Method for determining the trajectory of a vehicle

Abstract

A method for determining the trajectory of a vehicle traveling on a roadway, the method being implemented by a device comprising at least one computer, characterized in that it comprises: obtaining a set of data relating to the vehicle's path, including at least: high-definition map data, a vehicle route, detecting, from the data obtained, at least one intersection on the vehicle route, determining an intersection model from the detected intersection and the data set obtained, and determining, from the determined intersection model and the vehicle's route within the determined intersection model, a trajectory to be followed by the vehicle in the vicinity of the intersection and for crossing the intersection. Abstract: Figure 1

Description

Title of the invention: Method for determining the trajectory of a vehicle technical field

[0001] This disclosure relates to a method for determining the trajectory of a vehicle, more particularly, a connected autonomous or semi-autonomous vehicle. Prior art

[0002] It is well established that a vehicle equipped with a driver assistance system greatly improves the comfort and safety of road users. By using a set of sensors, such as cameras, such an autonomous or semi-autonomous vehicle can understand its surroundings and act accordingly. It can thus warn the driver of dangerous situations or make driving easier.

[0003] For example, accident prevention systems are known. Similarly, systems that allow for the safe overtaking of slower vehicles on highways, thereby maintaining a more stable driving speed and improving traffic flow, are known.

[0004] However, the array of sensors onboard the vehicle does not always allow an autonomous or semi-autonomous vehicle to understand the entirety of its surrounding environment. Typically, road markings such as road signs may be faded, or a road sign may be obscured by a large vehicle or damaged.

[0005] These problems limit the use of driver assistance systems in autonomous or semi-autonomous vehicles in urban journeys, where signage is both more complex and more difficult than in the countryside or on fast roads such as highways.

[0006] More specifically, in complex intersections, the layouts are specific and non-standardized. There are thus a multitude of different intersections, the rules of which are often specified by several signs. Conversely, one sometimes finds intersections where the signage is absent or missing. Under these conditions, the information provided by the various sensors in the car does not always allow for a complete understanding of the intersection.

[0007] Under these conditions, autonomous or semi-autonomous vehicles based on prior art are not sufficiently efficient.

[0008] In specific cases such as driving an autonomous or semi-autonomous vehicle on a highway, it is also known to use map data High definition is used to allow for precise driving adapted to the geometry of the highway. However, this use is limited to very simple cases such as highway driving. Summary

[0009] The present disclosure improves the situation by proposing a method to determine the trajectory that an autonomous or semi-autonomous vehicle should take when approaching and crossing an intersection, in particular by adding high-definition mapping data to the information from the sensors.

[0010] A method is proposed for determining the trajectory of a vehicle traveling on a roadway, the method being implemented by a device comprising at least one computer, including: - obtaining a set of data relating to the vehicle's journey, including at least: • high-definition mapping data, • a vehicle route, - detection, based on the data obtained, of at least one intersection on the vehicle's route, - determining an intersection model from the detected intersection and the resulting dataset, - a determination, based on the determined intersection model and the vehicle's route in the determined intersection model, of a trajectory to be followed by the vehicle in the vicinity of the intersection and to cross the intersection.

[0011] In one embodiment, the resulting dataset further includes data from sensors onboard the vehicle, such as an image acquisition device, and the method further comprises: - a verification of the optimization of the vehicle's trajectory based on data from onboard sensors, and - a correction of said trajectory, if said trajectory is not optimal.

[0012] Alternatively, the resulting dataset includes at least: - data relating to the vehicle's destination, - data relating to a departure lane of a vehicle route, such as that the road in which it is geographically positioned, and the vehicle's route is deduced from data relating to the vehicle's destination and departure lane, and from high-definition map data.

[0013] In one variant, determining an intersection model from the detected intersection includes: - access to a database of crossing histories of intersections to determine if there is historical data on crossing the intersection for that intersection, and • If historical crossing data exists for said intersection, the intersection model corresponds to the detected intersection, • if there is no historical crossing data for said intersection, the determination of a category of the intersection from among a plurality of predetermined categories.

[0014] In one variant, the determination of an intersection category is carried out by applying a predefined grouping model, previously trained on a learning database comprising at least historical vehicle trajectories of crossing intersections from high-definition map data, and preferably augmented with static and dynamic data of these intersections from on-board sensors.

[0015] In one embodiment, the method includes strengthening the determination of the trajectory, by correcting or manually validating said determined trajectory.

