It has been decided, in addition, that the length L of the clothoid curve remains constant in this sub-step.

Optionally, when the quality of the received signal forms part of the characteristics associated with the lane profile, an intermediate sub-step (not shown) of aggregating the input data from the camera and/or other sensors and/or aggregating the input data with memory points is added between the movement compensation sub-step (not shown) and the projection step P, this aggregation sub-step allowing to select the value with the lowest error rate and, if necessary, for example in the case of a momentary failure of a sensor or invalidity of the values received by a sensor (for example, when entering a tunnel when the lamp has not been turned on), to allow to select the clothoid from the memory by replacing the erroneous or invalid data of one or more sensors.

Then, a step P of projecting a reference point O of the own vehicle on each representation of the lane contour generated in the preceding step G (i.e. on the updated clothoid C) is started, wherein this step is explained using the following figure. Preferably, the projection of the points may be correlated with the speed of the own vehicle VEH, in order to reduce the frequency of high-speed sampling. In fact, projecting once per meter at a speed of 90km/h has no descriptive sense and would unnecessarily fill the memory. Therefore, the sampling rate associated with the projection step P is preferably speed dependent in order to avoid calculating unnecessary points.

If the projection in step P cannot be performed (e.g. the vehicle is not moving, the vehicle is moving too far and has crossed the entire representation of the lane contour without any continuity with the current, a trajectory change under the currently unknown lane contour, etc.) without being able to add a supplement to such a representation of the lane contour, a new lane contour representation, preferably with a new identifier, will start in the next iteration. Once the projection is complete, i.e. point M' is obtained, a step T of generating a drawn line of the representation of the crossing lane contour from the stored points is performed. Preferably, the drawn line is a point cloud, comprising stored projected points. After the first calculation step involving the movement of the vehicle VEH, there is therefore at least one stored projection point, i.e. a plot of a crossing lane contour (i.e. a plot of a crossing point), is at least one cloud consisting of two points, of which the first point is the origin M of the initial clothoid C₀And the second point is a point M' projected onto the initial clothoid C after the host vehicle moves, and the clothoid is updated at the same time.

Then, the step M of storing and updating the memory occurs sequentially or simultaneously with the drawing step T. This first new projected point is therefore recorded in the memory of the computer supporting the method, for example an ECU-ADAS (electronic control unit-advanced driver assistance system) driving assistance computer.

In the case that a plurality of lane contours (for example, the right lane boundary and the left lane boundary of the own vehicle VEH) are detected, the steps of the method are carried out simultaneously.

For each new cycle, at least one point is added to the drawn line thus completed in step T and to the memory in step M, considering that the memory is limited to, for example, 50 projected points by FIFO (first in first out) logic.

The execution of the method is organized, for example, in such a way that each cycle lasts for a predetermined maximum duration, for example 40ms, and the painted line is generated or updated in 40 ms. This execution rate enables the method to interact with other driving assistance methods, and in particular to provide a drawn line of a representation of a lane contour for other driving assistance services.

Fig. 3 shows a general flow chart using the method and more particularly of the communication between the output data of the previously described method and another driving assistance method, in particular for representing on a human-machine interface the association of an object detected behind the vehicle with the lane drawn in the drawing step T. In this example, the step E-0 of using at least one sensor (more specifically, a camera shot toward the front of the vehicle) to acquire data is taken as input. The received data is used in step D for detecting lane contours (ground markings such as lines or lane edges). Steps D to M of the method as described above are performed and are characterized using the dashed arrow between step D and the storing step M. At the same time, the data received from the sensors (camera, radar, etc.) in step E-0 may contain data associated with the object and result in the object detection step E-1 being performed. The sensors used to detect objects may be different from those used for lane contour detection and may even be located elsewhere in the vehicle. In step T, one or more drawn lines crossing the ground mark are generated in conjunction with the detection of the object, so that for each object detected behind the reference point of the host vehicle and belonging to one of the lanes delimited by the drawn representation of the crossing lane contour, it is possible to proceed to an object-lane association step E-2.

