DRIVER ASSISTANCE FOR A MOTOR VEHICLE

Abstract

A driver-assistance method for a motor vehicle of interest, in which the vehicle of interest detects the third-party vehicles which are present at an initial instant in its environment is disclosed. During a first prediction cycle, an order of priority is assigned to the third-party vehicles which are detected at the initial instant and to the vehicle of interest, corresponding to an order in which the vehicles in the set follow one another in the travel zone starting from a vehicle detected in a position which is furthest ahead of the vehicle of interest. For each selected vehicle in the set, taken in the order of priority, another vehicle in the set is identified which is able to be a primary target vehicle for this selected vehicle. A manoeuvre which is in progress for the selected vehicle is estimated on the basis at least of the identified primary target vehicle.

Description

TECHNICAL FIELD

The present invention generally relates to the field of motor vehicles, and more specifically to driver assistance for a motor vehicle.

TECHNOLOGICAL BACKGROUND

In order to increase road safety, certain motor vehicles, referred to as semi-autonomous motor vehicles, are equipped with partial automation systems or advanced driver-assistance systems (known by the acronym ADAS), in particular with systems carrying out, instead of the driver, the sideways control and/or the lengthways control of the vehicle, or at the very least alerting the driver of a potentially dangerous situation in order to make it possible for him/her to react in time. Provision is also made for making motor vehicles completely autonomous, that is to say without a driver.

In order to make it possible for an autonomous or semi-autonomous vehicle (called a “vehicle of interest” below) to detect dangerous situations and to react accordingly in order to avoid or reduce the risk of accidents, the on-board driver-assistance system on this vehicle must be capable not only of detecting all the dynamic objects (called “third-party vehicles” below) which are present in the immediate environment of the vehicle, such as other motor vehicles (cars, lorries, motorcycles), but also of predicting the future motion of these third-party vehicles.

As described, for example, in the document entitled “A survey on motion prediction and risk assessment for intelligent vehicles” (Lefévre et al., Robomech Journal 2014.1:1 http://www.robometechjournal.com/content/1/1/1), known trajectory prediction methods are based on a motion model chosen from among the following three types of motion models:

• • physics-based motion models;
  • manoeuvre-based motion models; and
  • interaction-sensitive motion models.

Physics-based motion models are the simplest because they consider that the future motion of a vehicle depends only on the laws of physics. These models are highly dependent on the accuracy of the on-board sensors on the vehicle of interest and assume that the third-party vehicle does not change state (speed or direction). Consequently, these models do not make it possible to reliably predict the trajectory of a third-party vehicle for a long-term prediction, for example beyond two seconds.

Manoeuvre-based motion models are a little more sophisticated than physics-based motion models because they also take into account the manoeuvre which the driver of a third-party vehicle intends to perform. Nevertheless, the vehicle of interest and the third-party vehicles the manoeuvres of which are estimated are considered to be moving independently of one another, which may lead to erroneous interpretations of certain road situations and affect the risk assessment for the vehicle of interest.

Interaction-sensitive motion models are currently the most advanced because they take into account the fact that the motion of a vehicle may be influenced by the motion of the other vehicles which are present in the road scene. Most of these models use dynamic Bayesian networks which make it possible to consider pairwise dependencies between several moving vehicles. These models make reliable and longer-term projections possible, but are incompatible with the need to assess risk in real time for the vehicle of interest because they require significant computing time resources to be able to estimate, for all the possible pairs of vehicles, all the potential trajectories of the vehicles.

SUMMARY OF THE INVENTION

One aim of the present invention is to overcome the limitations of the prior art by proposing, in particular, a simplified method making it possible to predict, reliably and inexpensively in terms of computing time, at least one future position of each vehicle which is present in a road scene in which the vehicle of interest is moving.

Another aim of the invention is to use this simplified method to quickly predict the trajectory of a plurality of vehicles which are moving in the environment of the third-party vehicle over a longer prediction time than the methods using a physics-based motion model.

Consequently, one subject of the present invention is a driver-assistance method for a motor vehicle of interest, comprising:

