Method for modeling tactical environment of motor vehicle
By using high-definition maps and geolocation devices in automatic vehicles, dividing and decomposing interactive navigation lane groups and building tactical environment models, the problem of failure to effectively represent interactive lanes in the prior art is solved, and the automatic vehicles are realized All-round tactical environment modeling and decision-making support.
Patent Information
- Application Number
- CN202280089765.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2021-12-22
- Filing Date
- 2022-12-22
- Publication Date
- 2024-11-12
AI Technical Summary
The existing grid-occupying method of automatic vehicle navigation lanes is only aimed at navigation lanes moving on automatic vehicles, and fails to effectively represent other navigation lanes interacting with it, limiting the formulation of tactical decisions.
Through automatic motor vehicles equipped with high-definition maps and geolocation devices, determine the current location and possible road groups, divide the pass lane groups according to the predefined interactive mode groups, and break them into continuous, spliced fixed-pitch units, and build tactics Environmental model to support mid-term decision-making.
A comprehensive modeling of the tactical environment of automatic vehicles is achieved, including all navigation lanes interacting with it, improving the accuracy and reliability of tactical decision-making, and simplifying the processing of subsequent decision-making modules.
Smart Images

Figure CN118946780A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a method for modeling a tactical environment of a motor vehicle. The invention also relates to a device for modeling a tactical environment of a motor vehicle. The invention also relates to a computer program for implementing the method mentioned. The invention finally relates to a recording medium on which such a program is recorded. Background Art
[0002] Smart or autonomous vehicles need to continuously make decisions based on the situation they find themselves in. In the literature, three levels of decision making by an autonomous car are identified:
[0003] - first level decisions, called operational decisions, which involve decisions made within a very limited first time window, for example in the order of milliseconds to 1 second,
[0004] - a second level of decision making, known as tactical decision making, which involves medium-term decisions, such as those made within a time interval greater than one second and limited to approximately 10 seconds,
[0005] - The third level of decision making, known as strategic decision making, involves long-term decisions, such as decisions made over a period of time greater than 10 seconds relative to the present moment.
[0006] Operational decisions are made at high frequency based on a fine and precise representation of the vehicle's immediate environment. Strategic decisions, for their part, are made at low frequency and require only high granularity of the vehicle's environment over a range of perhaps more than a kilometer. Both types of environment representations, fine for operational decisions or high granularity for strategic decisions, are already widely used. On the other hand, environment representations dedicated to tactical decision making are less common.
[0007] From document US 2018 / 0322777 A1 a method is known which enables the generation of an occupancy grid of a navigation lane for an automated vehicle based on a perception system. However, this solution has disadvantages. In particular, the occupancy grid is only implemented for the navigation lane on which the automated vehicle is traveling and only represents the lane portion that is located facing the vehicle. Summary of the invention
[0008] The object of the present invention is to provide a device and a method for modeling a tactical environment for a motor vehicle, which overcomes the above-mentioned disadvantages and improves the device and method for modeling a tactical environment for a motor vehicle. In particular, the present invention enables the realization of a device and a method that are simple and reliable and that recognize all navigation lanes that must be taken into account in the modeling of the tactical environment of an automated vehicle.
[0009] To this end, the present invention relates to a method for modeling the environment of an autonomous vehicle equipped with a data storage device with a high-definition map and a geo-positioning device, the tactical environment representing autonomous vehicle environmental data that may influence tactical decision making involving mid-term decisions. The method comprises the following steps:
[0010] a first step for determining the current position of the automated motor vehicle on the traffic lane based on data provided by the geo-positioning device and the high-definition map, and determining a set of possible roads, each possible road starting from the current position and being within a specified time range or within a specified distance range from the current position,
[0011] - a second step comprising a first sub-step of determining, for each designated road of the set of possible roads, a first set of traffic lanes directly interacting with the designated road according to an interaction pattern defined in a set of predefined interaction patterns, the set of predefined interaction patterns enabling classification of driving situations requiring tactical decision making, and
[0012] a second sub-step of determining a second traffic lane group that directly interacts with at least one lane of the first traffic lane group according to an interaction mode defined in the predefined interaction mode group,
[0013] The third step is to decompose the traffic lanes of the first group and the traffic lanes of the second group into continuous, spliced and fixed-distance units.
[0014] In one implementation, the automated motor vehicle comprises a navigation module and moves on a route planned by the navigation module, and the determination of the set of possible roads comprises the step of selecting roads that are compatible with the planned route.
[0015] Interaction mode groups can include:
[0016] - keep the automated motor vehicle in its lane, and / or,
[0017] - a lane change of the automated motor vehicle to a lane adjacent to its lane of travel, and / or
[0018] - the merging of the lane of automated motor vehicles with another lane of traffic, and / or
[0019] - The lane of the automated motor vehicle is crossed by at least one other lane of traffic.
