Method for supervising the autonomous driving capabilities of a set of motor vehicles

Abstract

The invention relates to a method for supervising the autonomous driving capabilities of a set of vehicles, the method being characterised in that it comprises: • a step of constructing, by a centralised supervision system, an overall model of a traffic environment of the vehicles of the set; • a step of transmitting, by a first vehicle of the set to the centralised supervision system, a planned route and a current state of its autonomous driving capabilities; • a step of detecting, by the centralised supervision system, incompatibility between the overall model, the current state of the autonomous driving capabilities of the first vehicle and the planned route of the first vehicle; • a step of transmitting, by the centralised supervision system to the first vehicle, an indication of incompatibility between the planned route and the current state of the autonomous driving capabilities of the first vehicle.

Description

Description Title of the invention: Method for supervising the autonomous driving capabilities of a set of motor vehicles.

[0001] The invention relates to a method for supervising the autonomous driving capabilities of a set of motor vehicles.

[0002] Smart vehicles are booming, including vehicles equipped with advanced driver assistance systems (ADAS) and autonomous navigation.

[0003] However, autonomous navigation capabilities are limited to a limited number of situations, depending on the vehicle's capabilities.

[0004] Furthermore, the situations in which the vehicle is likely to act autonomously must be formally defined first to ensure the safety of passengers and other road users.

[0005] Current systems rely on the autonomous vehicle self-assessing the suitability of its autonomous driving capabilities for the traffic conditions in which it will operate, with traffic conditions only being able to be assessed within a limited time and space horizon.

[0006] This solution has drawbacks. In particular, the autonomous vehicle's self-assessment of its traffic conditions does not allow it to sufficiently anticipate the occurrence of situations that cannot be managed by the autonomous vehicle under the conditions for which it was designed.

[0007] The aim of the invention is to provide a device and a method for supervising autonomous driving capabilities which overcome the above drawbacks and improve the devices and methods for supervising autonomous driving capabilities known from the prior art. In particular, the invention makes it possible to produce a device and a method which are simple and reliable and which allow the supervision of autonomous driving capabilities over an extended time and space horizon.

[0008] To this end, the invention relates to a method for supervising the autonomous driving capabilities of a set of vehicles, the autonomous driving capabilities of each of said vehicles grouping together a set of driving conditions that said vehicle is capable of managing autonomously, each vehicle being equipped with a means of perceiving its environment, and a means of communication with a centralized supervision system for the vehicles of the set, the method comprising • a step of construction by the centralized supervision system of a global model of a vehicle circulation environment of the whole, the model global integrating data from the perception means of the vehicles of the whole, • a step of transmission, by a first vehicle of the set to the centralized supervision system, of a planned route and a current state of its autonomous driving capabilities, • a step of detection, by the centralized supervision system, of an incompatibility between the global model, the current state of the autonomous driving capabilities of said first vehicle and the planned route of said first vehicle, • a step of transmission, by the centralized supervision system, to the first vehicle, of an indication of an incompatibility between the planned route and the current state of the autonomous driving capabilities of said first vehicle.

[0009] In one embodiment, the method further comprises a step of transmitting the global model by the centralized supervision system to each vehicle in the set.

[0010] In one embodiment, the method further comprises monitoring, by the first vehicle, - a first compatibility between data from the perception means of the first vehicle and a current state of the autonomous driving capabilities of said first vehicle, the monitoring being carried out over a first time and space horizon determined by the perception means of said first vehicle, and - a second compatibility between the global model transmitted by the centralized supervision system and the current state of the autonomous driving capabilities of said first vehicle, the monitoring being carried out over a second time and space horizon, optionally, the second time and space horizon being more extensive than the first time and space horizon.

[0011] In one embodiment, the method further comprises a step of transmission by the first vehicle to the centralized supervision system of an indication of an incompatibility between the planned route and the current state of the autonomous driving capabilities of said first vehicle.

