EP3065989A2

EP3065989A2 - Device for detecting the lateral position of a pedestrian relative to the trajectory of the vehicle

Info

Publication number: EP3065989A2
Application number: EP14825394.1A
Authority: EP
Inventors: Stéphanie AMBROISE
Original assignee: Renault SAS
Current assignee: Renault SAS
Priority date: 2013-11-04
Filing date: 2014-10-30
Publication date: 2016-09-14
Also published as: WO2015063422A3; WO2015063422A2; FR3012784A1; CN105830131A; FR3012784B1

Abstract

The invention proposes a system (1) for assisting the driving of a vehicle (2) comprising a camera (10) and comprising an optical recognition device capable of recognising an obstacle (3) in the field of vision (4) of the camera (10). The device (1) comprises a trajectory estimation unit (24) configured to define a right-hand trajectory boundary curve (5) and a left-hand trajectory boundary curve (6), which can be superposed onto the image of the camera (10), and comprises a lateral discrimination unit (25) configured to deliver at least two separate warning values, the warning value depending on the position of the image of the obstacle (3) relative to the right- and left-hand trajectory boundaries.

Description

Device for detecting the lateral position of a pedestrian in relation to the trajectory of the vehicle

The subject of the invention is the systems for assisting the driving of a vehicle, and more particularly the systems for assisting in the identification of obstacles or persons in the path of the vehicle. If the driver is helped in his driving decisions by the bright reflective floor markings on the roadway, he does not always have the time to analyze the nature of the objects or people on the road. other of the roadway, in particular in bad visibility situation as in time of fog or night.

Different night vision systems are available on a motor vehicle. Such systems can detect pedestrians at night and highlight them through different human-machine interfaces (HMI). The HMI can, for example, highlight the image area representing the pedestrian on a screen, or a pictogram symbolizing a pedestrian can appear on the dashboard. The US patent application 2009 023 11 16 of the Toyota company indicates the position of the pedestrian detected with respect to the vehicle by projecting with the aid of LEDs, on the windshield of the vehicle, which project light beams supposed to frame the image. pedestrian as seen by the driver.

The accuracy of this signaling is relative because the projected image of the pedestrian is not superimposed on the same location of the windshield depending on the size and position of the driver, and even to assume that the light beams surround the actual position of the driver. pedestrian, the driver must still analyze the situation, to estimate if the pedestrian will be to the right or left of the trajectory of the vehicle, and in which direction the driver will have to modify the trajectory of the vehicle. US Patent 2010 025 35 26 discloses a method of signaling an obj and or a pedestrian detected in front of the vehicle. The document proposes to indicate to the driver the direction of movement of the object and with respect to a direction transverse to the direction of the vehicle, but does not indicate the means by which the set of motion is determined.

The aim of the invention is to propose a driving assistance system which makes it possible not only to identify the position of a possible pedestrian with respect to the windshield of the vehicle or with respect to the digitized image on a screen, but which allows to give directly to the driver an information concerning the side towards which he will have to deviate his trajectory.