[0020] In one implementation, the automated motor vehicle comprises a set of perception devices, and the value of the fixed distance is determined according to the accuracy of the set of perception devices and / or the accuracy of the geolocation device, in particular the fixed distance value is comprised between 10 centimeters and 500 centimeters.
[0021] The third decomposition step may include, for each lane of interest of the first and / or second sequence, dividing the center line of the lane of interest into center line segments whose lengths are equal to the fixed spacing, and decomposing the lane road surface into quadrilateral-shaped splicing units with the center line segments as center lines.
[0022] The invention also relates to a device for modeling the tactical environment of an autonomous vehicle equipped with data storage means with high-definition maps and with geo-positioning means. The device comprises hardware and / or software elements implementing the method defined above.
[0023] The invention also relates to a computer program product comprising program code instructions recorded on a computer-readable medium for implementing the steps of the method defined above when the program is run on a computer. The invention also relates to a computer program product that can be downloaded remotely from a communication network and / or recorded on a computer-readable data medium and / or can be executed by a computer, comprising instructions that, when the program is executed by a computer, direct the computer to implement the method defined above.
[0024] The invention also relates to a computer-readable data recording medium on which a computer program including program code instructions for implementing the method defined above is recorded. The invention also relates to a computer-readable recording medium including instructions, which, when executed by a computer, direct the computer to implement the method defined above.
[0025] The invention also relates to a signal on a data carrier carrying a computer program product as defined above. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] The accompanying drawings show, as examples, how the device for modeling a tactical environment and the method for modeling a tactical environment according to the present invention are implemented and how the method for modeling a tactical environment is executed.
[0027] Figure 1 Represents a mobile vehicle equipped with modeling equipment for a tactical environment.
[0028] Figure 2 It is an alternative representation of a motor vehicle equipped with modeling equipment for a tactical environment.
[0029] Figure 3 An example of traffic lane modeling to be considered for tactical decision making of a motor vehicle is shown.
[0030] Figure 4 The figure is a flow chart of the execution mode of the modeling method of the tactical environment according to the present invention.
[0031] Figure 5 The selection process for the first and second sequence lanes of interest is shown.
[0032] Figure 6 An example of searching for first and second sequence lanes of interest based on a designated position of a motor vehicle is shown.
[0033] Figure 7 A lane of interest associated with a first interaction mode for a motor vehicle departing from the designated location is described.
[0034] Figure 8 A lane of interest associated with a second interaction mode for a motor vehicle departing from the designated location is described.
[0035] Fig. 9 A first and second sequence of lanes of interest are shown in relation to different types of interactions that a motor vehicle may face.
[0036] Fig.10 A connection is described between a first representation of the lanes of interest in the form of an interaction graph between lanes and a second representation of the lanes of interest in the form of a geometric representation of each lane of interest.
[0037] Fig.11 An example of a first sequence of lanes of interest and a second sequence of lanes of interest is provided in the event of a motor vehicle entering a roundabout.
[0038] Fig.12 A first example of discretization of the lane of interest is shown.
[0039] Fig.13 A first step of a second example of discretization of lanes of interest is shown.
[0040] Fig.14 A second step of a second example of discretization of lanes of interest is shown.
[0041] Fig.15 The third step of the second example of discretization of lanes of interest is shown.
[0042] Fig.16 The third and fourth examples focusing on lane discretization are shown.
[0043] Fig.17 The lane of interest discretization related to the maneuver of a motor vehicle entering a roundabout is shown. DETAILED DESCRIPTION
[0044] An example of a motor vehicle 100 equipped with an implementation of a modeling device for a tactical environment of a motor vehicle is described later with reference to Figure 1 is described.
[0045] The motor vehicle 100 may be any type of motor vehicle, in particular a coach, a van, a truck or a public transport vehicle, such as a bus or a shuttle. According to the described embodiment, the motor vehicle 100 is an autonomous vehicle and will be indicated as such in the following description.
[0046] The illustration is therefore given in a non-limiting manner. In particular, the motor vehicle may be a non-automatic vehicle equipped with a driver assistance system, in particular a driver assistance system corresponding to automation greater than or equal to level 2, that is to say corresponding to partial automation of the vehicle.
[0047] In the remainder of this document, the term "tactical environment of the automated vehicle 100" refers to environmental data of the automated vehicle 100 that may affect tactical decision making. This may involve environmental data that may affect decisions made within time intervals greater than 1 second and limited to a minute or even several minutes.
[0048] The automatic vehicle 100 moves in a specified environment. The automatic vehicle 100 includes a driving assistance system 70 and an automatic control device 20 of the automatic vehicle 100 .