[0012] In one embodiment, the first vehicle comprises a set of systems, and the method according to the invention comprises - an update, by said first vehicle, of the current state of its autonomous driving capabilities based on a state of the systems of the set of systems, followed by - transmission of a current state of its autonomous driving capabilities to the centralized supervision system.

[0013] In one embodiment, the method further comprises a step of transmission by the first vehicle to the centralized supervision system of a request for information relating to an area of ​​the global model.

[0014] In one embodiment, the step of building the global model comprises receiving information from a meteorological center and / or an emergency service and / or a municipality.

[0015] In one embodiment, the step of transmitting to the first vehicle an incompatibility between its planned route and its autonomous driving capabilities comprises a transmission by the centralized supervision system of an alternative route compatible with the current state of the autonomous driving capabilities of said first vehicle.

[0016] In one embodiment, the method comprises, following transmission of information by the supervision system to the first vehicle, - a step of observation by the supervision system of an effect induced on the behavior of said first vehicle by the transmitted information, - an automatic learning step by the supervision system taking into account the information transmitted and the induced effect.

[0017] In one embodiment, the transmitted information is a global model of a traffic environment of the first vehicle and / or an alternative route applicable by the first vehicle and / or an indication of an incompatibility between the planned route and the current state of the autonomous driving capabilities of the first vehicle.

[0018] The attached drawing represents, by way of example, an embodiment of a supervision device according to the invention and a mode of execution of a supervision method according to the invention.

[0019] [Fig.l] [Fig.l] illustrates a high-level architecture of a supervision device according to the invention.

[0020] [Fig.2] [Fig.2] is a flowchart of a mode of execution of a supervision method according to the invention.

[0021] [Fig.3] [Fig.3] illustrates a functional architecture of a supervision device according to the invention.

[0022] An example of an embodiment of a device for monitoring the autonomous driving capabilities of a set of motor vehicles is described below with reference to Figures 1 to 3.

[0023] In the rest of the document, the acronym ODD (relating to the English expression "Operational Design Domain") is used to name the set of situations and conditions for which an autonomous system or vehicle has been designed, and in particular the situations and conditions for which its operation and safety must be guaranteed. For example, an ODD may be limited to autonomous navigation on a highway, in sunny weather, without changing lanes, the lanes being delimited by road markings.

[0024] An ODD of an autonomous vehicle is advantageously defined by a taxonomy of data among which - data relating to road infrastructure, in particular the types of roads that can be used by the autonomous vehicle (motorway, national, departmental, etc.), the types of developments that can be taken into account by the autonomous vehicle (roundabout, interchange), road markings of traffic lanes necessary for the implementation of autonomous driving, - weather data, for example whether or not the autonomous vehicle can move autonomously when it is raining, snowing, or foggy, - light conditions, in particular the possibility for an autonomous vehicle to move at night on an unlit road, - data relating to the types of vehicles or pedestrians likely to be located in the traffic environment of the autonomous vehicle.

[0025] The ODD of a system or vehicle may change during vehicle use, for example, if the operation of equipment or a system fitted to the vehicle deteriorates. This is referred to as an ODD modification. For example, a vehicle's ODD may include driving in snowy weather. Then, while moving, the autonomous vehicle may detect a lack of tire grip on snow. In this case, the autonomous vehicle modifies its ODD to remove the possibility of driving in snowy weather.

[0026] In addition, the acronym OD (relating to the expression Operational Domain) is used to name the environment in which an autonomous vehicle actually operates at a given time. An autonomous vehicle normally circulates in an OD compatible with its ODD. However, it is inevitable that occasionally an OD is modified and therefore no longer compatible with the vehicle's ODD. For example, roadworks or the presence of unexpected obstacles (pedestrians on the highway) can modify an OD. All these potential changes must be detected to guarantee the safety of the vehicle and its environment. In the rest of the document, a modification of the OD making it incompatible with the ODD is called "ODD exit".

