Assisting the driving of an automotive vehicle when approaching a speed breaker

Abstract

A driving assistance method for a motor vehicle (1) when approaching a speed bump comprises, according to the invention: detecting and tracking at least one other moving vehicle (4.sub.1) in front of the motor vehicle (1) based on processing images captured by a camera (10) on board the motor vehicle (1); establishing a temporal profile of the estimated distance between the motor vehicle (1) and the at least one detected and followed other moving vehicle (4.sub.1); detecting an anomaly area in the temporal profile; and estimating a distance d.sub.bump between the motor vehicle (1) and a speed bump (3) on the basis of the estimated distance between the motor vehicle (1) and the at least one other vehicle (4.sub.1) at a time separate from the times corresponding to the detected anomaly area.

Claims

1. A driving assistance method for a motor vehicle when approaching a speed bump, comprising: detecting and tracking at least one other moving vehicle in front of the motor vehicle based on processing images captured by a camera on board said motor vehicle; establishing a temporal profile of the estimated distance between said motor vehicle and said at least one detected and followed other moving vehicle; detecting an anomaly area in said temporal profile; and estimating a distance d.sub.bump between said motor vehicle and a speed bump on the basis of a distance between the motor vehicle and said at least one other vehicle, estimated at a time t.sub.bump separate from the times corresponding to the detected anomaly area.

2. The method as claimed in claim 1, wherein the distance d.sub.bump is estimated using the relationship
d.sub.bump=D(t.sub.bump)−d.sub.parc wherein D(t.sub.bump) is the distance between the motor vehicle and said at least one other vehicle estimated at said time t.sub.bump, and d.sub.parc corresponds to the distance covered by the motor vehicle starting from said time t.sub.bump.

3. The method as claimed in claim 2, wherein the detection and tracking step comprises detecting and tracking at least one other vehicle traveling in the same direction as said motor vehicle, and wherein said time t.sub.bump is chosen before a time corresponding to the start of the detected anomaly area.

4. The method as claimed in claim 2, wherein the detection and tracking step comprises detecting and tracking at least one other vehicle traveling in the opposite direction to said motor vehicle, and wherein said time t.sub.bump is chosen after a time corresponding to the end of the detected anomaly area.

5. The method as claimed in claim 1, further comprising: correcting a position of said motor vehicle with respect to a high-definition geographical map on board said motor vehicle on the basis of a position of a marker corresponding to said speed bump and pre-recorded in said map, and of the estimated distance d.sub.bump.

6. The method as claimed in claim 5, wherein an elevational movement of said motor vehicle is estimated on the basis of an inertial sensor on board said motor vehicle.

7. The method as claimed in claim 5, wherein an elevational movement of said motor vehicle is estimated on the basis of analyzing optical flows over a plurality of successive image portions captured by said on-board camera.

8. The method as claimed in claim 1, further comprising: estimating an elevational movement of said motor vehicle, and wherein said distance d.sub.bump is estimated only if no elevational movement greater than a predefined threshold and concomitant with a time corresponding to the detected anomaly area is estimated.

9. A driving assistance system for a motor vehicle when approaching a speed bump, comprising an on-board processing module configured to perform the method as claimed in claim 1.

Description

(1) The invention will be better understood upon reading the following detailed description, given with reference to the appended figures, in which:

(2) FIG. 1 schematically illustrates the principle of the invention on the basis of plan views (a) to (d) corresponding to four successive times of a situation of a motor vehicle approaching a speed bump;

(3) FIG. 2 shows a temporal profile of estimated distances between the motor vehicle and another detected vehicle in the situation illustrated in FIG. 1;

(4) FIG. 3 shows a sequence of steps able to be implemented in a driving assistance system according to the invention.

