Method for modeling a navigation environment for an automated vehicle - Patents.com

Abstract

A method for modelling the navigation environment of a vehicle (100) equipped with environmental perception means (1), a decision module (2) and autonomous control means (20) for autonomously controlling the vehicle, characterized in that it comprises the steps of: - defining (E1) a global environmental model (M_ENV_G) of the vehicle as a structured set of information built from data provided by the environmental perception means (1); - receiving (E2) from the decision module (2) a request (RDP) for information related to a decision (DP) to be taken by the autonomous control means (20) for controlling the vehicle.

Description

Summary of the Invention

[0001] The present invention relates to a method for modelling a navigation environment of an automotive vehicle. The present invention further relates to a device for modelling a navigation environment of an automotive vehicle. The present invention also relates to a computer program implementing the above-mentioned method. Finally, the present invention relates to a storage medium storing such a program.

[0002] Autonomous vehicles are constantly required to make decisions regarding the current situation of the autonomous vehicle. Therefore, determining the current situation of the autonomous vehicle is essential for decision making. The current situation of the vehicle is traditionally obtained by means for perceiving the vehicle's environment, such as lidar, radar or camera. These provide a significant amount of data, most of which is not useful for describing the current situation of the vehicle for the purpose of making decisions. Another part of these data may contain errors or inaccuracies. Finally, automated vehicles are being fitted with an ever-increasing number of environmental perception means, but some data for describing the situation may be missing.

[0003] In particular, a method for adapting an autonomous vehicle to the current situation by identifying important information contained in data originating from the vehicle's perception means is known from document US 10860022. However, this solution has drawbacks, especially with regard to managing the consistency and reliability of the information.

[0004] The object of the present invention is to provide a device and a method for modeling the navigation environment of a motor vehicle, which overcomes the above-mentioned drawbacks and improves the devices and methods for modeling the navigation environment of a motor vehicle known from the prior art. In particular, the present invention makes it possible to provide a simple and reliable device and method, which allows a reliable, consistent and adequate modeling of the navigation environment of a motor vehicle.

[0005] For this purpose, the invention relates to a method for modelling the navigation environment of a vehicle equipped with environmental perception means, a decision-making module and autonomous control means for autonomously controlling the vehicle, comprising the following steps: - defining an overall environment model of the vehicle as a set of structured information built from data provided by the environment perception means; - receiving from a decision-making module a request for information related to a decision to be taken by the autonomous control means for controlling the vehicle; - determining, from a set of structured information defining a model of the environment, a selection of information to be taken into account for making said decision; - taking into account information of said selection from an overall environment model; - determining an integrity index associated with each item of information in said selection; - providing a selective environment model containing information of the selection and a corresponding completeness index to the decision-making module; The present invention relates to a method comprising:

[0006] The method may comprise the steps of obtaining at least one first item of information from said selection of information using a first environmental perception means and using a second environmental perception means, and an integrity index associated with said at least one first item of information may be determined in response to consistency between data provided by the first perception means and the second perception means.

[0007] The method may comprise the steps of obtaining at least one second item of information from said selection of information using environmental perception means at a first time point and at a second time point after the first time point, and an integrity index associated with said at least one second item of information may be determined in response to consistency between data provided by the perception means at the first time point and at the second time point.

[0008] The decision to be made may comprise a given action to be applied by the autonomous vehicle at a given deadline, which may be short-term, medium-term, or long-term; For the same given action, the decision step is: - if the given deadline is short-term, it is possible to determine a first selection of information to be taken into account, or - if the given deadline is medium-term, a second selection of information to be taken into account can be determined, or - if the given deadline is long term, a third selection of information to be taken into account can be determined; The first selection, the second selection, and the third selection may be different from each other.

[0009] The method may comprise the sub-step of providing said decision-making module with a selective environment model comprising a set of information characterising the situation of the vehicle after applying the decision to be taken.

[0010] The method is: - constructing a finite list of decisions to be taken into account by the method; - constructing a selection of the information to be taken into account for each decision of a finite list of decisions and constructing the order in which said information must be taken into account; - configuring at least one method for taking into account each item of information to be taken into account; It can be provided with:

[0011] The invention also relates to a device for modelling a navigation environment of an autonomous vehicle, the autonomous vehicle being equipped with autonomous control means, the device comprising hardware and / or software elements implementing the method defined above, in particular hardware and / or software elements designed to implement the method according to the invention, and / or the device comprising means for implementing the method defined above.