[0049] The driving assistance system 70 sends a movement control C to the automatic control device 20. The movement control C may include a longitudinal movement control and / or a lateral movement control. The longitudinal movement control may include a first torque command for an engine group of the vehicle and / or a second torque command for a vehicle brake actuator. The lateral movement control includes a rotation angle of a steering wheel of the automatic vehicle 100.
[0050] The driving assistance system 70 comprises in particular a modeling system 10 of the tactical environment of the automated vehicle 100 and a decision module 3. The role of the system 10 is to construct a modeling M_ENV_T of the tactical environment of the automated vehicle, which modeling will be sent to the decision module 3. This modeling M_ENV_T enables the decision module 3 to make one or more tactical decisions. These decisions will then be converted into movement control C of the automated vehicle 100.
[0051] The system 10 for modeling the tactical environment of the autonomous vehicle 100 includes, in particular, the following elements:
[0052] - a data storage device in which is recorded a high-resolution map including a description of the road network on which the automated vehicle 100 is travelling Figure 1 ,
[0053] - a geo-positioning system 2 which enables Figure 1 On the positioning automatic vehicle 100,
[0054] - an environmental data structure module 4, necessary for building a tactical environment model M_ENV_T of the autonomous vehicle 100,
[0055] - the environment perception system 5 of the autonomous vehicle 100, which provides data to the environment data structure module 4,
[0056] A computing unit 6 , comprising a microprocessor 61 , a local electronic memory 62 and a communication interface 63 , enabling the microprocessor 61 to receive data from the geographic positioning system 2 and to send the LGM modeling of the lane of interest to the environment data structure module 4 of the automated vehicle 100 .
[0057] land Figure 1 It provides a geometric, topological, and semantic description of the road network. Figure 1 It contains the information required to determine, from the position of the automated vehicle 100 determined by the geolocation system 2 , all possible paths whose starting point is determined by said position.
[0058] land Figure 1 The described road network consists of roads broken down into traffic lanes. The traffic lanes themselves are broken down into lane segments; Figure 1 Associate an identifier with each lane segment.
[0059] In the remainder of this article, the term "lane" or "traffic lane" refers to the Figure 1 A defined lane segment associated with a unique identifier.
[0060] The geo-location of the automated vehicle 100 by the geo-location system 2 includes determining the identifier of the lane in which the automated vehicle 100 is traveling, and the position of the vehicle on that lane segment.
[0061] land Figure 1 In addition, the centerline of each lane is defined. The position of the automated vehicle 100 is thus determined as follows:
[0062] -Lane Identifier N ,as well as
[0063] - The automatic vehicle 100 is identified by an identifier Id N Lane V N The one-dimensional position on the center line of a (e.g., the abscissa of a curve).
[0064] According to the processing method described later in this article, the center line of the lane can be modeled by a center "polyline", that is, a line composed of a series of straight line segments.
[0065] Figure 2 A high level description of the process implemented in the driver assistance system 70 regarding tactical level decision making is provided. The process is broken down into perception X1 of the environment of the automated vehicle 100, followed by structuring of the perception X2, followed by decision making based on the structured perception X3.
[0066] Therefore, based on the perception X1 of the tactical environment of the automatic vehicle 100, in particular when receiving the data D-ENV from the perception device group 5, the driving assistance system 70 creates a structured model M_ENV_T of the tactical environment of the automatic vehicle 100 and sends the structured model M_ENV_T to the decision module 3.
[0067] The construction of the structured model M_ENV_T is performed in two stages:
[0068] - a first step of creating, by the microprocessor 61 , a geometric description LGM of a set of navigation lanes that may play a role in the formulation of tactical decisions,
[0069] - A second step of constructing the environment data D_ENV by means of the module 4 from the LGM geometric description.
[0070] The geometric description LGM (named after the acronym “Lane Grid Map”) is determined based on the instantaneous position of the automatic vehicle 100 at a given time T. In the following part of this document, the LGM geometric description is named LGM grid. As will be seen in the following part of this document, the determination of the LGM grid includes: the determination 61A of the position of the automatic vehicle 100, followed by the determination 61B of the possible road group of the automatic vehicle 100, followed by the determination 61C of the navigation lanes to be considered in the construction of the LGM grid.
[0071] The LGM grid is a non-central and discrete representation of the navigation lane in the length direction and along the navigation lane (in the form of a grid composed of a series of interconnected cells).
[0072] In an alternative implementation not described herein, the lane may also be decomposed into cells in the width direction. In this case, the LGM grid will be a non-central and discrete representation (a grid consisting of a series of interconnected cells) of the navigation lane in the length and width directions.
[0073] The allocentric representation uses a globally fixed reference frame, in contrast to the egocentric representation which uses a reference frame that moves with the vehicle.
