Vehicle lateral behavior decision method and device, electronic equipment and storage medium

Abstract

Embodiments of the present disclosure disclose a vehicle lateral behavior decision method and device, electronic equipment and storage medium, the method obtains a position sampling directed graph of a current vehicle, for each sampling point, constructs a behavior sequence dictionary of the sampling point according to the lateral behavior of the trajectory segment connected between the sampling point and the next layer sampling point, and then obtains a to-be-searched behavior sequence, and according to the behavior sequence dictionary of each sampling point, finds out a current target trajectory in the position sampling directed graph that meets the to-be-searched behavior sequence, the method solves the problem that the decision result does not meet the habits of human drivers due to a small amount of fluctuations in perception information in the prior art, achieves the purposes of stabilizing vehicle behavior and improving behavior consistency, and greatly reduces the consumption of computing power and improves the trajectory query efficiency.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of automatic driving, and particularly relates to a vehicle lateral behavior decision method and device, an electronic device and a storage medium. BACKGROUND

[0002] In actual scenarios, the information obtained by the perception module of a vehicle is often inaccurate and fluctuant, and thus the prediction information determined by the prediction module based on the perception information is also fluctuant. The current decision planning method is to find the optimal trajectory of the vehicle in the current planning period based on the latest perception information and prediction information.

[0003] However, when the perception information and the prediction information are fluctuant, the optimal trajectory obtained is also fluctuant. In practice, the decision of a human driver often has relatively strong consistency and will not easily change his / her behavior due to a small amount of fluctuation in the perception information. This leads to the result of the decision based on the perception information and the prediction information not conforming to the decision result of the human driver. SUMMARY

[0004] To solve the above technical problems or at least partially solve the above technical problems, the embodiments of the present disclosure provide a vehicle lateral behavior decision method and device, an electronic device and a storage medium, which solve the problem in the prior art that the decision result does not conform to the habit of a human driver due to a small amount of fluctuation in the perception information.

[0005] In a first aspect, the embodiments of the present disclosure provide a vehicle lateral behavior decision method, which comprises:

[0006] obtaining a position sample directed graph of a current vehicle, wherein the position sample directed graph is composed of a start layer sample point, at least one intermediate layer sample point and at least one end layer sample point, and the connection between adjacent layer sample points forms a trajectory segment;

[0007] for each sample point in the position sample directed graph, constructing a behavior sequence dictionary of the sample point according to the lateral behavior of the trajectory segment connecting the sample point and the next layer sample point, wherein the behavior sequence dictionary comprises at least one behavior sequence and a sample point identifier set corresponding to each behavior sequence, the behavior sequence is composed of the lateral behavior of each trajectory segment in the trajectory with the sample point as the starting point, and the sample point identifier set comprises the next layer sample point satisfying the behavior sequence between the sample point;

[0008] obtaining a to-be-searched behavior sequence, and determining a current target trajectory conforming to the to-be-searched behavior sequence in the position sample directed graph based on the behavior sequence dictionary of the sample point.

[0009] In a second aspect, the embodiments of the present disclosure further provide a vehicle lateral behavior decision device, which comprises:

[0010] a directed graph obtaining module configured to obtain a position sampling directed graph of the current vehicle, wherein the position sampling directed graph comprises a start layer sampling point, at least one intermediate layer sampling point, and at least one end layer sampling point, and connections between adjacent layer sampling points form trajectory segments;

[0011] a dictionary determining module configured to, for each sampling point in the position sampling directed graph, construct a behavior sequence dictionary of the sampling point according to lateral behaviors of trajectory segments connected between the sampling point and a next layer sampling point, wherein the behavior sequence dictionary comprises at least one behavior sequence and a sampling point identifier set corresponding to each behavior sequence, the behavior sequence is composed of lateral behaviors of each trajectory segment in a trajectory with the sampling point as a starting point, and the sampling point identifier set comprises next layer sampling points satisfying the behavior sequence and connected with the sampling point;

[0012] a trajectory searching module configured to obtain a to-be-searched behavior sequence, and determine a current target trajectory conforming to the to-be-searched behavior sequence in the position sampling directed graph based on the behavior sequence dictionary of the sampling point.

[0013] In a third aspect, the embodiments of the present disclosure further provide an electronic device, which comprises one or more processors, a storage device configured to store one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the vehicle lateral behavior decision method as described above.

[0014] In a fourth aspect, the embodiments of the present disclosure further provide a computer readable storage medium, which stores a computer program, and when the program is executed by a processor, the vehicle lateral behavior decision method as described above is implemented.

[0015] The vehicle lateral behavior decision method provided by the embodiment of the present disclosure can obtain a position sampling directed graph of a current vehicle, for each sampling point in the position sampling directed graph, construct a behavior sequence dictionary of the sampling point according to the lateral behavior of a trajectory segment connected between the sampling point and a next layer sampling point, and then obtain a to-be-searched behavior sequence, and according to the behavior sequence dictionary of each sampling point, find out a current target trajectory in the position sampling directed graph that meets the to-be-searched behavior sequence. This method can take the behavior sequence decided at the last moment as the to-be-searched behavior sequence and then evaluate and score after trajectory query, or first evaluate and score various behavior sequences and then perform trajectory query based on the to-be-searched behavior sequence with the best score, thereby solving the problem in the prior art that a decision result does not meet the habits of human drivers due to a small amount of fluctuations in perception information, achieving the purposes of stabilizing vehicle behavior and improving behavior consistency, and the method realizes dictionary-based trajectory query, without the need to count all possible behavior sequences and then query from the behavior sequences, thereby greatly reducing the consumed computing power and improving the trajectory query efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0016] The above and other features, advantages, and aspects of the embodiments of the present disclosure will become more apparent by describing in detail the following specific embodiments in conjunction with the accompanying drawings. Throughout the drawings, the same or similar reference numerals refer to the same or similar elements. It should be understood that the drawings are schematic, and the original and elements are not necessarily drawn according to the scale.

[0017] Figure 1 A decision schematic diagram of an existing technology in the embodiment of the present disclosure;

[0018] Figure 2 A trajectory deviation accumulation schematic diagram in the embodiment of the present disclosure;

[0019] Figure 3 A flowchart of a vehicle lateral behavior decision method in the embodiment of the present disclosure;

[0020] Figure 4 A schematic diagram of a position sampling directed graph in the embodiment of the present disclosure;

[0021] Figure 5 A schematic diagram of a position type in the embodiment of the present disclosure;

[0022] Figure 6 A lane schematic diagram in the embodiment of the present disclosure;

[0023] Figure 7 A schematic diagram of a position sampling directed graph in the embodiment of the present disclosure;

[0024] Figure 8 A trajectory schematic diagram in the embodiment of the present disclosure;

[0025] Figure 9 A schematic diagram of a position sampling directed graph in an embodiment of the present disclosure;

[0026] Figure 10 A schematic diagram of a target output segment in an embodiment of the present disclosure;

[0027] Figure 11 A structural schematic diagram of a vehicle lateral behavior decision device in an embodiment of the present disclosure;

[0028] Figure 12 A structural schematic diagram of an electronic device in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0029] Embodiments of the present disclosure will be described in more detail with reference to the drawings. Although some embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments set forth herein, but rather, these embodiments are provided so that the present disclosure can be more thoroughly and completely understood. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0030] It should be noted that the terms "first", "second", and the like mentioned in the present disclosure are only used to distinguish different devices, modules or units, and are not intended to limit the order or interdependence of the functions performed by these devices, modules or units.