To this end, the invention proposes a device for assisting the driving of a vehicle comprising a camera and comprising an optical recognition device capable of recognizing an obstacle, in particular a pedestrian or an obstacle identified as potentially accidentogenic, in a vehicle. the field of view of the camera. The device includes a trajectory estimation unit configured to define a right boundary curve and a trailing edge boundary curve superimposable to the camera image, and includes a lateral discrimination unit configured to output at least two image values. The warning value is a function of the position of the obstacle image with respect to the right and left trajectories. The two distinct warning values can be two different values assigned, according to the position of the image of the obstacle, to the same variable in memory. The two warning values can, according to another variant of realization, correspond to two different memory variables, which can each take several values. According to a particular variant embodiment, the right and left limit curves can be confused, and correspond for example to the predictable trajectory of a central point of the front bumper of the vehicle. According to other embodiments, the right and left trajectories limits may for example correspond to traj ectoires predictable from the right and left edges of the vehicle, or may still correspond to these trajectories expanded a constant margin of safety. The trajectory estimation unit can determine globally each right or left limit curve as a function of the instantaneous angle of the wheels of the vehicle, or as a function of parameters making it possible to go back to this angle of the steering wheels. According to another embodiment, the trajectory estimation unit can determine each right or left limit curve taking into account recent variations in the angle of the wheels and extrapolating the variations of this angle of the wheels, for example to determine the portions of the trajectories most distant from the vehicle in the field of the camera. By obstacle or pedestrian is meant here an element detected by the image analysis system, with a particular signature, for example an infra-red type and / or a form factor signature and / or a typical motion type. , and / or a proximity or intersection with an area identified as the image of the roadway. The forms detected as pedestrians may for example include a standing human being, a cyclist or a person in a wheelchair, a dog, a deer, a boar, a tree leaning towards the road ..., or, as other variants , on the contrary, to selectively understand standing and / or moving human beings. According to some embodiments, the objects or pedestrians taken into account by the device may be limited to objects or pedestrians close to the roadway or objects or pedestrians encroaching on the roadway.

By feet of the pedestrian or obstacle, the following are the points of contact with the ground of the pedestrian or obstacle, or failing that, the bottom of the form detected, the "bottom" being understood according to the vertical axis of the image corresponding to the vertical lines of the filmed scene.

The device may be connected to an estimator of the angle of the steering wheels of the vehicle, and the trajectory estimating unit may be configured to define a right limit curve and a left limit curve of trajectory according to the angle of the steering wheels. The steering wheel angle estimator may for example be connected to a steering wheel angle sensor of the vehicle and to a driving angle estimator. the speed of the vehicle, and use the history of these two values to calculate the angle of the wheels. According to a variant embodiment, the right limit and the left trajectory limit can be determined according to the steering wheel angle, for example for a vehicle without power steering. The trajectories or portions of trajectories can be calculated, or preferably mapped, for example depending on the angle of the wheels, to allow a rapid analysis of the situation when the pedestrian is detected.

Preferably, the optical axis of the camera is at the time of estimation of the warning values, in a vertical plane comprising the direction of advance of the vehicle, the optical axis being oriented in a partially plunging direction . In a partially plunging direction, here is meant a direction directed towards the ground, without being perpendicular to the plane. In this way, a portion of the road in front of the vehicle is obtained in the image of the camera, as well as the elements bordering the road in the vicinity of the vehicle. In the image obtained, one of the coordinates of the image, typically a vertical ordinate of the image, can reflect both the distance of an object and in a horizontal plane in front of the vehicle, and the height of the image. an obj and s ensibly vertical. The other coordinate represents a dimension s horizontally horizontal, transverse to the traj ectoire of the vehicle.

The lateral discrimination unit may be configured to determine a right intersection point, which is calculated as an intersection of the right trajectorial boundary, with a foot line of the obstacle. The foot line is determined by the recognition system as bordering the image of the obstacle in a direction of the image corresponding to a horizontal axis transverse to the direction of advance of the vehicle. The lateral discrimination unit may further be configured to determine a left intersection point, as an intersection between the left trajectorial boundary and the same foot line of the obstacle. The lateral discrimination unit can for example calculate the abscissae of the two points of intersection, their ordinate being that of the foot line of the obstacle, therefore known.

The lateral discrimination unit may be configured to compare the abscissae of the right and left intersection points, with at least one abscissa characteristic of the detected obstacle. The abscis characteristic of the detected obstacle can correspond for example to the abscissa of a barycenter of the image of the obstacle. According to another variant embodiment, the lateral discrimination unit takes into account both a rightmost abscissa and an extreme left abscissa of the obstacle, the right and left extremum abscissae being determined by the recognition system such as those of FIG. two vertical lines bordering the image of the obstacle. Vertical lines are lines of the camera image corresponding to the vertical lines of the camera's field of view.

