阅读说明：本技术 用于估算车辆在地图上的位置的方法 (Method for estimating the position of a vehicle on a map ) 是由 F·李 P·博尼法特 J·伊巴涅斯-古茨曼 于 2019-02-12 设计创作，主要内容包括：本发明涉及一种用于估算车辆在地图上的精确位置的方法,该方法包括：-通过地理定位系统获取该车辆的至少一个地理定位位置(P<Sub>0</Sub>)的步骤,-在该地图上对该车辆进行预定位的步骤,以及-粒子滤波步骤,在该步骤中,对该车辆的、称为粒子(P<Sub>i</Sub>)的可能位置进行处理。根据本发明,该粒子滤波步骤包括：-在该地图上分布粒子的预备步骤、以及然后在该地图上更新这些粒子的步骤,-计算每个粒子的似然性的步骤,-选择一组有限的粒子的步骤,以及-在与所选粒子的似然性和所选粒子的数量相关的指标下降到阈值以下时对该地图上的粒子进行重采样的步骤。(The invention relates to a method for estimating the precise position of a vehicle on a map, the method comprising: -acquiring at least one geolocation position (P) of the vehicle through a geolocation system 0 ) A step of pre-positioning the vehicle on the map, and a particle filtering step in which particles (P) of the vehicle are identified i ) Is processed. According to the invention, the particle filtering step comprises: -A preliminary step of distributing particles on the map, and then a step of updating the particles on the map, -a step of calculating the likelihood of each particle, -a step of selecting a limited set of particles, and-a step of resampling the particles on the map when an index related to the likelihood of the selected particles and the number of selected particles falls below a threshold.)

1. Method for estimating the precise position (P) of a vehicle (10) on a map_p) The method of (1), the method comprising:

-acquiring at least one geolocation position (P) of the vehicle (10) through a geolocation system₀) In the step (2) of (a),

-prepositioning the vehicle (10) on the map in its geo-located position (P)₀) A step of (a) and

-a particle filtering step, in which the so-called particles (P) for the vehicle are filtered_i) Is processed to determine the precise location (P) of the vehicle (10) on the map_p)，

Characterized in that the particle filtering step comprises:

-according to the geo-located position (P) of the vehicle (10) on the map₀) Distributing particles (P) on the map_i) And then updating the particles (P) on the map_i) In the step (2) of (a),

-calculating each particle (P) based at least on data from the map_i) Likelihood (w) of_i) In the step (2) of (a),

-selecting a limited set of particles (P)_i) A step of

-in contact with selected particles (P)_i) Likelihood (w) of_i) And selected particles (P)_i) Is related to the number of indicators (N)_eff) For particles (P) on the map when falling below a threshold value_i) And performing resampling.

2. Estimation method according to claim 1, wherein each particle (P) is calculated only on the basis of data from the map_i) Likelihood (w) of_i)。

3. Estimation method according to claim 1, wherein each particle (P) is also calculated from the data from the sensors (P) under the condition that these data are considered reliable_i) Likelihood (w) of_i) The sensors enable the vehicle (10) to sense its surroundings.

4. Estimation method according to one of the preceding claims, wherein, in the particle filtering step, the precise position (P) is_p) Is based on each particle (P)_i) Likelihood (w) of_i) From the selected finite set of particles (P)_i) Is selected among them.

5. Estimation method according to one of the preceding claims, wherein, in the case of the map storing data relating to road sections, each particle (P) is associated with a closest road section (P)_i) To calculate the particle (P)_i) Likelihood (w) of_i)。

6. Estimation method according to one of the preceding claims, wherein, in the preliminary step, the particles (P) are subjected to_i) Distributed over geographically located locations (P) of the vehicle (10)₀) In a central disc.

7. Estimation method according to claim 6, wherein the radius of the disc is based on the geolocation position (P) assigned to the vehicle (10)₀) The level of protection (HPL).

8. Estimation method according to one of the preceding claims, wherein, in the selection step, the particles (P) are based on_i) A geographical location (P) with the vehicle (10)₀) The distance between these particles is chosen.