[0031] The names of the messages or information exchanged between the devices in the embodiments of the present disclosure are only for illustrative purposes, and are not intended to limit the scope of the messages or information.

[0032] Before the vehicle lateral behavior decision method provided by the embodiments of the present disclosure is described in detail, the technical problems solved by the method are described.

[0033] In the prior art, the information perceived in the actual scene is often inaccurate and fluctuating, and the existing decision planning method is to find the optimal trajectory based on the latest perception information and prediction information within the current planning period. When the perception information and prediction information fluctuate, the optimal trajectory obtained will also change accordingly.

[0034] For example, Figure 1 A decision schematic diagram of the prior art in an embodiment of the present disclosure. As Figure 1 shown, the first frame of perceived obstacle speed is low, and according to a series of safety, comfort, executability and other evaluation criteria, the Figure 1The decision in the first frame is to immediately change lanes. In the second frame, if a fluctuation is detected in the obstacle, based on the same evaluation metric, the optimal behavior is to first maintain the lane and then change lanes.

[0035] However, the actual decision-making process of human drivers is different from... Figure 1 The decision-making process shown is different. Human drivers' decisions tend to be more consistent, and they will not easily change their behavior due to slight fluctuations in perceived information.

[0036] To stabilize the vehicle's lateral behavior, the positional deviation of each point between the trajectory decided in the previous frame and the trajectory decided in the current frame can be compared and accumulated. Figure 2 As shown, Figure 2 This is a schematic diagram illustrating trajectory deviation accumulation in one embodiment of the present disclosure. However, this approach does not address the root cause (trajectory behavior) and cannot effectively solve the problem that decision-making results caused by fluctuations in perceived information do not conform to human driver habits.

[0037] Therefore, in order to solve the above problems, this disclosure provides a vehicle lateral behavior decision-making method. This method can analyze all possible lateral behaviors based on the position sampling directed graph and quickly filter out trajectories that match the search behavior sequence based on the search behavior sequence.

[0038] The method provided in this disclosure has specific application scenarios including but not limited to: 1. By determining the behavior sequence dictionary of each sampling point in the location sampling directed graph, traversing all possible behavior sequences, evaluating and scoring all possible behavior sequences, and comparing to obtain the optimal behavior sequence as the behavior sequence to be searched, and then querying the current target trajectory that matches the behavior sequence to be searched from the location sampling directed graph, so as to stabilize the vehicle behavior and improve behavior consistency; 2. First, find the behavior sequence of the current vehicle at the previous moment, use it as the behavior sequence to be searched, and then mark the trajectory that matches the behavior sequence to be searched from the behavior sequence dictionary of each sampling point, and give it a higher score when evaluating and scoring, thereby stabilizing the vehicle behavior and improving behavior consistency.

[0039] Figure 3 This is a flowchart illustrating a vehicle lateral behavior decision-making method according to an embodiment of this disclosure. The method can be executed by a vehicle lateral behavior decision-making device, which can be implemented in software and / or hardware and can be configured in an electronic device. Figure 3 As shown, the method may specifically include the following steps:

[0040] S110. Obtain the current vehicle's position sampling directed graph, wherein the position sampling directed graph consists of a starting layer sampling point, at least one intermediate layer sampling point, and at least one ending layer sampling point, and the connection between adjacent layer sampling points forms a trajectory segment.

[0041] In this embodiment of the disclosure, the current vehicle can be understood as a self-driving vehicle. The position sampling directed graph can be a lattice sampling mesh graph, which is used to describe the various possible behaviors derived from the current vehicle starting from the current position.

[0042] For example, Figure 4 This is a schematic diagram of a position sampling directed graph according to an embodiment of the present disclosure, such as... Figure 4 As shown, a directed graph of position sampling can be composed of two elements: sampling points and trajectory segments. Sampling points are divided into initial layer sampling points, intermediate layer sampling points, and final layer sampling points. A smooth line connecting sampling points of two adjacent layers (including but not limited to polynomial trajectories, spline curves, spiral curves, etc.) can form a trajectory segment.

[0043] Here, a sampling point can be understood as the vehicle's position. Specifically, the current position of the vehicle can be used as the starting layer sampling point. Starting from the current position, the position of the current vehicle at each time moment is sampled until the ending time, resulting in intermediate layer sampling points and termination layer sampling points. Then, based on the connection between adjacent layer sampling points, a trajectory segment is formed, resulting in a directed position sampling graph.

[0044] After obtaining the position sampling directed graph, the lateral behavior of each trajectory segment in the position sampling directed graph can be defined, so as to facilitate the subsequent construction of a behavior sequence dictionary based on the lateral behavior.

[0045] Optionally, before constructing the behavior sequence dictionary of the sampling points based on the lateral behavior of the trajectory segments connected between the sampling points and the next layer of sampling points, the method further includes: for each trajectory segment, determining the lateral behavior of the trajectory segment based on the position type of the adjacent layer sampling points connecting the trajectory segment.

[0046] The location type can reflect the positional relationship between the sampling point and the lane, and can be determined based on the area occupied by the sampling point in the lane.

[0047] like Figure 5 As shown, Figure 5FIG. 1 is a schematic diagram of one position type in an embodiment of the present disclosure, where the position type can be OUTSIDE_LANE_LEFT, OUTSIDE_LANE_OVERLAP_LEFT, INSIDE_LANE_OVERLAP_LEFT, INSIDE_LANE, INSIDE_LANE_OVERLAP_RIGHT, OUTSIDE_LANE_OVERLAP_RIGHT, or OUTSIDE_LANE_RIGHT.

[0048] In the embodiment, for each trajectory segment, the position type of the two adjacent layer sampling points connected with the trajectory segment can be determined according to the area occupied by the two sampling points in the lane.

[0049] For example, if the sampling point is located in the left outside lane area, the position type of the sampling point is OUTSIDE_LANE_LEFT; if the sampling point is located in the right outside lane area, the position type of the sampling point is OUTSIDE_LANE_RIGHT; if the sampling point is located outside the left overlap area, the position type of the sampling point is OUTSIDE_LANE_OVERLAP_LEFT; if the sampling point is located inside the left overlap area, the position type of the sampling point is INSIDE_LANE_OVERLAP_LEFT; if the sampling point is located in the lane center area, the position type of the sampling point is INSIDE_LANE; if the sampling point is located inside the right overlap area, the position type of the sampling point is INSIDE_LANE_OVERLAP_RIGHT; if the sampling point is located outside the right overlap area, the position type of the sampling point is OUTSIDE_LANE_OVERLAP_RIGHT; if the sampling point is located in the right outside lane area, the position type of the sampling point is OUTSIDE_LANE_RIGHT.

[0050] Further, the lateral behavior of the trajectory segment can be determined according to the position types of the adjacent layer sampling points connected with the trajectory segment. The lateral behavior of the trajectory segment can be understood as the behavior of the current vehicle in the lateral direction along the trajectory segment.

[0051] In the embodiment of the present disclosure, the lateral behavior of the trajectory segment in different types of lanes can have different definitions. Specifically, the lateral behavior of the trajectory segment can be determined in combination with the lane type of the associated lane of the sampling point.