According to an alternative embodiment, the lateral discrimination unit calculates only a central line of trajectory instead of a right and left limit of trajectory, calculates an intersection of the foot line with the center line of the line. traj ectoire, and then compares the position of this inters ection with abscissa characteristics of the detected obstacle, by checking if this abscis is at a minimum distance to the right or left of the inters ection with the central line. This embodiment is in fact equivalent to that comparing the characteristic abscissa with the positions of the two intersection points along the trajectory to the right and left of the vehicle.

The lateral discrimination unit may, for example, emit a first signal if the left abscis is located to the right of the right intersection point, and emit a different signal if the right extremal abscissa is to the left of the point d left intersection. According to another variant embodiment, the lateral discrimination unit may for example emit a first signal if the right intersection point is between the abscis and its right and left extremals, and transmit a different signal if the point d The left intersection is between the right and left extreme abscissae. According to still In another variant embodiment, the lateral discrimination unit may emit a first signal if the left extreme abscissa is to the right of the right intersection point, at a distance less than one of the right intersection point. , and emit a different signal if the right extremal abscissa is to the left of the left intersection point, at a distance less than the threshold of the left intersection point. By right and left, relative positions are understood along the axis of the line of feet of the obstacle, that is to say along an axis of the camera corresponding to a horizontal direction perpendicular to the direction of vehicle advancement.

According to an advantageous embodiment, the lateral discrimination unit is configured to adapt the warning values as a function of the distance of the right intersection point with the abscissa characteristic of the obstacle, or as a function of the distance of the left intersection point with the characteristic abscissa of the obstacle. By distance from the point to the abscissa, we mean the distance from the point to the vertical line passing through this abscissa.

Preferably, the device comprises a warning interface configured to transmit, according to the values of the estimated warnings, light, tactile or audible signals, which are transmitted on a right side of the driver of the vehicle for at least one value. of the warning value, and which are issued on a driver's left side for at least one other value of the warning value. Thus, if the warning variable corresponds for example to a single variable that can take at least three distinct values, a signal is emitted on the left side of the driver when the variable acquires a first predefined value, a signal is transmitted on the right side. of the driver when the variable acquires a second predefined value, and no signal is transmitted as long as the variable is equal to a third predefined value. In the embodiments in which the warning values correspond to two variables each of which may take more values, a signal is emitted on the left side of the driver when a first of the two variables acquires a predefined value, and is transmitted on the right side of the driver when the second of the two variables acquires the predefined value. Several predefined values of the same variable can then cause signals of different intensity on the same side of the driver

According to an advantageous embodiment, the device comprises a camera capable of detecting images in the infrared range.

According to another aspect, the invention proposes a motor vehicle equipped with a driving assistance device as described above.

The invention furthermore proposes a method of assisting the driving of a motor vehicle, in which the positions of at least two lines of paths, determined in particular according to the steering angle of the vehicle, are compared in a common reference frame. vehicle, and at least one characteristic element of an image obtained by a camera pointing towards the front of the vehicle, and is sent to the driver of the vehicle, at least two different signals depending on whether the element is located to the right or left of the two lines of trajectory.

Other objects, features and advantages of the invention will become apparent on reading the following description, given solely by way of nonlimiting example and with reference to the appended drawings in which:

FIG 1 is a schematic representation of a vehicle equipped with a driver assistance device according to the invention,

FIG. 2 is a simplified representation of the taking into account of the geometrical elements used by a driver assistance device according to the invention, and

FIG. 3 is an example of a simplified algorithm for operating a driving assistance device according to the invention.

As illustrated in FIG. 1, a vehicle 2 rolls on a roadway limited by a straight edge of roadway 7 and by a left edge of roadway 8. The vehicle follows a path delimited by a straight edge of trajectory 5 and by an edge left of trajectory 6. In Figure 1, a pedestrian 3 is on the road, in the vicinity of the right edge of trajectory 5. The right and left limits of the trajectory may arbitrarily be defined for example as the trajectories of the outer casing of the trajectories of the wheels 9 of the vehicle, or may, depending on the variant embodiments, be defined as the outer casings of the trajectories of one or more points of the vehicle bodywork .

