Method for automatically shifting a vehicle to one side of its traffic lane

Abstract

One aspect of the invention relates to a method for automatically shifting a given vehicle to one side of a lane on which the given vehicle is traveling, the method comprising the following steps: - recording a data set comprising, for each vehicle of a set of vehicles and for at least one shift of the vehicle to one side of its traffic lane carried out manually by a driver of the vehicle, a set of factors applying to the vehicle at the time of the shift and at least one shift characteristic of the shift; - determining, from the recorded data set, a plurality of given sets of factors for which a shift is applied and at least one associated shift characteristic; - when a given set of factors of the plurality of given sets of factors applies to the given vehicle, implementing, by a function for controlling the positioning of the vehicle in the lane implemented on the given vehicle, an automatic shift of the given vehicle to one side of the lane, the shift having the shift characteristic associated with the given set of factors.

Description

DESCRIPTION TITLE: Method for automatically shifting a vehicle to one side of its lane TECHNICAL FIELD OF THE INVENTION

[0001] The technical field of the invention is that of safety functions in vehicles, and more particularly that of functions for controlling the positioning of vehicles in their lanes.

[0002] The present invention relates to a method for shifting a vehicle to one side of its lane, and in particular to an automatic method for shifting a vehicle to one side of its lane. The invention also relates to a vehicle and a computer program for implementing the method. TECHNOLOGICAL BACKGROUND OF THE INVENTION

[0003] The lane positioning control function allows a vehicle to automatically move to one side of its lane to create space between lanes for motorcycles or emergency vehicles to pass. This function is activated, in particular, when a vehicle is detected between lanes or in the presence of a traffic jam.

[0004] The current problem with this function is that the applied offset is set by the manufacturer and therefore does not adapt to the driver's requests, who must therefore make the offset maneuvers manually, especially if certain characteristics of the offset, for example its amplitude, do not suit him or if he wishes to make an offset in situations other than those predefined.

[0005] Therefore, there is a need for a function capable of applying an automatic shift to the vehicle at a time and with the characteristics desired by the driver. SUMMARY OF THE INVENTION

[0006] The invention offers a solution to the problems mentioned above, by allowing the vehicle to be automatically shifted in the circumstances and with the characteristics that the driver would have chosen to apply the shift himself.

[0007] A first aspect of the invention relates to a method for automatically shifting a given vehicle to one side of a lane on which the given vehicle is traveling, comprising the following steps: Recording a dataset comprising, for each vehicle in a set of vehicles and for at least one shift of said vehicle to one side of its lane manually applied by a driver of said vehicle, a set of factors applying to said vehicle at the time of said shift and at least one characteristic of said shift; Determination, from the recorded dataset, of a plurality of given sets of factors for which a shift is applied and at least one associated shift characteristic; When a given set of factors from the plurality of given sets of factors applies to the given vehicle, implemented by a vehicle lane positioning control function implemented on the given vehicle, an automatic shift of the given vehicle to one side of the lane, said shift having the shift characteristic associated with said given set of factors.

[0008] Thanks to the invention, data relating to the circumstances under which, and the characteristics with which, a driver (if the vehicle combination consists solely of the vehicle in question) or a plurality of drivers (if the vehicle combination consists of a plurality of vehicles) manually perform lane changes are recorded. This data is then processed to configure the vehicle lane positioning control function implemented on the vehicle in question, enabling it to perform automatic lane changes with similar characteristics under similar circumstances.

[0009] In addition to the characteristics mentioned in the preceding paragraph, the process according to the invention may have one or more additional characteristics from among the following, considered individually or according to all technically possible combinations.

[0010] According to one embodiment, the set of factors applicable to the vehicle includes the type of lane on which the vehicle is traveling, the state of traffic, the presence of specific infrastructure on the lane, the weather conditions, the speed of the vehicle, the type of lane user(s) close to the vehicle, the type of lane lines and / or the type of lane edges.