[0012] The invention further relates to an autonomous vehicle comprising a device for modelling the navigation environment of said autonomous vehicle as defined above.

[0013] The invention also relates to a computer program product comprising program code instructions stored on a computer readable medium for implementing the steps of the method defined above when said program runs on a computer. The invention also relates to a computer program product which may be downloaded from a communication network and / or stored on a computer readable data medium and / or executed by a computer, comprising instructions which, when executed by a computer, cause said computer to implement the method defined above.

[0014] The invention further relates to a computer readable data storage medium storing a computer program comprising program code instructions for implementing the method defined above. The invention also relates to a computer readable storage medium comprising instructions which, when executed by a computer, cause the computer to implement the method defined above.

[0015] The invention also relates to a signal carried on a data medium, carrying a computer program product as defined above.

[0016] The accompanying drawings show, by way of example, an embodiment of a device for modelling the navigation environment of a motor vehicle. [Brief description of the drawings]

[0017] [Figure 1] FIG. 2 shows a vehicle equipped with a modeling device. [Diagram 2] FIG. 1 illustrates a schematic diagram of different levels of decisions to be made by an autonomous vehicle. [Diagram 3] FIG. 1 illustrates the sequencing of interactions between components of a modeled device. [Figure 4] FIG. 1 illustrates a flow chart of an embodiment of a modeling method. [Diagram 5] FIG. 1 illustrates a first situation in which the modeling method implements prediction. [Figure 6] FIG. 1 illustrates a second situation in which the modeling method implements prediction. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0018] An example of a motor vehicle 100 equipped with an embodiment of a device for modelling a navigation environment of a motor vehicle is described below with reference to FIG.

[0019] Autonomous vehicle 100 may be any type of automotive vehicle, particularly a passenger car, a utility vehicle, a truck, or even a public transportation vehicle such as a bus or shuttle. According to the described embodiment, automotive vehicle 100 is an autonomous vehicle and will be referred to as an "autonomous vehicle" throughout the remainder of this specification.

[0020] Therefore, this example is provided in a non-limiting manner. In particular, the automated vehicle may be a non-autonomous vehicle equipped with a driver assistance system, in particular a driver assistance system corresponding to a level of autonomy greater than or equal to level 2, in other words corresponding to partial autonomy of the vehicle.

[0021] The autonomous vehicle 100 travels on a given route IT in a given environment. The autonomous vehicle 100 comprises a driving assistance system 10 and an autonomous control means 20 for autonomously controlling the autonomous vehicle 100.

[0022] The driving assistance system 10 transmits a movement command C for the autonomous vehicle 100 to the autonomous control means 20. The movement command C may include a longitudinal movement command and / or a lateral movement command. The longitudinal movement command may include a first torque setpoint for the vehicle's powertrain and / or a second torque setpoint for the vehicle's brake actuators. The lateral movement command includes a rotation angle of the steering wheel of the autonomous vehicle 100.

[0023] The driving assistance system 10 comprises, in particular, the following elements: - an environmental perception means 1; - Decision-making module 2; a computation unit 3 comprising a microprocessor 31, a local electronic memory 32 and a communication interface 33 enabling the microprocessor to communicate with the perception means 1 and with the decision-making module 2; Equipped with.

[0024] Optionally, the driving assistance system 10 may comprise a human-machine interface 4 intended for displaying or sending notifications related to the consistency of the data received from the perception means 1 .

[0025] Throughout the remainder of this document, the term "data" is used to represent the raw data received from the environmental perception means 1, and the term "information" is used to represent interpretations built based on aggregations of the data.

[0026] The interpretation is carried out in a computation unit 3. The interpretation makes it possible, for example, that objects of interest (vehicles, pedestrians, obstacles, road signs) are identified in the environment of the ego-vehicle.

[0027] The environmental perception means 1 are the following means: - Means for observing the state of the vehicle11, and / or - means 12 for perceiving the environment close to the vehicle, and / or - Means for geolocating the vehicle13, and / or - Means 14 for linking the vehicle with the road infrastructure and / or with other vehicles, and / or - Means for connecting to the Internet15 It may comprise all or part of the above.