9. Estimation method according to one of the preceding claims, wherein, in the updating step, the particles (P) are caused to be present_i) The movement is performed on the map only based on information relating to the dynamic characteristics of the vehicle.

10. Estimation method according to one of the preceding claims, wherein, in the resampling step, the particles (P) are aligned using a low variance technique_i) Resampling is performed.

11. A vehicle (10) comprising:

-means for storing a map, the map being stored,

-a geographical positioning system, and

-a computer designed for pre-positioning the vehicle (10) on the map,

characterized in that the computer is designed to implement a method for estimating the precise position (P) of the vehicle (10) on the map as claimed in one of the preceding claims_p) The method of (1).

Background

In order to ensure the safety of autonomous vehicles and partially automated vehicles, it is necessary to have a wide knowledge of the surroundings in which these vehicles move.

In practice, a vehicle perceives its surroundings in two different ways, namely:

-a geographical positioning system using a map and a vehicle, and

use of external sensitivity sensors (camera, radar or lidar sensors, etc.).

Companies developing maps are currently studying so-called "high-definition" maps which make it possible to obtain highly detailed information about road network characteristics (lane width, ground markings, road signs, etc.).

These maps are embedded into vehicles equipped with a geolocation system, allowing these vehicles to locate themselves at estimated locations in longitude and latitude on the map.

Unfortunately, it has been observed that this position is not always highly accurate and reliable, which then results in the vehicle being positioned outside of the route on which it is actually traveling. This problem may prove particularly dangerous in the case of autonomous vehicles that use this information for self-guidance.

To overcome this problem, one solution known from document DE 102013217060 comprises adjusting the position of the vehicle on the map taking into account information from external sensitivity sensors of the vehicle.

This solution uses the so-called "particle filter" approach. In this method, particles corresponding to possible positions of the vehicle are conceived and processed in an attempt to find an accurate position of the vehicle. In the above document, the filter thus selects these particles one by one according to the position of the lane marker observed by the onboard camera on the vehicle until only a single particle is found that may correspond to the actual position of the vehicle.

This solution presents two drawbacks.

A first drawback of this solution is that its reliability depends to a large extent on the visibility of the road marking. In particular, it should be understood that this solution gives unreliable results in the case of insufficient visibility or no road markings.

A second drawback of this solution is that only a single particle can be selected with great care, and therefore if there is an error, the motor vehicle is unable to determine this, which may prove to be very dangerous.

Disclosure of Invention

In order to overcome the above-mentioned drawbacks of the prior art, the present invention proposes a novel method for determining whether and if so where the exact position of a motor vehicle on a map can be determined.

More specifically, according to the invention, a method as defined in the introduction is proposed, wherein the particle filtering step comprises:

-a preliminary step of distributing particles on the map according to the geo-located position of the vehicle on the map, and then a step of updating the particles on the map,

-a step of calculating a likelihood of each particle based at least on data from the map,

-a step of selecting a limited set of particles, and

-a step of resampling the particles on the map when an indicator relating to the likelihood of the selected particles and the number of selected particles falls below a threshold.

Thus, with the present invention, the likelihood of each particle is calculated based on data from the map. Thus, this calculation may not be dependent on data read by the vehicle's external susceptibility sensors. Therefore, it can be independent of the visibility of the ground mark and the visibility of the vehicle surroundings. Therefore, the reliability of the results it provides is independent of external conditions.

Also by means of the invention, the resampling step is not performed automatically, thereby avoiding depletion of particles, which could potentially lead to elimination of the correct particles.