[0011] According to an embodiment compatible with the previous embodiment, the recording step is only carried out if the manually applied offset is achieved via a vehicle trajectory modulation function implemented on the vehicle.

[0012] Thus, the shift manually applied by the driver is controlled by the trajectory modulation function which allows for a continuous but safe shift, meaning that the shift may not be carried out if certain danger conditions apply to the vehicle.

[0013] According to an embodiment compatible with previous embodiments, the registration step is only carried out if a predetermined set of conditions applies to said vehicle.

[0014] Thus, only data relating to a manual shift considered to have been carried out under reasonable conditions are recorded.

[0015] According to a first embodiment compatible with previous embodiments, the determination step includes a substep of calculating at least one factor shift characteristic for each factor, the factor shift characteristic corresponding to the average of the recorded shift characteristics corresponding to each occurrence of said factor in the dataset, the shift characteristic associated with a given set of factors being the lowest factor shift characteristic among the factor shift characteristics corresponding to the factors included in the given set of factors.

[0016] Thus, if at least one factor in the given set was not recorded, no shift is applied, and if several factors associated with different factor shift characteristics were recorded, the shift with the weakest factor shift characteristic is applied, which allows to only perform an automatic shift in circumstances and with characteristics desired by the driver.

[0017] According to a second embodiment compatible with the previous embodiments except the first embodiment, the determination step includes a substep of prediction, from a set of factors and by a machine learning algorithm, of a lag probability, and at least one characteristic probability associated with a characteristic value for each lag characteristic, said set of factors being part of the plurality of given factor sets if the predicted lag probability and at least one characteristic probability meet at least one predefined condition, the associated lag characteristic depending on the predicted lag probability, at least one characteristic probability and the predicted associated characteristic value.

[0018] Thus, the algorithm determines under what circumstances the driver is most likely to want to make a shift and with what characteristics and automatic shifts are actually made if the predicted probabilities are sufficiently high and with a shift characteristic dependent on the predicted probabilities.

[0019] According to a sub-variant of the previous implementation variant, the determination step includes a prior sub-step of supervised training of the algorithm from the recorded dataset.

[0020] According to an embodiment compatible with the previous embodiments, the method according to the invention includes a step of configuring the vehicle positioning control function in the lane, to make it capable of implementing, when a given set of factors from the plurality of given sets of factors applies to the given vehicle, an automatic shift of the given vehicle towards one side of the lane, the automatic shift having the associated shift characteristic.

[0021] A second aspect of the invention relates to a vehicle comprising means for implementing the process according to the invention.

[0022] A third aspect of the invention relates to a computer program product comprising instructions which, when the program is executed by a computer, lead the computer to implement the steps of the process according to the invention.

[0023] The invention and its various applications will be better understood by reading the following description and examining the accompanying figures. BRIEF DESCRIPTION OF THE FIGURES

[0024] The figures are presented for illustrative purposes only and are in no way limiting to the invention. Figure 1 is a synoptic diagram illustrating the sequence of steps of a process according to the invention. Figure 2 shows a schematic representation of a vehicle on which a fourth step of the process according to the invention is applied, as a function of time. DETAILED DESCRIPTION

[0025] Unless otherwise specified, the same element appearing on different figures has a unique reference.

[0026] The invention relates to a method for automatically shifting a given vehicle into a traffic lane in which the vehicle is traveling, i.e. without intervention from a driver of the vehicle.

[0027] The sequence of steps in process 100 is illustrated in Figure 1.

[0028] A first step 101 of process 100 is to record a set of data.

[0029] The dataset is recorded for at least one shift made by a vehicle included in a group of vehicles, towards one side of its lane of travel and exerted manually by a driver of said vehicle, i.e. by exerting a torque on the steering wheel of said vehicle.

[0030] According to a first variant embodiment, the set of vehicles includes only the given vehicle.

[0031] According to a second variant embodiment, the set of vehicles comprises a plurality of vehicles, which may or may not include the given vehicle.