[0028] The means 11 for observing the state of the vehicle may comprise means for observing a data network internal to the vehicle, for example of the CAN bus type. Data transmitted via the internal data network may comprise, for example, instantaneous measurements of the speed and / or acceleration and / or jerking of the autonomous vehicle 100, as well as the angle and rate of rotation of the steering wheel, the state of the brake or accelerator pedal, etc.

[0029] The means 12 for perceiving the environment near the vehicle may comprise a radar and / or a lidar and / or a camera. Other embodiments of the perceiving means 12 may be envisaged.

[0030] The means 13 for geolocating the vehicle may comprise, for example, a digital navigation map and data originating from a GPS type location system, enabling the autonomous vehicle to locate on the digital navigation map and thus access information relating to the road network, in particular to the topology and geometric description of the road network. The information relating to the road network may further comprise semantic information, which may for example be the presence of signs indicating rules to be followed on a section of the road (maximum speed, no overtaking) or indicating the presence of dangers (animals, risk of landslides, etc.).

[0031] The means 14 for linking the vehicle with the road infrastructure and / or with other vehicles also enable data to be obtained regarding a set of objects perceived by the road infrastructure equipment and / or other vehicles, the data including in particular data for classifying the objects and their dimensions, their position and associated speed.

[0032] The internet connection 15 also allows contextual data such as weather, road conditions and traffic conditions to be obtained.

[0033] To control the autonomous vehicle 100, and in particular to determine the movement of the autonomous vehicle 100, the decision-making module 2 makes decisions continuously. These decisions include some decisions that are intended to be implemented through commands C that are transmitted to autonomous control means 20 for autonomously controlling the autonomous vehicle.

[0034] In general, the term "decision" as it relates to an action may be defined as the result of deliberation involving a voluntary act to perform or not perform said action.

[0035] According to this definition, - Actions related to the decision, - The decision itself, which corresponds to the choice as to whether or not to carry out an action A distinction is made between:

[0036] In this document, the actions involved in the decision may be, for example, overtaking vehicles, going in the same direction, braking to avoid obstacles, changing their route.

[0037] To decide on a choice as to whether or not to perform an action, the decision-making module 2 can send an information request to the computation unit 3 .

[0038] Throughout the remainder of this document: - The decision that is the subject of the information request is called the "decision to be taken" An information request, also called an information request RDP, is a request for information that concerns an action related to a decision, said information enabling a decision to be taken by the decision-making module 2.

[0039] In the embodiment illustrated in FIG. 2, at each time instant t, the decision-making module 2 decides between three levels: - a first decision-making level N1, called the operational level, which concerns short-term decisions, for example decisions to be made between the current time t and a future time t+1 seconds; a second decision-making level N2, called the tactical level, which concerns mid-term decisions, for example decisions to be made between the time t+1 s and the time t+10 min; - A third decision-making level N3, called the strategic level, which concerns long-term decisions, i.e. decisions that have to be made over more than 10 minutes relative to the current time. It is possible to determine the decision to be made according to:

[0040] The first decision-making level N1 determines the responsiveness of the vehicle with respect to its immediate environment, for example, precisely following a trajectory or avoiding obstacles. The decision-making module 2 determines first level decisions continuously. These are decisions that must be made within a few hundred milliseconds at most.

[0041] The second decision-making level N2 concerns maneuvers, such as the commitment of the autonomous vehicle 100 at an intersection depending on priority rules and the presence of other road users traveling across or approaching said intersection. These decisions are made periodically and correspond to a well-defined set of situations.

[0042] The third decision level N3 concerns the overall movement of the vehicle. For example, in the case of an autonomous vehicle, the third decision level N3 may involve determining the route that connects point A to point B. These are decisions that depend directly on the mission of the vehicle and are made fairly infrequently (once at the start of the mission and occasionally during the mission).

[0043] FIG. 3 illustrates the sequence of decisions to be taken by the autonomous vehicle, in other words the sequencing of interactions between the computational unit 3, the decision-making module 2 and the autonomous control means 20 for controlling the autonomous vehicle 100.

[0044] At a given time t, the decision DP to be taken is defined by a decision-making module 2 depending in particular on the route IT along which the autonomous vehicle is commanded to travel.

[0045] The decision DP to be taken involves at least one exchange of information between the decision-making module 2 and the computation unit 3 before being transmitted in the form of a command C to the autonomous control means 20 .

