Automatic driving long-short time decision data labeling method and device based on conflict area, electronic equipment and storage medium

Abstract

The invention discloses an automatic driving long-short time decision data labeling method and device based on a conflict area, electronic equipment and a storage medium, and the method comprises the steps: in response to own vehicle track data and other vehicle track data, dividing a decision scene to obtain a game type label, and calculating the conflict area information of an own vehicle and other vehicles; marking long and short time decision information according to the game type label and the conflict area information of the own vehicle and the other vehicle; and obtaining a group of decision label data according to the game type label, the conflict area information of the own vehicle and the other vehicle, and the long and short time decision information. According to the decision data labeling method provided by the invention, a richer game scene is provided, so that the labeling data not only labels corresponding decision behaviors, but also labels time margins and space margins of specific robbing and giving.

Description

Technical Field

[0001] The present application relates to the field of autonomous driving technology, and in particular to a method, device, electronic device, and storage medium for labeling long- and short-term decision data for autonomous driving based on a conflict area. Background Art

[0002] Decision-making at intersections is a challenging issue in autonomous driving. This is because there are a variety of traffic participants at intersections, and it is very likely that multiple traffic participants will be competing with the vehicle at the same time, making rule-based intersection decision-making and planning complex and difficult to implement.

[0003] Most decision-making models in related technologies use learning-based methods for intersection traffic flow decisions. However, these methods rely on high-quality training data, while few methods address how to label decision data, especially those that integrate with traditional planning methods. For example, the ST diagrams required for speed planning make it difficult to improve the upper limit of traditional planning while also ensuring the lower limit of learning-based decision-making systems. Summary of the Invention

[0004] The embodiments of the present application provide a method, device, electronic device, and storage medium for labeling long- and short-term decision data for autonomous driving based on a conflict area, so as to provide a data labeling and processing method for a decision model.

[0005] The embodiments of this application adopt the following technical solutions:

[0006] In a first aspect, an embodiment of the present application provides a method for labeling long- and short-term decision data for autonomous driving based on a conflict area, wherein the method includes:

[0007] Based on the trajectory data of the ego vehicle and the trajectory data of other vehicles, the decision-making scenarios are divided, the game type labels are obtained, and the conflict area information between the ego vehicle and other vehicles is calculated;

[0008] Marking long-term and short-term decision information according to the game type label and the conflict area information between the vehicle and the other vehicle; and

[0009] A set of decision label data is obtained according to the game type label, the conflict area information between the vehicle and the other vehicle, and the long- and short-time decision information.

[0010] In a second aspect, an embodiment of the present application further provides a decision model, in which a decision label is obtained by using the long- and short-term decision data labeling method for autonomous driving based on the conflict area described in the first aspect.

[0011] In a third aspect, an embodiment of the present application further provides a decision-making method, wherein the method includes:

[0012] The decision model in the second aspect is adopted, and the input of the decision model includes obstacle decision information of multiple historical frames, current observation information, vehicle path planning information, vehicle current speed and vehicle acceleration information, and the learning target of the decision model is the long-term and short-term decision information at the current moment.

[0013] In a fourth aspect, an embodiment of the present application further provides a device for labeling long- and short-term decision data for autonomous driving based on a conflict area, wherein the device comprises:

[0014] A division and calculation module is used to divide the decision scenarios into game type labels and calculate the conflict area information between the self-vehicle and the other vehicles in response to the trajectory data of the self-vehicle and the trajectory data of the other vehicles;

[0015] a marking module, configured to mark long-term and short-term decision information according to the game type label and the conflict area information between the vehicle and the other vehicle; and

[0016] The decision label module is used to obtain a set of decision label data according to the game type label, the conflict area information between the vehicle and the other vehicle, and the long-term and short-term decision information.

[0017] In a fifth aspect, an embodiment of the present application further provides an electronic device, comprising: a processor; and a memory arranged to store computer-executable instructions, wherein the executable instructions, when executed, enable the processor to perform the above method.

[0018] In a sixth aspect, an embodiment of the present application further provides a computer-readable storage medium, which stores one or more programs. When the one or more programs are executed by an electronic device including multiple applications, the electronic device executes the above method.

[0019] At least one of the above technical solutions adopted in the embodiments of the present application can achieve the following beneficial effects: providing a solution for labeling long- and short-term decision data for autonomous driving based on conflict zones. In response to the trajectory data of the own vehicle and the trajectory data of other vehicles, the solution divides the decision scenarios to obtain game type labels and calculates the conflict zone information between the own vehicle and other vehicles. Then, based on the game type labels and the conflict zone information between the own vehicle and other vehicles, the solution labels the long- and short-term decision information. Finally, based on the game type labels, the conflict zone information between the own vehicle and other vehicles, and the long- and short-term decision information, a set of decision label data is obtained. The obtained decision label data can be used as training data for the subsequent decision model. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0021] Figure 1 This is a schematic diagram of the implementation process of labeling long- and short-term decision data for autonomous driving based on conflict areas in an embodiment of the present application;

[0022] Figure 2 This is a flow chart of a method for labeling long- and short-term decision-making data for autonomous driving based on conflict areas in an embodiment of the present application;

[0023] Figure 3 This is a reverse game type label for a decision scenario based on the long- and short-term decision data of autonomous driving in a conflict area in an embodiment of the present application;

[0024] Figure 4 This is a same-direction interactive game type label for a decision scenario based on the long- and short-term decision data of autonomous driving in a conflict area in an embodiment of the present application;

[0025] Figure 5 This is a same-direction intersection game type label for a decision scenario based on the long- and short-term decision data of autonomous driving in a conflict area in an embodiment of the present application;

[0026] Figure 6 This is a car-following game type label for a decision scenario based on the long- and short-term decision data of autonomous driving in a conflict area in an embodiment of the present application;

[0027] Figure 7 Schematic diagram of the conflict area calculation process in the method for labeling long- and short-term decision data for autonomous driving based on conflict areas in an embodiment of the present application;

[0028] Figure 8 A schematic diagram of a long-term decision in a method for labeling long-term and short-term decision data for autonomous driving based on a conflict area in an embodiment of the present application;

[0029] Figure 9 This is a schematic diagram of short-term decision-making in the method for labeling long- and short-term decision data for autonomous driving based on conflict areas in an embodiment of the present application;

[0030] Figure 10 This is a structural diagram of a device for labeling long- and short-term decision data for autonomous driving based on a conflict area in an embodiment of the present application;

[0031] Figure 11 This is a schematic structural diagram of an electronic device in an embodiment of the present application. DETAILED DESCRIPTION

[0032] To make the purpose, technical solutions, and advantages of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the specific embodiments of this application and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0033] The following describes in detail the technical solutions provided by various embodiments of the present application in conjunction with the accompanying drawings.