[0074] Each cell extends over a length equal to a fixed discretization spacing PDIS and over the entire width of the navigation lane. In one implementation, the fixed spacing PDIS value is determined according to the accuracy of the perception device group 5 and / or the accuracy of the geolocation device 2, in particular the fixed spacing value is comprised between 10 cm and 500 cm. The fixed spacing PDIS value may be comprised within other value ranges.
[0075] Figure 3An example of an LGM grid associated with an automated vehicle 100 is provided. The right-pointing arrow indicates the path that the automated vehicle 100 will follow. In this case, the LGM grid includes modeling of the following lanes:
[0076] - the lane 151 on which the automated vehicle 100 is travelling and the lane 152 along which it will follow after turning right,
[0077] a lane 153 adjacent to the lane of travel of the automated vehicle 100 , which is located immediately to the left and in the same direction of travel as the lane 151 ,
[0078] - Lane 154 coming from the left, which will cross the path of the automated vehicle 100 .
[0079] Other navigation lanes are not modeled in the LGM grid because they have no impact on the decision made by the automated vehicle 100 at that given moment.
[0080] The processing performed in module 4 consists in constructing the environment data D_ENV according to the geometry of the LGM grid. Therefore, in the tactical environment model M_ENV_T only the information related to the discretized lanes in the LGM grid will be considered. Moreover, the environment information will advantageously be structured according to the LGM grid cells.
[0081] At time T, the position of the automated vehicle 100 determines the beginning of a set of possible lanes E that may be taken by the automated vehicle 100. At time T, when the automated vehicle moves on any of the roads in the set E, the LGM grid is a geometric description of the set of possible lanes that may be considered when deciding whether to execute a decision.
[0082] In the remainder of this document, the term "road" is used to refer to a continuous sequence of lanes forming a specified route, which starts from the current position of the automated vehicle 100 and is compatible with the geometric, topological and semantic data of the road network.
[0083] As described later herein, in an implementation of the present invention, navigation data is not considered when searching a set of possible roads. In other words, the modeling of the environment of the automated vehicle 100 to be used for tactical decision making can be performed without regard to the route predetermined by the navigation module of the automated vehicle 100. The search of all possible roads is then limited only by distance or time range.
[0084] For example, all possible roads within a specified distance radius around the current position of the automatic vehicle 100 may be determined, and the specified distance may be, for example, one kilometer. Alternatively, all possible roads that the automatic vehicle 100 can travel around its current position within a specified time period, which may be, for example, 10 minutes, may be determined.
[0085] The environment modeling system 10 may also include a communication network 7, such as a CAN network, of the automated vehicle 100. The communication network 7 may provide the microprocessor 61 with movement data of the automated vehicle 100, such as speed, acceleration and jerk.
[0086] The microprocessor 61 can also access the shared database 8 via the cellular network, which includes contextual information such as map updates, work zone notifications, etc. . . .
[0087] In an implementation of the invention, the computer 61 enables the execution of software comprising the following modules communicating with each other:
[0088] - a determination module 611, which communicates with the geo-positioning device and the HD map, determines the lane of travel of the automated vehicle 100 and determines a set of possible roads,
[0089] - a determination module 612, which determines a set of lanes of interest of a first sequence and a set of lanes of interest of a second sequence based on a set of predetermined operations,
[0090] - a decomposition module 613, decomposing the lanes of interest of the first sequence and the lanes of interest of the second sequence into continuous, spliced and fixed-distance segments,
[0091] A transmission module 614, which cooperates with the environment data construction module 4 to transmit the modeling of the first sequence of lanes of interest and the second sequence of lanes of interest.
[0092] The execution method of the control method of the automatic vehicle is referred to below Figure 4 The method comprises four steps from E1 to E4.
[0093] In a first step E1 , the traffic lane of automated vehicle 100 is determined.
[0094] Thanks to the land Figure 1 and the information provided by the geographic positioning device 2 to determine the vehicle's location on the ground Figure 1 The current position on the lane can therefore determine the lane identifier Id N and the identifier Id N The one-dimensional position on the associated lane centerline. The term "centerline" refers to an imaginary line that divides a traffic lane into two half-lanes along the longitudinal direction of the lane.
[0095] In the following section of this article,
[0096] - with the identifier Id N The associated lane is named "VId N Lane",
[0097] -Id N The center line of the lane is named "LIdN Center Line",
[0098] -Automatic Vehicle 100 in LId N The position on the center line is named "PId N Location".
[0099] At a given time T in tactical environment modeling, the autonomous vehicle 100 is located at position PId N Place.
[0100] The search is then performed at time T by the automated vehicle 100 based on the position PId N , use the road group E within a specified distance or time range. Figure 1 The topological information provided is used to estimate all roads corresponding to the possibilities provided by the road network.
[0101] In the implementation of the present invention, the selection of roads can be implemented according to the navigation data. Therefore, some roads may be rejected because they are determined to be inconsistent with the planned route.