(5) Hereinafter and with reference to FIG. 1, it is assumed by way of non-limiting example that a motor vehicle 1 is moving on a road 2 and is approaching a speed bump 3 situated on the road 2. In the road situation that is shown, another vehicle 4.sub.1, hereinafter called other vehicle, is moving on the road 2 in front of the motor vehicle 1, in the example in the same direction as the motor vehicle 1. In views (a) and (b) of FIG. 1 corresponding to two successive times t.sub.1 and t.sub.2, the two vehicles 1 and 4.sub.1 are situated before the speed bump 3. In the following view (c) corresponding to a time t.sub.3, the other vehicle 4.sub.1 is currently driving over the speed bump 3. Lastly, view (d) corresponds to a time t.sub.4 at which the other vehicle 4.sub.1 has already passed the speed bump 3, whereas the motor vehicle 1 is still before this speed bump.

(6) The principle of the invention is based on the fact that it is possible for the motor vehicle 1, as will now be explained, to deduce the presence of the speed bump 3 and to dynamically estimate the distance D.sub.BUMP therefrom on the basis of the detection of the other vehicle 4.sub.1, performed through image processing.

(7) To this end, the motor vehicle 1 is equipped with a camera 10 with known calibration parameters and able to capture successive images of scenery. The camera 10 is preferably located at a position in the vehicle that best corresponds to what the driver sees, for example centered in the windshield inside the passenger compartment. The motor vehicle 1 furthermore comprises a processing module 11 forming, with the camera 10, a system for detecting in particular the presence, ahead of the motor vehicle 1, of other vehicles, such as the other moving vehicle 4.sub.1. The front camera 10 thus captures the images of the road scene situated in front of the vehicle 1 and provides these images to the image processing module 11 of the system.

(8) For the road situation illustrated in FIG. 1, the image processing module 11 will be able to detect and track the other moving vehicle 4.sub.1 in a step referenced S.sub.1 in FIG. 3.

(9) In this respect, it is recalled that any other vehicle detected through image processing delivered by a single camera is generally delivered in the form of a surrounding box that defines an image area representative of the detected vehicle. This surrounding box has the general shape of a rectangle, in particular with a horizontal lower border or lower limit, and an upper horizontal border or upper limit. A 3D position of the detected obstacle is then estimated, generally using the lower limit of the corresponding surrounding box. The estimation is conventionally based on what is called the flat world scenario, in which the detected vehicles move only over a single horizontal plane, and use the intrinsic calibration parameters (focal length, pixel size) and extrinsic calibration parameters (line of sight angle with respect to the horizontal) of the camera 10 to estimate the distance D between the motor vehicle 1 and a detected vehicle on the basis of the vertical position of the lower limit of the associated surrounding box.

(10) The module 11 will then conventionally have to establish (step S.sub.2 in FIG. 3), for each detected and tracked other vehicle (in this case the other vehicle 4.sub.1 for the road situation in FIG. 1), a temporal profile of the distance D between the motor vehicle 1 and the other vehicle. In FIG. 1, D(t.sub.1) thus denotes the distance between the two vehicles 1 and 4.sub.1 in view (a), D(t.sub.2) denotes the distance between the two vehicles 1 and 4.sub.1 in view (b), D(t.sub.3) denotes the distance between the two vehicle 1 and 4.sub.1 in view (c), and D(t.sub.4) denotes the distance between the two vehicles 1 and 4.sub.1 in view (b).

(11) One example of a temporal profile P.sub.1 of the distances between the motor vehicle 1 and the detected other vehicle 4.sub.1 in the specific road configuration of FIG. 1 is shown in FIG. 2.

(12) It is observed that the profile P.sub.1 contains a part in which the distance is substantially constant, reflecting the fact that the motor vehicle 1 and the detected other vehicle 4.sub.1 are traveling in the same direction, at substantially the same speed. The profile P.sub.1 furthermore contains an anomaly area A around the time t.sub.3 at which the other vehicle 4.sub.1 drives over the speed bump 3. This anomaly is linked to the flat world scenario used in the estimation of the distance between the motor vehicle 1 and the detected other vehicle 4.sub.1. Specifically, between the time at which the other vehicle 4.sub.1 begins to travel over the speed bump 3 and the time when the other vehicle 4.sub.1 leaves this speed bump 3, elevational differences of the other vehicle 4.sub.1 occur that are not taken into account in the conventional estimation of the distance between this other vehicle 4.sub.1 and the motor vehicle 1.