[0034] In the related art, (1) there is a lack of generally accepted decision data annotation rules, and insufficient attention is paid to the attributes of different game behaviors. For example, when overtaking vehicle A, the method of overtaking a left-turning vehicle by the self-driving vehicle is different from that of overtaking a right-turning vehicle by the self-driving vehicle. This is mainly because for the right-turning vehicle, the self-driving vehicle needs to leave a certain safety distance for the right-turning vehicle when overtaking.

[0035] In related technologies, (2) decision labels often only mark the corresponding decision behavior. For example, vehicle A gives way and vehicle B cuts in, but the specific time margin and space margin for cutting in and giving way are not marked, which is insufficient for guiding traditional planning methods.

[0036] In related technologies, (3) decision labels often only mark the final decision behavior, such as giving way to vehicle A or overtaking vehicle B. However, when making actual decisions based on prior expert data, the driver may slow down and observe before overtaking vehicle A, and then speed up after discovering that vehicle A has no intention of overtaking. The existing labeling method loses the complex intermediate game process between the vehicle itself and other vehicles, and is difficult to cope with the complex short-term uncertainty of intersections.

[0037] like Figure 1 As shown in the figure, decision data labeling mainly includes: obtaining obstacle (other vehicle) trajectory and ego vehicle trajectory, calculating conflict area, labeling other vehicle game type, long-term decision labeling, and short-term decision labeling.

[0038] The embodiment of the present application provides a method for labeling long-term and short-term decision data of autonomous driving based on conflict areas, such as Figure 2 As shown, a flow chart of a method for labeling long- and short-term decision data for autonomous driving based on a conflict area in an embodiment of the present application is provided. The method includes at least the following steps S210 to S240:

[0039] In step S210 , in response to the trajectory data of the own vehicle and the trajectory data of the other vehicle, a decision scenario is divided, a game type label is obtained, and conflict area information between the own vehicle and the other vehicle is calculated.

[0040] The vehicle's trajectory data and the trajectory data of other vehicles (obstacles) are acquired through relevant sensor information. Specifically, the acquired vehicle trajectory data includes, but is not limited to, positioning information obtained from the vehicle's onboard positioning module, from which the vehicle's lateral position, longitudinal position, heading angle, velocity, acceleration, and timestamp at each moment are extracted, and the vehicle's length and width information is added. It is understood that the vehicle's length and width information can be acquired as a priori information.

[0041] Furthermore, obtaining other vehicle trajectory data includes but is not limited to obstacle data obtained from the vehicle-mounted perception module, namely, including timestamp, lateral position, longitudinal position, heading angle, speed, length and width information.

[0042] Preferably, the self-vehicle data and the other-vehicle data are stored as serial data according to timestamps. Starting from the first timestamp, the trajectory data of the self-vehicle and the other-vehicle for a period of time T is extracted according to the time step. For example, 50 consecutive timestamps (within 5 seconds) are used, and the cumulative displacement S of the self-vehicle at each timestamp is calculated. The trajectories of the self-vehicle and the other-vehicle together constitute a set of annotated data. The first point of the trajectory is the starting point, recorded as the current position, and the last point is the end point, recorded as the end position. If the speed of the other vehicle at the current position is 0, it indicates that the vehicle is stationary and is filtered out. A set of annotated data must include at least one other vehicle, otherwise the set of data is filtered out. In other words, it is necessary to filter out scenes with stationary obstacles or no other vehicles.

[0043] Based on the above trajectory data, decision-making scenarios can be divided to obtain game type labels. That is, the game type labels include but are not limited to dividing the decision-making scenarios into reverse, same-direction interaction, same-direction intersection, and following vehicle scenarios based on the positional relationship between the vehicle and other vehicles to obtain game type labels.

[0044] The above trajectory data can also be used to calculate the conflict area information between the vehicle and the other vehicle, that is, the area where the vehicle conflicts with the other vehicle (the physical position overlaps) can be calculated based on the trajectory of the vehicle and the trajectory of the other vehicle.

[0045] Step S220 , marking long- and short-time decision information according to the game type label and the conflict area information between the vehicle and the other vehicle.

[0046] Decision data is annotated according to the divided game type labels and the conflict area information between the vehicle and the other vehicle, and long-term decision information and short-term decision information are obtained and annotated.

[0047] Step S230 , obtaining a set of decision label data according to the game type label, the conflict area information between the vehicle and the other vehicle, and the long- and short-time decision information.

[0048] The calculated conflict area information, game type label, other vehicle information, and long- and short-time decision information are integrated into a set of decision label data, which are divided into following vehicle decision labels and non-following vehicle decision labels.

[0049] Through the above method, the game type label is obtained and the conflict area information between the self-vehicle and the other vehicle is determined, so as to facilitate the labeling of long-term and short-term decision information, and then obtain a set of decision label data based on the long-term and short-term decision information.

[0050] Through the above method, the ego vehicle decision is labeled based on the conflict area, and the game scene is labeled according to the calculation of the conflict area, which enriches the relevant representation of the game type between the ego vehicle and the obstacle.

[0051] Unlike related technologies, which lack generally accepted decision data labeling rules and pay insufficient attention to the attributes of different game behaviors, the above method labels long- and short-term decision information based on the game type label and the conflict area information between the vehicle and the other vehicle, thereby improving the attention paid to the attributes of different game behaviors.

[0052] Unlike related technologies, where decision labels often only identify the corresponding decision behavior, the above method generates a set of decision label data based on the game type label, the conflict area information between the vehicle and the other vehicle, and the long- and short-term decision information. In addition to the decision behavior, the game type and decision offset are added, and the data is divided into following and non-following decision labels, thus overcoming the shortcomings of only labeling a single decision behavior.

[0053] Unlike related technologies, decision labels often only capture the final decision behavior. This labeling method misses the complex intermediate game between the ego vehicle and other vehicles, making it difficult to address the complex short-term uncertainty at intersections. This method divides the decision process into short-term and long-term decision-making, characterizing the short-term uncertainty and long-term stability of the decision, respectively, and supplementing the description of the intermediate decision process.

[0054] In one embodiment of the present application, the response to the own vehicle trajectory data and the other vehicle trajectory data is used to divide the decision scenarios and obtain the game type label, including: determining the area when the own vehicle and the other vehicle collide based on the vehicle trajectory data and the other vehicle trajectory data, and dividing the decision scenarios into: reverse scenario, same-direction interaction scenario, same-direction intersection scenario and following vehicle scenario based on the positional relationship between the own vehicle and the other vehicle, to obtain the game type label.

[0055] exist Figure 3-Figure 6 In the figure, position A and position C respectively indicate that the self-vehicle and the other vehicle enter the conflict area, and position B and position D respectively indicate that the self-vehicle and the other vehicle exit the conflict area. Figure 3 As shown in , it represents the reverse game type label, that is, the self-car and the other car drive in the opposite direction and interact. Figure 4 As shown in , it represents the same-direction interactive game type label, that is, the self-vehicle and the other vehicle are traveling in the same direction and interacting. Figure 5 As shown, it represents the same-direction intersection game type label, that is, the self-vehicle and the other vehicle are traveling in the same direction and overtaking occurs. Figure 6 As shown, it represents the car-following game type label, that is, the car and the other car are traveling in the same direction and following each other.