[0102] According to these circumstances, in the case of level 3 and above automated vehicles, the planned route may be a route determined by the mission planning module. Alternatively, in the case of level 0, level 1 or level 2 automated vehicles, the planned route may be defined by the navigation system. The navigation system should then preferably show the guidance function according to the defined route, and gradually show the evolution of the route according to the update of traffic data on the route.
[0103] Figure 5 The process of step E2 based on the road group E determined in step E1 is shown. In step E2, each possible road C in group E is implemented k To iterate over the possible road set E,
[0104] - A first sub-step E21 for determining a first set of traffic lanes V1 k , is called the designated road C k The first sequence of associated lanes of interest is associated with the designated road C according to one of the interaction modes of the predefined interaction mode group MI. k Directly interact, then
[0105] - A second sub-step E 22 of determining a second set of traffic lanes V2, called second sequence of lanes of interest k , which interacts with the first sequence focus lane group V1 according to one of the interaction modes of the predefined interaction mode group MI k At least one lane of the vehicle directly affects each other.
[0106] These lanes of interest are thus defined relative to the subjective point of view of the automated vehicle 100, and relative to a set of predefined interaction modes MI, which will be further described herein. In fact, the lanes of interest only include navigation lanes to be considered when deciding to perform any operation that fits into the set of predefined interaction modes MI.
[0107] The first sequence of lanes of interest correspond to lanes that directly interact with the vehicle road (they therefore share a common navigation space). For the automated vehicle 100, it is crucial to evaluate information related to these areas, especially if the vehicle does not have priority.
[0108] Second sequence lanes of interest include lanes that interact indirectly with the automated vehicle. They interact directly (first sequence) with first sequence lanes of interest previously defined for the vehicle. Information related to second sequence lanes of interest is less important but can be useful in providing data related to first sequence lanes.
[0109] In the described implementation, the predefined interaction mode group MI defines four possible interaction modes, enabling classification of driving situations requiring tactical decision making:
[0110] a first mode MI1 , called lane following, in which the automated vehicle 100 remains in its lane of travel,
[0111] a second mode MI2, called lane change, in which the automated vehicle 100 joins a lane adjacent to its lane of travel,
[0112] a third mode MI3, called lane merging, in which the lane of the automated vehicle 100 merges with another lane of traffic,
[0113] A fourth mode MI4, referred to as lane crossing mode, in which the traffic lane of the automated vehicle 100 crosses at least one other traffic lane.
[0114] exist Figure 6 In the example shown, at time T of tactical environment modeling, the automated vehicle 100 is located at position POS(T) on the leftmost lane 120 of a three-lane road.
[0115] The position POS(T) enables determination, in step E1 , of a set E comprising two possible roads for the automated vehicle 100 :
[0116] -First road C1, such as Figure 7 As shown, it corresponds to the lane following interaction mode MI1, and
[0117] - The second road C2, such as Figure 8As shown, it corresponds to the lane change interaction mode MI2 , which lane change can be performed only toward the lane 121 located on the right side of the automatic vehicle 100 .
[0118] In step E2 , an iteration of sub-steps E21 and E22 is applied to each of the roads C1 , C2 defined in step E1 .
[0119] Figure 7 An iteration of sub-steps E21 and E22 is shown for determining a first and a second sequence of lanes of interest respectively associated with road C1. The automated vehicle 100 is kept in its lane of travel. Thus,
[0120] In sub-step E21 , lane 120 is determined to be the only lane of interest of the first sequence V associated with road C1 11 ,
[0121] In sub-step E22, lane 121 is determined to be the only lane V of interest of the second sequence associated with road C1 21 , since lane 121 is the only lane that may interact with lane 120 according to one of the predetermined interaction modes MI.
[0122] Figure 8 An iteration of sub-steps E21 and E22 is shown for determining a first and a second sequence of lanes of interest, respectively associated with road C2. The automated vehicle 100 performs a lane change towards lane 121. Thus,
[0123] In sub-step E21 , lane 121 is determined to be the only lane of interest of the first sequence V associated with road C2 12 ,
[0124] In sub-step E22, lanes 120 and 122 are determined as two lanes of interest of the second sequence V associated with road C2. 22 , since they may interact with lane 121 according to one of the predetermined interaction modes MI.
[0125] Figures 6 to 8 The example shown involves a simple road configuration that limits the automated vehicle to a very small number of possible roads.
[0126] according to Fig. 9 As an additional perspective, the first and second sequence lanes of interest can also be described based on the type of interaction encountered by the automated vehicle 100. Note that Fig. 9 The lane of interest identified in the table is only a single road considered for vehicle 100, indicated by a thick arrow. Fig. 9, the first sequence of attention lanes are indicated by shading on the image corresponding to the first sequence, and the second sequence of attention lanes associated with the first sequence lanes framed in black are indicated by shading on the image corresponding to the second sequence.