According to the variant embodiments, whether the trajectories are defined from trajectories of points of the wheels or trajectories of points of the vehicle body, it is possible to define an enlarged secure trajectory of arbitrary width with respect to the trajectories actually described. by physical points of the vehicle. It is considered in the rest of the text, that the trajectory of the vehicle is defined in the same plane as the roadway. The trajectory is therefore most often in a substantially horizontal plane, or at least in a plane parallel to a plane defined by the axes of the vehicle wheels and passing through the points of contact with the road of the vehicle wheels.

The vehicle comprises an electronic control unit 13, a man-machine interface 12, guide wheels 9, a camera 10, a wheel speed sensor 11, a steering wheel 14.

The electronic control unit 13 comprises a trajectory estimation unit 24 and a lateral discrimination unit 25. The trajectory estimation unit 24 is connected to or comprises a map 15 and is connected either to an estimator of angle ψ of the vehicle wheel is directly to a steering wheel angle sensor which transmits a steering angle φ value and a linear estimator of the vehicle speed, for example to a wheel speed sensor 11 of the vehicle.

The lateral discrimination unit 25 is linked to the man-machine interface 12, to the trajectory estimation unit 24 and to an image analysis system receiving the images transmitted by the camera 10.

The camera 10 is arranged relative to the vehicle 2 so as to be able to film towards the front of the vehicle. Typically, the camera 10 receives the light rays in visible light, in light infrared or in other wavelengths from a cone 4 of light beam, representing the field of view of the camera. At least a portion of the light directions of the cone 4 are diagonally oriented towards the front of the vehicle, so as to record in the image of the camera, a portion of the roadway located towards the front of the vehicle .

The camera 10 transmits the stored image values to an electronic control unit 13. The electronic control unit 13 also receives information from the steering wheel 14 of the vehicle, for example in the form of a steering wheel angle data φ. It receives a value of linear speed of forward v of the vehicle from, for example, the wheel speed sensor 11 of the vehicle. From the steering angle of the vehicle, the electronic control unit 13 is able to calculate the angle of the steering wheels of the vehicle relative to the longitudinal axis of the vehicle.

According to certain embodiments, the electronic control unit 13 can directly receive the information concerning the angle of the steering wheels of the vehicle instead of the steering angle φ and the forward speed v.

The electronic control unit 13 is connected to the map 15 making it possible, from the angle made by the wheels with the longitudinal axis Z of the vehicle, to construct two theoretical limits of trajectories superimposable on the image of the camera 10 By convention, in the image illustrated in FIG. 1, Z is called the axis corresponding to the usual vehicle advancement direction when the vehicle is traveling in a straight line, Y is referred to as the vertical direction when the vehicle is traveling on level ground or else Y is the direction perpendicular to the mean plane defined by the axes of the wheels and is called X the transverse direction of the vehicle, that is to say a horizontal direction perpendicular to the usual direction of advance of the vehicle.

FIG. 2 schematically illustrates an image captured by the camera 10 of FIG. 1, in which curves 19 and 20 representing the left and right limits 5 of the FIG. trajectory as they could project if they were visible in the image digitized by the camera 10. The image 16 of the camera 10 is represented in an x axis system, y. The direction of the x-axis, or x-axis, corresponds to the direction of lines which could be visible by the camera at the front of the vehicle and which would be oriented in the transverse direction X of the vehicle.

The y direction, or ordinate axis, corresponds to the other axis of the camera image perpendicular to the x direction, and corresponds to both the projection directions of the vertical lines seen by the camera in front of the vehicle and the lines parallel to the direction of advance Z of the vehicle. The image 16 illustrated in FIG. 2 is an image that has already undergone a series of filtering and analysis by known image analysis methods. These filtering and analyzes make it possible in particular to determine a contour 17 of the image of the pedestrian 3 shown in FIG. 1, and possibly lines (not shown in FIG. 2) corresponding to the physical limits of the roadway, and / or lines corresponding to the markings on the roadway and on its edges.