[0027] It is important that an ODD change or an ODD exit is detected, so that the autonomous vehicle's path is modified to make the vehicle's OD compatible with its ODD.

[0028] [Fig.l] illustrates an overall architecture of a device for supervising autonomous driving capabilities according to the invention. The device 10 comprises a centralized supervision system 50 communicating with a local supervision system 60 embedded in each of the autonomous vehicles 101, 102, 103 of a set 100 of autonomous vehicles.

[0029] In the rest of the document, - the centralized supervision system 50 may be called “system 50” or “centralized system 50”, and - the local supervision system 60 may be called “system 60” or “local system 60”.

[0030] The set of vehicles 100 may also be referred to as “fleet 100” in the remainder of the document.

[0031] Each of the motor vehicles 101, 102, 103 may be a motor vehicle of any type, including a passenger vehicle, a utility vehicle, a truck, or a public transport vehicle such as a bus or a shuttle. According to the embodiment described, a motor vehicle of the fleet 100 is an autonomous vehicle and will be referred to as an "autonomous vehicle" in the remainder of the description.

[0032] This illustration is made for non-limiting purposes. In particular, a motor vehicle in fleet 100 could be a non-autonomous vehicle, equipped with a driving assistance system, in particular a driving assistance system corresponding to a level greater than or equal to level 2 of autonomy, i.e. corresponding to partial autonomy of the vehicle.

[0033] Each vehicle of the fleet 100 is equipped with at least one perception means 601 with which it perceives an OD called “local OD”. Each vehicle of the fleet 100 is further equipped with a means of communication 611 with the centralized supervision system 50.

[0034] Each vehicle of the fleet 100 further comprises a set of subsystems including, for example, a vehicle movement control system, a decision-making system, or a perception data processing system. Advantageously, each vehicle comprises a means for monitoring the state of each subsystem, in order to ensure that the current state of the vehicle is consistent with its ODD.

[0035] The centralized supervision system 50 can be seen as a control tower whose role is to facilitate the operation and navigation of the vehicles of the fleet 100, by means of monitoring the latter, as well as by collecting information provided by a set of information sources, in particular - information transmitted by fleet operators 70, - information from a weather data server 80, - information concerning the road network, which may be provided by municipal and / or regional services. - information from the emergency services of the police and / or fire brigade 90.

[0036] In addition, the centralized supervision system 50 is capable of receiving and processing information transmitted and updated regularly by each of the vehicles in the fleet 100, including in particular - a state of the vehicle, for example its position and speed, - a local OD perceived by the vehicle at a current time, - an indication of the compatibility of the vehicle's current F ODD with the local OD perceived at the current time, and - information on events relating to the local OD calling into question this compatibility.

[0037] The centralized supervision system 50 is able to use the information collected - in particular the local ODs and the information received from the different sources - to generate and update a global OD. It then transmits the updates of the global OD to the vehicles of the fleet 100.

[0038] In addition, from the global OD, the centralized supervision system 50 is able to verify that the ODD of each supervised vehicle is compatible with the portion of the global OD on which it is moving (or is about to move). In particular, the verification of the compatibility of the ODD of a vehicle with the global OD can relate to the entire route planned for the vehicle.

[0039] A mode of execution of a supervision method according to the invention is described below with reference to [Fig.2],

[0040] In a first step E1, the centralized supervision system 50 constructs a global model OD_global of a traffic environment of the vehicles 101, 102, 103 of the set 100.

[0041] In the rest of the document, the global model OD_global is called “OD global”.

[0042] The first step El includes an initialization of the global OD based on a digital map stored in memory. The digital map includes in particular digital maps of the road network describing the type of road (motorway, city, etc.), the type of infrastructure encountered (intersections, traffic lights, level crossings, etc.). These are static elements, which change only very rarely.