[0127] The first two rows of the table, LL1 and LL2, correspond to the previous reference Figures 5 to 7 Describe the situation.
[0128] The first row LL1 in the table describes the first and second sequence lanes of interest when the automated vehicle remains in its lane of travel. The first column CC1 of the row represents the first sequence lane of interest, which is the road the vehicle follows. The lane or lanes adjacent to the first sequence lane of interest constitute the second sequence lane of interest (see the second column CC2 of the row).
[0129] The second row LL2 in the table describes the first and second sequence of lanes of interest when the automated vehicle changes lanes to a lane adjacent to its lane of travel. The first column CC1 of the row represents the first sequence of lanes of interest, which is the lane to which the lane change is performed. Lanes adjacent to the first sequence of lanes of interest constitute the second sequence of lanes of interest (see the second column CC2 of the row).
[0130] The third row LL3 of the table describes the first and second sequence focus lanes when the lanes merge. The first column CC1 of the row represents the first sequence focus lane, which is the lane that merges with the vehicle road. All lanes that directly interact with the first sequence focus lane are second sequence focus lanes (see the second column CC2 of the row).
[0131] The fourth row LL4 in the table describes the first and second sequence focus lanes when the lanes intersect. The first column CC1 of this row represents the first sequence focus lane, which is the lane that intersects the vehicle road. All lanes that directly interact with one of the first sequence focus lanes are second sequence focus lanes (see the second column CC2 of this row).
[0132] Finally, when all paths C in group E k After implementing the above iterative steps E21 and E22, a set of graphs Gk of the first and second sequences of traffic lanes interacting with the automated vehicle 100 at a specified time T is obtained. As an example, Fig.10 Such a map Gk is shown, which particularly includes a first-level attention lane V1j and two second-level attention lanes V2m, V2p that can interact with the first-level attention lane V1j. Fig.10 It shows the fact that each lane of interest of the graph Gk (that is to say each node of the graph Gk) is then represented geometrically and in a discretized manner in the LGM grid.
[0133] As an example, Fig.11The concept of the first-sequence lane of interest and the second-sequence lane of interest is shown in the case where the automated vehicle 100 enters a roundabout. The road followed by the automated vehicle implements a lane merge between the passing lane borrowed by the road Ck and the first-level lane of interest V1j (corresponding to the outer lane of the roundabout). The passing lane V2m adjacent to the road Ck is a second-sequence lane of interest, as is the outer lane of the roundabout V2p.
[0134] In a third step E3, a discrete representation of the set of navigation lanes determined in step E2 is created in the form of a grid. In the remainder of this document, this representation is named "lane grid map" or LGM grid.
[0135] To this end, the first sequence lane of interest and the second sequence lane of interest are decomposed into continuous, spliced and fixed-pitch segments. In the subsequent parts of this paper, the continuous, spliced and fixed-pitch segments are named “units”.
[0136] Each lane is therefore decomposed into units according to a fixed discrete spacing PDIS. For example, Fig.12 As shown, the lane centerline L10 is divided according to a fixed spacing PDIS, and cells C1 to C8 are bounded by the width of the traffic lane and by straight line segments spaced PDIS apart from each other and perpendicular to the centerline L10. On the lane portion V10 shown, the centerline L10 is straight and the cells are rectangular.
[0137] Figures 12 to 15 An implementation is shown for decomposing a series of multiple lanes 201, 202, 203, 204 with non-aligned center lines into contiguous and fixed-spaced cells. In this implementation, the cells are quadrilateral and not systematically rectangular.
[0138] The broken lines of lanes 201, 202, 203, and 204 are Fig.13 It is shown in:
[0139] - polyline P1 of lane 201 connects point 301 to point 302,
[0140] - polyline P2 of lane 202 connects point 302 to point 303,
[0141] - polyline P3 of lane 203 connects point 303 to point 304, and
[0142] The polyline P4 of lane 204 connects point 304 to point 305 .
[0143] Note that, Fig.13This description is based on decomposing the series of lanes 201 to 204 into a plurality of polylines, each polyline consisting of a single segment, each polyline, in this example, comprising only two points (ie one point at each end of the polyline).
[0144] Alternatively, Fig.13 It can be interpreted as representing a unique polyline which includes a plurality of polyline segments P1 , P2 , P3 , P4 and is associated with a unique lane consisting of lane segments 201 to 204 .
[0145] In this case, the method described below will be applied to each point 301, 302, 303, 304, 305 of the polyline, not just to the end points of the polyline.
[0146] The problem of decomposing lanes into cells involves the definition of cells with constant spacing while ensuring that the cells are joined to each other at any junction of two adjacent lanes, in particular at junctions of two adjacent lanes whose center polylines meet at an angle different from 180 degrees.
