Intent recognition methods, apparatus, equipment, media and program products

Abstract

This application provides an intent recognition method, apparatus, device, medium, and program product. It determines a target curb based on the lane where the target vehicle is located, identifies road participants, including those located at the outer edge of the target road, based on the location of the target curb, and predicts the behavioral intent of the target road participants based on the road participant category and the curb category corresponding to the target curb. This yields a target behavioral intent characterizing whether the target road participant will cross the target curb. Thus, by combining curb categories that characterize the traversability of the curb with road participant categories corresponding to different movement patterns, accurate prediction of the target road participant's behavioral intent can be achieved under comprehensive consideration of the actual scenario, avoiding judgment errors. Based on this behavioral intent, vehicle safety can be controlled, effectively reducing collision risks and improving traffic efficiency.

Description

Technical Field

[0001] This application relates to the field of vehicle driving technology, specifically to an intent recognition method, device, equipment, medium, and program product. Background Technology

[0002] As autonomous driving technology moves from theoretical research to large-scale engineering applications, Home-zone Parking Assist (HPA) and City Navigate on Autopilot (CNOA) have become two key functions that determine product competitiveness. Together, they support the automated driving experience of vehicles in complex scenarios such as cities and parks.

[0003] In CNOA applications, vehicles face various scenarios, especially in corner cases—complex driving scenarios such as internal park roads, intersections without lane markings, areas connecting residential entrances to main urban roads, rural roads, and areas around zebra crossings. The behavioral intent of dynamic agents near the curb is difficult to determine. These agents are data models or software objects created based on road participants, including pedestrians, non-motorized vehicles (e.g., two-wheeled vehicles), and motorized vehicles. Since dynamic agents may cross the curb or intrude into the vehicle's current lane, mispredicting their behavioral intent can lead to collisions due to insufficient avoidance or excessive avoidance resulting in reduced traffic efficiency. Therefore, accurately determining the behavioral intent of dynamic agents near the curb is a pressing issue. Summary of the Invention

[0004] Based on the defects and shortcomings of the prior art, this application proposes an intent recognition method, device, equipment, medium and program product, which can accurately predict whether a vehicle has the intention to cross the curb, thereby reducing the risk of collision and ensuring vehicle traffic efficiency.

[0005] According to a first aspect of this application, an intent recognition method is provided, comprising:

[0006] Determine the target curb based on the lane where the target vehicle is located;

[0007] The target road participants are determined based on the location of the target curb, and the target road participants include road participants located outside the target curb.

[0008] Based on the road participant category corresponding to the target road participant and the road edge category corresponding to the target road edge, the behavioral intention of the target road participant is predicted to obtain the target behavioral intention of the target road participant. The target behavioral intention represents whether the target road participant will cross the target road edge.

[0009] According to a second aspect of the embodiments of this application, an intent recognition device is provided, comprising:

[0010] The first determining module is used to determine the target curb based on the lane where the target vehicle is located;

[0011] The second determining module is used to determine the target road participant based on the location of the target curb, wherein the target road participant includes road participants located outside the target curb;

[0012] The prediction module is used to predict the behavioral intention of the target road participant based on the road participant category corresponding to the target road participant and the road edge category corresponding to the target road edge, so as to obtain the target behavioral intention of the target road participant, which represents whether the target road participant will cross the target road edge.

[0013] According to a third aspect of the embodiments of this application, an electronic device is provided, including a memory and a processor;

[0014] The memory is connected to the processor and is used to store programs;

[0015] The processor is used to implement the intent recognition method as described in the first aspect by running a program in the memory.

[0016] According to a fourth aspect of the embodiments of this application, a storage medium is provided, on which a computer program is stored, and when the computer program is run by a processor, it implements the intent recognition method as described in the first aspect.

[0017] According to a fifth aspect of the embodiments of this application, a computer program product is provided, wherein the computer program instructions, when executed by a processor, cause the processor to implement the intent recognition method as described in the first aspect.

[0018] The aforementioned intention recognition methods, devices, equipment, media, and program products can determine the target curb based on the lane where the target vehicle is located. Based on the location of the target curb, road participants, including those located on the outer edge of the target road, are identified as target road participants. The behavioral intention of the target road participants is predicted based on their corresponding road participant category and the curb category, thus obtaining the target behavioral intention representing whether the target road participant will cross the target curb. Compared to judging the behavioral intention solely based on whether a road participant is located inside or outside the curb, this method, which considers both curb category and road participant category, can more accurately determine the behavioral intention of the target road participant in a more realistic scenario. This avoids misjudgments of behavioral intention due to missing curb information or differences in the movement characteristics of different road participants. It facilitates safe driving based on accurately determined road participant behavioral intentions, or allows for vehicle control based on these intentions, effectively reducing collision risks and improving traffic efficiency. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating an intent recognition method according to an embodiment of this application.

[0021] Figure 2 This is a schematic diagram illustrating the positional relationship between a target road participant and a target curb, as provided in an embodiment of this application.

[0022] Figure 3 This is a schematic diagram illustrating different curb categories as provided in an embodiment of this application.