[0043] The first step E1 comprises an update of the global OD by integrating data OD101, OD102, OD103 from the perception means of the vehicles 101, 102, 103 of the set 100. In other words, the first step comprises an integration of information sent back by the vehicles of the fleet 100. This information mainly concerns events that may call into question the proper functioning of the vehicles of the fleet 100. For example, if a vehicle of the fleet perceives an element whose presence is improbable, or inducing a non-nominal adaptation on the part of the autonomous navigation system or the driver, the information is immediately transmitted to the centralized system 50. For example, the information transmitted may concern the presence of a bicycle or a pedestrian on the highway, the deactivation of traffic lights, the presence of a lorry delivery requiring its bypass by driving on an oncoming lane, etc. These events are systematically geolocated on the map.

[0044] The first stage 11 also includes an update of the global OD based on information from services to which the centralized system has subscribed, the services being able to allow it to access information on the weather, on the state of traffic, on works or events taking place in certain areas, etc. The information can also include occurrences of events that can impact traffic conditions, such as an accident, the passage of emergency services (police, firefighters).

[0045] In one embodiment, step E1 may comprise an aggregation of information relating to events transmitted by different vehicles of the fleet 100, the aggregation being able to be done on the basis of the temporality of the events reported by the vehicles of the fleet to the centralized system 50. For example, a roundabout may be easily negotiable in the case of light traffic, and difficult to negotiate in the case of heavy traffic. If the same information is reported several times by different vehicles, an aggregation of the information relating to the entry onto the roundabout may be used by an algorithm executed by the centralized system 50, to model, for example, the overall OD as a function of a density of traffic circulating at the roundabout, and / or as a function of any other parameter, for example a meteorological criterion.This information is then integrated into the map, the update of which is shared with the fleet 100, by transmitting the global OD to the vehicles of the fleet 100. In other words, step E1 includes a transmission of the updates of the global model global OD to the vehicles of the fleet 100.

[0046] Step E1 may further comprise the implementation of an event manager whose role is to merge the events transmitted by the different information sources, including in particular the vehicles of the fleet 100 and the information services. After being merged, the information is then projected onto the digital map in order to obtain a mapping of the overall OD, that is to say a description of the overall OD which can be located, for example by zone, or contextualized according to the road axes.

[0047] Advantageously, in step E1, following receipt of the global OD, the vehicles of the fleet can transmit requests for information to the centralized system 50, in particular when they need information concerning an area of ​​the global OD whose state depends on certain parameters not controlled locally by the vehicle.

[0048] In parallel with step El, or following step El, in a step E2, at least a first vehicle 101 of the fleet 100 transmits its planned route as well as a current status of its ODD to the centralized system 50.

[0049] Prior to the transmission of its planned itinerary and its updated SDG, step E2 advantageously comprises a sub-step of monitoring by the first vehicle 101, - a first compatibility between data from the perception means 601 of the first vehicle and a current state of the autonomous driving capabilities ODD 101 of the first vehicle, the monitoring being carried out over a first time and space horizon determined by the perception means 601 of the first vehicle, and - a second compatibility between the global model transmitted by the centralized supervision system, i.e. the global OD, and the current state of the autonomous driving capabilities ODD101 of the first vehicle, the monitoring being carried out over a second time and space horizon.

[0050] Advantageously, the second temporal and spatial horizon is more extensive than the first temporal and spatial horizon. Indeed, the first temporal and spatial horizon is limited by the capabilities of the perception means 601 of the first vehicle 101, while the second temporal and spatial horizon is determined by the geographical and temporal scope of all the data sources 70, 80, 90, 101, 102, 103 of the centralized system 50, the data sources comprising perception means of each vehicle of the fleet 100, and data from different services (meteorological data, traffic data, data linked to municipal events, works) which may cover a very large area.