(13) This anomaly is reflected in a rapid increase in the estimated distance, followed by a rapid decrease.

(14) According to the invention, the module 11 will detect (step S.sub.3 in FIG. 3) the anomaly area A in the temporal profile.

(15) To this end, the module 11 may store the distances estimated for the other vehicle 4.sub.1 over a sliding window, for example of the order of 10 seconds, and then calculate statistical operators over the sliding window, such as the average and variance, which will increase and then decrease within a very short time corresponding to the anomaly area A.

(16) As a variant, the anomaly area A may be detected using a machine learning approach by applying a support vector machine (SVM) algorithm using a learning base in which examples of signatures specific to the anomaly areas are stored.

(17) In one preferred embodiment, it is ensured at the same time that a detected anomaly area actually corresponds to an elevational movement linked to the other vehicle 4.sub.1 and not to an elevational movement linked to the motor vehicle 1. To this end, there is advantageously provision that the module 11 implements a step (not shown) of estimating an elevational movement of the motor vehicle 1, and that said distance d.sub.bump is estimated only if no elevational movement greater than a predefined threshold and concomitant with the time corresponding to the detected anomaly area A is estimated.

(18) The elevational movement of the motor vehicle 1 may for example be estimated on the basis of an inertial sensor on board the motor vehicle 1. As a variant, the elevational movement of the motor vehicle 1 is estimated on the basis of analyzing optical flows over a plurality of successive image portions captured by the on-board camera 10. In the latter case, it is considered that it is the motor vehicle 1 that undergoes an elevational movement when the various analyzed optical flows have the same variations.

(19) Once the anomaly area A has been detected, it is then possible for the motor vehicle 1 firstly to conclude as to the presence of the speed bump 3 and secondly to dynamically estimate (step S.sub.4 in FIG. 4) the distance d.sub.bump between the motor vehicle 1 and this speed bump 3 by using a distance between the motor vehicle 1 and the other vehicle 4.sub.1 estimated at a time t.sub.bump separate from the times corresponding to the detected anomaly area.

(20) An estimation of the distance d.sub.bump between the motor vehicle 1 and the speed bump 3 may in particular be calculated using the following relationship:
d.sub.bump=D(t.sub.bump)−d.sub.parc

(21) wherein: D(t.sub.bump) is the distance between the motor vehicle 1 and the other vehicle 4.sub.1 as estimated at the time t.sub.bump, and d.sub.parc corresponds to the distance covered by the motor vehicle 1 starting from said time t.sub.bump.

(22) It is made possible to calculate the covered distance by saving the state of the vehicle 1 (in particular its longitudinal speed and its yaw speed) in the sliding window of 10 seconds, or by saving the data of the mileage counter of the motor vehicle 1.

(23) In the case of the road situation in FIG. 1, in which the vehicles 1 and 4.sub.1 are traveling in the same direction, the time t.sub.bump is preferably chosen before a time corresponding to the start of the detected anomaly area A (see FIG. 2).

(24) The invention is however also applicable if the detected other vehicle is traveling in the opposite direction to the motor vehicle 1. In this case, the time t.sub.bump is preferably chosen after a time corresponding to the end of the detected anomaly area A.

(25) The method and the system according to the invention thus make it possible to detect the presence of a speed bump and to estimate the distance between the motor vehicle using only the detection of other vehicles present on the road, thereby making it possible to overcome problems linked to conditions of being unable to see a signpost that is supposed to warn of this speed bump.

(26) If the motor vehicle 1 is furthermore a partly automated or fully automated vehicle, the method may advantageously comprise a step S.sub.5 of correcting a position of the motor vehicle 1 with respect to an HD geographical map on board the motor vehicle 1 on the basis of a position of a marker corresponding to said speed bump 3 and pre-recorded in said map, and of the estimated distance d.sub.bump.