• • a detection step during which an on-board system on said motor vehicle of interest detects a plurality of third-party vehicles which are present at an initial instant in the environment of the motor vehicle of interest, in a multi-lane travel zone;
  • a first cycle of prediction, by said on-board system, of a first relative position and of a first relative speed of each vehicle O_k in said set, in said first frame of reference and for a first prediction instant following said initial instant, said first prediction cycle comprising:
    • a step of storing, in a database of said on-board system, initial data for each vehicle O_k in a set of K vehicles comprising the third-party vehicles which are detected at the initial instant and the vehicle of interest, said initial data comprising information relating to the current traffic lane, a current relative position and a current relative speed of each vehicle O_k in said set at the initial instant and in a first frame of reference linked to the vehicle of interest;
    • a sorting step during which an order of priority is assigned to the K vehicles O_k in the set, the order of priority being determined in accordance with the position and with the traffic lane of each vehicle O_k in the set which are stored in the database, and corresponding to an order in which the vehicles O_k in the set follow one another in said travel zone starting from a vehicle detected in a position which is furthest ahead of the vehicle of interest;
    • a step of selecting each vehicle O_k in said set in the predetermined order of priority, and, for each selected vehicle O*_k in said set:
      • a sub-step of identifying another vehicle in said set which is able to be a primary target vehicle for the selected vehicle O*_k;
      • a sub-step of estimating a manoeuvre which is in progress or about to be performed by the selected vehicle O*_k on the basis at least of the identified primary target vehicle, of information relating only to the selected vehicle O*_k and of information relating to the current environment of the selected vehicle O*_k;
    • a sub-step of estimating the movement of the selected vehicle O*_k between the initial instant and said first prediction instant on the basis of the manoeuvre which is estimated for the selected vehicle O*_k;
  • the first relative position and the first relative speed of the selected vehicle O*_k for said first prediction instant being predicted on the basis of the estimated movement.

In one possible embodiment, the sub-steps are carried out on the basis of position and speed data which are transformed into a two-dimensional frame of reference which is linked to the selected vehicle O*_k.

The estimated manoeuvre is preferably chosen from among a predefined set of possible manoeuvres. The predefined set of possible manoeuvres for a selected vehicle O*_k may comprise: keeping in its current lane; it changing lane to the left lane; it changing lane to the right lane; it stopping in its current lane.

In one possible embodiment, the sub-step of identifying a potential primary target vehicle for the selected vehicle O*_k comprises searching for a vehicle in the set located in the same lane, ahead of and closest to the selected vehicle O*_k.

In one possible embodiment, the information relating only to the selected vehicle O*_k comprises, for example, its sideways movement, and/or an on or off state of one of its indicators, and/or a history of its stored positions.

In one possible embodiment, the information relating to the current environment of the selected vehicle O*_k comprises the type of marking lines of the traffic lane in which the selected vehicle O*_k is located and/or the occupancy of the traffic lanes which are adjacent to the current traffic lane of the selected vehicle O*_k, and/or the current speed limit assigned to the traffic lane.

In one possible embodiment, the method comprises a second cycle of prediction, by said on-board system, of a second relative position {X; Y; θ}_k,2 and of a second relative speed {V_X, V_Y, {dot over (θ)}}_k,2 of each vehicle O_k in said set, in said first frame of reference and for a second prediction instant following said first prediction instant, the second prediction cycle comprising:

• • a first step of updating the database, in which the initial relative position and relative speed data are replaced by the first relative position {X; Y; θ}_k,1 and the first relative speed {V_X; V_Y; {dot over (θ)}}_k,1 which are predicted in the first prediction cycle for each vehicle O_k in the set; and
  • the sorting step, the selection step and the sub-steps of the first prediction cycle, in which the initial instant has been replaced by the first prediction instant, and the first prediction instant has been replaced by the second prediction instant.

The method may comprise a number N, which is greater than 2, of successive prediction cycles, each n^th prediction cycle making it possible to predict an n^th relative position {X; Y; θ}_k,n and an n^th relative speed {V_X; V_Y; {dot over (θ)}}_k,n of each vehicle O_k in said set, in said first frame of reference and for an n^th prediction instant following a preceding prediction instant, each n^th prediction cycle comprising:

• • a step of updating the database, in which the relative position and relative speed data are replaced by the relative position and the relative speed which are predicted in the preceding prediction cycle for each vehicle O_k in the set; and
  • the sorting step, the selection step and the sub-steps of the first prediction cycle, in which the initial instant has been replaced by the preceding prediction instant, and the first prediction instant has been replaced by the n^th prediction instant.

The successive prediction instants are preferably separated by a constant time step.

In one possible embodiment, N is equal to 33 and the constant time step is equal to 200 ms.

Another subject of the invention is an on-board driver-assistance system on a vehicle of interest, configured to implement the method according to the invention.

BRIEF DESCRIPTION OF THE DRAWINGS

The invention will be better understood in view of the following description, given with reference to the appended figures, in which:

FIG. 1 schematically illustrates, in plan view, an example of a road scene serving to illustrate the principles of the invention;

FIG. 2 schematically shows trajectories predicted by a method in accordance with the invention for the example of the road scene of FIG. 1;

FIG. 3 shows steps which are able to be implemented according to a first embodiment of a detection method which is in accordance with the invention.