[0052] In one example, the lateral behavior of the trajectory segment is determined according to the position types of the adjacent layer sampling points connected with the trajectory segment, including:

[0053] In a case where the associated lane of the adjacent layer sampling point is the current lane of the current vehicle, the lateral behavior corresponding to the position type of the adjacent layer sampling point is queried in the first behavior mapping table; in a case where the associated lane of the adjacent layer sampling point is the lane-changing lane of the current vehicle, the lateral behavior corresponding to the position type of the adjacent layer sampling point is queried in the second behavior mapping table.

[0054] The lateral behavior in the first behavior mapping table includes lane keeping, left lane borrowing, right lane borrowing, lane borrowing left keeping, lane borrowing right keeping, lane borrowing left returning, and lane borrowing right returning, and the lateral behavior in the second behavior mapping table includes lane keeping, left lane-changing preparation, right lane-changing preparation, left lane changing, and right lane changing.

[0055] Specifically, the associated lane of the sampling point can be the lane corresponding to the sampling point, and the type of the associated lane can be the current lane or the lane-changing lane. For example, Figure 6 FIG. 1 is a schematic diagram of a lane according to an embodiment of the present disclosure, where lane A is the lane in which the current vehicle travels, i.e., the current lane, and lane B and lane C are the lanes in which the current vehicle needs to perform a lane-changing action to enter, i.e., the lane-changing lanes.

[0056] For the current lane, the lateral behavior of the trajectory segment can be lane keeping (LANE KEEPING), left lane borrowing (LANE BORROW LEFT), right lane borrowing (LANE BORROW RIGHT), lane borrowing left keeping (LANE BORROW LEFT KEEPING), lane borrowing right keeping (LANE BORROW RIGHT KEEPING), lane borrowing left returning (LANE BORROW RETURNING FROM LEFT), or lane borrowing right returning (LANE BORROW RETURNING FROM RIGHT).

[0057] Specifically, in a case where the associated lane of the adjacent layer sampling point is the current lane, the corresponding lateral behavior can be queried from the first behavior mapping table according to the position types of the two adjacent layer sampling points (i.e., the start point and the end point in the trajectory segment) constituting the trajectory segment. The first behavior mapping table includes the position types of the sampling points and the corresponding lateral behaviors. For example, Table 1 shows a first behavior mapping table, where the first row represents the position type of the end sampling point in the trajectory segment, and the first list represents the position type of the start sampling point in the trajectory segment.

[0058] Table 1: A first behavior mapping table

[0059]

[0060]

[0061] For the lane-changing lane, the lateral behavior of the trajectory segment can be: the lateral behavior includes lane keeping (LANE KEEPING), left lane-changing preparation (LANE_CHANGE_PREPARE_ON_LEFT), right lane-changing preparation (LANE_CHANGE_PREPARE_ON_RIGHT), lane changing to the left (LANE_CHANGING_TO_LEFT) or lane changing to the right (LANE_CHANGING_TO_RIGHT).

[0062] Specifically, in the case where the associated lane of the adjacent layer sampling point is a lane-changing lane, the corresponding lateral behavior can be queried from the second behavior mapping table according to the position type of the two adjacent layer sampling points constituting the trajectory segment. The second behavior mapping table includes the position type of each sampling point and the corresponding lateral behavior. For example, Table 2 shows a second behavior mapping table, where the first row represents the position type of the end sampling point in the trajectory segment, and the first table represents the position type of the start sampling point in the trajectory segment.

[0063] Table 2: A second behavior mapping table

[0064]

[0065]

[0066]

[0067] By the above embodiment, the lateral behavior under different lane types is defined respectively, compared with the uniform lateral behavior definition for all lanes, the embodiment can improve the accuracy of the lateral behavior of the trajectory segment, and further make the trajectory search more accurate.

[0068] S120, for each sampling point in the position sampling directed graph, constructing a behavior sequence dictionary of the sampling point according to the lateral behavior of the trajectory segment connected between the sampling point and the next layer sampling point.

[0069] The behavior sequence dictionary includes at least one behavior sequence and a sampling point identifier set corresponding to each behavior sequence, the behavior sequence is composed of the lateral behavior of each trajectory segment in the trajectory starting from the sampling point, and the sampling point identifier set includes the next layer sampling point satisfying the behavior sequence between the sampling point.

[0070] It should be noted that the behavior sequence dictionary is a data structure constructed by the embodiments of the present disclosure in order to facilitate determination of all behavior sequences in the position sampling directed graph and subsequent trajectory search. If the behavior sequence dictionary is not used, the trajectory search needs to be implemented by first traversing all possible trajectories in the position sampling directed graph.

[0071] Figure 7 A schematic diagram of a position sampling directed graph in the embodiments of the present disclosure is shown, in which the starting layer is represented by the 0th layer, the intermediate layer is assumed to have only one layer, represented by the 1st layer, and the termination layer is represented by the 2nd layer. The number of sampling points in the starting layer is 1, which can be represented by n(0, 0). The number of sampling points in the intermediate layer is 3, which can be represented by n(1, 0), n(1, 1), and n(1, 2). The number of sampling points in the termination layer is 3, which can be represented by n(2, 0), n(2, 1), and n(2, 2).

[0072] For example, if the behavior sequence dictionary is not used, the flow of traversing all possible trajectories (3 x 7 = 21 trajectories) is as follows: Figure 7

[0073] The behavior sequence of the trajectory composed of n(0, 0), n(1, 0), and n(2, 0) (including the lateral behavior of each trajectory segment in the trajectory) is: [LANE_BORROW_LEFT, LANE_BORROW_LEFT_KEEPING]; the behavior sequence of n(0, 0), n(1, 0), and n(2, 1) is: [LANE_BORROW_LEFT, LANE_BORROW_RETURNING_FROM_LEFT]; the behavior sequence of n(0, 0), n(1, 0), and n(2, 0) is: [LANE_KEEPING, LANE_BORROW_LEFT]; the behavior sequence of n(0, 0), n(1, 1), and n(2, 1) is: [LANE_KEEPING, LANE_KEEPING]; the behavior sequence of n(0, 0), n(1, 1), and n(2, 2) is: [LANE_KEEPING, LANE_KEEPING]; …; the behavior sequence of n(0, 0), n(1, 2), and n(2, 2) is: [LANE_KEEPING, LANE_KEEPING].

[0074] The above traversal method, although it counts all behavior sequences, can easily cause an explosion of computing power. When the number of sampling points increases, the number of trajectory segments increases, which in turn causes a linear increase in computing power consumption. Furthermore, if all trajectories that meet a specific behavior sequence are to be queried from the above-mentioned traversal method, each trajectory needs to be judged to see if it meets the specific behavior sequence, which is a very time-consuming process. Figure 7 ​For example, 21 times need to be judged, the trajectory query efficiency is low and the required computing power is more.

[0075] Therefore, in order to solve the problem of computing power explosion caused by the above traversal statistics, the embodiment of the disclosure provides a data structure, that is, a behavior sequence dictionary, which describes each sampling point through the behavior sequence dictionary.