Fig. 4 shows a flow chart of a projection step P of the method, in which the detected lane contour is considered to be represented by the clothoid C generated in step G. This projection step P is relatively long and is time limited by a timer to exit the step after a duration of, for example, 7 milliseconds.

The newton-raphson method chosen for searching for the projected point uses an iterative method consisting of several interlocking loops. The initialization sub-step P _ ini of the method involves defining a first point (s0, i, where i ═ 0) from which the search will start for initializing the recursions belonging to clothoid C. Preferably, the first point is an identification point of the clothoid C, for example calculated using fresnel integration, which point corresponds to the centre of the clothoid in terms of length:

that is, s0, 1 equals L/2, and to start the recursion, n equals 1. For a better understanding of fig. 4, the sub-steps (P _ ini) are not shown in this preferred embodiment, but in such a way as to present a clearer sequence of i iterations, where s0, i, ═ iL/2; where i is between O and 2, and the first initialization value is i ═ 0.

In practice, this projection step P comprises an iterative step of searching for the orthogonal projection point of the reference point O associated with the vehicle VEH on the representation (in this case the clothoid C).

The search for a projection point then involves the actual length I: l ∈ R, the convolution curve C (l) ∈ R²In order to find L ∈ R, so that C (L ∈), or else relates to the orthogonal projection of a reference point O associated with the vehicle VEH on the representation (i.e. in the present case on the clothoid C).

To search for a projected point, the function f is defined as:

[Math.5]

f(l)＝<O-C(l)，C′(l)>。

namely:

[Math.6]

f(l)＝(O-x(l))*cosΨ_0，proj+(O-y(l))*sinΨ_0，proj。

the newton-raphson optimization method was then used to find the zero point for f. The principle of the newton-raphson method is based on the definition of a series of successive approximations of the equation f (x) 0, x being the abscissa from which the approximate solution is estimated₀And on the basis thereof by plotting a curve f at x₀The tangent at (x) 1, which intersects the abscissa axis, is calculated using the tangent equation y ═ f' (xn) (x-xn) + f (xn). When y is 0, the tangent line intersects the abscissa axis, and therefore when f '(xn) (x-xn) + f (xn) ═ 0, f' (xn) (x-xn) ═ f (xn) x-xn ═ f (xn) f '(xn), that is, x ═ xn-f (xn) f' (xn), is also indicated. This therefore yields the following recursive relationship: xn +1 ═ xn-f (xn) f' (xn). Thus, x1 is obtained, and the steps are repeated until the loop converges. In order to converge the loop, the derivative of f cannot be cancelled.

Thus, the derivative of f is calculated as:

[Math.7]

f′(t)＝<O-C(t)，C″(t)>-||C′(t)||²

=[(O-x(l)).(-sinΨ_0，proj)+(O-y(l)).cosΨ_0，proj]^-1。

this is generated using Math 1 to 4, where c (t) is the curvature of the clothoid at t, as previously defined:

[Math.8]

f′(t)＝<O-C(t)，(-sin(Ψ(t))，cos(Ψ(t)))*c(t)>-1。

thus, the search sub-step Pff 'corresponds to the block (brick) of the iterative sequence of the newton-raphson method in which f and f' are calculated, where n is initially set to n ═ 1, and where the clothoid C is first evaluated at said first identified point (i.e. the point of length s) where the search starts, in this case this point corresponding to s0, 0 at the first iteration, and then the coordinates of the clothoid at the point s are calculated: m is a group of₀And the angle of the clothoid at point s: psi_0，projPsi(s) toAnd rate of curvature change of the clothoid at point s: c. C_0，proj＝c₁(s); f(s) is then evaluated according to equation Math 6; and the derivative of f is evaluated according to the equation Math 8: f'(s).