Other advantageous and non-limiting features of the method according to the invention are as follows:

-calculating a likelihood for each particle based only on data from the map;

-calculating the likelihood of each particle also from the data from sensors that make the vehicle perceive its surroundings, under conditions that the data are considered reliable;

-in the particle filtering step, the precise location is selected from among the selected set of finite particles according to a likelihood of each particle;

-in the case where the map stores data relating to road segments, calculating the likelihood of each particle from the position of the nearest road segment relative to the particle;

-in the preliminary step, distributing the particles in a disc centred on the geo-located position of the vehicle;

-the radius of the disc is determined according to a level of protection assigned to the geographical positioning position of the vehicle;

-in the selecting step, selecting the particles according to their distance from the geo-located position of the vehicle;

-in the updating step, moving the particles on the map only according to information related to the dynamic characteristics of the vehicle;

-in the resampling step, resampling the particles using a low variance technique.

The invention also relates to a vehicle comprising:

-means for storing a map, the map being stored,

-a geographical positioning system, and

a computer designed to implement a method for estimating the precise position of the vehicle on the map as mentioned above.

Detailed Description

A better understanding of the content of the invention and how it may be carried into effect will be provided with reference to the following description of the accompanying drawings, given by way of non-limiting example.

In the drawings:

figure 1 is a diagram showing the various steps of the method according to the invention;

figure 2 is a plan view of the vehicle travelling on a road;

figure 3 is a schematic view of particles distributed on a map;

FIG. 4 is a schematic view of two particles located in the vicinity of two consecutive road sections, and

fig. 5 is a schematic view of four particles located near four road sections.

Fig. 2 shows a motor vehicle 10 in the form of a car driving on a road section with four traffic lanes V1, V2, V3, V4.

In the remainder of the description, more particular attention will be paid to the positioning of this motor vehicle 10 on a map, but the invention will not be limited to this example. It will therefore be particularly applicable to the location of any land, sea, air or space vehicle on a map.

The motor vehicle 10 considered here generally comprises a chassis, a drive train, a steering system, a braking system, and an electronic and/or computerized processing unit (hereinafter referred to as a computer).

The computer is connected to so-called "proprioceptive" sensors which can accurately measure the speed of the vehicle and the yaw rate of the vehicle.

Preferably, the computer is also connected to so-called "external sensitivity" sensors that can sense the direct surroundings of the motor vehicle 10 (these sensors can be cameras, radar sensors, lidar sensors, etc.).

The computer is also connected to a geolocation system that can evaluate the geolocation position P, here defined by latitude and longitude, of the vehicle 10₀. This may be, for example, a GPS system.

It will be considered here that this geolocation system is also designed to transmit to the computer an item of data called "horizontal protection level HPL". This data, well known to those skilled in the art, corresponds to a geo-located position P₀The uncertainty of the measurement of. The value of which varies, for example, according to the number of satellites from which the geolocation system receives data, the quality of the signal reception, the quality of the geolocation system used, etc.

Following the same idea, it will also be considered here that this geographic positioning system is designed to transmit to the computer a covariance matrix associated with this same uncertainty.

The motor vehicle 10 under consideration may be semi-automated so that its computer can trigger emergency braking, for example, when the driver has not noticed a hazard and has not taken appropriate action by himself.

However, in the remainder of this disclosure, it will be considered that the motor vehicle 10 is autonomous and that the computer is designed to control the drive train, steering system and braking system of the vehicle.

The computer then comprises a computerized memory which records data used in the context of the autonomous control of the vehicle and records data used in the context of the method described in particular below.

The computerized memory particularly stores a computerized application consisting of a computer program comprising instructions that are executed by the processor to allow the computer to carry out the method described below.

The computerized memory also stores so-called "high definition" topographical maps.

This map stores a large amount of data.

Which first includes information relating to the topography of the road. In this case, this terrain is stored in the form of road sections (or "links"). In this case, each road section is defined as a portion of only one traffic lane of the road, which portion is characterized by being constant over its entire length (the shape of the ground markings along this road section is the same, the width of this road section is constant, etc.).

The map also stores other data characterizing each road segment, including the width of the roadway, the shape of the ground markings on both sides of the roadway, the position and shape of each marking on the road segment at the roadside, an identifier of the previous road segment and an identifier of the next road segment, and so on.

In order to estimate the precise position P of the motor vehicle 10 on a map_pWhile the computer-implemented method includes two major operations, a particle filtering operation 100 and a scene selection operation 200 (see fig. 1).