[0076] Specifically, for each sampling point in the location sampling directed graph, it respectively has a corresponding behavior sequence dictionary. The behavior sequence dictionary of the sampling point is composed of at least one behavior sequence and a sampling point identifier set corresponding to each behavior sequence, wherein the behavior sequence includes the lateral behavior of each trajectory segment in the trajectory starting from the sampling point, and the sampling point identifier set includes the next layer sampling point satisfying the behavior sequence with the sampling point.

[0077] For example, the lateral behavior of the i-th trajectory segment is defined as δ i , and the behavior sequence of a trajectory composed of N trajectory segments is Δ=[δ1, δ2...δ N ], as shown in Figure 8 , which is a trajectory diagram in the embodiment of the disclosure. Figure 8

[0078] The principle of defining two behavior sequences as equal is that each lateral behavior contained therein is equal, which can be expressed as Δ1=Δ2. If all valid elements of Δ1 are equal to the corresponding elements of Δ2 except invalid elements, it is considered that the latter contains the former, which is expressed as Δ1∈Δ2.

[0079] For example, Δ1=[none, none, LANE_KEEPING, LANE_KEEPING], Δ2=[LANE_KEEPING, LANE_KEEPING, LANE_KEEPING, LANE_KEEPING], which satisfies the relationship of Δ1∈Δ2.

[0080] The key-value pair of the behavior sequence Δ and the sampling point identifier set Β (the former is the key and the latter is the value) is defined as: Γ=Δ:Β. Define S={Γ}, S is the dictionary of “behavior sequence-sampling point identifier set”, that is, the behavior sequence dictionary (the dictionary is a data structure, and only one element with equal keys is retained, which can be realized by a hash table), to record the behavior sequence of each sampling point.

[0081] Define S(n) as the behavior sequence dictionary of the n-th sampling point, which includes: starting from the n-th sampling point, to the terminal layer sampling point in the location sampling directed graph, all possible behavior sequences, and the sampling point identifier set corresponding to each behavior sequence.

[0082] ​In the embodiments of the present disclosure, the behavior sequence dictionary of each sampling point can be obtained by starting from the terminal layer sampling points in the position sampling directed graph, adding and merging the topological relations of each trajectory segment, and recording the topological relation set.

[0083] For example, for any sampling point n, assuming that it is the starting point of L trajectory segments, and the ending points of the L trajectory segments are [k1, k2,... kL], the behavior sequence dictionary of the sampling point n is: L

[0084]

[0085] In a specific embodiment, for each sampling point in the position sampling directed graph, the behavior sequence dictionary of the sampling point is constructed according to the lateral behavior of the trajectory segment connecting the sampling point and the next layer sampling point, including the following steps:

[0086] Step 121, determining the behavior sequence dictionary of the terminal layer sampling point.

[0087] Step 122, determining the terminal layer sampling point as a processed sampling point, determining the previous layer sampling point of the terminal layer sampling point as a to-be-processed sampling point, determining the behavior sequence dictionary of the to-be-processed sampling point based on the behavior sequence dictionary of the processed sampling point and the lateral behavior of the trajectory segment connecting the to-be-processed sampling point and the processed sampling point.

[0088] Step 123, taking the to-be-processed sampling point as a new processed sampling point, taking the previous layer sampling point of the to-be-processed sampling point as a new to-be-processed sampling point, and returning to execute the step of determining the behavior sequence dictionary of the to-be-processed sampling point until the behavior sequence dictionary of the starting layer sampling point is obtained.

[0089] In step 121, the behavior sequence dictionary of the terminal layer sampling point can be determined first. Since the terminal layer sampling point is the last layer sampling point in the position sampling directed graph, the trajectory starting from the terminal layer sampling point does not exist, and therefore the behavior sequence dictionary of each terminal layer sampling point can be empty.

[0090] Further, each previous layer sampling point of the terminal layer sampling point can be taken as a to-be-processed sampling point, the terminal layer sampling point is taken as a processed sampling point, the behavior sequence dictionary of the to-be-processed sampling point is obtained based on the behavior sequence dictionary of the processed sampling point and the lateral behavior of the trajectory segment connecting the to-be-processed sampling point and the processed sampling point, so as to determine the behavior sequence dictionary of each sampling point from the terminal layer.

[0091] ​For example, based on the behavior sequence dictionary of the processed sampling point and the lateral behavior of the trajectory segment between the to-be-processed sampling point and the processed sampling point, the behavior sequence dictionary of the to-be-processed sampling point can be determined by repeatedly performing the following operations from the first trajectory segment between the to-be-processed sampling point and each processed sampling point until the last trajectory segment is reached:

[0092] On the basis of each behavior sequence in the behavior sequence dictionary of the processed sampling point connected by the trajectory segment, the lateral behavior of the trajectory segment is added to obtain each to-be-written behavior sequence, and it is determined whether each to-be-written behavior sequence already exists in the behavior sequence dictionary of the to-be-processed sampling point.

[0093] For the to-be-written behavior sequence that does not exist in the behavior sequence dictionary of the to-be-processed sampling point, a sampling point identifier set corresponding to the to-be-written behavior sequence is generated according to the identifier of the processed sampling point, and the to-be-written behavior sequence and the corresponding sampling point identifier set are written into the behavior sequence dictionary of the to-be-processed sampling point.

[0094] For the to-be-written behavior sequence that exists in the behavior sequence dictionary of the to-be-processed sampling point, the identifier of the processed sampling point is added to the sampling point identifier set corresponding to the same behavior sequence as the to-be-written behavior sequence.

[0095] After obtaining the behavior sequence dictionary of each to-be-processed sampling point, the to-be-processed sampling point can be taken as a new processed sampling point, and the previous layer sampling point of the to-be-processed sampling point can be taken as a new to-be-processed sampling point, and step 122 is returned to be executed until the behavior sequence dictionary of the starting layer sampling point is obtained.

[0096] Figure 9 A schematic diagram of a position sampling directed graph in an embodiment of the present disclosure is shown in FIG. 1. Figure 9 Taking the trajectory segment 1 between the starting layer sampling point n and the sampling point k_1 as an example, the determination process of the behavior sequence dictionary S(n) of the starting layer sampling point n is described: for the trajectory segment 1, on the basis of each behavior sequence in the behavior sequence dictionary S(k_1) of the sampling point k_1, the lateral behavior of the trajectory segment 1 is added to obtain each to-be-written behavior sequence, if the to-be-written behavior sequence does not exist in S(n), the to-be-written behavior sequence is written into S(n), and the corresponding sampling point identifier set is constructed according to the identifier of the sampling point k_1; if the to-be-written behavior sequence exists in S(n), the identifier of the sampling point k_1 is added to the corresponding sampling point identifier set.

[0097] By steps 121-123, the behavior sequence dictionary of each sampling point can be determined from the last layer of sampling points in the position sampling directed graph, layer by layer, until the behavior sequence dictionary of the starting layer sampling point is obtained. Each behavior sequence in the behavior sequence dictionary of the starting layer sampling point can be understood as all possible behavior sequences in the position sampling directed graph. Compared with the traversal statistics, the traversal statistics needs to traverse the trajectory segment between the sampling points in column 0 and column 1 multiple times, while the method provided by the embodiment does not need to traverse repeatedly, greatly reduces the calculation complexity, reduces the required computing power, and further improves the decision efficiency of the vehicle lateral behavior.