[0051] In order to evaluate the first compatibility (between its ODD and data from its means of perception), the first vehicle 101 uses data from a localization system to locate itself on a digital map of the road network. It then correlates the data from its means of perception with the information from the digital map of the road network. The first vehicle thus constructs a representation of the world with which it interacts, called “World Model” in the remainder of the document. The World Model can take different forms, depending on the level of abstraction with which the environment is represented. At a very low abstract level, the representation can be made in the form of an occupancy grid, of polygons representing the free space in front of the vehicle.At a more abstract level, the perceived objects can be represented for example by a simple point, to which semantic labels are associated, the relevant information being the representation of the interactions between these objects and the first vehicle.

[0052] The information structured in a World Model format is then converted into a format representing an OD, i.e. a specification of the minimum capabilities that an autonomous vehicle must have to navigate in the area described by the World Model. This processing is called "OD distinction".

[0053] For example, if pedestrians are present in the environment of the autonomous vehicle, the presence of pedestrians will be described in World Model. When processing distinction of the OD, it will be specified in the current OD of the vehicle that the vehicle must be able to identify pedestrians in its environment.

[0054] The OD thus determined by the first vehicle is a local OD, since it is limited by the range of the first vehicle's means of perception.

[0055] It is then verified that each minimum capacity described in the local OD is included in the ODD of the first vehicle. If this is not the case, the first vehicle transmits to the centralized system 50 an indication of a first incompatibility between the local OD and the ODD of the first vehicle.

[0056] In addition, the local OD can advantageously be enriched by data from a global OD previously transmitted by the centralized system 50 to the first vehicle. In this case, monitoring of the second compatibility is implemented, between the global OD, and the current state of the autonomous driving capabilities ODD of the given vehicle, the monitoring being carried out over a second time and space horizon. If an incompatibility is detected, the first vehicle transmits to the centralized system 50 an indication of a second incompatibility between the current global OD and the current ODD of the first vehicle.

[0057] In addition, the at least one first vehicle 101 comprises a set of systems and, upstream of the transmission of its planned route and its updated ODD, in step E2 the at least one first vehicle implements - an update of the current state of its autonomous driving capabilities based on a state of the systems of the set of systems, - followed by a transmission of a current state of its autonomous driving capabilities ODD to the centralized supervision system 50.

[0058] Indeed, the first vehicle has a representation of its nominal ODD corresponding to a state of the vehicle's ODD when each of the systems contributing to the autonomous navigation system is able to operate nominally.

[0059] In an advantageous embodiment, each system in the set of systems is capable of establishing at any time a report on its operating capacities, which it transmits to the local supervision system.

[0060] For example, in its nominal operation, a multi-sensor perception system must be able to perceive and classify pedestrians. The loss of one of the camera-type sensors can call into question the classification of perceived objects, and therefore the ability to identify pedestrians. Thus, when the multi-sensor perception system of the first vehicle detects the loss of a sensor, it transmits a message to the local supervision system, which modifies the ODD of the first vehicle to remove the ability to identify a pedestrian.

[0061] Then, this real-time ODD update is used to evaluate whether the ODD is compatible with the current OD and the planned route of the first vehicle.

[0062] Thus, the self-monitoring treatments implemented by the first vehicle in step E2 can allow it to detect an ODD departure, i.e. an incompatibility between its ODD and its OD or its planned route. As far as possible, the first vehicle will modify its route so as to make it compatible with its ODD.

[0063] However, bringing the route of the first vehicle into compliance with its ODD is not always possible. In this case, step E2 comprises a sub-step of transmission by the vehicle to the centralized supervision system 50 of an indication of an incompatibility between the planned route and the current state of its autonomous driving capabilities.

[0064] Then in a step E3, the centralized supervision system 50 detects an incompatibility between - the overall OD, and / or - the current state of the ODD autonomous driving capabilities of the first vehicle, and / or - the planned route of the first vehicle.

[0065] Indeed, thanks to - on the one hand the construction and maintenance of the global OD by the centralized system 50 in step El, - and on the other hand, the transmission of the updates of the ODDs of the vehicles of the fleet to the centralized system 50 implemented in step E2, in step E3, the centralized system 50 can supervise the compatibility of the ODD of each of the vehicles with the global OD, that is to say the compatibility of the ODD over the entire operating domain of the vehicles and over a longer time horizon than the local supervision implemented in each of the vehicles allows.