[0016] In one variant, said vehicle is equipped with an autonomous or semi-autonomous driving system, characterized in that the setpoint trajectory of the driving system is the determined trajectory.

[0017] Alternatively, the trajectory is further determined according to a predefined driving profile.

[0018] In one variant, the vehicle is said to be "connected", that is to say, equipped with a vehicle communication system enabling it to communicate and receive a set of information, characterized in that the data set obtained includes information received by said vehicle communication system.

[0019] Alternatively, the vehicular communication system is a V2X type device or a variant of the V2X type such as V2V or V2N.

[0020] In one embodiment, the vehicle is equipped with at least one sensor characterized in that the data set includes information from said at least one sensor.

[0021] In one variant, the data set includes static data, such as horizontal or vertical signage, the geometry of traffic lanes, or the arrangement of certain elements related to the road or traffic lanes.

[0022] Alternatively, the dataset includes dynamic data, such as traffic lane congestion, the status of a traffic light, the precise position of the vehicle in its lane, or the time.

[0023] According to another object, a computer program product is also proposed, including code instructions for implementing the process according to the preceding description, when executed by a computer. Brief description of the drawings

[0024] Other features, details and advantages will become apparent from reading the detailed description below and from analyzing the accompanying drawings, in which: Fig. 1

[0025] [Fig.1] shows a schematic representation of the process according to one embodiment. Fig. 2a

[0026] [Fig.2a] schematically illustrates a 2D representation of a model of a first intersection in which the process takes place according to an embodiment. Fig. 2b

[0027] [Fig.2b] schematically illustrates a 2D representation of a second model of the first intersection. Fig. 3

[0028] [Fig.3] schematically illustrates a 2D representation of a model of a second intersection. Fig. 4

[0029] [Fig.4] schematically illustrates a vehicle of the process according to one embodiment. Description of the implementation methods

[0030] In order to prepare for crossing an intersection with an autonomous or semi-autonomous vehicle, the vehicle must be able to determine in advance a trajectory to follow. Reference is now made to [Fig. 1], which schematically represents a flowchart of the process for determining the vehicle's trajectory according to one embodiment, and to [Fig. 4], which represents such a vehicle according to one embodiment of the process. This process is implemented by a vehicle trajectory determination device 400, comprising at least one computer 401, for example, of the processor, microprocessor, microcontroller, programmable logic circuit (FPGA, etc.), and / or specialized integrated circuit (ASIC, etc.) type. The computer 401 is advantageously embedded in the vehicle.Alternatively, the control unit 401 can be located remotely from the vehicle and communicate remotely with an on-board communication device 403 and / or vehicle components 404, 405, 406. The device may further include a memory 402, storing code instructions executed by the control unit 401 to implement the process. The memory 402 may also store one or more pre-trained models, described in more detail below, to determine the type of intersection. that a vehicle is approaching, and a trajectory to follow at that intersection. 402 memory can be non-volatile electronic memory, an optical hard drive, etc.

[0031] In order to determine a trajectory of a vehicle in an intersection, it is necessary first to determine in which intersection the vehicle must follow the trajectory.

[0032] An intersection is a point where two or more roads join or cross, regardless of the angle(s) of their axes. When the order of priority for vehicles is not obvious at the intersection, their passage is often regulated by traffic signals. Such an intersection is considered complex when it includes specific and non-standardized features indicated by traffic signals. Complex intersections thus differ from roundabouts or Indonesian-style intersections.

[0033] The method first comprises obtaining, by the computer 401, a set of data relating to the route that the vehicle will take or is taking. This set advantageously includes high-definition map data. This data can be retrieved from a memory 402 local to the vehicle or from a server, if necessary, the vehicle being equipped with a communication device 403 configured to communicate at least with the server.

[0034] High-definition maps (HD maps) are very precise maps that include details absent from traditional, standard-definition maps. These maps include identification of the different traffic lanes. They can be accurate to the decimeter level and often include traffic sign information. Their precision makes it possible to differentiate the lanes that make up a road, rather than the different roads themselves, which is the usual limitation of traditional maps.

[0035] High-definition maps can, for example, include descriptive elements of traffic rules within lanes, such as information concerning the direction of travel, as well as priority or reserved lanes like bus or taxi lanes. They also include various elements related to the road and lanes, such as signs, whether on the side or above the road, streetlights, physical barriers such as concrete barriers or guardrails, etc. These elements are particularly helpful for the precise geolocation of the vehicle within the lane, especially in urban canyons where GNSS geolocation is severely degraded.