[0098] In Figure 7 The position sampling directed graph is taken as an example to illustrate the process of determining the behavior sequence dictionary of each sampling point. Figure 7 There are three layers of sampling points in the position sampling directed graph (i.e., corresponding to three columns), and column 2 is the terminal layer sampling point of the position sampling directed graph. First, the behavior sequence dictionary of all terminal layer sampling points can be set to empty, and then the position type of the sampling point can be determined according to the position relationship between the sampling point and the lane, and the lateral behavior of all trajectory segments in the position sampling directed graph can be determined.

[0099] The sampling point in column i and numbered j in the graph is denoted as n(i, j). Table 3 shows the lateral behavior of the trajectory segment between the sampling points in column 0 and column 1, and Table 4 shows the lateral behavior of the trajectory segment between the sampling points in column 1 and column 2.

[0100] Table 3 Lateral behavior of trajectory segment between sampling points in column 0 and column 1

[0101]

[0102] Table 4 Lateral behavior of trajectory segment between sampling points in column 1 and column 2

[0103]

[0104] Further, based on the dynamic programming idea, the behavior sequence dictionary can be determined from the terminal layer sampling point. Table 5 shows the behavior sequence dictionary of the terminal layer sampling point, Table 6 shows the behavior sequence dictionary of the sampling point in column 1, and Table 7 shows the behavior sequence dictionary of the sampling point in column 0.

[0105] Table 5 Behavior sequence dictionary of terminal layer sampling point

[0106]

[0107] Table 6 Behavior sequence dictionary of sampling point in column 1

[0108]

[0109] Table 7 Behavior sequence dictionary of column 0 sampling points

[0110]

[0111]

[0112] From the table 7, from n(0,0), there are four behavior sequences in the position sampling directed graph, and the behavior sequence dictionary also records the sampling point identifier set under each behavior sequence.

[0113] In S130, a to-be-searched behavior sequence is acquired, and a current target trajectory conforming to the to-be-searched behavior sequence is determined in the position sampling directed graph based on the behavior sequence dictionary of the sampling points.

[0114] In the embodiments of the present disclosure, after obtaining the behavior sequence dictionary of each sampling point, a specific behavior sequence can be searched to find a current target trajectory conforming to the specific behavior sequence.

[0115] The to-be-searched behavior sequence is a specific behavior sequence for which a corresponding trajectory needs to be searched. In the embodiments of the present disclosure, all possible behavior sequences can be displayed first, and then all possible behavior sequences are evaluated and scored based on evaluation indexes such as safety, comfort, and executability, and the behavior sequence with the optimal score is determined as the to-be-searched behavior sequence, and then a current target trajectory conforming to the optimal behavior sequence is queried.

[0116] For example, the to-be-searched behavior sequence is acquired, including: determining a behavior sequence score corresponding to each behavior sequence in the behavior sequence dictionary of the starting layer sampling point, and determining the behavior sequence with the optimal behavior sequence score as the to-be-searched behavior sequence.

[0117] Correspondingly, after determining the current target trajectory conforming to the to-be-searched behavior sequence in the position sampling directed graph, the current target trajectory conforming to the to-be-searched behavior sequence is further output.

[0118] That is, the behavior sequences in the behavior sequence dictionary of the starting layer sampling point reflect all possible behavior sequences in the position sampling directed graph, each behavior sequence can be evaluated and scored in combination with the safety, comfort, and executability of the behavior sequence, the behavior sequence with the optimal behavior sequence score is determined as the to-be-searched behavior sequence, and then a current target trajectory conforming to the to-be-searched behavior sequence is queried in the position sampling directed graph based on the behavior sequence dictionary of each sampling point, which is output as the decision result of the lateral behavior of the current vehicle.

[0119] In the example, all possible behavior sequences can be evaluated and scored, and the optimal behavior sequence is compared, and the trajectory search is performed, thereby providing a fast analysis service of all possible lateral behaviors in the position sampling directed graph for the decision planning module of the vehicle, and providing a fast screening service of trajectories conforming to a specific behavior sequence.

[0120] In the embodiments of the present disclosure, in addition to showing all possible behavior sequences and then evaluating and scoring all possible behavior sequences, the decision result output at the previous time can also be used as a behavior sequence to be searched to query the current target trajectories conforming to the decision result at the previous time, and then the optimal trajectory is output as the decision result at the current time by evaluation and scoring.

[0121] For example, in another example, the behavior sequence to be searched is obtained, including: obtaining a historical target trajectory output at the previous time, and determining the behavior sequence to be searched according to the historical target trajectory output at the previous time.

[0122] Correspondingly, after determining the current target trajectories conforming to the behavior sequence to be searched in the position sampling directed graph, the method further includes: adding a preset mark to each current target trajectory in the position sampling directed graph, wherein the preset mark is used to increase the trajectory score of the current target trajectory; and determining the corresponding trajectory score of each trajectory in the position sampling directed graph, and outputting a trajectory with the optimal trajectory score.

[0123] That is, the behavior sequence corresponding to the decision result at the previous time (the historical target trajectory output at the previous time) can be used as the behavior sequence to be searched. Further, based on the behavior sequence dictionary of each sampling point, the current target trajectories conforming to the behavior sequence to be searched are queried in the position sampling directed graph, and a preset mark is added to each current target trajectory.

[0124] Further, each trajectory in the position sampling directed graph is evaluated and scored in combination with the safety, comfort and executability of the behavior sequence, wherein the trajectory with the preset mark can be increased by the corresponding trajectory score, and the specific value of the increase is not limited in the embodiments. Finally, the trajectory with the optimal trajectory score can be output as the decision result of the lateral behavior of the current vehicle.

[0125] In the above example, by querying the current target trajectories conforming to the decision result at the previous time, a higher score is given when evaluating and scoring, so as to stabilize the vehicle behavior and improve the consistency of the behavior.

[0126] In the embodiments of the present disclosure, based on the behavior sequence dictionary of each sampling point, the current target trajectories conforming to the behavior sequence to be searched can be queried from the behavior sequence dictionary of the starting layer sampling point.

[0127] In a specific embodiment, based on the behavior sequence dictionary of the sampling points, the current target trajectory conforming to the behavior sequence to be searched is determined in the position sampling directed graph, comprising the following steps:

[0128] Step 131, determining the starting layer sampling point as the sampling point to be queried, and querying the sampling point identifier set corresponding to the behavior sequence to be searched in the behavior sequence dictionary of the sampling point to be queried;

[0129] Step 132, determining the trajectory segment between the sampling point in the queried sampling point identifier set and the sampling point to be queried as the target output segment;

[0130] Step 133, updating the behavior sequence to be searched, and taking the sampling point in the queried sampling point identifier set as the new sampling point to be queried, returning to execute the step of querying the sampling point identifier set corresponding to the behavior sequence to be searched until the sampling point in the queried sampling point identifier set is the terminal layer sampling point;

[0131] Step 134, generating the current target trajectory corresponding to the behavior sequence to be searched based on all the target output segments.

[0132] Specifically, the starting layer sampling point can be taken as the sampling point to be queried, and then the sampling point identifier set corresponding to the same behavior sequence of the behavior sequence to be searched is found in the behavior sequence dictionary of the sampling point to be queried.