[0046] The exchange of information is firstly, sending from the decision-making module 2 to the calculation unit 3 an information request RDP for information related to the decision to be taken DP resulting from the decision-making module 2, and then secondly, sending from the calculation unit 3 to the decision-making module 2 a selective environmental model M_ENV_S(t), the selective environmental model M_ENV_S(t) being defined at a time instant t as a function of the decision DP to be taken, and then - optionally, thirdly, sending from the calculation unit 3 to the decision-making module 2 an alternative environmental model M_ENV_S(t+Δt), the alternative environmental model M_ENV_S(t+Δt) being defined at the time instant t+Δt as a function of the decision DP to be taken, and then fourthly, sending commands C from the decision-making module 2 to the autonomous control means 20 for moving the vehicle so as to apply the decision DP; Equipped with.

[0047] Throughout the remainder of this document, an information request related to a decision to be taken, transmitted by the decision-making module 2 to the computation unit 3, is called an "information request RDP".

[0048] The computation unit is capable of receiving information requests RDP arising from the decision-making module and creating a specific environmental model M_ENV_S of the autonomous vehicle 100, which comprises the information required to evaluate the decision DP to be taken by the decision-making module 2 for the current situation of the vehicle.

[0049] Throughout the remainder of this document, the term "vehicle situation" is used to denote the selection of an element of the driving scene among all elements that can be perceived via the perception means 1. The element can be another road user, an object, a surface, a navigation path, etc. The elements of the driving scene are selected depending on their relevance with respect to the decision DP to be made. The situation also depends on the route IT that the vehicle must follow. For example, when crossing an intersection, the vehicle situation differs depending on whether the vehicle must go straight or turn left. Indeed, depending on the direction the vehicle must take, the vehicle situation will not take into account the same user located in the zone of the intersection. The vehicle situation can also include relationships between various elements of the driving scene. For example, the situation can include a connection between an obstacle and a nearby vehicle, which may then change its trajectory to avoid the obstacle, which may interfere with the decision of the autonomous vehicle 100.

[0050] The particular environment model M_ENV_S contains the information required to model the vehicle's situation for the decision DP to be taken. The analysis of the information request RDP by the computation unit 3 therefore involves determining a selection of information or parameters to be taken into account. Based on the data derived from the perception means 1 and from the selection of parameters to be taken into account, the computation unit 3 builds a selective environment model M_ENV_S of the environment of the autonomous vehicle 100.

[0051] In an embodiment of the present invention, the computer 31 comprises the following modules in communication with each other: a module 311 for defining an overall environment model M_ENV_G of the autonomous vehicle 100, in communication with the environmental perception means 1, with the decision-making module 2 and with the human-machine interface 4; - a module 312 for receiving an information request RDP related to a decision DP to be taken, communicating with the decision-making module 2; - a module 313 for determining a selection BP of the information or parameters to be taken into account, - a module 314 for obtaining said selection of information in order to form a selective environment model M_ENV_S; - a module 315 for determining an integrity index associated with each item of information of the environment model M_ENV_S, and a step 316 of transmitting the selection of information M_ENV_S in collaboration with the decision-making module 2; This enables software including

[0052] A mode for carrying out the method for controlling an autonomous vehicle is described below with reference to figure 3. The method comprises six steps E1 to E6.

[0053] In a first step E1, an overall environment model M_ENV_G of the autonomous vehicle 100 is defined, comprising a set of information determined from data provided by the environmental perception means.

[0054] The overall environment model M_ENV_G has the following objectives: - providing a sufficiently accurate and complete set of information about the environment of the autonomous vehicle to enable the decision-making module 2 to operate, in particular to enable the decision to be taken to be identified; - Ensure consistency of information It is constructed according to

[0055] For this purpose, a calculated completeness index IC is assigned to each item of information of the model M_ENV_G. The calculation of the index IC can take into account an initial completeness index of the data, the initial completeness index being provided by the means for perceiving said data.

[0056] The index IC also takes into account the consistency of the data: in particular, for each data that may be provided by at least two distinct and independent perception means, the index IC is determined depending on the difference between the respective measurements of each of the perception means and / or depending on the reliability index assigned to each of the measurements.