[0036] The data set also includes a vehicle route, that is, the sequence of roads it will take. The route can thus be directly communicated to the computer 401, for example via communication with a server.

[0037] Alternatively, the vehicle or the on-board computer integrates a 404 satellite geolocation data receiver, for example of the GNSS or GPS type, as follows The GNSS receiver is called a GNSS receiver, and the dataset simply includes the destination provided by the driver and a starting point for a route, such as the position of the 404 GNSS receiver, which can then deduce a route. Similarly, the 401 computer receives the destination via communication with a server or from the driver, and the starting point from a simple onboard 404 GNSS beacon, and then calculates the route, advantageously using high-definition maps.

[0038] The method then includes the detection 120 of the intersection in which to determine the trajectory, from the dataset.

[0039] It should be noted that it is possible for the intersection to be detected 120 times before the vehicle even approaches it, or even from the very beginning of the journey. For example, for an autonomous or semi-autonomous vehicle configured so that the driver does not have to intervene, the route is necessarily known from beginning to end. It is thus possible, using high-definition map data, to detect 120 times in advance the intersections that will be encountered during the journey along the planned route.

[0040] In one embodiment, the intersection is detected 120 when the vehicle is physically approaching it. The resulting dataset 110 may also include information from any sensors onboard the vehicle. The vehicle can then retrieve information about the road to make detection more reliable. In addition to the vehicle's position retrieved by satellite and compared with high-definition maps, images from onboard image acquisition devices 405, such as onboard cameras, and processed by image processing algorithms for such cameras 405 for recognizing markings or signage elements, can be transmitted to the computer 401 to obtain information about the markings or signage that define the intersection. Furthermore, lidars 406 can also inform the computer of the 2D or 3D geometry of the intersection.

[0041] Thus, the map data includes a sufficient majority of information about the environment of the route's intersections. The various sensor data can, where appropriate, augment the information present on the high-definition map, but are limited to the physical environment close to the vehicle equipped with them.

[0042] It should also be noted that sensor data can be used, for example, to update map data when content detected by the sensors differs from the map data, such as when a road sign has been modified. The computer can then update the map information stored locally in the vehicle or transmit this information to a server using the communication device.

[0043] The data in the dataset can be categorized into two groups: - static data, which is data that does not change or hardly changes, such as the geometry of road lanes, or certain signaling elements such as road markings or traffic signs, as well as - dynamic data, which changes over the time of a journey, such as the congestion of a traffic lane, the position of the vehicle or certain signaling elements such as a traffic light.

[0044] Certain static data from the dataset, such as some map data, can advantageously be pre-recorded in the vehicle for access at any time. However, this requires keeping them regularly updated to avoid providing the computer with outdated information. For example, an update can be performed automatically if the vehicle is connected to the internet via a mobile communication device or Wi-Fi.

[0045] Advantageously, the vehicle can be equipped with a vehicle communication system 403 in order to receive data via said communication system. For example, equipped with a V2X type communication system (for "Vehicle-to-Everything" in English), the vehicle could communicate and receive some of the data.

[0046] V2X vehicular communication systems can be adapted to several specific communication scenarios, such as when communication is with a network server in V2N (for "Vehicle-to-Network"), with another vehicle in V2V (for "Vehicle-to-Vehicle"), or with a pedestrian in V2P (for "Vehicle-to-Pedestrian" in English), the road infrastructure in V2I (for "Vehicle-to-Infrastructure" in English), etc.

[0047] With a server, via the internet in V2N, transmitting up-to-date map data, it is possible to obtain map data including dynamic data, for example the congestion of a lane of the route in real time.

[0048] With another vehicle connected in V2V, it is also possible to obtain data concerning the physical environment of the third vehicle, or for example the time spent to recently cross an intersection on the route.

[0049] By means of these vehicular communication systems, the method advantageously includes obtaining 110 data concerning an intersection, in real time, optionally from sensors, even before the vehicle is physically there.

[0050] Obtaining 110 a set of varied and numerous data advantageously allows for reliable and complete detection of an intersection.

[0051] Advantageously, high-definition map data makes it possible to know the exact geometry of an intersection, which greatly facilitates its detection 120.

[0052] After its detection 120, the method comprises the determination 130 of a model of intersection corresponding to the detected intersection 120. This model is thus a numerical representation of the detected intersection 120 which can be translated into several forms.