[0133] Further, the trajectory segment between the sampling point in the queried sampling point identifier set and the sampling point to be queried is marked as the target output segment. It should be noted that the purpose of containing the sampling point identifier set in the behavior sequence dictionary is to facilitate the behavior sequence as the key to find the corresponding value, i.e. the sampling point, and then to determine the target output segment according to the found sampling point.

[0134] Further, the sampling point in the queried sampling point identifier set is taken as the new sampling point to be queried, and the i-th behavior in the behavior sequence to be searched is set to be empty, where i is the layer number of the sampling point to be queried in the position sampling directed graph in this round, and then returning to execute the step of querying the sampling point identifier set corresponding to the behavior sequence to be searched in the behavior sequence dictionary of the sampling point to be queried until the sampling point in the queried sampling point identifier set is the terminal layer sampling point. Finally, the target output segments of each round of queries are combined to obtain the current target trajectory.

[0135] Following Figure 7 the position sampling directed graph shown in the drawing, it is assumed that the behavior sequence Δ s to be searched is [LANE KEEPING, LANE KEEPING].

[0136] In the first round of searching, the behavior sequence to be searched Δs ′ is [LANE KEEPING, LANE KEEPING]. The starting layer sampling point is taken as a to-be-queried sampling point, denoted as: curr_to_check = [n(0, 0)]. Δ s ′ is searched from the behavior sequence dictionary of n(0, 0), and a key-value pair {[LANE KEEPING, LANE KEEPING]: (n(1, 1), n(1, 2))} is found, indicating that there is a trajectory segment in the position sampling directed graph that meets the to-be-searched behavior sequence, and the ending points of the trajectory segment are n(1, 1) and n(1, 2) respectively. The trajectory segment between n(0, 0) and n(1, 1) and the trajectory segment between n(0, 0) and n(1, 2) are taken as target output segments. As shown in Figure 10 Figure 10 FIG. 1 is a schematic diagram of a target output segment in an embodiment of the present disclosure, and the thick trajectory segment in the figure is the target output segment.

[0137] In the second round of searching, n(1, 1) and n(1, 2) are taken as new to-be-queried sampling points, and the to-be-searched behavior sequence Δ s ′ is updated to [none, LANE KEEPING]. Δ s ′ is searched from the behavior sequence dictionary of n(1, 1), and a key-value pair {[none, LANE KEEPING]: (n(2, 1), n(2, 2))} is found. The trajectory segment between n(1, 1) and n(2, 1) and the trajectory segment between n(1, 1) and n(2, 2) are taken as target output segments. Further, Δ s ′ is searched from the behavior sequence dictionary of n(1, 2), and a key-value pair {[none, LANE KEEPING]: (n(2, 1), n(2, 2))} is found. The trajectory segment between n(1, 2) and n(2, 1) and the trajectory segment between n(1, 2) and n(2, 2) are taken as target output segments. In the second round of searching, (n(2, 1), n(2, 2))} found is a terminal layer sampling point, so the searching can be stopped, and all target output segments are combined to obtain the current target trajectory. At this time, all trajectories meeting the behavior sequence [LANE KEEPING, LANE KEEPING] have been found.

[0138] Through the above steps 131-134, the trajectory search based on the behavior sequence dictionary is realized. By taking the to-be-searched behavior sequence as a key, the corresponding value is found from the behavior sequence dictionary, and then the target output segment is obtained to construct the current target trajectory. All trajectories meeting the to-be-searched behavior sequence can be extracted from the position sampling directed graph. Compared with the traversal of all trajectories and the judgment of each trajectory, the calculation complexity is greatly reduced, the required computing power is reduced, and the decision efficiency of the vehicle lateral behavior is provided.​

[0139] The vehicle lateral behavior decision method provided by the embodiment can use the behavior sequence decided at the previous moment as the to-be-searched behavior sequence to evaluate and score after trajectory searching, or first evaluate and score various behavior sequences to search for a trajectory based on the to-be-searched behavior sequence with the optimal score, thereby solving the problem that the decision result does not conform to the habits of human drivers due to a small amount of fluctuations in perception information in the prior art, achieving the purposes of stabilizing vehicle behavior and improving behavior consistency, and realizing dictionary-based trajectory searching, without the need to count all possible behavior sequences and then query from the behavior sequences, thereby greatly reducing the consumed computing power and improving the trajectory searching efficiency.

[0140] Figure 11 FIG. 1 is a structural schematic diagram of a vehicle lateral behavior decision device according to an embodiment of the present disclosure. Figure 11 As shown in the figure, the device comprises a directed graph acquisition module 210, a dictionary determination module 220, and a trajectory searching module 230, wherein:

[0141] The directed graph acquisition module 210 is configured to acquire a position sampling directed graph of a current vehicle, wherein the position sampling directed graph is composed of a starting layer sampling point, at least one intermediate layer sampling point, and at least one terminal layer sampling point, and a connection between adjacent layer sampling points forms a trajectory segment.

[0142] The dictionary determination module 220 is configured to, for each sampling point in the position sampling directed graph, construct a behavior sequence dictionary of the sampling point according to a lateral behavior of a trajectory segment connected between the sampling point and a next layer sampling point, wherein the behavior sequence dictionary comprises at least one behavior sequence and a sampling point identifier set corresponding to each behavior sequence, the behavior sequence is composed of lateral behaviors of trajectory segments in a trajectory with the sampling point as a starting point, and the sampling point identifier set comprises a next layer sampling point satisfying the behavior sequence and connected with the sampling point.

[0143] The trajectory searching module 230 is configured to acquire a to-be-searched behavior sequence, and determine a current target trajectory conforming to the to-be-searched behavior sequence in the position sampling directed graph based on the behavior sequence dictionary of the sampling point.

[0144] The vehicle lateral behavior decision device provided by the embodiment of the present disclosure can execute the steps in the vehicle lateral behavior decision method provided by the method embodiment of the present disclosure, and has the execution steps and beneficial effects which are not described herein.

[0145] Figure 12 This is a schematic diagram of the structure of an electronic device according to an embodiment of this disclosure. See below for details. Figure 12 It shows a schematic diagram of a structure suitable for implementing the electronic device 500 in the embodiments of this disclosure. Figure 12 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0146] like Figure 12 As shown, the electronic device 500 may include a processing device (e.g., a central processing unit, a graphics processor, etc.) 501, which can perform various appropriate actions and processes to implement the methods of the embodiments described herein, based on a program stored in a read-only memory (ROM) 502 or a program loaded from a storage device 508 into a random access memory (RAM) 503. The RAM 503 also stores various programs and data required for the operation of the electronic device 500. The processing device 501, ROM 502, and RAM 503 are interconnected via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0147] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts, thereby implementing the vehicle lateral behavior decision-making method as described above. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 509, or installed from a storage device 508, or installed from a ROM 502. When the computer program is executed by the processing device 501, it performs the functions defined in the methods of embodiments of this disclosure.

[0148] It should be noted that the computer readable medium in the above disclosure can be a computer readable signal medium or a computer readable storage medium or any combination of the two. The computer readable storage medium may, for example, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or apparatus, or any combination of the above. More specific examples of the computer readable storage medium can include, but are not limited to, an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, the computer readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device or apparatus. In the present disclosure, the computer readable signal medium can include a data signal carried in a baseband or as a part of a carrier wave, which carries computer readable program code. Such a propagated data signal can take many forms, including but not limited to an electromagnetic signal, an optical signal or any suitable combination of the above. The computer readable signal medium can also be any computer readable medium other than the computer readable storage medium, which can send, propagate or transmit a program for use by or in conjunction with an instruction execution system, device or apparatus. The program code contained in the computer readable medium can be transmitted by any suitable medium, including but not limited to a wire, a cable, a RF (radio frequency) or the like, or any suitable combination of the above.