The pedestrian 3 of Figure 1 can be detected for example as a particular contrast element and / or form factor. Thus, the contour 17 of the image of the pedestrian 3 has been defined in the image 16.

The image analysis system has also defined an outside contour of the pedestrian, here in the form of a rectangle 18, which makes it possible to define a left side xi and a right side x ₂ of the contour 17 of the silhouette of the pedestrian 3. The left side corresponds for example to the coordinate on the left side of the rectangle 18 and the right side corresponds to the coordinate on the right side of the rectangle 18. It is possible to envisage embodiments in which the silhouette 17 of the pedestrian would be associated not with an outline 18 but for example to an ellipsoidal contour, the left and right limit coordinates x1 and x2 can then be defined as the abscissa of the ellipse points lying furthest to the x-axis and being the most right along the x-axis. The lowest point of frame 18 allows define a foot line 23 of the pedestrian silhouette 17. The foot line 23 may for example be defined as the line parallel to the x axis and passing through the lowest point, along the y axis, from the frame 18.

The ordinate yi of this line of feet 23 of the pedestrian 17 is here reported on the y-axis. For the understanding of the invention, two curves 19 and 20 have been reported on the filtered analyzed image 16. The curve 19 represents a left-hand trajectory edge of the vehicle 2 and the curve 20 represents a right-hand trajectory edge of the vehicle 2. To implement the invention, it is not necessary to display at any time on a screen the filtered analyzed image, or to superimpose lines 19 and 20 thereon.

However, the invention requires an analysis of the image detected by the camera to identify the possible pedestrian and also requires knowing the coordinates that would have in this image of the camera the right and left edges of the vehicle trajectory , which are in Figure 2 shown respectively by the curves 20 and 19. The curves 20 and 19 may for example be stored in the form of mapping 15 as a first function "edge R" and a second function "edge L". These functions corresponding to the curves 20 and 19 are mapped for example along the x-axis, as a function of a steering wheel angle ψ, and as a function of an ordinate y in the plane of FIG. 16.

According to the variants, the mapping can take into account the angle of the steered wheels, according to other embodiments, the map can directly take into account a vehicle speed and a steering wheel angle φ of the vehicle. According to still other embodiments, a unit for calculating the angle of the steered wheels calculates this angle ψ of the steered wheels from a history of speeds v of the vehicle and a history of angles φ at the wheel of the vehicle. The lateral discrimination unit 25 analyzes whether the pedestrian silhouette 17 is to the right or left of the path delimited by the lines 20 and 20, and, depending on the position of the silhouette 17, sends to the driver of the vehicle a signal indicating a pedestrian is to the right of the vehicle's path or a pedestrian is to the left of the vehicle's path, possibly allowing the driver to deviate slightly from the path of the vehicle even before he has actually seen the pedestrian.

The signal may be for example a light signal, a sound signal, a vibrating signal sent depending on the position of the pedestrian, either on the right side of the driver, or on the left side of the driver, thus making the interpretation of the signal more intuitive, especially if the driver does not know where his right side and his left side are. To determine whether the pedestrian 3 is to the right or left of the path of the vehicle delimited by the lines 5 and 6, the discrimination unit 25 determines for example the coordinates of the intersection points-respectively 21 and 22- of the feet line 23 of the pedestrian, with the curves 19 and 20 respectively representing the left edge of the vehicle trajectory and the right trajectory edge of the vehicle.

If the silhouette 17 or its outline 18 are to the right of the point 22 or straddling this point 22, the discrimination unit 25 sends a signal that is transmitted to the right of the driver of the vehicle. If the silhouette 17 or its outline 18 are to the left of the point 21 or straddling this point 21, the discrimination unit 25 sends to the HMI 12 a signal that is emitted to the left of the driver of the vehicle.