[0053] For example, the determined model 130 may be a simplified or detailed version of a set of characteristic data for the intersection. It may advantageously include a 2D or 3D reconstruction of the detected intersection 120 and its surroundings, allowing a trajectory to be located within it. It may also advantageously include all the traffic rules in the intersection and its surroundings.

[0054] The characteristics of the determined model 130 can be mainly classified into two categories: - geometry characteristics, which include the physical limits of the lanes and the road, elements related to the road, etc. - and the rule characteristics, which include traffic rules related to different types of signage elements, intersection topology, etc.

[0055] Advantageously, it is possible to use a database of intersection crossing histories to determine 130 a model of the detected intersection 120. Indeed, by aggregating the intersection crossing history data of the same vehicle and advantageously of other vehicles, a large database and intersection model database can be created against which the detected intersection data 120 can be compared.

[0056] If in the database there is historical data for crossing the detected intersection 120, the associated intersection model then corresponds, in fact, to the detected intersection 120.

[0057] If, on the contrary, there is no history of crossing the detected intersection 120 in the database, the process then includes the determination 131 of a category to which to attach the detected intersection 120 among the associated intersection models in the database.

[0058] The association of the detected intersection 120 with a category can be carried out on the similarity of the detected intersection 120 with the intersections associated with the crossing histories of the database.

[0059] Alternatively, it is possible to compare the intersections with each other on the basis of all their characteristics, and to categorize them thus by their similarity.

[0060] Indeed, for example, by grouping intersections with similar geometry, topology, and identical traffic and priority rules within them, it is possible to determine the category to which the detected intersection 120 belongs, and thus, to determine an intersection model corresponding to the intersection. For example, the determined model may be an average of the models intersections of the category, or a weighted average based on similarity to the detected intersection 120.

[0061] Advantageously, the categorization and matching of intersections in the database and the detected intersection 120 can be performed using a clustering machine learning algorithm. Intersections can thus be grouped primarily based on the intersection's geographic location, but also using static data such as the markings and geometry of the intersection, as well as the resulting traffic rules. Furthermore, grouping can also be performed using dynamic data, such as lane congestion at the intersection or simply the time of passage. Such an algorithm effectively categorizes historical data once trained on this type of data.

[0062] Once the intersection model 130 is determined, the method includes determining 140 the trajectory from the model. As a preliminary step, the computer determines a starting lane for a trajectory and a destination that the vehicle must follow in the intersection, in order to comply with the route. This essentially involves first determining which lane the vehicle will use to enter the intersection and in which direction it must travel to follow the overall route.

[0063] Generally, if the vehicle is near the detected intersection, the starting lane of the trajectory is the lane in which the vehicle is positioned.

[0064] In a variant where the detection of the intersection would be predictive (where the vehicle is not yet physically close to it), a departure lane can be automatically determined in the vicinity of the intersection, based on the planned route and the conventions of the traffic rules.

[0065] From the route, the computer deduces the route to take to exit the intersection, in order to follow the route. Advantageously, the exact lane to exit the intersection can be known or determined at the same time.

[0066] The process finally includes determining 140 a trajectory to be followed by the vehicle to cross the intersection.

[0067] From the intersection model it is possible to determine 140 a trajectory starting in a starting lane and in the direction of the destination road.

[0068] The trajectory however corresponds to the passage from one lane (departure) to another (exit or arrival) and in particular embodiments, several trajectories are possible in order to cross the intersection by following the route, in particular if the destination road has several lanes.

[0069] In one variant, the trajectory may also include a lane change to get to the starting lane.

[0070] Alternatively, the determination of the trajectory 140 may also depend on a predetermined driving profile chosen by the driver. For example, a comfortable driving profile may tend to avoid overtaking compared to a normal driving profile when it is not really necessary. Conversely, a sporty driving profile may more frequently employ overtaking trajectories to save travel time.

[0071] Advantageously, the trajectory can be determined 140 using the intersection crossing history database for the detected intersection 120.

[0072] Furthermore, in variations where, for example, the destination road has several lanes, the method may advantageously include a check 150 of the determined trajectory before the vehicle takes it. Indeed, during the time elapsed between obtaining the information 110 and determining 140 the trajectory, the vehicle's environment may have changed, and the initially determined trajectory 140 may not be optimal. In particular, the exit lane of the determined trajectory may be less optimal than another possible lane.