[0056] In one embodiment of the present application, the calculation of the conflict area information between the own vehicle and the other vehicle includes: calculating the current position of the own vehicle, the current position of the other vehicle, the terminal position of the own vehicle, and the terminal position of the other vehicle; determining whether there is a conflict between the starting and ending positions of the own vehicle and the other vehicle based on the current position of the own vehicle, the current position of the other vehicle, the terminal position of the own vehicle, and the terminal position of the other vehicle, and calculating whether the other vehicle and the own vehicle are in opposite directions; using position A and position C to represent the entry of the own vehicle and the other vehicle into the conflict area, calculating the first conflict point between the other vehicle and the own vehicle starting from the current position of the other vehicle, taking the moment before this point as position C, recording the position of position C and the displacement C_dis and time C_time traveled by the other vehicle to reach position C, and ignoring the other vehicle if there is no conflict point; calculating the first conflict point between the own vehicle and the other vehicle starting from the current position of the own vehicle, taking the moment before this point as position A, recording the position of position A and the displacement A_dis traveled by the own vehicle to reach position A, and ignoring the other vehicle if there is no conflict point.

[0057] like Figure 7 As shown, we first need to calculate whether there is a collision between our vehicle and the other vehicle at our current position, the other vehicle's current position, our final position, and the final position of the other vehicle. We also need to calculate whether the other vehicle and our vehicle are moving in opposite directions (dividing other vehicles into vehicles moving in the same direction and vehicles moving in the opposite direction). For vehicles moving in the opposite direction, all vehicles moving in the opposite direction are moving in the opposite direction. In other words, by calculating the conflict area, we can determine whether there is any overlap between the trajectory at the beginning and the end.

[0058] Please continue to refer to Figure 7 Then, continue calculating position C. Starting from the other car's current position, calculate the point of first collision between the other car and the ego car. This is the point immediately before that collision, which is position C, at which the other car can just give way to the ego car. Record the position of point C, as well as the displacement C_dis and time C_time it takes the other car to reach point C. If there is no collision, ignore the other car.

[0059] Please continue to refer to Figure 7 Next, calculate point A. Starting from the current position of the ego vehicle, calculate the first collision point between the ego vehicle and the other vehicle. The moment before this point is point A, at which the ego vehicle can just give way to the other vehicle. Record the position of point A and the displacement A_dis that the ego vehicle travels to reach point A. If there is no collision point, ignore the other vehicle.

[0060] In one embodiment of the present application, the method further includes: position B and position D respectively represent the vehicle and the other vehicle driving out of the conflict area, and for the oncoming other vehicle, determine whether a conflict occurs at the current position of the other vehicle or the end position of the vehicle; if a conflict occurs, the vehicle does not meet the overtaking condition and position B does not exist; otherwise, starting from the end point of the vehicle, the first point where a conflict occurs with the oncoming other vehicle is calculated in reverse order, and the next point after this point is used as position B, and record whether point B exists and the displacement B_dis traveled by the vehicle to reach point B.

[0061] like Figure 7 As shown, calculate the oncoming vehicle's position, point B. First, determine whether there is a conflict between the other vehicle's current position or the vehicle's final destination. If so, the vehicle does not meet the overtaking condition and point B does not exist. Otherwise, starting from the vehicle's final destination, calculate the first point where there was a conflict with the oncoming vehicle (i.e., the last point of conflict). The point after this point is point B, at which point the vehicle does not conflict with the other vehicle. Record the existence of point B and the displacement B_dis traveled by the vehicle to reach point B.

[0062] In one embodiment of the present application, the method also includes: for other vehicles traveling in the same direction, starting from the position of the own vehicle point A, calculating frame by frame whether the own vehicle's trajectory conflicts with the other vehicle; if there is no conflict, the point is position B. At this time, the own vehicle overtakes the other vehicle, and the decision scenario is same-direction interaction; if there is a conflict, it is determined whether the own vehicle at this position can drive in front of the other vehicle and meet the safety distance. If so, the point is recorded as position B, and the decision scenario is same-direction intersection; if there is a conflict and it is determined that the own vehicle is in front of the other vehicle and meets the safety distance and the current position of the other vehicle is on the own vehicle's trajectory, then the decision scenario is following the vehicle; if there is a conflict and the safety conditions are not met, the conflict point is on the extension line of the own vehicle's path, and position B is calculated based on the horizontal and vertical safety distances. If the lateral safety of the own vehicle and the other vehicle is met first, the decision scenario is same-direction interaction; otherwise, it is same-direction intersection, and it is recorded whether position B exists and the displacement B_dis traveled by the own vehicle to reach position B.

[0063] Please continue to refer to Figure 7 , and then calculate the position B of the other car in the same direction.

[0064] Starting from point A, the vehicle's trajectory is calculated frame by frame to see if there is a conflict with the other vehicle. If there is no conflict, the point is point B, at which point the vehicle can just overtake the other vehicle, and the scene is a same-direction interaction.

[0065] If there is a conflict, determine whether the ego vehicle is ahead of the other vehicle at this position and meets the safety distance. If so, the ego vehicle can safely overtake the other vehicle. This is recorded as point B, and the scene is a same-direction intersection.

[0066] If the other vehicle's current position is on the vehicle's trajectory, the scenario is car following.

[0067] If there is a conflict and the safety conditions are not met, the conflict point will be on the extension line of the ego vehicle's path. Based on the lateral and longitudinal safety distances, the position B is calculated. If the lateral safety conditions between the ego vehicle and the other vehicle are met first, the scene is a same-direction interaction; otherwise, it is a same-direction intersection. The existence of position B and the displacement B_dis traveled by the ego vehicle to reach position B are recorded.

[0068] In one embodiment of the present application, the method further includes: for the oncoming vehicle, determining whether a conflict occurs at the terminal position of the other vehicle; if a conflict occurs, the other vehicle does not meet the overtaking condition and position D does not exist; otherwise, starting from the terminal of the other vehicle, calculating in reverse order the first point where a conflict occurs with the own vehicle, the next point after the point is position D, the other vehicle does not conflict with the own vehicle, and recording whether position D exists and the displacement D_dis and time D_time traveled by the other vehicle to reach position D.

[0069] like Figure 7 As shown, we then calculate point D of the oncoming vehicle. First, we determine whether there is a conflict at the other vehicle's destination. If so, the other vehicle does not meet the overtaking condition and point D does not exist. Otherwise, we start from the other vehicle's destination and calculate in reverse order the first point where there is a conflict with the ego vehicle. The point after this point is point D, at which point the other vehicle does not conflict with the ego vehicle. We also record whether point D exists, as well as the displacement D_dis and time D_time it takes the other vehicle to reach point D.