[0147] In the example shown, a stitching unit is defined when switching between lane 201 and lane 202 , then between lane 202 and lane 203 , and then between lane 203 and lane 204 .
[0148] To this end, in one embodiment, a tangent interpolation method or a method known as lanelet can be used. In the subsequent parts of this article, the term "vecteur tangent" is used to name a vector, which is defined at a specified point of a polyline and is used to define a cell whose side is passed by the specified point.
[0149] Applied to lane 202, interpolation methods include, for example Fig.13 As shown, a first tangent vector 402 is defined at a first end of lane 202 (that is, at point 302 ), and a second tangent vector 403 is defined at a second end of lane 202 (that is, at point 303 ).
[0150] In one implementation, Fig.13 As shown, the direction of the first tangent vector 402 (at point 202) can be defined as the direction of the vector V202 connecting points 301 and 303; similarly, the direction of the second tangent vector 403 (at point 203) can be defined as the direction of the vector V203 connecting point 302 to point 304.
[0151] In the case where the lane is decomposed into cells starting at point 302 (eg, if lane 201 does not exist) (not shown), the direction of tangent vector 402 is defined as the direction of polyline P2.
[0152] Alternatively, in the case where the two lanes meet at point 302 (not shown), the direction of the tangent vector 402 is defined as the direction of the broken line P2.
[0153] Then, if Fig.14 As shown, a tangent vector may be defined for each intermediate point of the polyline P2, the intermediate point being a point between points 302 and 303, and the polyline P2 is divided into segments of length PDIS.
[0154] In this implementation, each tangent line at a point on the fold line depends on the tangent line at the previous point on the fold line and the tangent line at the next point on the fold line.
[0155] The tangent vectors thus determined are then used to define the cells of the LGM grid. Fig.15 As shown, in order to define the cell 310 of the lane 202, two consecutive points P of the polyline P2 are selected. A , P B .
[0156] Follow up
[0157] - perpendicular to the point P A defined at and passing through point P A The first segment S of the tangent A And, the first segment starts with point P A is centered and has a length equal to the width of lane 202, and
[0158] - perpendicular to the point P B defined at and passing through point P B The second segment S of the tangent B And, the first segment starts with point P B is centered and has the width of lane 202 as its length.
[0159] Note that, through point P A and P B The straight line is a center line of the quadrilateral defining the unit 310, and point P A and P B The tangent lines at are segment S A and S B The perpendicular bisector of .
[0160] Then segment S A The first end is connected to segment S B The two ends are located on the same side relative to the fold line P2. A The second end is connected to segment S B The second end.
[0161] The processing steps described above thus make it possible to model, step by step, the series of traffic lanes that make up the road.
[0162] However, there are lane configurations where the tangent interpolation method cannot be applied. This particularly concerns the case of lane divergence and intersection.
[0163] The first and second examples of lane configurations are given by Fig.16 Shows:
[0164] - a first configuration example where two lanes 501, 502 merge into a single lane 503,
[0165] A second configuration example in which one lane 504 is separated into two lanes 505 , 506 .
[0166] In the first configuration example, the problem concerns the direction chosen by the merging point or tangent lines of the two lanes 501, 502. The polylines 601, 602 of lanes 501 and 502, respectively, meet and meet polyline 603 of lane 503 at a first common point 701. The direction of the tangent line 801 at point 701 coincides with the direction of polyline 603.
[0167] In the second configuration example, the polyline 604 of the lane 504 is separated into two polylines 605 and 606 of the lanes 505 and 505 at the common point 702 . The direction of the tangent line 802 at the point 702 coincides with the direction of the polyline 604 .
[0168] Fig.17 The discretization of lanes of interest V1j, V2m, and V2p associated with the operation of the automated vehicle 100 entering a roundabout is shown (previously with reference to Fig.11 described above).
[0169] When all first and second sequence lanes of interest are discretized into cells, proceed to step E4, where the modeling of the first sequence lanes of interest and the second sequence lanes of interest is sent to the environmental data structure module 4. In module 4, the environmental data from the perception system 5 will be distributed in the LGM grid. For example, the data assigned to the cell may relate to the position of a vehicle located in the cell and having priority over the automatic vehicle 100, or the presence of a traffic light and its status. The data assigned to the cell may also relate to spatial information, in particular the spatial occupancy of each cell. In one implementation, the LGM grid cells may also be used to contain prediction data.
[0170] Finally, the method of modeling the tactical environment of the automatic vehicle 100 in the form of an LGM grid makes it possible to describe the tactical situation of the automatic vehicle 100 according to two levels of information. The first level of information is related to the geometric properties of the diagram. In fact, the sampling of the navigation lane in the cell makes it possible to know at any moment the distance separating the automatic vehicle 100 from a given cell. Moreover, the precision of this distance can be calibrated by the PDIS values assigned to the discretization intervals. The second level of information concerns the information associated with each cell, which makes it possible to characterize the space covered by the cell, that is to say the portion of the lane covered by the cell.