[0032] The dataset includes a set of factors applicable to said vehicle at the time of said offset and at least one offset characteristic of said offset.

[0033] The set of factors includes, for example, the type of road on which the vehicle is traveling, the state of the traffic, the presence of specific infrastructure on the road, the weather conditions, the speed of the vehicle, the type of road user(s) near the vehicle, the type of road lines and / or the type of road edges.

[0034] The set of factors is obtained, for example, via one or more sensors on the vehicle, for example one or more cameras on the vehicle.

[0035] The specific infrastructure is, for example, a roundabout, a tollbooth, or an intersection.

[0036] The offset characteristic is, for example, the side of the lane towards which the offset causes the vehicle to move and / or the magnitude of the offset.

[0037] According to one example of an implementation, only data relating to offsets applied manually via a trajectory modulation function of said vehicle, implemented on said vehicle, are recorded.

[0038] According to one embodiment, the trajectory modulation function can only be activated if a lane centering function implemented on the vehicle, which allows the vehicle to be automatically centered in its lane, is activated.

[0039] According to another embodiment example, only data relating to manually applied shifts made if a predetermined set of conditions applies to said vehicle are recorded.

[0040] The predetermined set of conditions is, for example, a minimum lane width, a maximum vehicle speed, and a given lane type.

[0041] The first step 101 can be carried out periodically, over a predetermined period or distance travelled, or continuously.

[0042] A second step 102 of process 100 consists of processing the dataset recorded in the first step 101 to determine a plurality of given factor sets for which a shift is applied and the associated shift characteristic(s).

[0043] According to a first embodiment, the second step 102 includes a substep 1021 of calculating at least one factor shift characteristic for each factor that can be applied to the vehicle.

[0044] The number of factor shift features is equal to the number of shift features associated with each given set of factors to be determined.

[0045] The factor shift characteristic is, for example, the average of the shift characteristics recorded in the first step 101 and corresponding to each occurrence of said factor in the dataset recorded in the first step 101, i.e., corresponding to each set of factors recorded in the first step 101 including said factor.

[0046] Thus, for a given factor, if said factor does not appear in the dataset recorded in the first step 101, the factor shift characteristic is zero.

[0047] For example, if during the first step 101, no driver has ever applied an offset on a divided highway, then as soon as this factor applies to the given vehicle, no offset is applied.

[0048] The plurality of given factor sets then includes each factor set that does not include any factor exhibiting a zero shift characteristic.

[0049] The shift characteristic associated with a given set of factors is, for example, the weakest factor shift characteristic among the factor shift characteristics corresponding to the factors included in the given set of factors.

[0050] According to a second embodiment, the second step 102 includes a first substep 1022 consisting of training in a supervised manner a machine learning algorithm to enable it to predict, from a set of factors, a probability that a shift will be applied, called the shift probability, as well as at least one corresponding characteristic probability associated with a characteristic value, per shift characteristic.

[0051] The machine learning algorithm is, for example, a neural network, more specifically a recurrent neural network.

[0052] Supervised training allows the algorithm to be trained on a predefined task, by updating a set of algorithm parameters in such a way as to minimize the error between the output data provided by the algorithm and the true output data or ground truth, i.e. what the algorithm should provide as output to fulfill the predefined task on a certain input data.

[0053] A training database therefore contains input data, each associated with a real output data.

[0054] The training database is the dataset recorded in the first step 101 and the machine learning algorithm is designed to predict, from a set of factors, a lag probability and at least one feature probability associated with a feature value, by lag feature, and the dataset therefore includes, for a plurality of factor sets, the fact that a lag occurs and the associated lag feature.

[0055] The second step 102 includes a second substep 1023 consisting, for each set of factors in a plurality of sets of factors, of using the machine learning algorithm trained in the first substep 1022, to predict a lag probability and at least one feature probability associated with a feature value for each lag feature, from said set of factors.