[0057] The consistency check also concerns the temporal changes of the respective data: for example, the sudden appearance or disappearance of objects, as well as highly unlikely trajectories of elements of the scene are detected. For example, the consistency check may concern the magnitude of the speed of movement of vehicles or pedestrians.

[0058] The consistency check may also refer to the operational limits of the driver assistance system 10 in a broader sense. Indeed, the driver assistance system 10 is defined to operate in a traffic environment that checks very specific criteria, formalized in the literature called ODD (Operational Design Domain). For example, the driver assistance system 10 may be designed to operate in a controlled environment in which no pedestrians are expected to move. In this case, the decision-making module 2 is not designed to be able to interact with pedestrians, but the perception system is able to detect them. Therefore, the presence of a pedestrian is detected in step E1 as not fitting the traffic environment for which the autonomous vehicle 100 was designed.

[0059] Detecting an incompatibility in the data or information may trigger a warning message to be sent to the decision making module 2 and / or to the human machine interface 4. The warning message may specify the incompatible data items.

[0060] Thereby, in step E1, each time data is received from the perception means, the overall environmental model M_ENV_G is refined and updated, which environmental model M_ENV_G is then stored in the memory 32.

[0061] As long as no information request RDP is sent by the decision-making module 2, the method loops back to step E1 of defining the overall environment model M_ENV_G.

[0062] When an information request RDP has been sent by the decision-making module 2, the method transitions to a step E2 of receiving an information request related to a decision to be taken for controlling the vehicle.

[0063] In step E2, the content of the received information request RDP is analyzed.

[0064] Advantageously, a predefined list LD of decisions is stored in the memory 32. In fact, there is a finite number of decisions that the vehicle can take. For example, at a tactical level, the decisions can be to cross an intersection, to change lanes, to stay in the same lane, etc.

[0065] In one embodiment, the list LD identifies each decision to be taken using an identifier iDP, and the information request RDP advantageously includes the decision identifier iDP.

[0066] The information requirement RDP may also include items of information that specify the decision level, ie, tactical, operational, or strategic.

[0067] The information request RDP may also include information related to the route IT to be followed by the autonomous vehicle 100.

[0068] The different information contained in the information request RDP can be stored in the memory 32 and used when performing the following steps E3 to E6.

[0069] In step E3, a selection of information required to make the decision DP is determined, the selection being a subset of the set of information contained in the overall environment model M_ENV_G.

[0070] Advantageously, the memory 32 comprises a correspondence table between, on the one hand, an identifier iDP of a decision to be taken and, on the other hand, a list containing the information required to take the decision represented by the index iDP, which enables a list BP of information required for the decision-making module 2 to take the decision DP to be determined.

[0071] Throughout the remainder of this document, said list of information BP is called "blueprint". A blueprint BP is associated to each decision DP to be made, in other words to each decision identifier iDP. The blueprint BP formalizes and structures each item of information for which the decision-making module 2 needs to evaluate a decision and then send commands to the autonomous control means 20 for controlling the autonomous vehicle 100. A blueprint can be considered as a blank form, each field of which has to be completed, and the fields to be completed are specific to each blueprint, in other words to each decision DP to be made.

[0072] Blueprints may depend on the level of the decision (tactical, operational or strategic). In particular, for a given decision, one blueprint may exist for each decision level.

[0073] For example, in the case of the decision to enter an intersection, the first tactical level blueprint BP1 defines the information required to make the decision to enter the intersection at the time the vehicle arrives at the intersection. This blueprint defines the following information: - traffic conditions in the driving lane of the autonomous vehicle 100; - a list of relevant zones (e.g. alternative lanes for entering an intersection) and the traffic conditions in these zones; - a list of preferred vehicles for the autonomous vehicle 100; - a list of non-priority vehicles for the autonomous vehicle 100; - The presence of traffic lights and the status of these lights may include.

[0074] Further referring to the case of deciding to enter an intersection, a second blueprint BP2 different from the first blueprint BP1 may be defined for the operational level, and optionally a third blueprint BP3 different from the first and second blueprints BP1, BP2 may be defined for the strategic level. For example, blueprint BP3 (strategic level) may take into account congested road sections or blocked roads that may impair the route calculated at the start of the mission.