[0070] In one embodiment of the present application, the method also includes: for other vehicles traveling in the same direction, starting from the position point C of the other vehicle, calculating frame by frame whether the trajectory of the other vehicle conflicts with the own vehicle; if there is no conflict, the point is the position point D, and the decision scenario is the same-direction interaction; if there is a conflict, it is determined whether the other vehicle is in front of the own vehicle and meets the safety distance; if it meets the requirements, the point is recorded as the position point D, and the decision scenario is the same-direction intersection. If it is determined that the other vehicle can drive in front of the own vehicle and meets the safety distance requirements while meeting the current position of the other vehicle on the trajectory of the own vehicle, the decision scenario is following the vehicle; if there is a conflict and the safety conditions are not met, the conflict point is on the extension line of the other vehicle's path, and the position point D is calculated based on the horizontal and vertical safety distances; if the lateral safety of the own vehicle and the other vehicle is met first, the decision scenario is the same-direction interaction; otherwise, it is the same-direction intersection, and the existence of the position point D and the displacement D_dis and time D_time traveled by the other vehicle to reach the position point D are recorded.

[0071] like Figure 7As shown, when calculating the D point of the other car in the same direction, starting from the C point of the other car, calculate frame by frame whether the trajectory of the other car conflicts with the ego car. If there is no conflict, the point is point D. At this time, the other car can just overtake the ego car, and the scene is the same-direction interaction; if there is a conflict, determine whether the other car is in front of the ego car and meets the safety distance. If it does, the other car can safely overtake the ego car, which is recorded as point D. The scene is the same-direction intersection. If the current position of the other car is on the trajectory of the ego car, the scene is following the car; if there is a conflict and the safety conditions are not met, the conflict point will be on the extension line of the other car's path. Point D is calculated based on the horizontal and vertical safety distances. If the lateral safety of the ego car and the other car is met first, the scene is the same-direction interaction, otherwise it is the same-direction intersection; record whether point D exists and the displacement D_dis and time D_time traveled by the other car to reach point D.

[0072] In one embodiment of the present application, the long-term decision information is marked based on the game type label and the conflict area information between the self-vehicle and other vehicles, including: when the displacement of the self-vehicle at C_time is less than the displacement A_dis of position A, the long-term decision label is yielding, otherwise it is overtaking; the offset label corresponding to the yielding label is the difference between the displacement of the self-vehicle to position A and the displacement at D_time, offset = A_dis–S_D_time, where S_D_time represents the displacement of the self-vehicle at D_time; the offset label corresponding to the overtaking label is the difference between the displacement of the self-vehicle at C_time and the displacement to point B, offset = S_C_time-B_dis, where S_C_time represents the displacement of the self-vehicle at C_time.

[0073] like Figure 8 As shown in , based on the conflict area information and the vehicle trajectory, the long-term decision is marked. Figure 8 In the ST diagram shown, for one scenario, when the ego vehicle's displacement at C_time is less than the displacement at point A, A_dis, the long-term decision label is "yield"; otherwise, it is "cut in." The offset label corresponding to the yield label is the difference between the ego vehicle's displacement to point A and the displacement at D_time, i.e., offset = A_dis – S_D_time, where S_D_time represents the ego vehicle's displacement at D_time.

[0074] In another case, the offset label corresponding to the overtaking label is the difference between the displacement of the ego vehicle at C_time and the displacement to point B, that is, offset = S_C_time - B_dis, where S_C_time represents the displacement of the ego vehicle at C_time.

[0075] The above decision labels not only mark the corresponding decision-making behaviors, but also mark the specific time margin and space margin for cutting in and giving way, thereby facilitating the use of downstream modules.

[0076] In one embodiment of the present application, the short-term decision information is marked based on the game type label and the conflict area information between the self-vehicle and other vehicles, including: setting the short-term decision time T1, and calculating the trajectory of the self-vehicle in the short-term decision-making process according to the self-vehicle maintaining a constant speed after T1; when the displacement of the self-vehicle at C_time in the short-term decision-making process is less than the displacement A_dis of position A, the short-term decision label is yielding, otherwise it is overtaking; the offset label corresponding to the yielding label in the short-term decision-making process is the difference between the displacement of the self-vehicle to position A and the displacement at D_time, offset = A_dis–S_D_time_new, where S_D_time_new represents the displacement of the self-vehicle at D_time in the short-term decision-making process; the offset label corresponding to the overtaking label in the short-term decision-making process is the difference between the displacement of the self-vehicle at C_time and the displacement to position B, offset = S_C_time_new-B_dis, where S_C_time_new represents the displacement of the self-vehicle at C_time in the short-term decision-making process.

[0077] like Figure 9 As described above, the short-term decision selects a short duration, T1, and calculates the displacement as if the vehicle maintains a constant speed after T1 (a tentative behavior of the vehicle attempting to maintain a constant speed). This replaces the displacements covered by D_time and C_time in the long-term decision, respectively, to obtain the short-term decision label and offset label. Because the trajectory after T1 differs from the original trajectory, the relationship between the displacement covered by C_time and the displacement of point A, as well as the offset, changes. This reflects the uncertainty of the decision-making process. The short-term decision supplements the description of the intermediate decision process. When the long-term decision time T is less than the short-term decision time T1 (for example, 2 seconds), the short-term decision is the same as the long-term decision.

[0078] For the following vehicle scenario, the long- and short-term decisions differ slightly from those in other scenarios. The decision labels are both "yield." The short-term decision offset is the difference between the current vehicle distance and the displacement of the ego vehicle at T1 equal to 0.5 seconds, i.e., (offset) = delta_S - S_0.5, where delta_S represents the current distance between the ego vehicle and the other vehicle, and S_0.5 represents the displacement of the ego vehicle in 0.5 seconds. The long-term decision offset is the current vehicle distance and the displacement of the ego vehicle at T1 equal to 1.0 seconds, i.e., (offset) = delta_S - S_1.0, where S_1.0 represents the displacement of the ego vehicle in 1.0 seconds. A positive (offset) offset indicates sufficient safety distance, while a negative (offset) offset indicates a potential collision.

[0079] In one embodiment of the present application, a set of decision label data is obtained based on the game type label, the conflict area information between the vehicle and other vehicles, and the long- and short-time decision information, including: a set of following vehicle decision labels are obtained based on the game type label, the conflict area information between the vehicle and other vehicles, and the long- and short-time decision information: A_dis=0, B_dis=current vehicle distance+longitudinal safety distance, C_dis=0, D_dis=0, C_time=0, D_time=0, non-reverse driving, non-interactive game, intersection game, following vehicle, current x-coordinate of obstacle, current y-coordinate of obstacle, current direction of obstacle, obstacle category, long-time decision type, long-time decision offset, short-time decision type, short-time decision offset.

[0080] The calculated conflict area information, game type label, other vehicle information, and long- and short-time decision information are integrated into a set of decision label data, which are divided into following vehicle decision labels and non-following vehicle decision labels.