[0066] The content of the current ODD of each vehicle is known at all times by the centralized system 50. The latter also has a history of modifications to the supervised ODD, which allows it to further predict events likely to occur, in particular disruptive events.

[0067] The centralized system 50 having a global representation of the OD of the area it covers, and the up-to-date ODD of each vehicle in the fleet 100, as well as the planned route for each of the vehicles in the fleet 100, it holds the information necessary to carry out global supervision of the ODD of each vehicle in the fleet 100.

[0068] Global ODD monitoring can be carried out using the same process as local ODD monitoring, except that the environmental data available to the centralized system 50 is more comprehensive than the information available to each vehicle when it locally monitors its ODD. Thanks to centralized monitoring, the estimation of an ODD output can be carried out earlier than if it were done locally, that is, individually by the vehicles of the 100 fleet.

[0069] Then in a step E4, the centralized supervision system transmits to the first vehicle an indication of an incompatibility between the planned route and the current state of the ODD autonomous driving capabilities of the first vehicle.

[0070] In other words, when a probable ODD exit is detected for a vehicle in the fleet, the centralized management system 50 informs the local system 60 of the vehicle concerned.

[0071] In one embodiment, step E4 comprises a transmission to the first vehicle 101 by the centralized supervision system of an alternative route compatible with the current state of the autonomous driving capabilities of the first vehicle 101.

[0072] Indeed, the centralized system 50 has the capacity to determine an alternative route of the first vehicle (if this route exists), the alternative route defining an OD compatible with the ODD of the first vehicle 101. This route is subsequently transmitted to the first vehicle 101, as is the description of the associated OD.

[0073] In one embodiment, following step E4, we continue with a step E5 comprising - a sub-step of observation by the supervision system of an effect induced, on the behavior of the first vehicle, by the information transmitted, - an automatic learning step by the supervision system taking into account the information transmitted and the induced effect.

[0074] The information transmitted may be a global model of a traffic environment of the vehicles of the first set, i.e. a global OD, and / or an alternative route applicable by the first vehicle and / or an indication of an incompatibility between a route planned by the first vehicle and the current state of the autonomous driving capabilities ODD101 of the first vehicle.

[0075] The sub-step of observing an induced effect may concern, for example, the occurrences of the following events: - a first event corresponding to the first vehicle remaining on its initial route despite an indication of an ODD exit transmitted by the centralized system 50, - a second event corresponding to a decision by the first vehicle to apply or not an alternative route transmitted by the centralized system 50, this decision possibly involving a driver of the first vehicle, - a third event corresponding to a possible modification by the first vehicle of the alternative route before application, - a fourth event corresponding to a transmission by the first vehicle to the centralized system 50 of a request for information on an area of ​​the global OD, - a fifth event corresponding to an indication by the first vehicle to the centralized system 50 of an incompatibility between the alternative route and the ODD of the first vehicle.

[0076] The machine learning sub-step can include building a model that integrates these different events. The model can also integrate data relating to the movement history of each vehicle, a history of ODD changes, traffic density, weather conditions, etc.

[0077] [Fig.3] summarizes, in the form of a functional diagram, the processing described in the steps of the supervision method according to the invention.

[0078] The centralized system 50 receives information relating to the supervised area and from different sources 501, including updated mapping, weather information, and / or optionally information from a municipality. The centralized system 50 further receives events from different sources 502, the events possibly including events related to emergency services (fire brigade, police), events concerning road traffic, or even events relating to the vehicles of the fleet 100.

[0079] An event manager 504 processes the events from the sources 502 to locate them in space and time. An information fusion module 505 integrates all the information and events received by the centralized system 50, so that they can be projected onto an OD map in the module 506, the module 506 using for this purpose information from a high-definition map 503 available in memory and information relating to the ODs of the vehicles in the fleet.