The scene selection operation 200 uses the results of the particle filtering operation 100, which means that it is performed after the particle filtering operation.

We can then start with a description of the first particle filtering operation 100.

This operation is performed recursively, that is, in a round-robin fashion and at fixed time increments.

This operation comprises three main steps.

A first step 101 includes for a computer to acquire various data via its connected sensors.

Thus, the computer obtains the geo-location position P of the motor vehicle 10₀And a horizontal protection level HPL associated with the geo-located location. These data are acquired by means of a geolocation system that provides latitude, longitude and a level of protection HPL.

The computer also obtains data relating to the dynamics of the motor vehicle 10. Thus, the computer acquires the speed V of the vehicle and its yaw rate Ψ.

The second step 102 is to pre-position the vehicle 10 on the map at the acquired geo-location position P₀The step (c).

The third step 103 is a particle filtering step, in which the so-called particles P for the vehicle_iTo determine the precise position P of the vehicle 10 on the map (or more precisely the possible attitude of the vehicle) is processed_p(or more precisely the precise attitude of the vehicle on the map).

Each particle P_iCan be defined by:

two coordinates x for defining the position of the particle in a cartesian reference system_i、y_i(these coordinates are linked to the acquired longitude and latitude),

a yaw angle for defining an angle formed by the particles with respect to a given direction, such as north, an

-particles P_iA map identifier associated therewith.

Fig. 3 to 5 show particles P in the form of isosceles triangles_iEach triangle having particles on the mapCenter M corresponding to the position of_iAnd an orientation corresponding to a yaw angle of the particle on the map.

As shown in fig. 1, the third particle filtering step 103 more precisely consists of a number of sub-steps, which can now be described in more detail.

The first sub-step 110 comprises: the initialization phase of determining whether the particle filter is present is the case, for example, when the motor vehicle 10 is started.

This situation may be considered a start-up scenario, in which case no particles have yet been generated.

The next substep 112 then comprises: at the geographic position P of a given vehicle 10₀Creating and distributing the particles P on the map_i。

For this purpose, particles P are introduced_iDistributed over geographically located locations P with vehicle 10₀Inside the central disk, the radius of the disk is in this case equal to the horizontal protection stage HPL.

These particles are more accurately distributed in a helical pattern with a constant angular offset. According to desired generation of particles P_iThe number of the helical elements and the particles P_iThe angular offset between.

This number is greater than 100 and preferably about 1000. In a manner that achieves sufficient accuracy without otherwise overloading the computer.

At this time, the particle P is not yet aligned_iOrientation is performed.

Each particle P is considered to be in the presence of errors affecting the geolocation system_iCorresponding to possible locations the vehicle may have.

As can be seen in fig. 3, some particles are located outside the road. This demonstrates the following fact: the particles are not confined to a map and they can move in two dimensions. The filter is therefore highly flexible and makes it possible to initially envisage a very large number of different solutions, the least of which will then be eliminated by the particle filter.

In the next substep 113, the meter is countedComputer-controlled per particle P_iAssociated with its nearest road section.

The method chosen here is a point-to-curve method. The method comprises subjecting each particle P to_iAssociated with the road segment with the closest euclidean distance.

By way of illustrative example, the particle P can therefore be seen in FIG. 4₁Associated with the road section AB.

At this time, the computer may make the particle P_iIn particular according to the orientation of the road section with which each particle is associated (and possibly also according to the dynamic characteristics of the vehicle).

The method then proceeds to substep 116, which will be further described.

As already described above, the first sub-step 110 comprises: it is determined whether there is an initialization phase for the particle filter.

It can now be imagined that this is not the case and that the process has been initiated in advance.

In this case, in sub-step 114, the computer updates the particle P on the map_i。

To this end, the particles P are caused to_iAll moving on the map according to information related to the dynamic characteristics of the vehicle.

The two data items, namely the vehicle speed V and the yaw rate Ψ of the motor vehicle 10, are used in particular for all particles P_iMove a given distance and redirect the particles at a given angle.