[0081] Following decision labels: [A_dis = 0, B_dis = current vehicle distance + longitudinal safety distance, C_dis = 0, D_dis = 0, C_time = 0, D_time = 0, non-reverse driving, non-interactive game, merging game, following vehicle, obstacle current x-coordinate, obstacle current y-coordinate, obstacle current orientation, obstacle category, long-term decision type, long-term decision offset, short-term decision type, short-term decision offset]. These following decision labels are used as training data for the decision model. Using a long- and short-term decision labeling method, the possible degree of trial and error is annotated, characterizing the complex intermediate game process between the ego vehicle and other vehicles.

[0082] In one embodiment of the present application, a set of decision label data is obtained based on the game type label, the conflict area information between the vehicle and other vehicles, and the long- and short-time decision information, including: based on the game type label, the conflict area information between the vehicle and other vehicles, and the long- and short-time decision information, a set of non-following vehicle decision labels are obtained: A_dis, B_dis, C_dis, D_dis, C_time, D_time, reverse driving flag, interactive game flag, intersection game flag, following vehicle flag, current x-coordinate of obstacle, current y-coordinate of obstacle, current direction of obstacle, obstacle category, long-time decision type, long-time decision offset, short-time decision type, short-time decision offset.

[0083] Non-following decision labels: [A_dis, B_dis, C_dis, D_dis, C_time, D_time, reverse driving flag, interactive game flag, merging game flag, following car flag, obstacle current x-coordinate, obstacle current y-coordinate, obstacle current orientation, obstacle category, long-term decision type, long-term decision offset, short-term decision type, short-term decision offset]. These non-following decision labels are used as training data for the decision model. Using a long- and short-term decision labeling method, the possible degree of trial and error is annotated, characterizing the complex intermediate game process between the ego vehicle and other vehicles.

[0084] An embodiment of the present application further provides a decision model, wherein a decision label is obtained by using the method for labeling long- and short-term decision data of autonomous driving based on a conflict area. The method for labeling long- and short-term decision data of autonomous driving based on a conflict area comprises:

[0085] In response to the trajectory data of the ego vehicle and the trajectory data of other vehicles, the decision-making scenarios are divided to obtain the game type labels and the conflict area information between the ego vehicle and other vehicles is calculated;

[0086] Marking long-term and short-term decision information according to the game type label and the conflict area information between the vehicle and the other vehicle; and

[0087] A set of decision label data is obtained according to the game type label, the conflict area information between the vehicle and the other vehicle, and the long- and short-time decision information.

[0088] An embodiment of the present application further provides a decision-making method, wherein the method includes:

[0089] The decision model is adopted, the input of which includes obstacle decision information of multiple historical frames, current observation information, vehicle path planning information, vehicle current speed and vehicle acceleration information, and the learning target of the decision model is the long-term and short-term decision information at the current moment.

[0090] When decision labels are actually used, the neural network model inputs include four frames of historical obstacle decision information and current observation information (excluding long- and short-term decision information, which is used as the true value), ego vehicle routing information, and current ego vehicle speed and acceleration information. Furthermore, the learning target of the neural network model used for decision model training is the current long- and short-term decision information.

[0091] The embodiment of the present application also provides an automatic driving long-term and short-term decision data labeling device 1000 based on the conflict area, such as Figure 10 , a schematic diagram of the structure of a device for labeling long- and short-term decision data of autonomous driving based on a conflict area in an embodiment of the present application is provided. The device 1000 for labeling long- and short-term decision data of autonomous driving based on a conflict area at least includes: a division and calculation module 1010, a labeling module 1020, and a decision labeling module 1030, wherein:

[0092] In one embodiment of the present application, the division and calculation module 1010 is specifically used to: respond to the own vehicle trajectory data and the other vehicle trajectory data, divide the decision scene to obtain the game type label and calculate the conflict area information between the own vehicle and the other vehicle.

[0093] The vehicle's trajectory data and the trajectory data of other vehicles (obstacles) are acquired through relevant sensor information. Specifically, the acquired vehicle trajectory data includes, but is not limited to, positioning information obtained from the vehicle's onboard positioning module, from which the vehicle's lateral position, longitudinal position, heading angle, velocity, acceleration, and timestamp at each moment are extracted, and the vehicle's length and width information is added. It is understood that the vehicle's length and width information can be acquired as a priori information.

[0094] Furthermore, obtaining other vehicle trajectory data includes but is not limited to obstacle data obtained from the vehicle-mounted perception module, namely, including timestamp, lateral position, longitudinal position, heading angle, speed, length and width information.

[0095] Preferably, the self-vehicle data and the other-vehicle data are stored as serial data according to timestamps. Starting from the first timestamp, the trajectory data of the self-vehicle and the other-vehicle for a period of time T is extracted according to the time step. For example, 50 consecutive timestamps (within 5 seconds) are used, and the cumulative displacement S of the self-vehicle at each timestamp is calculated. The trajectories of the self-vehicle and the other-vehicle together constitute a set of annotated data. The first point of the trajectory is the starting point, recorded as the current position, and the last point is the end point, recorded as the end position. If the speed of the other vehicle at the current position is 0, it indicates that the vehicle is stationary and is filtered out. A set of annotated data must include at least one other vehicle, otherwise the set of data is filtered out. In other words, it is necessary to filter out scenes with stationary obstacles or no other vehicles.

[0096] Based on the above trajectory data, decision-making scenarios can be divided to obtain game type labels. That is, the game type labels include but are not limited to dividing the decision-making scenarios into reverse, same-direction interaction, same-direction intersection, and following vehicle scenarios based on the positional relationship between the vehicle and other vehicles to obtain game type labels.

[0097] The above trajectory data can also be used to calculate the conflict area information between the vehicle and the other vehicle, that is, the area where the vehicle conflicts with the other vehicle (the physical position overlaps) can be calculated based on the trajectory of the vehicle and the trajectory of the other vehicle.

[0098] In one embodiment of the present application, the division and calculation module 1010 is specifically used to: mark the long-term and short-term decision information according to the game type label and the conflict area information between the vehicle and the other vehicle.

[0099] Decision data is annotated according to the divided game type labels and the conflict area information between the vehicle and the other vehicle, and long-term decision information and short-term decision information are obtained and annotated.

[0100] In one embodiment of the present application, the division and calculation module 1010 is specifically used to obtain a set of decision label data according to the game type label, the conflict area information between the vehicle and other vehicles, and the long- and short-time decision information.

[0101] The calculated conflict area information, game type label, other vehicle information, and long- and short-time decision information are integrated into a set of decision label data, which are divided into following vehicle decision labels and non-following vehicle decision labels.

