Assistance in driving on a fast road with carriageways separated by a safety rail

Abstract

A method for assisting in the driving of a vehicle on a fast road with carriageways separated by a safety rail in which the presence of the safety rail is detected is disclosed. The safety rail is modelled from measurements performed continuously by at least one laser scanner sensor mounted on the motor vehicle, with the determination of a confidence index associated with the detection by the laser scanner sensor, an automatic driving mode is activated when the confidence index I.sub.CONF is above a confidence threshold. This mode is maintained as long as a current confidence index associated with the detection is above the confidence threshold, and this mode is deactivated when the current confidence index passes below said confidence threshold. The density of traffic in front of the motor vehicle is estimated from images captured by an embedded camera.

Claims

1. A method for assisting in the driving of a motor vehicle on a fast road with carriageways separated by a safety rail comprising: detecting the presence of and modelling the safety rail from measurements performed continuously by at least one laser scanner sensor mounted on the motor vehicle, comprising determining a confidence index associated with the detection by the laser scanner sensor; activating an automatic driving mode when said confidence index is above a confidence threshold for at least a predefined time corresponding to a minimum rolling distance travelled by the motor vehicle; maintaining said automatic driving mode as long as a current confidence index associated with the detection is above said confidence threshold; deactivating said automatic driving mode when the current confidence index passes below said confidence threshold; and estimating a density of traffic in front of the motor vehicle from images captured by a camera embedded on said motor vehicle, wherein said confidence index is negatively correlated to the density of traffic, and wherein after the activating activation of said automatic driving mode, said current confidence index, which is taken into account in deactivating deactivation and maintaining said automatic driving model is a function of a combination of the confidence index associated with the detection of the presence of the safety rail and of the estimated traffic density, wherein the current confidence index is calculated as a sum of the confidence index and the estimated traffic density.

2. The method according to claim 1, wherein the confidence index associated with the detection by the laser scanner sensor is determined as a function of a number and a position of objects detected in front of the motor vehicle from the images captured by the camera.

3. The method according to claim 1, wherein the modelling of the safety rail uses the measurements performed continuously by the laser scanner sensor, from which measurements coinciding with objects detected in front of the motor vehicle from the images captured by the camera are removed.

4. The method according to claim 1, wherein, after the activating of said automatic driving mode, the current confidence index is greater than or equal to the confidence index.

5. A system for assisting in the driving of a motor vehicle on a fast road with carriageways separated by a safety rail, the system comprising: a laser scanner sensor; and a camera embedded on said motor vehicle, said system being configured to: detect the presence of and modelling the safety rail from measurements performed continuously by at least said laser scanner sensor, comprising the determination of a confidence index associated with the detection by the laser scanner; activate an automatic driving mode when said confidence index is above a confidence threshold for at least a predefined time corresponding to a minimum rolling distance travelled by the motor vehicle; maintain said automatic driving mode as long as a current confidence index associated with the detection is above said confidence threshold; deactivate said automatic driving mode when the current confidence index passes below said confidence threshold; and estimate a density of traffic in front of the motor vehicle from images captured by said camera, wherein said confidence index is negatively correlated to the density of traffic, and wherein, after activation of said automatic driving mode, said current confidence index, which is taken into account in the deactivating and maintaining of said automatic driving mode, is a function of a sum of the confidence index associated with the detection of the presence of the safety rail and of the estimated traffic density, wherein the current confidence index is calculated as a sum of the confidence index and the estimated traffic density.

6. The system according to claim 5, wherein, after the activation of said automatic driving mode, the current confidence index is greater than or equal to the confidence index.

Description

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

(2) FIG. 1, already described, describes possible steps for the activation of an automatic driving mode on detection of the presence of a safety rail separating the carriageways of a fast road;

(3) FIGS. 2a, 2b and 2c, described previously, illustrate three examples of points of impact obtained by a laser scanner sensor in three different rolling situations;

(4) FIG. 3 schematically represents the rate of the variations of a confidence index associated with the detection of a safety rail, as a function of the occupancy of the environment situated in front of the vehicle;

(5) FIG. 4 illustrates steps of a driving assistance method in a possible implementation conforming with the invention;

(6) FIG. 5 schematically represents the rate of the variations of the traffic density as a function of the occupancy of the environment situated in front of the vehicle;

(7) FIG. 6 schematically represents the rate of the variations of a current confidence index associated with the detection of a rail and calculated according to the principles of the invention.

(8) Referring to FIG. 4, and as already described previously in relation to FIG. 1, a driving assistance system for a motor vehicle taking a fast road with carriageways separated by a safety rail comprises a laser scanner sensor 20 and a front-end camera 30 embedded on the vehicle. The system also preferably uses information supplied by a navigation module 10, for example of GPS type, in particular the attributes delivered by such a navigation module 10 making it possible to know that the vehicle is indeed on a fast road with carriageways separated by a safety rail.

(9) All these data will be able to be combined in a step 40 in order to allow a detection of presence and a modelling of the safety rail, a detection with which it is possible to associate a confidence index I.sub.CONF. A formalism of the Dempster-Shafer type can be used, or any other learning algorithm such as Adaboost or SVM (acronym for Support Vector Machine), to deliver safety rail detection information with its associated confidence index.

(10) The modelling of the safety rail 2 uses the measurements performed by the laser scanner sensor 20 (points of impact), from which have preferably been removed the measurements coinciding with objects detected in front of the motor vehicle from the images captured by the camera. For this modelling, the abovementioned Ransac algorithm can be used or any other equivalent algorithm that makes it possible to determine a straight line which is best fitted to the points of impact retained.