It will be noted that this sub-step does not use the geolocation position P of the motor vehicle this time₀。

In the next substep 115, the computer causes each particle P to_iRe-associated with the road section.

More precisely, the computer determines which particles P are_iA new road section must be associated and it identifies this new road section.

To understand how the computer works, reference may be made to FIG. 4, in which two particles P are shown centered around points M1, M2₁、P₂And also shows a road section AB.

In this case, it is taken into account that, at the previous time increment, due to the two particles P₁、P₂In sub-step 114, a movement occurs, so that they are only associated with the same road section AB.

The computer then targets each particle P_iThe ratio r is determined to determine whether each particle should be associated with a new road segment.

This ratio r is calculated using the following equation:

if this ratio r is between 0 and 1, the particle P should not be altered_iAssociation with its original road section. For the particle P₁This is true here.

If this ratio is negative, the particle P should be altered_iAssociation with its road section. This particle should be more accurately associated with the previous road section or one of the previous road sections.

If this value ratio is strictly greater than 1, the particle P should be altered_iAssociation with its road section. This particle should be more precisely associated with the subsequent road section or one of the subsequent road sections.

Various situations may be encountered.

In the case of fig. 4 where the road section AB has only one successor BB', the particle P₂Associated with this successor (as long as the ratio r to this new road section is between 0 and 1, otherwise another successor is envisaged).

In the case of fig. 5 where the road section AB has a plurality of successors BC, BD, BE, the particle P considered at the previous time increment₂Cloned as many particles P as there are successors BC, BD, BE₂₁、P₂₂、P₂₃。

In view of the dynamic nature of the vehicle, it may also be provided to have fewer particle clones if certain successors may not be envisaged.

In another case, not shown in the figure, it may be necessary to associate the particle with another road section parallel to the road section with which it was associated at the previous time increment (this would happen especially when a certain vehicle laterally changes lanes, for example in order to overtake another vehicle). This is possible because the particles are not restricted to moving only on the same road section. This can be detected given the new position of the particle and the data stored in the map (ground marker information, lane width, etc.). In one variant, it is also conceivable to use an onboard camera on the vehicle to detect this.

In sub-step 116, which follows two sub-steps 115 and 113, the computer calculates each particle P_iLikelihood w of_i。

Here, the likelihood of a particle is given by its weight w_iTo indicate. The greater the weight of a particle, the more likely it is that the particle under consideration corresponds to the exact location of the motor vehicle 10.

This weight can be calculated in a number of different ways.

In the first embodiment, each particle P is calculated from only data from a map_iWeight w of_i。

More precisely, the weight is determined according to the euclidean distance between the considered particle and the road section with which it is associated (this weight is for example inversely equal to this distance).

In the second embodiment, each particle P is also calculated from the data from the external sensitivity sensor, under the condition that the data is considered reliable_iWeight w of_i。

In particular, it is conceivable to increase or decrease the weight of the particles under consideration according to the lateral information from the camera CAM of the vehicle. These cameras are effectively able to detect the ground marker lines and feed them back to the computer in the form of a polynomial model. The computer may then verify whether the shape of these lines corresponds to the shape of the ground marks recorded in the map, and adjust the weight of the particles accordingly.

It may be noted that the ground mark is not always detected by the camera. This may be due to conditions to which the sensor is difficult to adapt, such as for example insufficient light, wet road, marks being erased, etc. In these particular cases, the camera displays a low confidence to the computer, and then calculates the weights based only on the data provided by the map, as described in the first embodiment.

It will be observed here that the reliability criterion for a piece of data is typically provided by the sensor measuring the piece of data, for example in the form of a percentage (which percentage will then be used to determine whether the piece of data should be considered).

Regardless of which method is used, the method proceeds to sub-step 117: selecting a finite set of particles P_iTo eliminate the instantaneous geolocation position P from the motor vehicle 10₀Particles too far away.

To perform this substep, the computer obtains a new geographic position P of the motor vehicle 10₀And then it calculates each particle P_iAnd this instantaneous geographical positioning position P₀The distance between them.