[0102] In one embodiment of the present application, the partitioning and calculation module 1010 is specifically configured to:

[0103] Based on the vehicle trajectory data and the other vehicle trajectory data, an area where the own vehicle collides with the other vehicle is determined, and the decision scenarios are divided into reverse scenarios, same-direction interaction scenarios, same-direction intersection scenarios, and following vehicle scenarios according to the positional relationship between the own vehicle and the other vehicle, thereby obtaining the game type label.

[0104] In one embodiment of the present application, the partitioning and calculation module 1010 is specifically configured to:

[0105] Calculate the current position of the vehicle, the current position of the other vehicle, the final position of the vehicle, and the final position of the other vehicle;

[0106] Determine whether there is a conflict between the starting and ending positions of the self-vehicle and the other vehicle, and calculate whether the other vehicle and the self-vehicle are in opposite directions based on the current position of the self-vehicle, the current position of the other vehicle, the end position of the self-vehicle, and the end position of the other vehicle;

[0107] Points A and C represent the points where the ego vehicle and the other vehicle enter the conflict zone, respectively. Calculate the first conflict point between the other vehicle and the ego vehicle starting from the other vehicle's current position. The moment before this point is point C. Record the position of point C, as well as the displacement C_dis and time C_time it takes the other vehicle to reach point C. If there is no conflict point, ignore the other vehicle.

[0108] Starting from the current position of the ego vehicle, calculate the first collision point between the ego vehicle and the other vehicle. The moment before this point is point A. Record the position of point A and the displacement A_dis that the ego vehicle travels to reach point A. If there is no collision point, ignore the other vehicle.

[0109] In one embodiment of the present application, a conflict area calculation module is further included, which is used to:

[0110] Points B and D represent the vehicle and the other vehicle exiting the conflict zone, respectively. For the oncoming vehicle, determine whether there is a conflict at the other vehicle's current position or the vehicle's final position.

[0111] If a conflict occurs, the vehicle does not meet the overtaking condition, and point B does not exist;

[0112] Otherwise, start from the end point of the ego vehicle and calculate the first point where the collision occurs with the oncoming vehicle in reverse order. The next point after this point is point B. Record whether point B exists and the displacement B_dis that the ego vehicle travels to reach point B.

[0113] In one embodiment of the present application, a conflict area calculation module is further included, which is used to:

[0114] For other vehicles traveling in the same direction, the trajectory of the vehicle is calculated frame by frame starting from point A to see if there is a conflict with the other vehicle.

[0115] If there is no conflict, then this point is point B, and the decision scenario is a same-direction interaction;

[0116] If there is a conflict, determine whether the vehicle can drive in front of the other vehicle and meet the safety distance. If so, the point is recorded as point B, and the decision scenario is same-direction intersection;

[0117] If there is a conflict and the ego vehicle is judged to be ahead of the other vehicle and the safe distance is met, and the other vehicle's current position is on the ego vehicle's trajectory, the decision scenario is to follow the vehicle;

[0118] If there is a conflict and the safety conditions are not met, the conflict point is on the extension line of the ego vehicle's path. Point B is calculated based on the lateral and longitudinal safety distances. If the lateral safety conditions between the ego vehicle and the other vehicle are met first, the decision scenario is same-direction interaction.

[0119] Otherwise, it is considered as a same-direction intersection, and the existence of point B and the displacement B_dis traveled by the vehicle to reach point B are recorded.

[0120] In one embodiment of the present application, a conflict area calculation module is further included, which is used to:

[0121] For the oncoming vehicle, determine whether there is a conflict at the vehicle's final destination.

[0122] If a conflict occurs, the other car does not meet the overtaking conditions and point D does not exist;

[0123] Otherwise, start from the other car's end point and calculate the first point where the collision occurs with the self-car in reverse order. The next point is point D. Record whether point D exists and the displacement D_dis and time D_time traveled by the other car to reach point D.

[0124] In one embodiment of the present application, a conflict area calculation module is further included, which is used to:

[0125] For other vehicles traveling in the same direction, the trajectory of the other vehicle is calculated frame by frame starting from point C to see if it conflicts with the vehicle itself.

[0126] If there is no conflict, then the point is point D, and the decision scenario is same-direction interaction;

[0127] If there is a conflict, determine whether the other car is in front of the vehicle and meets the safe distance;

[0128] If it satisfies the requirement, the point is recorded as point D, and the decision scenario is the same-direction intersection.

[0129] If the other vehicle is judged to be able to drive in front of the ego vehicle while meeting the safety distance requirement and the other vehicle's current position is on the ego vehicle's trajectory, the decision scenario is to follow the vehicle.

[0130] If there is a conflict and the safety conditions are not met, the conflict point is on the extension line of the other vehicle's path, and point D is calculated based on the horizontal and vertical safety distances;

[0131] If the lateral safety of the self-vehicle and other vehicles is satisfied first, the decision scenario is same-direction interaction;

[0132] Otherwise, it is a same-direction intersection, and record whether point D exists and the displacement D_dis and time D_time traveled by the other car to reach point D.

[0133] In one embodiment of the present application, the marking module 1020 is further configured to:

[0134] Mark out long-term decision-making information, including:

[0135] When the displacement of the ego vehicle in C_time is less than the displacement of point A A_dis, the long-term decision label is yielding, otherwise it is cutting in;

[0136] The offset label corresponding to the yield label is the difference between the vehicle's displacement from point A to point A and the displacement traveled at D_time: offset = A_dis – S_D_time, where S_D_time represents the displacement traveled by the vehicle at D_time.

[0137] The offset label corresponding to the grabbing label is the difference between the displacement of the vehicle at C_time and the displacement to point B, offset=S_C_time-B_dis, where S_C_time represents the displacement of the vehicle at C_time.

[0138] In one embodiment of the present application, the marking module 1020 is further configured to:

[0139] Mark out short-term decision-making information, including:

[0140] Set the short-term decision time T1 and calculate the trajectory of the ego vehicle during the short-term decision process based on the ego vehicle maintaining a constant speed after T1;

[0141] If the displacement of the ego vehicle in C_time is less than the displacement of point A A_dis during the short-term decision process, the short-term decision label is yielding, otherwise it is overtaking;

[0142] The offset label corresponding to the yield label in the short-term decision process is the difference between the vehicle's displacement from point A to point A and the displacement traveled at D_time, offset = A_dis - S_D_time_new, where S_D_time_new represents the displacement traveled by the vehicle at D_time in the short-term decision process.

[0143] For the short-term decision-making process, the offset label corresponding to the overtaking label is the difference between the displacement of the vehicle at C_time and the displacement to point B, offset = S_C_time_new-B_dis, where S_C_time_new represents the displacement of the vehicle at C_time during the short-term decision-making process.

[0144] In one embodiment of the present application, the decision label module 1030 is further configured to:

[0145] According to the game type label, the conflict area information between the vehicle and other vehicles, and the long- and short-time decision information, a set of following decision labels is obtained: A_dis = 0, B_dis = current vehicle distance + longitudinal safety distance, C_dis = 0, D_dis = 0, C_time = 0, D_time = 0, non-reverse driving, non-interactive game, intersection game, following vehicle, current x-coordinate of obstacle, current y-coordinate of obstacle, current direction of obstacle, obstacle category, long-time decision type, long-time decision offset, short-time decision type, short-time decision offset.