[0075] The list of required information, or blueprint, can include predictions. Throughout the remainder of this document, the term prediction is used to represent a predictable event that may affect a decision to be made by autonomous vehicle 100. For example, a prediction may relate to a predictable behavior of a nearby vehicle that may affect the behavior of autonomous vehicle 100. The term "prediction" therefore does not only relate to trajectory predictions. The term "prediction" may relate, for example, to a prediction of the intent of a nearby vehicle, or a prediction of the route of a nearby vehicle.

[0076] The list of information required to make the decision DP may include a set of information characterizing the situation of the vehicle after applying the decision DP to be made.

[0077] The prediction can then take into account the route IT provided by the decision-making module 2 in relation to the decision DP to be taken.

[0078] The situation shown in FIG. 5 illustrates the advantage of including predictions in the blueprint associated with the decision DP to be made.

[0079] Figure 5 shows a first diagram (located on the left side of the diagram) showing the autonomous vehicle 100 at a time t as it moves towards an intersection that it must cross in a straight line, and a second diagram (located on the right side of the diagram) showing the autonomous vehicle 100 at time t+Δt, when the autonomous vehicle 100 has almost reached the intersection; Includes.

[0080] Shown on each diagram are the three zones Z1, Z2, Z3 of interest for the autonomous vehicle 100, as well as the visibility Visi(t) or Visi(t+Δt) of the autonomous vehicle over these three zones at time t and time t+Δt, respectively.

[0081] In the first diagram showing the situation of the autonomous vehicle at time t, a building located on the corner of the intersection obscures the visibility Visi(t) over the relevant zone Z3 located to the right of the autonomous vehicle 100.

[0082] In the second diagram showing the situation of the autonomous vehicle at t+Δt, the zone Z3 in question to the right of the autonomous vehicle is now in the visibility zone Visi(t+Δt).

[0083] Thereby, although information about the relevant zone Z3 is missing at time t, a prediction at t+Δt enables the decision-making module to ensure that all relevant zones are covered by the perception means 1 when the autonomous vehicle 100 reaches the intersection.

[0084] Prediction also enables the decision-making module to anticipate possible events after applying the decision DP.

[0085] For example, in the situation illustrated in FIG. At the time t of evaluating the decision to be taken, which is shown by the diagram located on the left, the navigation path of the autonomous vehicle 100 is completely clear and a second vehicle is traveling in the oncoming lane, At prediction time t+Δt, a second vehicle moves into the lane of autonomous vehicle 100 to avoid an obstacle, creating a collision risk.

[0086] This means that the situation is safe at time t, but decision-making module 2 knows that a second vehicle will enter the lane of autonomous vehicle 100, possibly avoiding an obstacle, and it is reasonable to estimate when this risk is likely to occur.

[0087] When the selection of information associated with the decision DP to be taken, in other words the blueprint, has been determined, the method proceeds to step E4.

[0088] In step E4, said selection of information is obtained from data originating from the environmental perception means, in other words, a blueprint, a type of form containing the fields to be taken into account, is taken into account based on the information of the environmental model M_ENV_G.

[0089] In an advantageous embodiment, the data is obtained according to the same method regardless of the blueprint.

[0090] In this embodiment, a set of identity functions is defined. By way of non-limiting example, the identity functions may be: - Identifying the zone in question; - identifying a vehicle preceding the autonomous vehicle 100 on the lane of the autonomous vehicle 100; - identifying one or more vehicles that have priority over the autonomous vehicle 100 at the intersection; - Predicting the trajectory of nearby vehicles, etc. can be defined respectively for

[0091] Each unit function uses information from the overall environment model M_ENV_G to build a structured representation of a portion of the autonomous vehicle 100's current situation.

[0092] To complete the blueprint, a selection of unit functions needs to be executed, each unit function allowing information about the blueprint to be taken into account. Additionally, unit functions may need to be executed in a well-defined order. For example, to take into account the representation of an intersection at the tactical level, the zone in question may need to be identified before identifying road users that have priority over the autonomous vehicle 100.

[0093] For this purpose, a loadsheet is associated with each blueprint to describe the order for executing the unit functions taking into account the data of the blueprint. A supervisor (or orchestrator) executes the unit functions in the order defined by the loadsheet.