If this distance is greater than the horizontal protection level HPL, the corresponding particle P will be present_iWeight w of_iSet to zero, which will thus allow automatic elimination of this particle thereafter.

If not, the corresponding particle P is not modified_iWeight w of_i。

In the next substep 118, the computer determines whether a particle P on the map is required_iResampling is performed.

To this end, the computer uses the index_effThe index is based on the particle P_iWeight w of_iAnd particles P_iIs calculated from the number of the first and second electrodes.

If this index is_effFalls below a predetermined threshold (which is stored in the computer's read-only memory), the computer detects the particles P on the map_iResampling is performed. If not, then the particle P is made_iRemaining in its state.

As is well known, resampling comprises: particles (hereinafter referred to as original particles) are considered in their entirety, and new particles are extracted from this original set.

To resample the particles, the computer may use a conventional method in which the computer will resample the original particles P_iRandomly extracting a predefined number of new particles from the set to each particle P_iProbability of and the particle P_iWeight w of_iIs in direct proportion. However, this approach typically results in particle depletion because very high weight particles are always drawn.

In this case, instead, the computer preferably uses a so-called "low variance" resampling method. In particular, this approach may facilitate maintaining a good distribution of particles on the map. The method comprises the following steps: from the primary particle P randomly_iExtracting a predefined number of new particles from the set, to each particle P_iHas a probability that the particle P is_iWeight w of_iBut this time not in proportion to this weight.

At this point, the computer may resume the execution of the sub-steps 114 to 118 cycle until particles are obtained that are all located near the same point that will be considered to correspond to the exact position P of the motor vehicle 10 on the map_p。

However, this is not the option chosen here. Thus, as already explained above, once sub-step 118 is completed, a scene selection operation 200 is provided.

This scenario selection operation 200 is implemented once the particle filtering operation 100 has converged and has given a finite number of solutions (e.g., the particles are grouped around a point where the number is below a predetermined threshold).

This scene selection operation 200 is implemented recursively, that is, in a round-robin fashion and at fixed time increments. The scene selection operation comprises several successive steps.

In a first step 201, the computer selects a "scene".

To this end, the computer takes into account the particles P in the respective sets_iAt each one ofWithin the set, the particles are all associated with the same traffic lane (or, as a variant, with the same road section).

The benefit of studying the scenes is that it will then be possible to select all the most likely scenes, so that it will be possible to first retain the correct scene from the selected scenes and second verify the validity of each selected scene.

The scenario may be expressed in the form of an assertion, such as "the vehicle is on a lane of travel referenced … …".

To obtain a better understanding of what scenes correspond within the meaning of this disclosure, FIG. 3 groups particles together into eight sets Z each corresponding to a scene₁、Z₂、Z₃、Z₄、Z₅、Z₆、Z₇、Z₈。

By way of example, set Z₁The particle in (1) corresponds to the scene "vehicle is located on the road R₁On the right traffic lane ".

Set Z₂The particle in (1) corresponds to the scene "vehicle is located on the road R₁Left lane of traffic ".

Set Z₃The particle in (1) corresponds to the scene "vehicle is located on the road R₂Left lane of traffic ".

Set Z₄Corresponds to "the vehicle is located in it with the road R₁And R₂At the ring intersection between junctions ", etc.

Each scene may also be in a vector, considering that a number "J" of scenes is found (in fig. 3, J ═ 8)Is expressed in the form of a vector whose components correspond to the particles P in this scene_iAnd by the weight w of the particles_iAnd (4) weighting.

The computer may assign to each scene a "confidence index" equal to the particle P of that scene_iWeight w of_iAnd (4) summing.

In the second stepAt 202, the computer will determine a covariance matrix for each sceneAnd the geo-location position P of the vehicle 10₀Of the covariance matrix ∑^-1(X_GNSS)。

In particular, operating such covariance matrices may characterize the uncertainty associated with each scene and with the geolocation position P provided by the geolocation system₀An associated uncertainty.