FIG. 3 illustrates an example of an operating algorithm that can be used by the discrimination unit 25 to determine whether the pedestrian 3 is to the right or to the left of the path of the vehicle 2. In the particular case of corresponding embodiment in FIG. 3, the discriminating unit 25 generates WarnR and WarnL variables which are 0 when no lateral detection signal is to be sent to the driver, which is 1 when the pedestrian is at the right at the same time. outside the trajectory, either to the left outside the trajectory of the vehicle, and which are worth 2 when the silhouette of the pedestrian straddles the left edge 19 or overlaps the right edge 20 of the trajectory. A WarnM variable can also be used, which is for example maintained at zero as long as the pedestrian is not in the path of the vehicle, and which is brought to the value 1 for example if the discrimination unit 25 detects that the pedestrian is in the path of the vehicle. Depending on the value of the WarnM variable, an emergency strategy can be set up, for example a braking strategy. The discrimination unit may for example cause the transmission of a first signal on the right of the driver when the WarnR variable is equal to 1, and cause the transmission of a stronger signal always to the right of the driver when the signal WarnR is equal to 2. Similarly, the discrimination unit 25 can cause the transmission of a signal to the driver's left when the variable WarnL is equal to 1, and cause the transmission of a stronger signal when the variable WarnL is equal to 2.

In cases where the discrimination unit detects that the silhouette of the pedestrian is between the lines 19 and 20, that is to say that the pedestrian is in the path of the vehicle, other strategies are put in place. place which are not detailed here and which may for example include an automatic braking of the vehicle if in addition another front sensor of the vehicle detects that the pedestrian is in the immediate vicinity of the vehicle.

As illustrated in FIG. 3, at an initialization step, the WarnR, WarnL and WarnM variables are set to zero.

In a step 32, the analysis system of the camera 10 performs a test to decide whether a pedestrian 3 is visible in the field of the camera. If the result is negative, it returns to the initialization step 31. If the result is positive, in a step 33 the image analysis system determines the extreme positions xi, x ₂ , along the x axis of the X, the silhouette of the pedestrian 17 and determines the low position yi along the y axis of the silhouette 17 of the pedestrian.

A series of tests 34 to 38 or 34 to 37 are then conducted to determine the position of the pedestrian with respect to the path of the vehicle. The test 34 checks whether the silhouette 17 of the pedestrian is to the left of the line 9 representing the left edge of trajectory. If the answer is positive, the variable WarnL is brought to the value 1 at a step 41.

If the answer is negative, we test at step 35 if the silhouette 17 is straddling the line 19. If it is the case, we bring the value WarnL to 2 at a step 42. If it is not no longer the case, at a step 36 is tested whether the silhouette 17 of the pedestrian, at its line of feet, is between the lines 19 and 20 that is to say if the pedestrian is in the path of the vehicle. If this is the case, the WarnM variable is switched to the value 1 at a step 39.

In the case of a negative answer to the test 36, a test 37 is made to check whether the pedestrian, at his line of feet, straddles the line 20 representing the right trajectory edge, and if this is the case, brings the WarnR variable to a step 43.

If the answer is negative, it is possible to carry out the test 38, or to conclude directly that the pedestrian, at his line of feet, is to the right of the right trajectory edge. In this case, one activates in a step 40 the variable WarnR to 1.

The operating mode described in FIG. 3 is only one of the possible modes of operation according to the invention. For example, it is possible to activate a signal only when the pedestrian is completely to the left or completely to the right of the trajectory, and to activate emergency procedures as soon as the pedestrian engages in the path of the vehicle.

In this case, when the silhouette 17 straddles one of the lines 19 and 20, an emergency strategy is already in place. It is also possible to envisage variant embodiments in which the position is not compared with respect to two trajectory edge lines, but with respect to four lines. A first line directly borders the trajectory on the right and a second line borders the trajectory a little to the right with a slight difference. Two other lines define a band to the left of the trajectory. The warning signal is then delivered when the portion of the pedestrian foot line is located either inside the left band or inside the right band bordering the theoretical trajectory of the vehicle. The invention is not limited to the embodiments described and can be declined in many variants. The coordinates of the left and right edge lines 19 and 20 of the trajectory can be calculated instead of being mapped.