[0056] The algorithm is, for example, capable of predicting three characteristic probabilities, a first characteristic probability associated with a first a characteristic value corresponding to the probability that the shift characteristic is greater than the first characteristic value, a second characteristic probability associated with a second characteristic value corresponding to the probability that the shift characteristic is greater than the second characteristic value, and a third characteristic probability associated with a third characteristic value corresponding to the probability that the shift characteristic is greater than the third characteristic value.

[0057] The first characteristic value is, for example, less than the second characteristic value, which is itself less than the third characteristic value, which is itself less than a fourth characteristic value.

[0058] For example, for a shift characteristic corresponding to the amplitude, the first characteristic value is, for example, equal to 0.1 m, the second characteristic value is, for example, equal to 0.2 m, the third characteristic value is, for example, equal to 0.4 m, the fourth characteristic value is, for example, equal to 0.5 m.

[0059] The factor set is part of the plurality of given factor sets, and therefore a shift is applied, if the shift probability and at least one predicted feature probability meet at least one predefined condition.

[0060] Considering the case of a single lag feature corresponding to the lag amplitude with three corresponding feature probabilities predicted by the algorithm, the predefined condition is, for example, that the feature probability associated with the first feature value is greater than 50% if the lag probability is less than 90% and that the feature probability associated with the first feature value is greater than 20% if the lag probability is greater than 90%.

[0061] The associated lag characteristic then depends on the lag probability, at least one characteristic probability, and the predicted associated characteristic value.

[0062] Considering the same case as before, the lag characteristic is, for example: the first characteristic value if the characteristic probability associated with the first characteristic value is greater than 50%, the characteristic probability associated with the second value of characteristic is less than 60% and the probability of shift is less than 90% or if the probability of characteristic associated with the first characteristic value is greater than 20%, the probability of characteristic associated with the second characteristic value is less than 40% and the probability of shift is greater than 90%;the third characteristic value if the characteristic probability associated with the first characteristic value is greater than 50%, the characteristic probability associated with the second characteristic value is greater than 60%, the characteristic probability associated with the third characteristic value is less than 80% and the lag probability is less than 90% or if the characteristic probability associated with the first characteristic value is greater than 20%, the characteristic probability associated with the second characteristic value is greater than 40%, the characteristic probability associated with the third characteristic value is less than 60% and the lag probability is greater than 90%;the fourth characteristic value if the characteristic probability associated with the first characteristic value is greater than 50%, the characteristic probability associated with the second characteristic value is greater than 60%, the characteristic probability associated with the third characteristic value is greater than 80% and the probability of lag is less than 90% or if the characteristic probability associated with the first characteristic value is greater than 20%, the characteristic probability associated with the second characteristic value is greater than 40%, the characteristic probability associated with the third characteristic value is greater than 60% and the probability of lag is greater than 90%.;

[0063] Step 2, 102, can be performed on the given vehicle or on a remote server.

[0064] A third step 103 of process 100 consists of configuring a vehicle positioning control function within the lane, implemented on the given vehicle, to enable it to implement, when a set of factors given the plurality of factor sets determined in step 2 102 applies to the given vehicle, an automatic shift of the given vehicle to one side of its lane, the automatic shift having the associated shift characteristic determined in step 2 102.

[0065] The vehicle positioning control function in the lane is, for example, the function that allows the vehicle to automatically move to one side of its lane to clear a space between lanes in certain predefined situations, for example so that motorcycles or emergency vehicles can pass.

[0066] According to one embodiment, the vehicle positioning control function in the lane can only be activated if the lane centering function is activated.

[0067] The vehicle positioning control function in the lane can, for example, be configured differently depending on the driver of the given vehicle, for example via the selection of a user profile when starting the given vehicle.

[0068] A fourth step 104 of the process 100 consists, for the vehicle positioning control function in the lane, in implementing an automatic shift of the given vehicle towards one side of its lane when a given set of factors from the plurality of given sets of factors determined in the second step 102 applies to the given vehicle.