[0146] In one embodiment of the present application, the decision label module 1030 is further configured to:

[0147] According to the game type label, the conflict area information between the vehicle and other vehicles, and the long- and short-time decision information, a set of non-following vehicle decision labels are obtained: A_dis, B_dis, C_dis, D_dis, C_time, D_time, reverse driving flag, interactive game flag, intersection game flag, following vehicle flag, current x-coordinate of the obstacle, current y-coordinate of the obstacle, current direction of the obstacle, obstacle category, long-time decision type, long-time decision offset, short-time decision type, and short-time decision offset.

[0148] It can be understood that the above-mentioned long- and short-time decision data labeling device for autonomous driving based on the conflict area can implement the various steps of the long- and short-time decision data labeling method for autonomous driving based on the conflict area provided in the aforementioned embodiment. The relevant explanations on the long- and short-time decision data labeling method for autonomous driving based on the conflict area are applicable to the long- and short-time decision data labeling device for autonomous driving based on the conflict area, and will not be repeated here.

[0149] Figure 11 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present application. Figure 11 At the hardware level, the electronic device includes a processor and, optionally, an internal bus, a network interface, and memory. The memory may include internal memory, such as high-speed random-access memory (RAM), or non-volatile memory, such as at least one disk drive. Of course, the electronic device may also include other hardware required for its services.

[0150] The processor, network interface, and memory can be interconnected via an internal bus, which can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 11 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0151] The memory is used to store programs. Specifically, the program may include program code, which includes computer operating instructions. The memory may include internal memory and non-volatile memory, and provides instructions and data to the processor.

[0152] The processor reads the corresponding computer program from the non-volatile memory into the internal memory and then runs it, forming a long-term and short-term decision-making data labeling device for autonomous driving based on the conflict area at the logical level. The processor executes the program stored in the memory and is specifically used to perform the following operations:

[0153] In response to the trajectory data of the ego vehicle and the trajectory data of other vehicles, the decision-making scenarios are divided to obtain the game type labels and the conflict area information between the ego vehicle and other vehicles is calculated;

[0154] Marking long-term and short-term decision information according to the game type label and the conflict area information between the vehicle and the other vehicle; and

[0155] A set of decision label data is obtained according to the game type label, the conflict area information between the vehicle and the other vehicle, and the long- and short-time decision information.

[0156] The above application Figure 2 The method performed by the device for labeling long- and short-term decision data for autonomous driving based on conflict zones disclosed in the illustrated embodiment can be applied to or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be completed by hardware integrated logic circuits in the processor or by software instructions. The above processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in conjunction with the embodiments of this application can be directly implemented and executed by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium well-known in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, etc. The storage medium is located in the memory, and the processor reads the information in the memory and, in conjunction with its hardware, completes the steps of the above method.

[0157] The electronic device may also perform Figure 2The method for executing the long-term and short-term decision data labeling device of the automatic driving based on the conflict area is realized in the automatic driving based on the conflict area long-term and short-term decision data labeling device. Figure 2 The functions of the illustrated embodiment will not be described in detail in the embodiments of the present application.

[0158] The embodiment of the present application also provides a computer-readable storage medium, which stores one or more programs, wherein the one or more programs include instructions, which, when executed by an electronic device including multiple application programs, can enable the electronic device to execute Figure 2 The method executed by the device for labeling long and short-term decision data of autonomous driving based on the conflict area in the embodiment shown is specifically used to perform:

[0159] In response to the trajectory data of the ego vehicle and the trajectory data of other vehicles, the decision-making scenarios are divided to obtain the game type labels and the conflict area information between the ego vehicle and other vehicles is calculated;

[0160] Marking long-term and short-term decision information according to the game type label and the conflict area information between the vehicle and the other vehicle; and

[0161] A set of decision label data is obtained according to the game type label, the conflict area information between the vehicle and the other vehicle, and the long- and short-time decision information.

[0162] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0163] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0164] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0165] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0166] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0167] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0168] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0169] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0170] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0171] The foregoing is merely an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A method for labeling long-term and short-term decision data for autonomous driving based on conflict areas, wherein: The method comprises: Based on the trajectory data of the ego vehicle and the trajectory data of other vehicles, the decision-making scenarios are divided, the game type labels are obtained, and the conflict area information between the ego vehicle and other vehicles is calculated; Marking long-term and short-term decision information according to the game type label and the conflict area information between the vehicle and the other vehicle; and A set of decision label data is obtained according to the game type label, the conflict area information between the vehicle and the other vehicle, and the long- and short-time decision information.

2. The method according to claim 1, wherein: The decision-making scenarios are divided in response to the own vehicle trajectory data and the other vehicle trajectory data to obtain the game type labels, including: Based on the own vehicle trajectory data and the other vehicle trajectory data, the area where the own vehicle collides with the other vehicle is determined, and the decision scenarios are divided into: reverse scenario, same-direction interaction scenario, same-direction intersection scenario, and following vehicle scenario according to the positional relationship between the own vehicle and the other vehicle, to obtain the game type label.

3. The method according to claim 2, wherein: The calculation of the conflict area information between the vehicle and other vehicles includes: Calculate the current position of the vehicle, the current position of the other vehicle, the final position of the vehicle, and the final position of the other vehicle; Determine whether there is a conflict between the starting and ending positions of the own vehicle and the other vehicle based on the current position of the own vehicle, the current position of the other vehicle, the end position of the own vehicle, and the end position of the other vehicle, and calculate whether the other vehicle and the own vehicle are in opposite directions; Position A and position C represent the points where the ego vehicle and the other vehicle enter the conflict zone, respectively. Calculate the first conflict point between the other vehicle and the ego vehicle starting from the other vehicle's current position. The moment before this point is position C. Record the position of point C, as well as the displacement C_dis and time C_time it takes the other vehicle to reach point C. If there is no conflict point, ignore the other vehicle. The first collision point between the ego vehicle and the other vehicle is calculated starting from the current position of the ego vehicle. The previous moment is taken as position A. The position of position A and the displacement A_dis traveled by the ego vehicle to reach position A are recorded. If there is no collision point, the other vehicle is ignored.

4. The method according to claim 3, wherein: The method further comprises: Positions B and D represent the vehicle and the other vehicle exiting the conflict zone, respectively. For the oncoming vehicle at position B, determine whether a conflict occurs at the other vehicle's current position or the vehicle's final position. If a conflict occurs, the vehicle does not meet the overtaking condition, and position B does not exist; Otherwise, starting from the end point of the ego vehicle, calculate the first point where the collision occurs with the oncoming vehicle in reverse order. The next point after this point is taken as position B. Record whether position B exists and the displacement B_dis that the ego vehicle travels to reach position B.