[0080] In parallel, in a module 507, the centralized system 50 compiles all the ODDs of the fleet 100. In a module 508, the centralized system 50 supervises the ODDs, i.e., checks the compatibility of the current OD of each vehicle (from the module 506) with the ODD of the vehicle. When an incompatibility between the OD and the ODD of a vehicle is detected, then in the module 509, the centralized system can propose an alternative route to the vehicle.

[0081] Then, in module 509, the centralized system 50 self-evaluates, i.e. it evaluates the performance of the ODD supervision. Then, in module 510, the centralized system 50 improves its supervision processing based on the results from module 509.

[0082] Meanwhile, the local systems 60 of the vehicles of the fleet 100 exchange information with the centralized system 50.

[0083] From the data from a perception module 601, a localization module 602 and a map 603, in a module 605, the local system 60 of the vehicle 101 constructs its World Model. The World Model is then transmitted to the module 607 in which the OD is “distinguished”, i.e. the capacities are determined. minimum requirements that the vehicle must have to navigate autonomously in the environment described by the World Model.

[0084] In addition, a nominal ODD of the vehicle 101 is available in a memory 604. From the nominal ODD, and data from a module 606 for monitoring the vehicle subsystems, the local system 60 maintains a current ODD of the vehicle.

[0085] The consistency of the current ODD with the current OD of the vehicle 101 is then checked in a module 609. When an inconsistency is detected, in the module 609 a message is transmitted informing the centralized system 50 of an incompatibility between the current OD of the vehicle 101 and its current ODD. In the module 609, the local system 60 further transmits the current OD of the vehicle to the centralized system 60 (whether or not it is compatible with the ODD).

[0086] In a module 610, the local system 60 manages the missions of the vehicle 101. In particular, it receives and possibly takes into account alternative routes transmitted by the centralized system 50, following an ODD exit.

[0087] Finally, the supervision system according to the invention makes it possible to extend the anticipation time of an ODD exit of an autonomous vehicle, by centralizing data from multiple sources in a system supervising all the vehicles of a fleet. Indeed, thanks to the centralization of the data, each vehicle of the fleet benefits in particular from the information transmitted by the other vehicles of the fleet, the supervision of ODD then being able to be carried out over a longer time and spatial horizon than if the supervision were carried out locally by each vehicle.

[0088] Extending the anticipation time for an ODD exit makes it possible to react further in advance of the ODD exit, for example by modifying the route of the autonomous vehicle, and thus avoid the vehicle becoming blocked.

[0089] The transmission of the ODD status by each of the vehicles in the fleet to the centralized system allows the centralized system to keep the ODD of all vehicles in the fleet up to date.

[0090] In addition, the transmission of the perceived OD and events perceived by each of the fleet vehicles also makes it possible to update a representation of the supervised OD. The update of the supervised OD also integrates a compilation and fusion of information acquired through remote services. All this information makes it possible to generate a mapped representation of the OD of the fleet's activity area. The updates of the supervised OD are transmitted to the fleet vehicles.

[0091] Furthermore, thanks to a centralization of information from the fleet vehicles, the supervision system according to the invention makes it possible to model the OD of certain zones according to parameters such as traffic density, time, etc.

[0092] The supervision system according to the invention also makes it possible to calculate, for a given vehicle, an alternative route (or any other recommendation) based on the compatibility of the vehicle's ODD and the OD of the areas which are to be traveled by the vehicle.

[0093] In addition, the supervision system according to the invention implements an evaluation and continuous improvement of the recommendations (in particular a change of route) that it sends to the vehicles of the fleet. To do this, it can use all the data from the data sources, as well as a behavior induced on the vehicles by the recommendations.