In a next output diagnostic sub-step P _ S, the output conditions of the two cycles i, n are checked, i.e. whether the following inequality is satisfied

[Math.9]

|f′(s)|＜ε，

Where ε is equal to a predetermined threshold value close to zero, e.g., about 10^-7. If the output condition is satisfied, L ═ s.

The selected output condition may be replaced by another, for example:

[Math.10]

|f(s)|＜ε，

or even:

[Math.11]

furthermore, if the output condition is not satisfied, the method proceeds to step P _ ds, where a movement ds is calculated on the curve of the clothoid C according to the newton-raphson method:

[Math.12]

preferably, in the next merging sub-step P _ cons, the initial verification checks whether the new point corresponding to the length s + ds lies on the clothoid, i.e. (x (s + ds), y (s + ds)) ∈ C, which means that for positive ds, ds becomes ds: min (L-s, ds), while for negative ds, it becomes ds: max (-s, ds). Another condition may be verified in this step, in which if the absolute value of ds is less than a predetermined threshold, for example 1cm, the loop is exited without considering this ds, since it is considered meaningless, and therefore it is preferable to continue the search.

In the sub-step P _ max of comparing the current value of n with a predetermined iteration threshold Nmax (for example equal to 10, or preferably equal to 5), if n is less than said iteration threshold Nmax, the method continues, otherwise the loop is exited, which means that convergence is not obtained and a new initialization point (S) will be reached_0，iWhere i ═ 2) repeats these sub-steps, the new initialization point corresponding to (x (l), y (l)), i.e. the end of the clothoid curve, i.e. s_0，2L, and so on, and if this does not yield any result, then the third and last initialization point is the start of the clothoid: (x (0), y (0)), i.e., s_0，3＝0。

There is no restriction on the selection of these three initialization points, more or fewer points may be selected, or even different points may be selected, requiring only a single initialization point.

By continuing the method, i.e. when n < Nmax, the sub-step P _ s + ds _ cons of merging said obtained projected points s + ds is performed. To do this, s + ds is calculated and becomes s, and in order to decide whether to use the solution s, a verification is also made by forcing s to 0 when new s becomes negative and L when new s becomes greater than L, to check that this new s does not exceed the clothoid [ 0; l ], and n is incremented by 1.

Then, in sub-step P _ C0, the start of the clothoid is updated. To this end, the clothoid C is evaluated at length s according to the equation Math 4: c(s), namely: (x(s), y(s), Ψ(s), c1), and if the point is different from the previous starting point of the clothoid (x (s))₀，y₀，Ψ₀，c₀，c₁) It will take the replacement of the previous starting point as the new starting point (x) of the clothoid C_0，s，y_0，s，Ψ_0，s，c_0，s，c₁)。

In a next verification sub-step P _ val, a verification is performed to check whether a line segment formed by the connection of the reference point O of the host vehicle VEH and the point M' corresponding to the length s on the clothoid now forms a right angle with the clothoid.

Therefore, the method comprises the following steps:

-calculating the value of f(s) and if the value is close enough to zero, for example less than for example a first predetermined verification threshold of 10cm, retaining the solution s;

another alternative method (possibly also in combination with the comparison of f(s) with the first predetermined verification threshold) involves calculating the cosine at the point of solution s:

[Math.13]

and if the cosine is sufficiently close to zero, i.e. less than a second predetermined verification threshold, e.g. 0.05, the solution s is retained as a projection: l ═ s;

otherwise, the solution s will be rejected and the loop is exited by incrementing i to restart the loop at the new initialization point in sub-step P _ ini.