5. The method according to claim 4, wherein: The method further comprises: For other vehicles traveling in the same direction, the trajectory of the vehicle is calculated frame by frame starting from the position A of the vehicle to see if there is a conflict with the other vehicle. If there is no conflict, the point is position B, and the decision scenario is same-direction interaction; If there is a conflict, determine whether the vehicle can drive in front of the other vehicle and meet the safety distance requirement. If so, the point is recorded as position B, and the decision scenario is same-direction intersection; If there is a conflict, and the vehicle can be judged to be able to drive in front of the other vehicle while maintaining a safe distance, and the other vehicle's current position is on the vehicle's trajectory, then the decision scenario is to follow the vehicle. If there is a conflict and the safety conditions are not met, the conflict point is on the extension line of the ego vehicle's path. Point B is calculated based on the lateral and longitudinal safety distances. If the lateral safety conditions between the ego vehicle and the other vehicle are met first, the decision scenario is a same-direction interaction; otherwise, it is a same-direction intersection. Record whether position B exists and the displacement B_dis traveled by the vehicle to reach position B.

6. The method of claim 4, wherein: The method further comprises: For the oncoming vehicle, determine whether there is a conflict at the vehicle's final destination. If a conflict occurs, the other car does not meet the overtaking conditions and position D does not exist; Otherwise, start from the other car's end point and calculate the first point where the collision occurs with the self-car in reverse order. The next point is position D. Record whether position D exists and the displacement D_dis and time D_time traveled by the other car to reach position D.

7. The method of claim 6, wherein: The method further comprises: For other vehicles traveling in the same direction, the trajectory of the other vehicle is calculated frame by frame starting from point C to see if it conflicts with the vehicle itself. If there is no conflict, the point is position D, and the decision scenario is same-direction interaction; If there is a conflict, determine whether the other car is in front of the vehicle and meets the safe distance; If it satisfies the requirement, the point is recorded as position D, and the decision scenario is the same-direction intersection. If the other vehicle is judged to be able to drive in front of the ego vehicle while meeting the safety distance requirement, and the other vehicle's current position is on the ego vehicle's trajectory, the decision scenario is to follow the vehicle. If there is a conflict and the safety conditions are not met, the conflict point will be located on the extension line of the other vehicle's path, at position D calculated based on the horizontal and vertical safety distances; If the lateral safety of the self-vehicle and the other vehicle is satisfied first, the decision scenario is same-direction interaction, otherwise it is same-direction intersection; Record whether the position D exists and the displacement D_dis and time D_time that the other car travels to reach the position D.

8. The method of claim 7, wherein: The long-term decision information is marked based on the game type label and the conflict area information between the vehicle and the other vehicle, including: When the displacement of the ego vehicle at C_time is less than the displacement of point A, A_dis, the long-term decision label is yielding, otherwise it is cutting in. The offset label corresponding to the yield label is the difference between the vehicle's displacement to position A and the displacement traveled at D_time: offset = A_dis – S_D_time, where S_D_time represents the displacement traveled by the vehicle at D_time. The offset label corresponding to the grabbing label is the difference between the displacement of the vehicle at C_time and the displacement to position B, offset = S_C_time - B_dis, where S_C_time represents the displacement of the vehicle at C_time.

9. The method of claim 7, wherein: The short-term decision information is marked based on the game type label and the conflict area information between the vehicle and the other vehicle, including: Set the short-term decision time T1 and calculate the trajectory of the ego vehicle during the short-term decision process based on the ego vehicle maintaining a constant speed after T1; If the displacement of the ego vehicle in C_time is less than the displacement of point A A_dis, the short-term decision label is yielding, otherwise it is overtaking. The offset label corresponding to the yield label in the short-term decision process is the difference between the displacement of the ego vehicle to position A and the displacement traveled at D_time, offset = A_dis - S_D_time_new, where S_D_time_new represents the displacement traveled by the ego vehicle at D_time in the short-term decision process; For the short-term decision-making process, the offset label corresponding to the grabbing label is the difference between the displacement of the vehicle at C_time and the displacement to position B, offset = S_C_time_new-B_dis, where S_C_time_new represents the displacement of the vehicle at C_time during the short-term decision-making process.

10. The method of claim 9, wherein: According to the game type label, the conflict area information between the vehicle and the other vehicle, and the long- and short-time decision information, a set of decision label data is obtained, including: According to the game type label, the conflict area information between the vehicle and other vehicles, and the long- and short-time decision information, a set of following decision labels is obtained: A_dis = 0, B_dis = current vehicle distance + longitudinal safety distance, C_dis = 0, D_dis = 0, C_time = 0, D_time = 0, non-reverse driving, non-interactive game, intersection game, following vehicle, current x-coordinate of obstacle, current y-coordinate of obstacle, current direction of obstacle, obstacle category, long-time decision type, long-time decision offset, short-time decision type, short-time decision offset.

11. The method of claim 9, wherein: According to the game type label, the conflict area information between the vehicle and the other vehicle, and the long- and short-time decision information, a set of decision label data is obtained, including: According to the game type label, the conflict area information between the vehicle and other vehicles, and the long- and short-time decision information, a set of non-following vehicle decision labels are obtained: A_dis, B_dis, C_dis, D_dis, C_time, D_time, reverse driving flag, interactive game flag, intersection game flag, following vehicle flag, current x-coordinate of the obstacle, current y-coordinate of the obstacle, current direction of the obstacle, obstacle category, long-time decision type, long-time decision offset, short-time decision type, and short-time decision offset.

12. A decision model, wherein A decision label is obtained by using the method for labeling long- and short-term decision data of autonomous driving based on a conflict area as described in any one of claims 1 to 11.

13. A decision-making method, wherein: The method comprises: The decision model as claimed in claim 12 is adopted, wherein the input of the decision model includes obstacle decision information of multiple historical frames, current observation information, vehicle path planning information, vehicle current speed and vehicle acceleration information, and the learning target of the decision model is the long-term and short-term decision information at the current moment.

14. A device for labeling long-term and short-term decision data for autonomous driving based on conflict areas, wherein: The device comprises: A division and calculation module is used to divide the decision-making scenarios into game type labels and calculate the conflict area information between the self-vehicle and the other-vehicle in response to the trajectory data of the self-vehicle and the trajectory data of the other-vehicle; a marking module, configured to mark long-term and short-term decision information according to the game type label and the conflict area information between the vehicle and the other vehicle; and The decision label module is used to obtain a set of decision label data according to the game type label, the conflict area information between the vehicle and the other vehicle, and the long-term and short-term decision information.

15. An electronic device comprising: processor; as well as A memory arranged to store computer executable instructions, which when executed cause the processor to perform the method of any one of claims 1 to 11.

16. A computer-readable storage medium storing one or more programs, which, when executed by an electronic device comprising a plurality of application programs, causes the electronic device to execute the method according to any one of claims 1 to 11.

Images