DESCRIPTION OF EMBODIMENT(S)

In order to give a concrete idea, the invention will now be described in the context of the non-limiting example of the road scene shown schematically in plan view in FIG. 1.

In this FIG. 1, a vehicle of interest O_I, having an advanced driver-assistance system (which is not shown) which is in accordance with the invention, is moving in a travel zone comprising three traffic lanes L₁, L₂, L₃. Five other vehicles O₁ to O₃ and O₅, O₆ are also moving in the environment of the vehicle of interest O_I. For the sake of simplicity, it is non-limitingly assumed that the travel zone corresponds to a portion of motorway with two lanes L₂, L₃, the lane L₁ corresponding to a feeder road, and that all the vehicles are travelling in the same direction (from left to right in FIG. 1), according to the French highway code (overtaking on the left and speed limited to 130 km/h). In the non-limiting example, the third-party vehicles O₁, O₅ and O₆ are all motor vehicles, the third-party vehicle O₂ is a motorcycle, and the third-party vehicle O₃ is a lorry. Furthermore, in this example, the lorry O₃ is stationary, and the brake lights of the third-party vehicle O₁ are on because its driver is braking. The nature of the third-party vehicles which are present in the environment of the vehicle of interest has no impact on the principles of the present invention. In other words, a third-party vehicle may be either a conventional vehicle, a semi-autonomous vehicle or an autonomous vehicle.

It is assumed below that the vehicle of interest O_I is equipped:

• • with various sensors (which are not shown) of different natures (for example, image sensors, Radar, Lidar) making it possible for it to detect, on the one hand, the third-party vehicles O₁ to O₃ and O₅, O₆ which are present in its environment, and, on the other hand, the information relating to the geometry of the road scene (in particular the marking lines, the traffic signs, etc.);
  • with an ADAS (which is not shown) making it possible for it to process the information detected and to take appropriate decisions (alerts to the driver of the vehicle of interest O_I and/or sideways and lengthways control of the vehicle of interest O_I).

A complete driver-assistance method which is in accordance with the invention consists in detecting the presence of the various third-party vehicles at an initial instant t₀ and in predicting, for the vehicle of interest O_I and for all the third-party vehicles the presence of which was detected at the initial instant t₀, the future trajectory (or predicted trajectory) over a predetermined total prediction period.

Below, the following notations will be used:

• • O_k represents a vehicle in a set of K vehicles comprising the vehicle of interest O_I and all the third-party vehicles detected at the initial instant t₀. Thus, in the example of FIG. 1, the set comprises in total six vehicles, namely the detected third-party vehicles O₁ to O₃ and O₅, O₆, and the vehicle of interest O_I, which will interchangeably bear the reference O_I or O₄ below;
  • P_k(0) represents the point associated with a vehicle O_k in the set of K vehicles at the initial detection instant t₀, this point P_k(0) being conventionally associated with:
    • the current relative position {X; Y; θ}_k,0 of the vehicle O_k in a first two-dimensional frame of reference linked to the vehicle of interest O_I (see FIG. 1), where θ represents the relative orientation, in radians, of the vehicle O_k with respect to the first frame of reference O_I;
    • the current relative speed {V_X, V_Y, {dot over (θ)}}_k,0 of the vehicle O_k in this same first two-dimensional frame of reference linked to the vehicle of interest O_I, where represents the variation, in radians/seconds, in the relative orientation of the vehicle O_k with respect to the first frame of reference O_I;
  • P_k(n) represents a point in the predicted trajectory for a vehicle O_k in the set of K vehicles, at a prediction instant t_k such that

t _k =t ₀ +nΔt

this point P_k(n) being also conventionally associated with:

• • the relative position {X; Y; θ}_k,n of the vehicle O_k predicted at the prediction instant t_k in the first frame of reference linked to the vehicle of interest O_I;
  • the relative speed {V_X, V_Y, {dot over (θ)}}_k,n of the vehicle O_k predicted at the prediction instant t_k in this same first frame of reference linked to the vehicle of interest O_I.

A predicted trajectory for each vehicle O_k taken from the set comprising the detected third-party vehicles and the vehicle of interest is thus formed, starting from an initial point P_k(0) measured at the initial instant t₀, by a succession of N points P_k(n) estimated successively with a time step Δt, in which n varies from 1 to N.