As shown in fig. 5, the plotting step T consumes the output of P _ C0, i.e. once the projection of the reference point O of the own vehicle on the initial clothoid C is obtained, while this last projected point M 'exists and belongs to the clothoid, this projected point M' with coordinates (x '0, y' 0) is defined as the new origin of the clothoid and the last point of the passing part of the historic ground mark. In a preferred embodiment, the step T of plotting the cross-lane profile beyond the projected point M 'and having a length dL-L' (see fig. 1b) comprises a sub-step of calculating an intermediate point in which the value of the clothoid C for a series of lengths ((i.e. the distance from the origin of the clothoid, dL-dL, dL-2dL, … dL-k dL, up to dL-k dL <0) is calculated as shown in fig. 5, for example using fresnel integration) with a discretization distance dL defined by the user, so in fig. 5 the points outside the clothoid dL-3dL <0 (x (dL-3dL), y (dL-3dL)) will not remain in this series of intermediate points. But it may also be a function of the speed of the own vehicle in order to reduce the frequency of high speed sampling, such as for example dl up to 5m at over 90km/h in the projection step P.

Preferably, the mapping step T comprises a sub-step aimed at geometrically ordering the obtained new set of projected points, for which a geometric comparison is performed to arrange them in a descending order, whether for the same lane contour identifier or without such an identifier comprising the series of intermediate points. For example, as shown in FIG. 6a, if the new set of projected points already has points from the passing point, i.e., at (x)₀,y₀) The latter points, then this means that "old transit point" and "new transit point" coexist in all these points: and then compares its last point with the first point in the "old pass points" (and more specifically, its second coordinate in a conventional cartesian coordinate system). For better understanding, points in the "old past points" are shown connected together by solid lines, while points in the "new" past points are shown connected together by dashed lines. For example, in FIG. 6a, when the last point (x (dL-2dL), y (dL-2dL)) in the "new pass point" is compared to the first point in the "old pass point", the old point (x) can be seen (x-2 dL)₀,y₀) Far behind (x (dL-2dL), y (dL-2dL)), "old transit points" are thus kept as a historical sequence: extension of "New pass Point". However, fig. 6b shows one of the old points: (x)_p1,y_p1) Not after (x (dL-2dL), y (dL-2dL)), so this one of the "old pass points" is eliminated, while the others are retained. These situations may occur, in particular, when reversing at a curve.

Preferably, the drawn line is in the form of a point cloud. Alternatively, another form of drawn line may be selected by connecting the registered points, for example in the form of a line.

Further optionally, as described above, a characteristic other than geometric coordinates is associated with each identified and stored clothoid and may be used to provide a different appearance for the drawn line depending on the characteristic. For example, the stored characteristic may be a property (e.g., continuous or discontinuous) of the ground mark represented by the clothoid at the projection point, as this characteristic may form part of the characteristic detected by the sensor and associated with the clothoid. Thus, if the drawn line is in the form of a line of two ground connection points, the properties of the line may be a function of the characteristics of the properties of the ground markers assigned to each projected point. This example is not limiting. The painted line may be virtual, i.e. it is not displayed on a Human Machine Interface (HMI), but is used as input data for the driving assistance algorithm, but it may also be displayed on an HMI intended to inform the driver, especially when they are backing up and the vehicle has no rear camera, and in this case it is preferably decided not to project on the passing point, but to keep the painted line and the memory unchanged.

In a subsequent step M of storing and updating the memory (or this step is performed simultaneously with the drawing step T), the new point s and the series of intermediate points are then stored, given that the maximum number of stored points is set to, for example, 50. In this way, the discretization as a function of speed allows a better optimization of the memory.

When the memory is full, the deletion of the past points furthest from the position of the own vehicle is deleted, this deletion being reliably performed thanks to the preparations already carried out in the sub-step of the drawing step T, which is intended to geometrically order said projected points.

Once these steps are completed, the method loops back to step G of generating or updating a clothoid, in which step the clothoid C is updated. In fact, once the projection of the reference point O of the host vehicle on the initial clothoid C is obtained, while this last projected point M 'exists and belongs to a clothoid, in the sub-step P _ C0, the coordinates are (x'₀,y’₀) This projected point M' of (a) is defined as the new origin of the clothoid and the last point of the historically passing portion of the ground mark. The new length L' is then preferably calculated in a step G of updating the clothoid, as indicated in the description of fig. 1 b. This sub-step of calculating the new length L' can also be performed at the end of the projection step (P), for example.