[0069] The automatic offset implemented then presents the offset characteristic(s) associated with said given set of factors determined during the second step 102.

[0070] The effects of the fourth step 104 on the given vehicle 200 as a function of time are illustrated in Figure 2.

[0071] In Figure 2, the given vehicle 200 is traveling on a dual carriageway with a road user in the other carriageway and in sunny weather. It is assumed that the set of factors applying to the given vehicle 200 is part of the plurality of given factor sets determined in step 2, and the vehicle positioning control function therefore implements a shift of the given vehicle 200 towards one side of the carriageway in which it is traveling, for example, to facilitate the possible passage of users between lines, with a shift characteristic corresponding to a shift amplitude A.

Claims

DEMANDS 1. Method (100) for automatically shifting a given vehicle (200) towards one side (2011) of a lane (201) on which the given vehicle (200) is traveling, comprising the following steps: Recording (101) of a dataset comprising, for each vehicle (200) in a set of vehicles (200) and for at least one shift of said vehicle (200) to one side (2011) of its lane (201) manually applied by a driver of said vehicle (200), a set of factors applying to said vehicle (200) at the time of said shift and at least one shift characteristic (A) of said shift; Determination (102) from the recorded dataset, of a plurality of given sets of factors for which a shift is applied and at least one associated shift characteristic (A); When a given set of factors from the plurality of given sets of factors applies to the given vehicle (200), implemented (104) by a control function for the positioning of the vehicle (200) in the lane implemented on the given vehicle (200), of an automatic shift of the given vehicle (200) towards one side (2011) of the lane (201), said shift having the shift characteristic (A) associated with said given set of factors.

2. A method (100) according to the preceding claim, wherein the set of factors applicable to the vehicle (200) includes the type of road (201) on which the vehicle (200) is traveling, the traffic conditions, the presence of specific infrastructure on the road (201), the weather conditions, the speed of the vehicle (200), and the type of road user(s) (201) near the vehicle. (200), the type of track lines (201) and / or the type of track edges (201).

3. A method (100) according to any one of the preceding claims, wherein the recording step (101) is performed only if the offset applied manually is done via a vehicle trajectory modulation function (200) implemented on the vehicle (200).

4. Method (100) according to any one of the preceding claims, wherein the recording step (101) is only carried out if a predetermined set of conditions applies to said vehicle (200).

5. A method (100) according to any one of the preceding claims, wherein the determination step (102) comprises a substep (1021) of calculating at least one factor shift characteristic for each factor, the factor shift characteristic corresponding to the average of the recorded shift characteristics corresponding to each occurrence of said factor in the dataset, the shift characteristic (A) associated with a given set of factors being the lowest factor shift characteristic among the factor shift characteristics corresponding to the factors included in the given set of factors.

6. A method (100) according to any one of claims 1 to 4, wherein the determination step (102) comprises a substep (1023) of predicting, from a set of factors and by a machine learning algorithm, a shift probability, and at least one characteristic probability associated with a characteristic value for each shift characteristic, said set of factors being part of the plurality of given factor sets if the predicted shift probability and at least one characteristic probability meet at least one predefined condition, the associated shift characteristic (A) depending on the predicted shift probability, at least one characteristic probability and the associated characteristic value.

7. Method (100) according to the preceding claim, wherein the determination step (102) includes a prior substep (1022) of supervised training of the algorithm from the recorded dataset.

8. A method (100) according to any one of the preceding claims, comprising a step (103) for configuring the vehicle (200) positioning control function within the lane (201), to enable it to implement, when a given set of factors from the plurality of given sets of factors applies to the given vehicle (200), an automatic shift of the given vehicle (200) towards one side (2011) of the lane, the automatic shift (201) having the associated shift characteristic (A).

9. A vehicle (200) comprising means for implementing the method (100) according to any one of the preceding claims.

10. Product: A computer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out the steps of the process (100) according to any one of the claims 1 to 8.

Images