(11) The confidence index I.sub.CONF associated with the detection by the laser scanner sensor 20 is advantageously determined as varying as a function of the number and/or of the position of objects detected in front of the motor vehicle from the images captured by the camera (see curve C.sub.1 of FIG. 3 which gives an example of variations of this confidence index I.sub.CONF as a function of the parameter O representative of the number and/or of the position of the obstacles detected by the camera in front of the vehicle). The confidence index can also take into account other information resulting from the processing of images, for example the identification of marking lines on the ground making it possible to know if the vehicle is moving on a lane more or less close to the safety rail.

(12) The activation of the driving mode is triggered by the system when the confidence index I.sub.CONF is above a confidence threshold I.sub.th for at least a predefined time corresponding to a minimum rolling distance D.sub.th travelled by the motor vehicle (steps 50 and 60).

(13) Once this mode is activated, to avoid an unwanted deactivation even though the vehicle is still on the fast road with traffic conditions intensifying, provision is made to complement the processing with the following steps:

(14) an estimation (step 70) of the density of traffic in front of the motor vehicle is performed from the images captured by the front-end camera 30, and more particularly the third-party vehicles detected by the processing of these images. Several parameters can be used, alone or in combination, to establish an estimation, denoted Dens hereinbelow and in the figures, of the traffic density:

(15) a first parameter P.sub.1 relates to the observation of the environment of the vehicle, linked to the presence of third-party vehicles in front of the vehicle 1. This first parameter is for example a function of the number of third-party vehicles detected over a given time window (for example for a minute) by using the images from the front-end camera 30, of an estimation of the overall average speed of these third-party vehicles, and of the speed limit on the fast road (obtained via the navigation module 10).

(16) A second parameter P.sub.2 relates to the movement behaviour of the vehicle 1, a function for example of its average speed observed over a given time window (for example for a minute), of the number of positive or negative accelerations observed, and of the speed limit on the fast road (obtained via the navigation module 10).

(17) A third possible parameter P.sub.3 relates to the movement behaviour of the vehicle 1 relative to the third-party vehicle closest to the vehicle 1. This third parameter is a function, for example, of the average speed of the vehicle 1 observed over a given time window (for example for a minute), of the number of positive or negative accelerations observed, of the average distance between the vehicle 1 and the closest third-party vehicle, and of the speed limit on the fast road (obtained via the navigation module 10).

(18) Other parameters can be envisaged, such as, in the case where some third-party vehicles detected are two-wheel vehicles, a parameter P.sub.4 relating to the behaviour of the two-wheel vehicles relative to the third-party vehicles detected, in particular a comparison of the speeds of these two-wheel vehicles with those of the third-party vehicles.

(19) The estimation Dens of the traffic density can then be obtained from a weighted sum of the different parameters used, for example according to the equation:

(20) $\mathrm{Dens}=\frac{\underset{k=1}{\overset{4}{.Math.}}{\alpha }_k{P}_k}{\underset{k=1}{\overset{4}{.Math.}}{\alpha }_k}$
in which α.sub.k represents the weighting coefficient associated with each parameter P.sub.k.

(21) The curve C.sub.2 represented in FIG. 5 illustrates the variations of the traffic density Dens as a function of the parameter O.

(22) At the end of the step 70, the estimated traffic density is used in the calculation of a new confidence index I′.sub.CONF associated with the detection of the safety rail, which is a function of a combination of the confidence index I.sub.CONF associated with the detection of presence of the safety rail and of the estimated traffic density.

(23) As an example, the combination of the confidence index I.sub.CONF associated with the detection of presence of the safety rail and of the estimated traffic density is a sum, which can be formulated mathematically by the expression
I′.sub.CONF=min(I.sub.CONF+Dens;1)

(24) In other words, because it is already known that a safety rail has been detected with a sufficient confidence index, and that consequently the automatic mode has been activated (step 60), the value of the confidence index is increased to avoid a premature deactivation of the automatic driving mode due to an increase in traffic density.

(25) The confidence index I′.sub.CONF calculated in the step 80 can possibly, in a step 90, be reduced by a value taking into account the distance travelled by the motor vehicle 1 during which the confidence index I.sub.CONF is low.

(26) It is this new value of the confidence index I′.sub.CONF which is taken into account in the comparison (step 100) with the confidence threshold I.sub.th to decide whether the automatic driving mode can be maintained (step 120) or, on the contrary, be deactivated (step 110) depending on whether the value of the confidence index I′.sub.CONF is above or below the confidence threshold.

(27) The curve C.sub.3 represented in FIG. 3 illustrates the variations of the confidence index I′.sub.CONF as a function of the parameter O, resulting from the sum of the curves C.sub.1 and C.sub.2. By comparing with the curve C.sub.1 of FIG. 3, it is observed that, by taking into account the estimation Dens of the traffic density in the calculation of the confidence index, a higher value of this confidence is maintained when the traffic intensifies, thus reducing the risk of an unwanted deactivation of the automatic driving mode.

(28) When the traffic once again becomes fluid, two possible cases can arise: either the detection of the safety rail is associated with a sufficient confidence index, such that the automatic driving mode will be maintained, or the confidence index becomes too low and leads to the deactivation of the automatic driving mode.

(29) Obviously, other conditions independent of the calculation of I′.sub.CONF may require an immediate deactivation of the automatic driving mode, such as the detection by the camera that the vehicle is crossing an exit line, entering a town or the presence of a traffic light, or the fact that the navigation predicts a context unfavourable to that of a road with separated carriageway.