[0149] The computer readable medium described above can be contained in the electronic device described above; or can exist separately and not be assembled into the electronic device. The computer readable medium described above carries one or more programs, which, when executed by the electronic device, cause the electronic device to perform the vehicle lateral behavior decision method provided by any of the embodiments described above.

[0150] Optionally, when the one or more programs are executed by the electronic device, the electronic device can also perform other steps described in the embodiments.

[0151] In the context of this disclosure, a machine-readable medium can be a tangible medium that contains or stores a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include but is not limited to an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of the machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0152] Scheme 1, a vehicle lateral behavior decision-making method, the method comprising:

[0153] Obtaining a position sampling directed graph of a current vehicle, wherein the position sampling directed graph is composed of a starting layer sampling point, at least one intermediate layer sampling point, and at least one terminal layer sampling point, and connections between adjacent layer sampling points form trajectory segments;

[0154] For each sampling point in the position sampling directed graph, constructing a behavior sequence dictionary of the sampling point according to lateral behaviors of trajectory segments connected between the sampling point and a next layer sampling point, wherein the behavior sequence dictionary comprises at least one behavior sequence and a sampling point identification set corresponding to each behavior sequence, the behavior sequence is composed of lateral behaviors of each trajectory segment in a trajectory with the sampling point as a starting point, and the sampling point identification set comprises next layer sampling points satisfying the behavior sequence and connected with the sampling point;

[0155] Obtaining a to-be-searched behavior sequence, and determining a current target trajectory in the position sampling directed graph that conforms to the to-be-searched behavior sequence based on the behavior sequence dictionary of the sampling point.

[0156] Scheme 2, according to scheme 1, the method of constructing a behavior sequence dictionary of each sampling point in the position sampling directed graph according to lateral behaviors of trajectory segments connected between the sampling point and a next layer sampling point, comprising:

[0157] Determining the behavior sequence dictionary of the terminal layer sampling point;

[0158] determining, as a processed sample point, the terminal layer sample point, determining, as a to-be-processed sample point, a previous layer sample point of the terminal layer sample point, determining, based on a behavior sequence dictionary of the processed sample point and a lateral behavior of a trajectory segment between the to-be-processed sample point and the processed sample point, a behavior sequence dictionary of the to-be-processed sample point;

[0159] returning to execute the step of determining the behavior sequence dictionary of the to-be-processed sample point, taking the to-be-processed sample point as a new processed sample point and taking a previous layer sample point of the to-be-processed sample point as a new to-be-processed sample point, until a behavior sequence dictionary of the starting layer sample point is obtained.

[0160] Scheme 3, the method according to scheme 1, the behavior sequence dictionary based on the sample point, determining a current target trajectory in the position sample directed graph that meets the to-be-searched behavior sequence, comprising:

[0161] determining, as a to-be-queried sample point, the starting layer sample point, and querying, in the behavior sequence dictionary of the to-be-queried sample point, a sample point identifier set corresponding to the to-be-searched behavior sequence;

[0162] determining, as a target output segment, a trajectory segment between a sample point in the queried sample point identifier set and the to-be-queried sample point;

[0163] updating the to-be-searched behavior sequence, taking the sample point in the queried sample point identifier set as a new to-be-queried sample point, and returning to execute the step of querying the sample point identifier set corresponding to the to-be-searched behavior sequence until the sample point in the queried sample point identifier set is a terminal layer sample point;

[0164] generating, based on all the target output segments, a current target trajectory corresponding to the to-be-searched behavior sequence.

[0165] Scheme 4, the method according to scheme 1, the to-be-searched behavior sequence, comprising:

[0166] determining, for each behavior sequence in the behavior sequence dictionary of the starting layer sample point, a corresponding behavior sequence score, and determining a behavior sequence with an optimal behavior sequence score as the to-be-searched behavior sequence;

[0167] correspondingly, after determining the current target trajectory in the position sample directed graph that meets the to-be-searched behavior sequence, further comprising:

[0168] outputting the current target trajectory that meets the to-be-searched behavior sequence.

[0169] Scheme 5, the method according to scheme 1, the to-be-searched behavior sequence, comprising:

[0170] acquire a historical target trajectory output at a previous time, and determine a to-be-searched behavior sequence according to the historical target trajectory output at the previous time;

[0171] Correspondingly, after determining the current target trajectory in the position-sampled directed graph that meets the to-be-searched behavior sequence, the method further includes:

[0172] In the position-sampled directed graph, a preset mark is added to each current target trajectory, where the preset mark is used to increase a trajectory score of the current target trajectory.

[0173] For each trajectory in the position-sampled directed graph, a corresponding trajectory score is determined, and a trajectory with an optimal trajectory score is output.

[0174] Scheme 6, the method according to scheme 1, before constructing the behavior sequence dictionary of the sampling point, the lateral behavior of the trajectory segment connected between the sampling point and the next layer sampling point, further includes:

[0175] For each trajectory segment, the lateral behavior of the trajectory segment is determined according to the position type of the adjacent layer sampling point connected to the trajectory segment.

[0176] Scheme 7, the method according to scheme 6, the lateral behavior of the trajectory segment is determined according to the position type of the adjacent layer sampling point connected to the trajectory segment, including:

[0177] In the case that the associated lane of the adjacent layer sampling point is the current lane of the current vehicle, in the first behavior mapping table, the lateral behavior corresponding to the position type of the adjacent layer sampling point is queried;

[0178] In the case that the associated lane of the adjacent layer sampling point is the lane-changing lane of the current vehicle, in the second behavior mapping table, the lateral behavior corresponding to the position type of the adjacent layer sampling point is queried;

[0179] The lateral behaviors in the first behavior mapping table include lane keeping, left lane borrowing, right lane borrowing, lane borrowing left keeping, lane borrowing right keeping, lane borrowing left returning, and lane borrowing right returning, and the lateral behaviors in the second behavior mapping table include lane keeping, left lane-changing preparation, right lane-changing preparation, left lane-changing, and right lane-changing.

[0180] Scheme 8, a vehicle lateral behavior decision device, comprising:

[0181] A directed graph acquisition module is configured to acquire a position-sampled directed graph of a current vehicle, where the position-sampled directed graph is composed of a starting layer sampling point, at least one intermediate layer sampling point, and at least one terminal layer sampling point, and a connection between adjacent layer sampling points forms a trajectory segment.

[0182] a dictionary determining module configured to, for each sampling point in the position-sampled directed graph, construct a behavior sequence dictionary of the sampling point according to lateral behaviors of trajectory segments connected between the sampling point and next-layer sampling points, wherein the behavior sequence dictionary comprises at least one behavior sequence and a sampling point identification set corresponding to each behavior sequence, the behavior sequence is composed of lateral behaviors of each trajectory segment in a trajectory with the sampling point as a starting point, and the sampling point identification set comprises next-layer sampling points satisfying the behavior sequence between the sampling point;

[0183] a trajectory searching module configured to acquire a to-be-searched behavior sequence, and determine a current target trajectory conforming to the to-be-searched behavior sequence in the position-sampled directed graph based on the behavior sequence dictionary of the sampling point.