[0094] For example, when the decision DP to be made concerns the commitment of the autonomous vehicle 100 at an intersection, the loadsheet may include the following order for executing the unit functions: - obtaining the state (position, velocity, etc.) of the autonomous vehicle 100; - obtaining the state of the visual field of the perceptual system; - obtaining information about the presence of other road users around the autonomous vehicle 100 as well as the status of other road users; - obtaining the zones in question at the intersection, their location and / or their spatial distribution; - Contextualizing the zone (i.e., is it visible, hidden, out of range, etc.); - contextualizing other road users present in the zone; - Identifying road users who should be given priority; - Identifying road users to whom the right of way should be yielded; - etc. can be determined.

[0095] The set of primitive functions is scalable, in other words, each of the functions can be modified without harming the overall architecture of the system. In addition, new functions can be added, for example to take new situations into account.

[0096] In a preferred embodiment, the step E4 of obtaining an integrity index and the step E5 of determining an integrity index are carried out successively for each unitary function. In other words, for each unitary function, the method loops back to steps E4 and E5. Each unitary function can, for example, calculate its integrity index from the integrity index associated with each item of information of the overall environment model M_ENV_G defined in step E1.

[0097] In this embodiment, for each step in the loadsheet, the orchestrator is able to estimate the quality and completeness of the information gathered by the unit functions executing said step.

[0098] Based on the integrity index returned by each atomic function, the orchestrator updates the overall integrity index.

[0099] Moreover, the orchestrator can analyze whether the completeness index returned by the unit functions executed at this stage of the loadsheet is high enough to continue executing the loadsheet. This analysis is essential because if one of the unit functions gathers information that is incomplete or not accurate enough, subsequent unit functions may be affected.

[0100] Depending on the situation, the lack of information integrity may be managed by the orchestrator in different ways: The load sheet may be fully executed and then the decision-making module 2 may be informed of the low level of completeness of the information via the Overall Completeness Index calculated by the orchestrator, or The orchestrator can apply conditions to stop the execution of a loadsheet, for example, if the overall completeness index falls below a predefined completeness threshold associated with the blueprint, the orchestrator can pause the progress of the loadsheet.

[0101] Upon completion of the loadsheet execution, the information gathered by the identity functions together forms the selective environment model M_ENV_S.

[0102] In step E6, the decision-making module 2 is supplied with a selective environment model M_ENV_S, the model M_ENV_S comprising the units and the overall completeness index associated with the data.

[0103] Finally, the modeling method according to the present invention implements an efficient and reliable interpretation of the situation of the autonomous vehicle 100 depending on the decisions to be made by the autonomous vehicle 100.

[0104] actual, - the modelling according to the invention takes into account only the information required for decision making, - Modelling ensures that the information identified as needed is as complete as possible and that its reliability is known; - The decisions to be made result in different models depending on the deadline by which the decision has to be made, in other words according to the decision-making level (operational, tactical or strategic); - Modeling can include prediction to anticipate possible events after applying the decision to be made; - The consistency of information is checked in various ways, namely the consistency of items of information originating from several sources, the time consistency of information and the consistency of information with the operating limits of the system.

[0105] In addition, the proposed modeling method is generic and can be adapted to the requirements of various driver assistance and / or autonomous driving systems. Indeed, the method can be configured at different levels: - the first overall construction level is implemented on the list LD of possible decisions, - the second level of organization per decision is implemented by associating a blueprint with each possible decision (in other words, a set of information to be taken into account), making it possible to use different blueprints for each decision level (operational, tactical and strategic) for the same decision; - the third configuration level is implemented by defining a set of unit functions, each unit function allowing an item of information to be obtained and associated with an integrity index; The fourth configuration level concerns the order for obtaining the information, in other words the order for performing the unit functions, and the calculation of the overall completeness index for the information. The fourth configuration level is performed by defining the loadsheet associated with each blueprint.

[0106] This modular architecture allows the modeling to be easily configured for each decision by changing the unit functions used to obtain the list of decisions and / or the blueprints associated with each decision and / or the loadsheets and / or information associated with each blueprint.

[0107] Of course, as can be seen from the above description, the implemented modeling must be understood in the practical sense of automatically understanding the vehicle's environment with the aim of assisting the vehicle's driving in a reliable, robust and optimized manner. Therefore, the implemented modeling is not an abstract or virtual structure. In other words, the present invention must be understood as a method and device for assisting the driving of an automated vehicle, whether the automated vehicle is autonomous or not, implementing the method for modeling this environment as described above in order to take into account the navigation environment in one or more autonomous decisions for guiding the vehicle and also for assisting the driving of the driver of the vehicle.