[0023] Figure 4 This is a schematic diagram of an image-based intent recognition process proposed in an embodiment of this application.

[0024] Figure 5 This is a schematic diagram of an intent recognition process proposed in an embodiment of this application.

[0025] Figure 6 This is a schematic diagram of the structure of an intent recognition device proposed in an embodiment of this application.

[0026] Figure 7 This is a schematic diagram of the structure of an electronic device proposed in an embodiment of this application. Detailed Implementation

[0027] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0028] As described in the background section, during autonomous driving, vehicles may encounter complex driving scenarios, such as roads within industrial parks, intersections without lane markings, and areas connecting residential entrances to main urban roads. In these situations, it is difficult to determine whether dynamic or moving road users near the curb have the intention to cross it. Incorrectly judging the intentions of road users may lead to collisions due to insufficient avoidance or excessive avoidance resulting in reduced traffic efficiency.

[0029] Building upon this foundation, the inventors further discovered that curb categories can reflect traversability to a certain extent, while different road participant categories can represent different movement patterns, or in other words, different road participant categories have different movement characteristics. By determining the target curb based on the lane where the target vehicle is located, and by identifying road participants located on the outer edge of the target road based on the location of the target curb, and by predicting the behavioral intentions of the target road participants based on the road participant category corresponding to the target road participant and the curb category corresponding to the target curb, a target behavioral intention that can accurately characterize whether the target road participant will cross the target curb can be obtained. This allows for accurate judgment of the behavioral intentions of the target road participants under a more realistic scenario, thereby avoiding misjudgments of behavioral intentions caused by missing curb information or differences in the movement characteristics of different road participants. This effectively ensures the safety of vehicle driving based on this intention, reduces collision risks, and improves traffic efficiency.

[0030] Based on the above concept, this specification provides an intent recognition method, which will be described exemplarily below with reference to the accompanying drawings.

[0031] Please see Figure 1 In one exemplary embodiment, an intent recognition method is provided, applied to any electronic device located in or outside a vehicle, capable of communicating with the vehicle. For example... Figure 1 As shown, the intent recognition method includes steps S101-S103:

[0032] S101: Determine the target curb based on the lane where the target vehicle is located.

[0033] The target vehicle is any vehicle in autonomous driving mode.

[0034] Specifically, the target curb is determined based on the boundary of the lane where the target vehicle is located.

[0035] The boundary can be, for example, a curb, a fence, and / or a green belt.

[0036] Acquire scene data, or environmental data, outside the target vehicle, and determine the boundary information of the lane where the target vehicle is located based on the scene data, and determine the target road edge based on the boundary information.

[0037] Among them, the scene data outside the target vehicle includes at least the scene data of the preset curb area of ​​the lane where the target vehicle is located.

[0038] Generally, data is collected by sensors from the environment outside the target vehicle, especially the preset curb area of ​​the lane where the target vehicle is located.

[0039] Depending on their location, the sensors used to collect scene data include vehicle-mounted sensors and / or roadside sensors. Vehicle-mounted sensors are those located on vehicles, while roadside sensors are those located along roads.

[0040] Depending on the type of sensor, sensors used to collect scene data include LiDAR, cameras, etc. Correspondingly, scene data, depending on the type of sensor, includes laser point cloud data, image data, etc.

[0041] In addition, the information on the boundary of the lane where the target vehicle is located includes the three-dimensional geometric features and visual features of that boundary.

[0042] For example, the sensors include a lidar and a camera. The point cloud data captured by the lidar is analyzed to determine the three-dimensional geometric features of the boundary of the lane where the target vehicle is located. The image data collected by the camera is semantically segmented and recognized to obtain the visual features of the boundary of the lane where the target vehicle is located.

[0043] For example, three-dimensional geometric features include height, width, and / or continuity, while visual features include color, texture, and / or extension trends.

[0044] That is, the target road edge is determined based on the information of the boundary of the lane where the target vehicle is located, namely the road structure features of the boundary, including three-dimensional geometric features, visual features, etc.

[0045] Furthermore, based on the three-dimensional geometric features of the boundary of the lane where the target vehicle is located, it is determined whether the boundary of the lane where the target vehicle is located is continuous, and the target curb is determined based on whether the boundary of the lane where the target vehicle is located is continuous.

[0046] If the boundary of the lane where the target vehicle is located is continuous, the detected boundary of the lane where the target vehicle is located will be directly identified as the target curb.

[0047] For example, by performing semantic segmentation on the images captured by the camera, the boundary of the lane where the target vehicle is located is identified, and the boundary of the lane where the target vehicle is located is determined as the target road edge. That is, the target road edge is the boundary of the lane where the target vehicle is located that can be identified in the images captured by the camera.

[0048] If the boundary of the lane where the target vehicle is located is discontinuous, a virtual boundary is constructed based on the extension trend of the boundary of the lane where the target vehicle is located to obtain the target road edge.

[0049] The extension trends include the direction of lane line extension and changes in road width.