As explained above, in this case, the geo-location position P of the vehicle 10 will be referenced by the geo-location system₀Associated covariance matrix ∑^-1(X_GNSS) In this case, the covariance matrix is a 2 × 2 matrix.

With respect to covariance matrix associated with each sceneThis is based on all particles P associated with this scene_iWeight w of_iIn this case, the covariance matrix is also the 2 × 2 matrix.

Then, at a given geographic position P provided by the geographic positioning system₀And taking into account the error associated with the measurement of this geolocation position, it is necessary to determine the degree of "coherence" for each scenario.

For this purpose, in step 203, a distance D called mahalanobis is used_MjThe mathematical object of (1), whose expression is as follows:

wherein, X_GNSSCorresponding to the vector geolocation position P₀”。

Mahalanobis distance is specifically an object that can take into account the covariance of the variables (that is, the uncertainty associated with each variable) to evaluate the coherence between the two uncertain cases.

Then, in step 204, it is provided to select a first limited (or even empty) set of scenes from among the scenes acquired in step 201.

For this purpose, for each mahalanobis distance D_MjProceed with chi fang (X)²) And (6) testing.

In practice, here, each mahalanobis distance D will be_MjIs compared with a critical threshold value to determine whether the scene under consideration is associated with the geographical positioning position P₀And (4) coherence.

If the scene under consideration is associated with the geographical positioning position P₀Coherent within the meaning of chi-squared test, the scenario is preserved.

On the contrary, if the scene under consideration is associated with the geographical positioning position P₀Irrelevant within the meaning of the chi-squared test, the scenario is rejected.

It will be noted here that if a certain scene is retained, this does not necessarily mean that this scene is true. Specifically, at this time, several scenes may be retained.

Conversely, if a scene is rejected, this does not necessarily mean that the scene is false. In particular, there may be a large error impact on the geolocation position P₀The case of the measurement of (1). In this case, scenes that are true may be rejected. As will become clear in the remainder of the disclosure, this will not otherwise affect the reliability of the method presented herein.

In a subsequent step 205, it is provided to select a second limited (or even empty) set of scenes from among the scenes selected in step 204.

It will be noted here that this second selection may be made before the first selection without otherwise affecting the progress of the method.

This second option includes: only the "possible" scene for which the particles P forming this scene are retained_iWeight w of_iThe associated indicator is greater than the determined threshold. The goal is specifically to eliminate scenarios where the chi-squared coherence test has been met but is unlikely.

For this purpose, the computer eliminates the need for a confidence index (conceivably, equal to the particle P in the scene considered)_iWeight w of_iSum) of the scenes below the determined threshold. In this case, this threshold value is unchanged and stored in the read-only memory of the computer.

At the end of these two scene selection steps, the computer has reserved a number N of not only coherent but also possible scenes.

In step 206, it is then provided to determine the availability or unavailability of each selected scene from this number N.

Three cases can be envisaged.

The first case is the case where the number N is equal to 1. In this case, since only one scenario remains, this scenario is considered reasonable and can be used to generate driving instructions for the autonomous vehicle. Thus, the computer can trust the driving instructions. In this case, the computer may then recognize that the highest weighted particle of this scene corresponds to the precise position P of the vehicle 10_p。

The second case is the case where the number N is strictly greater than 1. In this case, since several scenarios remain, none of the scenarios is considered to be able to be used to generate driving instructions for the autonomous vehicle. Thus, the method restarts in a loop until only a single scene is acquired.

The last case is the case where the number N is equal to 0. In this case, no particles are considered to be available for generating driving instructions for the autonomous vehicle, since no scenario has yet been preserved. Furthermore, the computer may advantageously deduce from this that there is a incoherence between the measurements performed by the geolocation system and the acquired scene, which may be due to problems affecting the geolocation system. In this case, step 207 is provided: an alert is transmitted to the driver and/or to a control unit of the vehicle in autonomous mode, so that the driver and/or the control unit can take the necessary measures (emergency stop, degraded mode control, etc.).