Claims

Claims

1. Method for supervising the autonomous driving capabilities (ODD) of a set (100) of vehicles (101, 102, 103), the autonomous driving capabilities (ODD101, ODD102, ODD103) of each of said vehicles grouping together a set of driving conditions that said vehicle is capable of managing autonomously, each vehicle being equipped with a means of perception (601) of its environment, and a means of communication (611) with a centralized supervision system (50) of the vehicles (101, 102, 103) of the set (100), characterized in that it comprises: • a step (El) of construction by the centralized supervision system (50) of a global model (OD_global) of a traffic environment of the vehicles of the set, the global model integrating data (OD101, OD102, OD103) from the perception means (601) of the vehicles (101, 102, 103) of the set (100), • a step (E2) of transmission, by a first vehicle (101) of the assembly (100) to the centralized supervision system (50), of a planned route and a current state of its autonomous driving capabilities (ODD 101), • a step (E3) of detection, by the centralized supervision system (50), of an incompatibility between the global model (OD_global), the current state of the autonomous driving capabilities (ODD 101) of said first vehicle (101) and the planned route of said first vehicle, • a step of transmission, by the centralized supervision system (50), to the first vehicle (101), of an indication of an incompatibility between the planned route and the current state of the autonomous driving capabilities (ODD 101) of said first vehicle (101).

2. Supervision method according to the preceding claim, characterized in that it further comprises a step of transmission of the global model (OD_global) by the centralized supervision system (50) to each vehicle (101, 102, 103) of the set (100).

3. Supervision method according to the preceding claim, characterized in which includes surveillance, by the first vehicle (101), - a first compatibility between data (0D101) from the perception means (601) of the first vehicle and a current state of the autonomous driving capabilities (ODD 101) of said first vehicle (101), the monitoring being carried out over a first time and space horizon determined by the perception means (601) of said first vehicle, and - a second compatibility between the global model (OD_global) transmitted by the centralized supervision system (50) and the current state of the autonomous driving capabilities (ODD 101) of said first vehicle, the monitoring being carried out over a second time and space horizon, optionally, the second time and space horizon being more extensive than the first time and space horizon.

4. Supervision method according to the preceding claim, characterized in that it further comprises a step of transmission by the first vehicle (101) to the centralized supervision system (50) of an indication of an incompatibility between the planned route and the current state of the autonomous driving capabilities of said first vehicle (101).

5. Supervision method according to one of the preceding claims, the first vehicle (101) comprising a set of systems, characterized in that it comprises - an update, by said first vehicle (101), of the current state of its autonomous driving capabilities (ODD101) as a function of a state of the systems of the set of systems, followed by - a transmission of a current state of its autonomous driving capabilities (ODD101) to the centralized supervision system (50).

6. Supervision method according to one of the preceding claims, characterized in that it comprises a step of transmission by the first vehicle (101) to the centralized supervision system (50) of a request for information relating to an area of ​​the global model (OD_global).

7. Supervision method according to one of the preceding claims, characterized in that the step of constructing the global model (OD_global) comprises receiving information from a meteorological center and / or an emergency service and / or a municipality.

8. Supervision method according to one of the preceding claims, characterized in that the step of transmission to the first vehicle (101) of an incompatibility between its planned route and its autonomous driving capabilities (ODD 101) comprises a transmission by the centralized supervision system (50) of an alternative route compatible with the current state of the autonomous driving capabilities (ODD 101) of said first vehicle (101).

9. Supervision method according to one of the preceding claims, characterized in that it comprises, following transmission of information by the supervision system (50) to the first vehicle (101), - a step of observation by the supervision system (50) of an effect induced on the behavior of said first vehicle (101) by the transmitted information, - an automatic learning step by the supervision system (101) taking into account the information transmitted and the induced effect.

10. Supervision method according to claims 8 and 9, characterized in that the information transmitted is a global model (OD_global) of a traffic environment of the first vehicle (101) and / or an alternative route applicable by the first vehicle (101) and / or an indication of an incompatibility between the planned route and the current state of the autonomous driving capabilities (ODD 101) of the first vehicle (101).

Images