In addition, a sub-step not shown in the figures can also be performed, which is intended to evaluate the error associated with a new origin of the clothoid curve, corresponding to the projection point M'. Of course, once the projected point M' is obtained, this sub-step may be performed in another step. The evaluation of the error corresponds to the calculation of the variance Σ (l) for the new origin of the updated clothoid C. This step involves propagating the error of the previous initial point, with a variance of Σ (0), which results in:

[Math.14]

∑(l)＝J∑(0)J^T，

wherein J is a function f_l(x₀，y₀，ψ₀，c₀，c₁) Jacobian of (x (l), y (l), ψ (l), c (l)), where x (l), y (l), ψ (l), c (l) are provided by equations Math 1 to 4. Thus, according to the example of FIG. 1b, x'₀X (l) and y'₀Y (l). This variance Σ (l) is recorded as an additional characteristic of the updated clothoid C, in view of which variance at the origin based on the clothoid can be calculated along the clothoid. The updated clothoid C is then stored geometrically with its associated characteristics.

When the vehicle is stationary, the memory is empty.

The method is advantageously used for driving assistance applications.

FIG. 7 illustrates a use case in which a visualization of a history plot line is provided to the driver. The drivers of the own vehicle VEH can display the vehicles behind them on the HMI screen, each assigned to a lane previously identified during the execution of the method. For example, in this figure, three objects behind the VEH of the own vehicle, for example, the vehicles VEH1, VEH2, and VEH3 have been detected by radar. Information such as the position and/or relative distance of each object with respect to the own vehicle, its type, kinematic data such as speed and acceleration, the presence of directional lights (in this case illustrated by the grey areas in the figure), etc. form part of the characteristics detected and associated with each object.

This method allows the history of four ground markers to be plotted, defining three lanes: the lane of the vehicle VEH and the adjacent lanes on the two sides. For such use of the display for the driver, an association step is added in the method, followed by a step of displaying on the screen said drawn lines crossing the lane contour, which uses the stored data already ordered and which steps can be performed during the method (for example at the end of the storing step M) or outside the method by simply calling up the stored data.

In an additional associating step, the detected lane (if present) is associated with each object. For example, for each object belonging to a region delimited by the outer contour of the lane (defined by its drawn line crossing the ground markings), the step of associating the object in question with one of the lanes is carried out. At the end of this association step, the objects are then associated with the lanes in which they were detected; thus, the first object VEH1 is associated with the adjacent right lane of the lane in which the host vehicle is traveling, the second object VEH2 is associated with the same lane of the host vehicle VEH, and the third object VEH3 is associated with the adjacent left lane of the lane in which the host vehicle VEH is traveling. These associations are based on position and/or relative distance information acquired by the sensors.

The display step then consumes the output of the associated step, as well as the data stored in step M of the method, such display on the HMI will help the driver of the own vehicle VEH to adjust his trajectory (i.e. stay on the lane, lane change, etc.) and his speed according to the dynamic behaviour of other objects, even if they do not have a back-up camera. In fact, it is necessary to know the lanes where the other vehicles are located, in order to best predict their behaviour and make decisions, not only on the basis of kinematic information but also on the basis of semantic information (such as driving rules). Thus, in fig. 7, the own vehicle VEH knows that it has to adjust its speed, for example, so that the vehicle VEH1 located in the blind spot on its right-hand adjacent lane can fall behind the own vehicle.

Fig. 8 shows a use case in which an improvement is proposed to the on-board functions, enabling them to act automatically and enabling one vehicle to follow another at a suitable distance (ACC, adaptive cruise control). In this embodiment, an association step is added to the method, in which a detected lane is associated with each object (if present), as explained in the preceding paragraph. This association step can be performed, for example, simultaneously in the ACC computer, in particular when it is only executed for the ACC (i.e. without HMI display), which then calls the stored data of the method.