[0073] The trajectory optimization is therefore verified in step 150: primarily using sensors, it is verified that the determined model 130 is not obsolete with respect to the vehicle's current environment. In particular, it is verified that the exit lane of the trajectory is the most fluid among all possible exit lanes. For example, it is verified 150 using cameras on board the vehicle that the second exit lane, determined 140, is indeed less congested than the first exit lane. And specifically, for a trajectory that includes a lane change, it is verified that this change can be carried out safely.

[0074] In the event that the check 150 indicates that the determined path is not optimal, the method advantageously includes the rectification 160 of the determined path 140. The rectification 160 thus modifies the determined path to a rectified path that is more optimal in the current environment. For example, if a determined path 140 includes overtaking a congested path, and the check 150 indicates that the initially congested path is no longer sufficiently so that the determined path 140 is no longer optimal, the rectification can modify the determined path to a rectified path that avoids the now obsolete and unnecessary overtaking.

[0075] Advantageously, the rectification 160 takes into account the viability of a trajectory, in particular if a third vehicle would prevent the vehicle from taking a more optimal rectified trajectory.

[0076] The verification 150 and the rectification 160 thus take into account the trajectory to be followed in the intersection in real time.

[0077] Even more advantageously, the trajectory can be determined 140 by the application of a reinforcement machine learning model, in which the driver evaluates the determined trajectory 140 advantageously rectified 160. For example, if the trajectory is incorrect, the driver can correct it by taking back the steering wheel, and reinforce the model by indicating the trajectory that should have been determined.

[0078] Optionally, the driver can also manually validate the determined trajectory, for example passively, i.e. not taking the wheel to modify it - since not correcting it amounts to approving it - or actively, with an interaction which consists of noting or approving it.

[0079] Machine learning, grouping, and reinforcement models can advantageously be pre-trained using intersection crossing history databases. For example, during a preliminary training phase, an intersection crossing history database is built using one or more vehicles whose trajectories are tracked, for example, by means of a GNSS beacon and high-definition map data, as well as an inertial measurement unit to measure the vehicle's speed, acceleration, and orientation. This database is then used to pre-train the learning models. The intersection crossing history databases can also include the state, or values, of sensors onboard a vehicle when crossing the various intersections.

[0080] Advantageously, crossing an intersection and reinforcing the model can update the intersection crossing database, so as to store the driver's driving habits for certain trajectories. Thus, as corrections are made to the same trajectory, the determined trajectory 140 will get closer and closer to the driver's usual trajectory.

[0081] The check 150 and the rectification 160 thus allow the vehicle to follow an optimal trajectory, including when the vehicle is in unexpected environmental conditions.

[0082] Alternatively, the reinforcement learning model can be advantageously optimized to determine a real-time trajectory when the dataset also includes onboard sensor data. In this way, verification 150 and rectification 160 are not necessary, since the trajectory is therefore always optimized for the real-time situation.

[0083] Reference is now made to [Fig. 2a], which schematically illustrates a 2D representation of an intersection model. In one embodiment, the method determines the trajectory for vehicle 201 to cross the intersection from its position in the right-hand lane 210 to reach one of lanes 220 and 221. face.

[0084] Several trajectories are possible among the lane changes at the intersection so that the vehicle can proceed straight ahead: - the vehicle can remain 202 in the right lane, that is to say, move from the right lane 210 before the intersection to the right lane 220 of the road opposite, or - the vehicle can change 203 from the right lane 210 before the intersection to the left lane 221 of the road opposite.

[0085] The trajectory is determined 140 according to the model which includes the traffic rules in the intersection: staying in the right lane is therefore the preferred trajectory.

[0086] In [Fig. 2b], the intersection model is different, although it is physically the same intersection. Indeed, the model takes into account lane congestion, and therefore differs from the model in [Fig. 2a] because the right-hand lane 210 is congested in the intersection.

[0087] The trajectory is also determined 140 according to the driver's driving comfort and the preferred trajectory is therefore overtaking 203 the congestion by passing in the left lane 221.

[0088] Advantageously, the intersection model includes the approaches to the intersection, so the determined trajectory 140 can anticipate the intersection. As illustrated in [Fig. 2b], the overtaking trajectory 231 can thus begin before the intersection so that the vehicle 201 proceeds directly into the left lane 211.

[0089] Reference is now made to [Fig. 3], which schematically illustrates a 2D representation of a model of a second intersection. In this embodiment of the method, vehicle 201 is headed for the left-hand road 320 and 321.

[0090] Similarly, two trajectories are possible to reach the road on the left: - a first 331 from the left lane 311 towards the right lane 320 of the destination road, and - a second 332 from the left lane 311 towards the left lane 321 of the destination road, the right lane 310 does not allow taking the left lane according to the traffic rules in the intersection.