The positions of the trajectory lines 19 and 20 may be chosen so as to correspond to an enlarged trajectory of the vehicle 2, that is to say to draw on the image if they are superimposed on the image of the camera a trajectory wider than that traveled by physical points of the vehicle.

For example, it is possible to widen a trajectory corresponding to lateral physical points of the vehicle by a few tens of centimeters, for example of a width of the order of 20 cm.

The driver assistance system according to the invention allows the driver to react in case of poor visibility even if he has not had time to identify himself the pedestrian at the edge of the road.

Claims

1. System (1) for assisting the driving of a vehicle (2) comprising a camera (10) and comprising an optical recognition device capable of recognizing an obstacle (3), in particular a pedestrian or an obstacle identified as potentially accidentogenic in the field of view (4) of the camera (10), characterized in that the device (1) comprises a trajectory estimation unit (24) configured to define a right limit curve (20) and a left boundary curve (19) of trajectory superimposed on the image (16) of the camera (10), and comprises a lateral discrimination unit (25) configured to deliver at least two distinct warning values (WarnL , WarnR), the value of the warning being a function of the position of the image (17) of the obstacle (3) with respect to the right (20) and left (19) trajectories.

A driving assistance system according to claim 1, connected to an estimator (13) of the angle (ψ) of the vehicle steered wheels, wherein the trajectory estimating unit (24) is configured to define a right limit curve (20) and a left limit curve (1 9) of trajectory according to the angle (ψ) of the steering wheels (9) of the vehicle.

A driving aid system according to any one of claims 1 or 2, wherein the optical axis of the camera is at the time of estimation of warning values (WarnL, WarnR), in a vertical plane (YZ) comprising the direction of advance of the vehicle (2), the optical axis being oriented in a partially plunging direction.

The driving assistance system according to claim 3, wherein the lateral discrimination unit (25) is configured to determine a right intersection point (22), such as an intersection of the right-of-way boundary ( 20), with a foot line (23) of the obstacle, the foot line being determined by the recognition system as bordering the image (17) of the obstacle following a direction (x) of the image corresponding to a horizontal axis (X) transverse to the vehicle traveling direction (Z), and the lateral discrimination unit (25) being configured to determine a left intersection point (2 1), as intersection between the left limit of trajectory (19) and the same foot line (23) of the obstacle.

The driving assistance system according to claim 4, wherein the lateral discrimination unit is configured to compare the abscis s (edge R (ψ, yi), edge L (ψ, yi)) of the right intersection points. (22) and left (2 1), with at least one abscissa (xl, x2) characteristic of the obstacle (3) detected.

The driving assistance system according to claim 5, wherein the lateral discrimination unit (25) is configured to adapt the warning values (WarnL, WarnR) according to the distance from the intersection point. right (22) with the abscissa characteristic of the obstacle, or as a function of the distance from the left intersection point (1 9) with the abscissa e (x *, x ₂ ) characteristic of the obstacle.

A driver assistance system according to any one of claims 1 to 6, comprising an alarm interface (12) configured to transmit, based on the estimated warning values (WarnL, WarnR), signals luminous, tactile or audible, which are transmitted on a right side of the driver of the vehicle for at least one value (WarnR = 1) of the warning value, and which have issued on a left side of the driver for at least one other value (WarnL = 1) of the warning value.

8. Driving assistance system according to any one of claims 1 to 7, comprising a camera (10) capable of detecting images in the infrared range.

9. Motor vehicle equipped with a driving assistance system according to any one of claims 1 to 7.

10. A method of assisting the driving of a motor vehicle (2), in which the positions of at least two predetermined trajectory lines (19, 20) are compared in a common coordinate system (x, y). in particular as a function of the steering angle (φ) of the vehicle, and minus one characteristic element (17) of an image obtained by a camera (10) pointing towards the front of the vehicle (2), and the driver of the vehicle is sent at least two different signals depending on whether the element is located to the right or to the left of the two trajectory lines (19, 20).