In an example of an implementation, the time step Δt is constant between each successive point of a predicted trajectory. By way of example, a time step Δt which is equal to 200 ms is chosen, and the number N is set equal to 33, this making it possible to make a trajectory prediction for each vehicle O_k over a total prediction time of 7 seconds counting from the initial instant t₀. FIG. 2 illustrates an example of the portions of the various predicted trajectories for the various vehicles O_k in accordance with the principles of the invention, which will be detailed below. For example, the predicted trajectory for the vehicle O₅ comprises the initial point starting from the point P₅(0), and a succession of predicted points such as the point P₅(5) at the prediction instant t₅ and the point P₅(10) at the prediction instant t₁₀, and the predicted trajectory for the vehicle O₆ comprises the initial point starting from the point P₆(0), and a succession of predicted points such as the point P₆(7) at the prediction instant t₇ and the point P₅(19) at the prediction instant t₁₉.

With reference to FIG. 3, a method 100 which is in accordance with the invention comprising at least one first prediction cycle making it possible for the on-board system on the vehicle of interest O_I (O₄) to predict, for each vehicle O_k, the first point P_k(1) on the basis of the point P_k(0) will now be described.

The method 100 comprises an initial detection step 110 during which the on-board system on the motor vehicle of interest O_I detects a plurality of third-party vehicles which are present at the initial instant t₀ in the environment of the motor vehicle of interest O_I, in a multi-lane travel zone. In the example of FIGS. 1 and 2, it is thus assumed that the on-board system on the vehicle of interest O_I detects the presence, at the initial instant t₀, of the third-party vehicles O₁ to O₃, O₅ and O₆.

In accordance with the notations indicated above, the aim of this first prediction cycle is to make it possible for the on-board system on the vehicle of interest O_I to predict the parameters associated with the first point P_k(1), namely the first relative position {X; Y; θ}_k,1 and the first relative speed {V_X; V_Y, {dot over (θ)}}_k,0 of each vehicle O_k, in the frame of reference associated with the vehicle of interest and for a first prediction instant t₁ following the initial instant t₀.

For this purpose, the first prediction cycle begins with a step 120 of storing, in a database of the on-board system, initial data for each vehicle O_k in a set of K vehicles comprising the third-party vehicles detected at the initial instant t₀ and the vehicle of interest O_I. The initial data comprise:

• • on the one hand, the parameters indicated above associated with each initial point P_k(0), namely a current relative position {X; Y; θ}_k,0 and a current relative speed {V_X; V_Y, {dot over (θ)}}_k,0 of each vehicle O_k in the set at the initial instant t₀ and expressed in the first frame of reference linked to the vehicle of interest O_I;
  • on the other hand, the current traffic lane of each vehicle O_k at the initial instant t₀, as estimated by the on-board system on the vehicle of interest O_I (conventional estimation consisting, for example, in identifying the traffic lane for which the sideways distance along the Y axis calculated between the centre of the vehicle of interest O_I and the centre of each traffic lane L₁, L₂ and L₃ is minimal).

Table 1 below gives an example of the contents of the database at the detection instant t₀ for the road scene shown in FIGS. 1 and 2:

TABLE 1
Initial working database
Vehicle	O1	O2	O3	Ol = O4	O5	O6
Point at t0	P1(0)	P2(0)	P3(0)	P4(0)	P5(0)	P6(0)
	k = 1	k = 2	k = 3	k = 4	k = 5	k = 6
{X; Y; θ}k, 0	{60; 0; 0}	{70; 3.5; 0}	{100; 0; 0}	{0; 0; 0}	{−20; 3.5; 0}	{−25; −3.5; 0}
{VX; VY; {dot over (θ)}}k, 0	{−16.1; 0; 0}	{5.5; 0; 0}	{−36.1; 0; 0}	{0; 0; 0}	{0.2; 0; 0}	{−0.5; +0.8; 0}
Lane No.	2	3	2	2	3	1

According to an important feature of the invention, the first prediction cycle continues with a sorting step 130 during which an order of priority is assigned to the K vehicles O_k in the set, the order of priority being determined in accordance with the position and with the traffic lane of each vehicle O_k in the set which are stored in the database, and corresponding to an order in which the vehicles O_k in the set follow one another in the travel zone starting from a vehicle detected in the position furthest ahead of the vehicle of interest O_I. In the case of the road scene given by way of example in FIGS. 1 and 2, and as indicated by the numbers placed at the rear of each vehicle, it is the lorry O₃ which appears as the first vehicle to be considered, followed by the motorcycle O₂, then by the vehicle O₁, by the vehicle of interest O₄, by the vehicle O₅ and finally by the vehicle O₆.