In fact, in the ACC function, the lateral and longitudinal control needs to know not only the target C _ ACC to be followed, but also information about objects VEH1, VEH2, VEH3 around the own vehicle VEH, which are allocated on key lanes (especially, the lane of the own vehicle and the adjacent two lanes). Information such as the position of each object, its type, kinematic data, the presence of directional lights (shown in the figure with grey areas) etc. will be used for the function of adjusting the trajectory and speed of the vehicle VEH according to the dynamic behavior of these objects. Further, depending on the context, the ACC function may replace the target to follow if the risk level associated with the first target is deemed to exceed the risk threshold of the host vehicle VEH (e.g., if the target C _ ACC is traveling slowly relative to the rear vehicle VEH2 in the lane of the host vehicle). Therefore, it is useful to know in which lane other vehicles are located in order to better predict their behavior not only from kinematic information but also from semantic information (especially in fig. 8, regarding VEH2 and C _ ACC).

These examples of application of the method according to the invention are illustrative and in no way limiting.

Such a method may be applied to automobiles for use and/or viewing of over-the-ground markings, but may also be applied in the field of robotics, for example.

Claims (12)

1. An on-board driving assistance method for a Vehicle (VEH), the method comprising:

-a step (D) of detecting a lane contour located in front of a reference point of the Vehicle (VEH);

-a step (G) of generating a representation (C) of the detected contour;

-a step (Δ) of defining a movement;

-a step (P) of projecting a reference point (O) of the vehicle onto the representation (C);

-a step (T) of drawing said crossing profile according to the projection points (M').

2. Method according to the preceding claim, wherein the representation (C) of the contour is a clothoid (C), or a polynomial of order 3, or a series of polynomials of order 3 or clothoid splines.

3. The method as claimed in any one of the preceding claims, wherein the projection step (P) comprises at least one iterative sub-step (P _ ff', P _ S, P _ ds _ cons, P _ S + ds _ cons), in particular by searching for a projected point of the reference point (O) of the Vehicle (VEH) on the representation (C) using a newton-raphson optimization method.

4. Method according to the preceding claim, wherein the projecting step (P) comprises a step of projecting from a first identified point(s) of the representation (C) of the contour_0,1) A sub-step (P _ ini) of initializing the at least one iterative search sub-step (P _ ff', P _ S, P _ ds _ cons, P _ S + ds _ cons) is started.

5. The method of any one of the preceding claims, wherein the projection is orthogonal.

6. A method as claimed in any one of the preceding claims, wherein the mapping step (T) comprises a sub-step of calculating intermediate points whose discretized distances are a function of the speed of the Vehicle (VEH).

7. The method of any one of the preceding claims, comprising a step (M) of storing at least one portion of the drawn line crossing the contour.

8. The method as claimed in any one of the preceding claims, wherein the reference point (O) of the Vehicle (VEH) corresponds to the center of a line segment formed by the rear axle of the Vehicle (VEH).

9. A method as claimed in any preceding claim, comprising the step of displaying the drawn line across the outline on a screen.

10. An in-vehicle driving assistance apparatus comprising:

-means for detecting a lane contour located in front of a reference point (O) of the Vehicle (VEH), in particular a sensor, and more particularly a front camera;

-means for generating a representation (C) of the contour;

-means for defining a movement;

characterized in that, this on-vehicle driving assistance apparatus further includes:

-means for projecting a reference point (O) of the Vehicle (VEH) onto the representation (C);

-means for mapping said crossing profile according to the projection points (M').

11. Motor Vehicle (VEH), characterized in that it comprises a device according to the preceding claim, said Vehicle (VEH) being in particular devoid of contour detection means located behind said reference point (O).

12. A computer program comprising instructions for carrying out the steps of the method according to any one of claims 1 to 9, when said program is executed on a computer.

Images