[0184] Scheme 9, an electronic device, comprising:

[0185] one or more processors;

[0186] a storage device configured to store one or more programs;

[0187] When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of schemes 1-7.

[0188] Scheme 10, a computer readable storage medium, having stored thereon a computer program, which is executed by a processor to implement the method according to any one of schemes 1-7.

[0189] The above description is merely preferred embodiments of the present disclosure and a description of principles of applied technologies. It should be understood by those skilled in the art that the disclosed scope of the present disclosure is not limited to the technical solutions formed by the specific combinations of the above technical features, and should also cover other technical solutions formed by any combinations of the above technical features or equivalent features without departing from the above disclosed concept. For example, the above technical features are replaced with technical features disclosed in the present disclosure (but not limited to) having similar functions to form technical solutions.

Claims

1. A method for vehicle lateral behavior decision-making, characterized in that, The method includes: Obtain the current vehicle's position sampling directed graph, wherein the position sampling directed graph consists of a starting layer sampling point, at least one intermediate layer sampling point, and at least one ending layer sampling point, and the connection between adjacent layer sampling points forms a trajectory segment; For each sampling point in the location sampling directed graph, a behavior sequence dictionary for the sampling point is constructed based on the lateral behavior of the trajectory segments connecting the sampling point to the next layer sampling point. The behavior sequence dictionary includes at least one behavior sequence and a set of sampling point identifiers corresponding to each behavior sequence. The behavior sequence is composed of the lateral behavior of each trajectory segment in the trajectory starting from the sampling point. The set of sampling point identifiers includes the next layer sampling points that satisfy the behavior sequence with respect to the sampling point. Obtain the sequence of behaviors to be searched, determine the starting layer sampling point as the sampling point to be queried, and query the set of sampling point identifiers corresponding to the sequence of behaviors to be searched in the behavior sequence dictionary of the sampling point to be queried. The trajectory segment between the sampling point in the queried sampling point identifier set and the sampling point to be queried is determined as the target output segment; Update the sequence of behaviors to be searched, and take the sample points in the queried sample point identifier set as the new sample points to be queried. Return to the step of querying the sample point identifier set corresponding to the sequence of behaviors to be searched, until the sample points in the queried sample point identifier set are the terminal layer sample points. Based on all target output segments, generate the current target trajectory corresponding to the sequence of behaviors to be searched.

2. The method according to claim 1, characterized in that, For each sampling point in the directed graph of the location sampling, a behavior sequence dictionary of the sampling point is constructed based on the lateral behavior of the trajectory segment connecting the sampling point to the next layer sampling point, including: Determine the behavior sequence dictionary of the terminal layer sampling points; The terminal layer sampling point is determined as the processed sampling point, and the sampling point of the previous layer of the terminal layer sampling point is taken as the sampling point to be processed. Based on the behavior sequence dictionary of the processed sampling point and the lateral behavior of the trajectory segment between the sampling point to be processed and the processed sampling point, the behavior sequence dictionary of the sampling point to be processed is determined. The sample point to be processed is taken as a new processed sample point, and the sample point of the previous layer of the sample point to be processed is taken as a new sample point to be processed. The process is repeated until the behavior sequence dictionary of the initial layer sample point is obtained.

3. The method according to claim 1, characterized in that, The process of obtaining the sequence of behaviors to be searched includes: For each behavior sequence in the behavior sequence dictionary of the initial layer sampling points, a corresponding behavior sequence score is determined, and the behavior sequence with the best behavior sequence score is determined as the behavior sequence to be searched. Accordingly, after determining the current target trajectory that matches the search behavior sequence in the directed graph at the location sampling point, the process further includes: Output the current target trajectory that matches the search behavior sequence.

4. The method according to claim 1, characterized in that, The process of obtaining the sequence of behaviors to be searched includes: Obtain the historical target trajectory output at the previous moment, and determine the sequence of behaviors to be searched based on the historical target trajectory output at the previous moment; Accordingly, after determining the current target trajectory that matches the search behavior sequence in the directed graph at the location sampling point, the process further includes: In the location-sampled directed graph, a preset marker is added to each current target trajectory, wherein the preset marker is used to increase the trajectory score of the current target trajectory; For each trajectory in the directed graph of the location sampling, determine the corresponding trajectory score, and output the trajectory with the best score.

5. The method according to claim 1, characterized in that, Before constructing the behavior sequence dictionary of the sampling points based on the lateral behavior of the trajectory segments connecting the sampling points to the next layer of sampling points, the following steps are also included: For each trajectory segment, the lateral behavior of the trajectory segment is determined based on the location type of the sampling points in the adjacent layers that connect the trajectory segment.

6. The method according to claim 5, characterized in that, Determining the lateral behavior of the trajectory segment based on the position type of the sampling points of adjacent layers connecting the trajectory segment includes: If the associated lane of the adjacent layer sampling point is the current lane of the current vehicle, query the lateral behavior corresponding to the location type of the adjacent layer sampling point in the first row mapping table; If the associated lane of the adjacent layer sampling point is the lane change lane of the current vehicle, query the second row mapping table for the lateral behavior corresponding to the location type of the adjacent layer sampling point; The first behavior mapping table includes lane keeping, left lane borrowing, right lane borrowing, lane borrowing left keeping, lane borrowing right keeping, lane borrowing left return, and lane borrowing right return. The second behavior mapping table includes lane keeping, left lane change preparation, right lane change preparation, left lane change, and right lane change.

7. A vehicle lateral behavior decision-making device, characterized in that, include: The directed graph acquisition module is used to acquire the current vehicle's position sampling directed graph, wherein the position sampling directed graph consists of a starting layer sampling point, at least one intermediate layer sampling point, and at least one ending layer sampling point, and the connection between adjacent layer sampling points forms a trajectory segment. A dictionary determination module is used to construct a behavior sequence dictionary for each sampling point in the directed graph of the location sampling, based on the lateral behavior of the trajectory segments connecting the sampling point and the next layer sampling point. The behavior sequence dictionary includes at least one behavior sequence and a set of sampling point identifiers corresponding to each behavior sequence. The behavior sequence is composed of the lateral behavior of each trajectory segment in the trajectory starting from the sampling point. The set of sampling point identifiers includes the next layer sampling points that satisfy the behavior sequence with respect to the sampling point. The trajectory search module is used to acquire the behavior sequence to be searched, determine the starting layer sampling point as the query sampling point, query the behavior sequence dictionary of the query sampling point to find the sampling point identifier set corresponding to the behavior sequence to be searched; determine the trajectory segment between the sampling point in the queried sampling point identifier set and the query sampling point as the target output segment; update the behavior sequence to be searched, and take the sampling point in the queried sampling point identifier set as the new query sampling point, return to the step of querying the sampling point identifier set corresponding to the behavior sequence to be searched, until the sampling point in the queried sampling point identifier set is the termination layer sampling point; and generate the current target trajectory corresponding to the behavior sequence to be searched based on all target output segments.

8. An electronic device, characterized in that, The electronic device includes: One or more processors; Storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-6.

Images