Table 2 below illustrates Table 1, the columns of which have been reordered according to the order of priority corresponding to the example of a road scene in FIGS. 1 and 2:

TABLE 2
Initial working database reordered according to order of priority
Vehicle	O3	O2	O1	Ol = O4	O5	O6
Point at t0	P3(0)	P2(0)	P1(0)	P4(0)	P5(0)	P6(0)
	k = 3	k = 2	k = 1	k = 4	k = 5	k = 6
{X; Y; θ}k, 0	{100; 0; 0}	{70; 3.5; 0}	{60; 0; 0}	{0; 0; 0}	{−20; 3.5; 0}	{−25; −3.5; 0}
{VX; VY; {dot over (θ)}}k, 0	{−36.1; 0; 0}	{5.5; 0; 0}	{−16.1; 0; 0}	{0; 0; 0}	{0.2; 0; 0}	{−0.5; +0.8; 0}
Lane No.	2	3	2	2	3	1
Order of	1	2	3	4	5	6
priority

The first prediction cycle continues with particular prediction processing being performed, by the on-board system of the vehicle of interest, on each of the vehicles O_k in the set (including the vehicle of interest O_I). More specifically, the on-board system on the vehicle of interest O_I selects (step 140) each vehicle O_k in the order of priority assigned in step 130. Each vehicle selected in the order of priority is then denoted O_k*. In the example of the road scene shown in FIGS. 1 and 2, the particular prediction processing will thus firstly be performed for the selected vehicle O₃* corresponding to the lorry O₃, then to the selected vehicle O₂* corresponding to the motorcycle O₂ and so on until the last vehicle O₆* from Table 2, corresponding to the vehicle O₆, is selected.

The particular processing performed on each selected vehicle O_k* essentially comprises the following sub-steps, which will be detailed more fully below:

• • a sub-step 150 of identifying another vehicle in said set which is able to be a primary target vehicle for the selected vehicle O*_k;
  • a sub-step 160 of estimating a manoeuvre which is in progress or about to be performed by the selected vehicle O*_k on the basis at least of the primary target vehicle identified in the sub-step 150, of information relating only to the selected vehicle O*_k and of information relating to the current environment of the selected vehicle O*_k; and
  • a sub-step 170 of estimating the movement of the selected vehicle O*_k between the initial instant t₀ and the first prediction instant t₁, on the basis of the manoeuvre which is estimated in the sub-step 160 for the selected vehicle O*_k.

In order to simplify the calculations, the sub-steps 150, 160 and 170 are preferably carried out, not on the basis of data expressed in the first frame of reference linked to the vehicle of interest O_I (except in the case where the processing relates to the vehicle O_I as selected vehicle), but of data transformed into a two-dimensional frame of reference linked to the selected vehicle O*_k. All the prediction processing is thus performed as if the on-board system on the vehicle of interest O_I was in fact on board the selected vehicle O*_k. The system must thus, for each prediction processing associated with a selected vehicle O*_k, transform beforehand (by rotation and translation) all the data stored in the reordered initial database into the frame of reference associated with the selected vehicle O*_k, and store this information in a temporary database which is representative of the selected vehicle O*_k.

By way of examples, Table 3 below gives the temporary database obtained when the selected vehicle O*_k for the processing according to the sub-steps 150 to 170 corresponds to the lorry O₃ (first processing according to the order of priority), and Table 4 below gives the temporary database obtained when the selected vehicle O*_k for the processing according to the sub-steps 150 to 170 corresponds to the vehicle O₃ (second processing according to the order of priority):

TABLE 3
Temporary database vehicle selected O3* = O3
Vehicle	O3* = O3	O2	O1	Ol = O4	O5	O6
{X; Y; θ}*k, 0	{0; 0; 0}	{−30; 3.5; 0}	{−40; 0; 0}	{−100; 0; 0}	{−120; 3.5; 0}	{−125; −3.5; 0}
{VX; VY; {dot over (θ)}}*k, 0	{0; 0; 0}	{41.6; 0; 0}	{20; 0; 0}	{36.1; 0; 0}	{36.3; 0; 0}	{35.6; 0; 0}
Lane No.	2	3	2	2	3	1

TABLE 4
Temporary database vehicle selected O2* = O2
Vehicle	O3	O* = O2	O1	Ol = O4	O5	O6
{X; Y; θ}*k, 0	{30; −3.5; 0}	{0; 0; 0}	{−10; −3.5; 0}	{−70; −3.5; 0}	{−90; 0; 0}	{−95; 0; 0}
{VX; VY; {dot over (θ)}}*k, 0	{−41.6; 0; 0}	{0; 0; 0}	{−21.6; 0; 0}	{−5.5; 0; 0}	{−5.3; 0; 0}	{−6; +0.8; 0}
Lane No.	2	3	2	2	3	1

In Tables 3 and 4 above:

• • {X; Y; θ}*_k,0 represents the current relative position of each vehicle O_k after being transformed into a two-dimensional frame of reference linked to the selected vehicle O*_k;
  • {V_X, V_Y, {dot over (θ)}}_k,0 represents the current relative speed of each vehicle O_k after being transformed into the two-dimensional frame of reference linked to the vehicle of interest O*_k.