Claims

1. A method for modeling a navigation environment of a vehicle (100) equipped with an environmental perception means (1), a decision-making module (2), and an autonomous control means (20) for autonomously controlling the vehicle, comprising: - defining a global environment model (M_ENV_G) of the vehicle as a set of structured information constructed from data provided by the environmental perception means (1) (step E1); - receiving a request (RDP) for information related to a decision (DP) to be made to control the vehicle by the autonomous control means (20), the request being transmitted by the decision-making module (2) (step E2); - determining a selection (BP) of information to be considered for making the decision (DP) from among the set of structured information defining the environmental model (step E3); - considering the selected information from the global environment model (M_ENV_G) (step E4); - determining a completeness index associated with each item of the selected information (step E5); - providing a selective environmental model (M_ENV_S) including the selected information and the corresponding completeness index to the decision-making module (2) (step E6). A method characterized by comprising the above steps.

2. The method according to claim 1, characterized by comprising the step of obtaining at least one first item of information from the selection of information using a first environmental perception means and using a second environmental perception means, and the completeness index associated with the at least one first item of information being determined according to the consistency between the data provided by the first perception means and the second perception means.

3. The method includes obtaining at least one second item of information from the selection of information using environmental perception means at a first time point and at a second time point after the first time point, and the integrity index associated with the at least one second item of information is determined according to the consistency between the data provided by the perception means at the first time point and at the second time point. A method for modeling a navigation environment of an autonomous vehicle according to claim 1 or 2, characterized in that.

4. The determination (DP) to be made includes a given action to be applied by the autonomous vehicle within a given deadline, and the given deadline can be short-term, medium-term, or long-term. For the same given action, the step (E3) of making the determination is - If the given deadline is short-term, determine a first selection (BP1) of information to be considered, or - If the given deadline is medium-term, determine a second selection (BP2) of information to be considered, or - If the given deadline is long-term, determine a third selection (BP3) of information to be considered Characterized in that The first selection, the second selection, and the third selection are different from each other. A method for modeling a navigation environment of an autonomous vehicle according to claim 1 or 2.

5. The method includes a sub-step of providing the decision-making module (2) with a selective environmental model (M_ENV_S) including a set of information characterizing the situation of the vehicle after applying the determination (DP) to be made. A method for modeling a navigation environment of an autonomous vehicle according to claim 1 or 2, characterized in that.

6. The method is - Constructing a finite list (LD) of decisions to be considered by the method; - For each decision in the finite list of decisions, constructing a selection (BP) of information to be considered and constructing the order in which the information must be considered; - Constructing at least one method for considering each item of information to be considered A method for modeling a navigation environment of an autonomous vehicle according to claim 1 or 2, characterized by including.

7. A device (10) for modeling a navigation environment of an autonomous vehicle, wherein the autonomous vehicle is equipped with an autonomous control means (20), and the device comprises hardware elements and / or software elements (1, 2, 3, 4, 10, 11, 12, 13, 14, 15, 20, 31, 32, 33, 311, 312, 313, 314, 315, 316) for implementing the method according to claim 1 or 2, in particular hardware elements and / or software elements (1, 2, 3, 4, 10, 11, 12, 13, 14, 15, 20, 31, 32, 33) designed to implement the method according to claim 1 or 2, and / or the device comprises means for implementing the method according to claim 1 or 2, device (10).

8. A computer program product comprising program code instructions stored on a computer-readable medium, the program code instructions being for performing the steps of the method according to claim 1 or 2 when the program runs on a computer, or a computer program product that is downloaded from a communication network and / or stored on a computer-readable data medium and / or executable by a computer, the computer program product comprising instructions for causing the computer to perform the method according to claim 1 or 2 when the program is executed by the computer, characterized in that it comprises instructions for causing the computer to perform the method according to claim 1 or 2 when the program is executed by the computer, computer program product.

9. A computer-readable data storage medium storing a computer program comprising program code instructions for performing the method according to claim 1 or 2, or a computer-readable storage medium comprising instructions for causing the computer to perform the method according to claim 1 or 2 when the computer-readable storage medium is executed by the computer.

10. A signal from a data medium carrying the computer program product according to claim 8.

Images