[0091] Taking into account the surroundings of the intersection in the model thus makes it possible to determine a more natural trajectory that respects traffic rules. Indeed, since the trajectory must start from the left lane 311, the first part 330 of the trajectory is to position oneself in said left lane 311, and finally to begin crossing 331 or 332 the intersection.

[0092] Advantageously, if, for example, the right lane 320 of the destination road is congested near the exit of the intersection, the trajectory is determined 140 taking this into account and the left lane 321 will then be preferred at the exit.

[0093] By reinforcement, when a driver usually passes through the intersection when it is congested, for example at the same times, the trajectory 332 determined 140 with the left lane 321 at the exit may be preferred by default.

[0094] Advantageously, the vehicle 201 may include a computer enabling the implementation of the various stages of the process. This computer may be embedded in the vehicle or located remotely.

[0095] In particular, the vehicle 201 equipped with a communication system, for example of type V2X, can transmit data to a remote computer which then responds with the results.

Claims

1.

2.

3. Claims Method for determining the trajectory of a vehicle traveling on a roadway, the method being implemented by a device comprising at least one computer, characterized in that it comprises: - obtaining a set of data relating to the vehicle's journey, including at least: • high-definition map data, • a vehicle route, - detection, from the data obtained, of at least one intersection on the vehicle's route, - a determination of an intersection model from the detected intersection and the obtained data set, - a determination, from the determined intersection model and the vehicle's route in the determined intersection model, of a trajectory to be taken by the vehicle around the intersection and to cross the intersection. Method according to claim 1, in which the data set obtained further comprises data from sensors on board the vehicle, such as an image acquisition device, characterized in that it further comprises: - a verification of the optimization of the said trajectory to be taken by the vehicle based on data from on-board sensors, and - a rectification of said trajectory, if said trajectory is not optimal. Method according to claim 1 or 2, wherein the obtained data set comprises at least: - data relating to the destination of the vehicle, - data relating to a departure route of a route of the vehicle, such as the lane in which it is geographically positioned, and the vehicle route is deduced from the data relating to the des- destination and departure lane of the vehicle, and high-definition map data.

4. A method according to any preceding claim, wherein determining an intersection pattern from the detected intersection comprises: accessing an intersection crossing history database to determine whether there is, for said intersection, intersection crossing history data, and • if there is crossing history data for said intersection, the intersection pattern corresponds to the detected intersection, • if there is no crossing history data for said intersection, determining a category of the intersection from a plurality of predetermined categories.

5. Method according to the preceding claim in which the determination of a category of the intersection is carried out by the application of a predefined grouping model, previously trained on a learning database comprising at least histories of vehicle trajectories of intersection crossings from high definition cartographic data, and preferentially augmented with static and dynamic data of these intersections from on-board sensors.

6. Method according to any one of the preceding claims, characterized in that it comprises a reinforcement of the determination of the trajectory, by the manual correction or validation of said determined trajectory.

7. Method according to any one of the preceding claims for which said vehicle is equipped with an autonomous or semi-autonomous driving system, characterized in that the set trajectory of the driving system is the determined trajectory.

8. Method according to any one of the preceding claims for which the trajectory is further determined according to a predefined driving profile.

9. A method according to any preceding claim for which the vehicle is said to be “connected”, that is to say equipped with a vehicular communication system allowing it to communicate and receive a set of information, characterized in that the set of data obtained includes information received by said vehicular communication system.

10. Method according to the preceding claim in which the vehicular communication system is a V2X type device or a variation of the V2X type such as V2V or V2N.

11. Method according to any one of the preceding claims for which the vehicle is equipped with at least one sensor characterized in that the data set comprises information from said at least one sensor.

12. Method according to any one of the preceding claims, characterized in that the data set comprises static data, such as horizontal or vertical signage, the geometry of traffic lanes, or the arrangement of certain elements related to the road or traffic lanes.

13. Method according to any one of the preceding claims characterized in that the data set comprises dynamic data, such as the congestion of a traffic lane, the state of a traffic light, the precise position of the vehicle in its traffic lane, or the time.

14. Computer program product, comprising code instructions for implementing the method according to any one of the preceding claims, when executed by a computer.

15. Device for determining the trajectory of a connected vehicle comprising at least one computer, characterized in that said determination device is configured to implement the method according to one of claims 1 to 13.

Images