On the basis of the transformed information stored in the temporary database which is representative of a selected vehicle, the on-board system will be able to identify (sub-step 150 mentioned above) whether there exists, in the set of K vehicles, a primary target for the selected vehicle O*_k. Such a potential primary target is conventionally identified by the on-board system of the vehicle of interest by searching for a vehicle in the set which, according to the transformed information stored in the temporary database which is representative of the selected vehicle O*_k, is located in the same lane, ahead of and closest to the selected vehicle O*_k. This search is conventionally performed by searching for the vehicle for which the estimated time to collision (or TTC) with the selected vehicle O*_k is the smallest.

In the example of a road scene shown in FIGS. 1 and 2, no primary target will be detected during the sub-step 150 carried out for the selected vehicle corresponding to the lorry O₃, to the motorcycle O₂ or to the vehicle O₆, because no vehicle is located directly ahead of and in the same lane as these vehicles. On the other hand, the primary target which will be detected at the end of the sub-step 150 will be:

• • the lorry O₃ when the vehicle selected for processing corresponds to the vehicle O₁;
  • the vehicle O₁ when the vehicle selected for processing corresponds to the vehicle of interest O₄; and
  • the motorcycle O₂ when the vehicle selected for processing corresponds to the vehicle of interest O₅.

The on-board system will then be able to estimate (sub-step 160 mentioned above) the manoeuvre which is in progress or about to be performed by the selected vehicle O*_k using, in particular:

• • the primary target (or the absence of primary target) identified in the previous sub-step 150 for the selected vehicle O*_k;
  • information relating only to the selected vehicle O*_k, for example:
    • its sideways movement, and/or
    • the on or off state of one of its indicators, and/or
    • a history of its stored positions;
  • information relating to the current environment of the selected vehicle O*_k, such as:
    • the type of marking lines (dotted or continuous) of the traffic lane in which the selected vehicle O*_k is located; and/or
    • the occupancy of the traffic lanes which are adjacent to the current traffic lane of the selected vehicle O*_k; and/or
    • the current speed limit assigned to the traffic lane (originating, for example, from a GPS system or from a digital map pre-recorded in the on-board system, or extracted from the image of a traffic sign captured by an on-board camera on the vehicle of interest).

The manoeuvre which is predicted for the selected vehicle O*_k preferably forms part of a predefined set of possible manoeuvres, such as:

• • KL_k, or keeping in its current lane: The selected vehicle O*_k continues to drive in its current travel lane;
  • LLC_k, or changing lane to the left lane: The selected vehicle O*_k will move to the adjacent lane to the left of its current travel lane;
  • RLC_k, or changing lane to the right lane: The selected vehicle O*_k will move to the adjacent lane to the right of its current travel lane;
  • S_k, or stopping in current lane: The selected vehicle O*_k will stop in its current travel lane.

The preceding manoeuvres are not limiting. Other manoeuvres better suited to other road configurations (for example, arriving at a roundabout or at a junction) may be envisaged without departing from the scope of the present invention.

Table 5 below gives, in particular, the primary targets, predicted manoeuvres, current lane, and target lane which are obtained for the various vehicles in the road scene illustrated in FIGS. 1 and 2 at the end of the sub-step 160 for the prediction instant t₁:

TABLE 5
Selected vehicle	Primary	Predicted
O*k	target	manoeuvre	Current lane	Target lane
O1	O3	KL1	L2	L2
O2	Ø	KL2	L3	L3
O3	Ø	S3	L2	L2
O4 = Ol	O1	KL4	L2	L2
O5	O2	KL5	L3	L3
O6	Ø	LLC6	L1	L2

The on-board system may then estimate (sub-step 170), in the frame of reference of the selected vehicle O*_k, the predicted movement in terms of position and of speed for the selected vehicle O*_k between the initial instant t₀ and the first prediction instant t₁ on the basis of the primary target (or the absence of primary target) and the estimated manoeuvre for the selected vehicle O*_k. Table 6 below gives an example of the movement thus estimated when the previous prediction processing has been performed, according to the order of priority, on all the vehicles in the set, selected in turn:

TABLE 6
Movement of each vehicle in its own frame of reference
Vehicle	O3	O2	O1	Ol = O4	O5	O6
{ΔX; ΔY; Δθ}*k	{0; 0; 0}	{1.1; 0; 0}	. . .	{1; 0; 0}	. . .	. . .
{ΔVX; ΔVY; Δ{dot over (θ)}}*k	{0; 0; 0}	{0; 0; 0}	. . .	{0; 0; 0}	. . .	. . .