[0171] The method according to the invention differs from the prior art, in particular document US2018 / 0322777A1, in which the occupancy grid defines only a simple space of interest corresponding to the space in front of the automated vehicle and the topology of the road is not taken into account when constructing the grid. In fact, first of all, the modeling method according to the invention takes into account the geometry of the navigation lane. Moreover, the LGM grid includes not only the navigation lane in which the automated vehicle travels, but also lanes that may interact with the lane of the automated vehicle. Therefore, in contrast to the prior art, the method according to the invention is not limited to the modeling of the lane located in front of the automated vehicle.
[0172] The modeling according to the invention has several advantages. Firstly it provides a compact illustration of the environment with which the vehicle interacts, since only the lane of interest is modeled. It involves a illustration that is dependent on the background environment, since it is focused on the area of tactical decision making.
[0173] Furthermore, the fineness of the modeling can be easily adjusted by modifying the discretization interval or by merging cells.
[0174] Furthermore, the modeling according to the present invention only requires a high-definition map database with one topological layer.
[0175] Finally, due to its geometric nature, the modelling according to the invention simplifies the processing subsequently performed in the decision module, in particular by facilitating the calculation of risk measures such as the delay before collision.
Claims
1. A method for modeling a tactical environment of an autonomous motor vehicle (100), the autonomous motor vehicle (100) being equipped with a data storage device having a high-definition map (1) and a geo-positioning device (2), the tactical environment indicating environmental data of the autonomous motor vehicle (100) being able to influence tactical decision making involving mid-term decisions, the method being characterized by comprising the following steps: - a first step (E1) for determining the current position of the automated motor vehicle (100) on the traffic lane based on data provided by the geolocation device (2) and the high-definition map (1) and determining a set of possible roads (E), each possible road starting from the current position and being within a specified time range or within a specified distance range from the current position, - a second step (E2) comprising a first sub-step (E21) of determining, for each designated road (Ck) of the set of possible roads (E), a first set of traffic lanes (V1k) directly interacting with the designated road (Ck) according to an interaction pattern defined in a set of predefined interaction patterns (MI), the set of predefined interaction patterns enabling classification of driving situations requiring tactical decision making, and The method further comprises a second sub-step (E22) of determining a second traffic lane group (V2k) directly interacting with at least one lane of the first traffic lane group (V1k) according to an interaction mode defined in a predefined interaction mode group (MI), A third step (E3) of breaking down the traffic lanes of the first group and the traffic lanes of the second group into continuous, spliced and fixed-distance (PDIS) cells.
2. The modeling method according to the preceding claim, wherein the autonomous motor vehicle (100) comprises a navigation module (9) and moves on a route planned by the navigation module (9), characterized in that The determination of the set of possible roads (E) comprises the step of selecting roads that are compatible with the planned route.
3. Modeling method according to one of the preceding claims, characterized in that The interaction mode group includes: - the automated motor vehicle (100) remains in its lane of travel, and / or, - a lane change of the automated motor vehicle (100) to a lane adjacent to its lane of travel, and / or - the merging of the lane of the automated motor vehicle (100) with another lane of traffic, and / or The lane of the automated motor vehicle (100) is crossed by at least one other traffic lane.
4. Modeling method according to one of the preceding claims, the autonomous motor vehicle (100) comprising a set of perception devices (5) characterized in that The fixed distance (PDIS) value is determined according to the accuracy of the set of sensing devices (5) and / or the accuracy of the geolocation device (2), and in particular the fixed distance value is comprised between 10 cm and 500 cm.
5. Modeling method according to one of the preceding claims, characterized in that The third decomposition step (E3) includes, for the first and / or second sequence of lanes of interest, splitting the center line of the lane of interest into spliced center line segments whose length is equal to the fixed spacing (PDIS), and decomposing the lane surface into spliced units in the form of quadrilaterals with the center line segments as median lines.
6. A tactical environment modeling device (10) for an automated motor vehicle (100), the automated motor vehicle being provided with a data storage device having a high-definition map (1) and a geo-positioning device (2), the device comprising hardware and / or software elements (1, 2, 4, 5, 6, 7, 8, 9, 61, 62, 63, 611, 612, 613, 614) for implementing a method according to any one of claims 1 to 5.
7. A computer program product, comprising program code instructions recorded on a computer-readable medium, for implementing the steps of the method according to any one of claims 1 to 5 when the program is run on a computer.
8. A computer-readable data recording medium having recorded thereon a computer program including program code instructions for implementing the method according to any one of claims 1 to 5.
9. A signal on a data carrier carrying a computer program product according to claim 7.