The preceding results are then transformed again (translation and rotation) in order to be expressed in the frame of reference linked to the vehicle of interest O_I, so that it is possible, on the basis of the calculated movement, to give the predicted position and speed for the prediction instant t₁ for each vehicle (step 180FIG. 3), this position and this speed being expressed in the frame of reference of the vehicle of interest O_I. For example, for the lorry O₃, the relative position {X; Y; θ}_3,1 and speed {V_X, V_Y, {dot over (θ)}}_3,1 which are obtained at the end of the first prediction cycle, associated with the point P₃(1), originate from the following calculations:

$\left(\begin{array}{c}\Delta X\\ \Delta Y\end{array}\right)=\left(\begin{array}{cc}\cos \left(\theta i\right)& -\sin \left(\theta i\right)\\ \sin \left(\theta i\right)& \cos \left(\theta i\right)\end{array}\right){\left(\begin{array}{c}\Delta X\\ \Delta Y\end{array}\right)}_{*3}$ $\Delta \theta =\Delta {\theta }_{*3}+\theta i$ ${\left(\begin{array}{c}X\\ Y\end{array}\right)}_{3,1}={\left(\begin{array}{c}X\\ Y\end{array}\right)}_{3,0}+\left(\begin{array}{cc}\cos \left(\Delta \theta \right)& -\sin \left(\Delta \theta \right)\\ \sin \left(\Delta \theta \right)& \cos \left(\Delta \theta \right)\end{array}\right)\left(\begin{array}{c}\Delta X\\ \Delta Y\end{array}\right)$ ${\theta }_{3,1}={\theta }_{3,0}+\Delta \theta$ ${\left(\begin{array}{c}\mathrm{Vx}\\ \mathrm{Vy}\end{array}\right)}_{3,1}={\left(\begin{array}{c}\mathrm{Vx}\\ \mathrm{Vy}\end{array}\right)}_{3,0}+\left(\begin{array}{cc}\cos \left(\Delta \stackrel{.}{\theta }\right)& -\sin \left(\Delta \stackrel{.}{\theta }\right)\\ \sin \left(\Delta \stackrel{.}{\theta }\right)& \cos \left(\Delta \stackrel{.}{\theta }\right)\end{array}\right)\left(\begin{array}{c}\Delta \mathrm{Vx}\\ \Delta \mathrm{Vy}\end{array}\right)$ ${\stackrel{.}{\theta }}_{3,1}={\stackrel{.}{\theta }}_{3,0}+\Delta \stackrel{.}{\theta }$

A second prediction cycle (which is not shown), which is similar to the first prediction cycle described above, may be carried out by the on-board system on the vehicle of interest O₁ in order to predict the points P_k(2) occupied by each of the K vehicles in the set at a second prediction instant t₂ separated from the first instant t₁ by the time step Δt, that is to say in order to estimate a second relative position {X; Y; θ}_k,2 and a second relative speed {V_X; V_Y; {dot over (θ)}}_k,2 of each vehicle O_k in said set, in said first frame of reference and for the second prediction instant. For this purpose, the second prediction cycle comprises:

• • a first step of updating the database, in which the initial relative position and relative speed data are replaced by the first relative position {X; Y; θ}_k,1 and the first relative speed {V_X; V_Y; {dot over (θ)}}_k,1 which are predicted in the first prediction cycle for each vehicle O_k in the set; and
  • the sorting step 130, the selection step 140 and the sub-steps 150-170 of the first prediction cycle, in which the initial instant t₀ has been replaced by the first prediction instant t₁, and the first prediction instant t₁ has been replaced by the second prediction instant t₂.

By generalizing the preceding principles, provision may be made for completing the method with a number N, which is greater than 2, of successive prediction cycles, each n^th prediction cycle making it possible to predict an n^th relative position {X; Y; θ}_k,n and an n^th relative speed {V_X; V_Y; {dot over (θ)}}_k,n of each vehicle O_k in the initial set, in the first frame of reference linked to the vehicle of interest and for an n^th prediction instant t₀ following a preceding prediction instant t_n-1 every n^th prediction cycle then comprising:

• • a step of updating the database, in which the initial relative position and relative speed data are replaced by the relative position {X; Y; θ}_k,n-1 and the relative speed {V_X; V_Y; {dot over (θ)}}_k,n-1 which are predicted in the preceding prediction cycle for each vehicle O_k in the set; and
  • the sorting step 130, the selection step 140 and the sub-steps 150-170 of the first prediction cycle, in which the initial instant t₀ has been replaced by the preceding prediction instant t_n-1, and the first prediction instant has been replaced by the n^th prediction instant t_n.

Claims

1. A driver-assistance method for a motor vehicle of interest, comprising: a detection step during which an on-board system on said motor vehicle of interest detects a plurality of third-party vehicles which are present at an initial instant in the environment of the motor vehicle of interest, in a multi-lane travel zone;a first cycle of prediction, by said on-board system, of a first relative position and of a first relative speed of each vehicle Ok in said set, in a first frame of reference linked to the vehicle of interest and for a first prediction instant following said initial instant, said first prediction cycle comprising:storing, in a database of said on-board system, initial data for each vehicle Ok in a set of K vehicles comprising the third-party vehicles which are detected at the initial instant and the vehicle of interest, said initial data comprising information relating to the current traffic lane, a current relative position and a current relative speed of each vehicle Ok in said set at the initial instant and in said first frame of reference;sorting, during which an order of priority is assigned to the K vehicles Ok in the set, the order of priority being determined in accordance with the position and with the traffic lane of each vehicle Ok in the set which are stored in the database, and corresponding to an order in which the vehicles Ok in the set follow one another in said travel zone starting from a vehicle detected in a position which is furthest ahead of the vehicle of interest;selecting each vehicle Ok in said set in the predetermined order of priority, and, for each selected vehicle O*k in said set by:identifying another vehicle in said set which is able to be a primary target vehicle for the selected vehicle O*k;estimating a manoeuvre which is in progress or about to be performed by the selected vehicle O*k on the basis at least of the identified primary target vehicle, of information relating only to the selected vehicle O*k and of information relating to the current environment of the selected vehicle O*k;estimating the movement of the selected vehicle O*k between the initial instant and said first prediction instant on the basis of the manoeuvre which is estimated for the selected vehicle O*k;wherein the first relative position and the first relative speed of the selected vehicle O*k for said first prediction instant is predicted on the basis of the estimated movement.

2. The method according to claim 1, wherein the sub-steps are carried out on the basis of position and speed data which are transformed into a two-dimensional frame of reference which is linked to the selected vehicle O*k.

3. The method according to claim 1, wherein the estimated manoeuvre is chosen from a predefined set of possible manoeuvres.

4. The method according to claim 3, wherein the predefined set of possible manoeuvres for a selected vehicle O*k comprises: keeping the selected vehicle in a current lane;changing a selected vehicle lane to the left lane;changing a selected vehicle lane to the right lane;stopping the selected vehicle in the current lane.

5. The method according to claim 1, wherein, identifying a potential primary target vehicle for the selected vehicle O*k comprises searching for a vehicle in the set located in the same lane, ahead of and closest to the selected vehicle O*k.

6. The method according to claim 1, wherein the information relating only to the selected vehicle O*k comprises a sideways movement, and/or an on or off state of one of its indicators, and/or a history of its stored positions.

7. The method according to claim 1, wherein the information relating to the current environment of the selected vehicle O*k comprises the type of marking lines of the traffic lane in which the selected vehicle O*k is located and/or the occupancy of the traffic lanes which are adjacent to the current traffic lane of the selected vehicle O*k and/or the current speed limit assigned to the traffic lane.

8. The method according to claim 1, further comprising: a second cycle of prediction, by said on-board system, of a second relative position and of a second relative speed of each vehicle Ok in said set, in said first frame of reference and for a second prediction instant following said first prediction instant, the second prediction cycle comprising: updating the database, in which the initial relative position and relative speed data are replaced by the first relative position and the first relative speed which are predicted in the first prediction cycle for each vehicle Ok in the set,wherein the sorting, the selection and the sub-steps of the first prediction cycle, in which the initial instant has been replaced by the first prediction instant, and the first prediction instant has been replaced by the second prediction instant.

9. The method according to claim 8, further comprising: a number N, which is greater than 2, of successive prediction cycles, each nth prediction cycle making it possible to predict an nth relative position and an nth relative speed of each vehicle Ok in said set, in said first frame of reference and for an nth prediction instant following a preceding prediction instant, each nth prediction cycle comprising:a step of updating the database, in which the relative position and relative speed data are replaced by the relative position and the relative speed which are predicted in the preceding prediction cycle for each vehicle Ok in the set; andwherein the sorting, the selection and the sub-steps of the first prediction cycle, in which the initial instant has been replaced by the preceding prediction instant, and the first prediction instant has been replaced by the nth prediction instant.

10. The method according to claim 9, wherein the successive prediction instants are separated by a constant time step.

11. The method according to claim 10, in which N is equal to 33 and the constant time step is equal to 200 ms.

12. An on-board driver-assistance system on a vehicle of interest, configured to implement the method according to claim 1.