[0050] For example, in scenarios where there are no physical curbs but the road layout is regular, such as the entrance of a residential area connecting to a main road, the end of a zebra crossing, or a straight section of road within the park, the core feature of the lane boundary is that the boundary is straight. If two discontinuous boundaries are detected, the endpoints connecting the two discontinuous boundaries are used as the core logic. The endpoints at both ends of the boundary are extracted and connected to form a virtual boundary, which is then used as the target curb. If only one side of the boundary is detected (such as a curb on one side of the entrance to the residential area, and no curb or no detection on the other side), the endpoint of the single-sided boundary is used as a reference. A fixed distance is set, and the boundary is extended outward by a fixed distance from the endpoint of the boundary according to the direction of the boundary. The detected boundary and the extended boundary form the required virtual boundary, which is then used as the target curb.

[0051] For example, for scenarios such as crossroads without lane lines, T-junctions, and rural road intersections, based on the intersection topology, such as perpendicular intersections and diagonal intersections, the boundary shape of the lanes matches the traffic area of ​​the intersection. Based on the detected boundary endpoints, a virtual boundary is constructed by combining the intersection topology and traffic rules, and the constructed virtual boundary is used as the target curb.

[0052] S102: Determine target road participants based on the location of the target curb.

[0053] Among them, the target road participants include road participants located outside the target curb.

[0054] The outer side of the target curb refers to the side of the target curb where the target vehicle is not located. Correspondingly, the inner side of the target curb refers to the side of the target curb where the target vehicle is located.

[0055] Specifically, scene data outside the target vehicle is acquired and analyzed to identify road participants in the scene outside the target vehicle. Based on the location of the target curb, road participants located outside the target curb are identified as target road participants.

[0056] S103: Based on the road participant category corresponding to the target road participant and the curb category corresponding to the target curb, predict the travel intention of the target road participant to obtain the target behavioral intention of the target road participant.

[0057] Among them, the target behavioral intention characterizes whether the target road participant will cross the target curb.

[0058] Specifically, the types of road participants include pedestrians, non-motorized vehicles, and motorized vehicles.

[0059] For example, taking bicycles as non-motorized vehicles and cars as motorized vehicles, the positional relationship between the target road user and the target vehicle can be as follows: Figure 2 As shown, the solid white line represents the target road edge, and the dashed white line is the center line of the lane where the target vehicle is located.

[0060] Of course, depending on the specific circumstances, road participant types can be classified more broadly, such as including pedestrians and non-pedestrians, or motor vehicles and non-motor vehicles, or more precisely, such as including pedestrians, bicycles, electric two-wheelers, electric tricycles, and cars. Alternatively, different road participant categories can be derived based on road attributes other than vehicle attributes.

[0061] In the embodiments of this application, the specific implementation of the scheme is described with road participant types including pedestrians, non-motorized vehicles, and motorized vehicles.

[0062] Specifically, curb categories include physical curbs and virtual curbs.

[0063] Similarly, based on different actual situations, curb types can be further subdivided, or classified into different curb categories according to different classification standards.

[0064] This embodiment describes the specific implementation of the solution by classifying road edges into physical road edges and virtual road edges.

[0065] Specifically, by analyzing and processing the scene data, the road participant type corresponding to the target road participant and the curb category corresponding to the target curb are determined.

[0066] For example, pedestrians, non-motorized vehicles and motorized vehicles can be distinguished by a target detection algorithm based on sensor fusion. For details on the specific implementation of this process, please refer to the prior art.

[0067] Subsequently, if the target curb is a physical curb, the target behavior intention is determined based on the road participant type corresponding to the target road participant. If the target curb is a virtual curb, the target behavior intention is directly determined to be to cross the target curb.

[0068] When the target curb corresponds to a physical curb, if the target road participant corresponds to a pedestrian, the target behavioral intention indicates that the target road participant will cross the target curb. If the target road participant corresponds to a motor vehicle or a non-motor vehicle, the target behavioral intention indicates that the target road participant will not cross the target curb.

[0069] For example, different categories of target curbs can be as follows: Figure 3 As shown. Figure 3 In (a), the white solid line represents the target curb that has a physical entity. Figure 3 In (b), one side of the curb on both sides of the vehicle is a white dashed line, indicating that the curb on that side does not have a physical entity, and the other side is a white solid line, indicating that the curb on that side has a physical entity. Figure 3 In (c), one side of the road curb on both sides of the vehicle is a solid white line, representing a curb with a physical entity. The middle part of the other side is a solid white line, and the other part is a black dashed line AB. The black dashed line AB represents a virtual curb without a physical entity, and the solid white line represents a physical curb with a physical entity.

[0070] In this embodiment, the curb category can reflect its traversability to a certain extent, while different road participant categories can represent different movement patterns to a certain extent, or in other words, different road participant categories have different movement characteristics. By determining the target curb based on the lane where the target vehicle is located, and by determining the road participants located on the outer edge of the target road based on the location of the target curb, the behavioral intention of the target road participants can be predicted based on the road participant category corresponding to the target road participant and the curb category corresponding to the target curb. This can accurately represent the target behavioral intention of the target road participant, enabling accurate judgment of the behavioral intention of the target road participant under the premise of being closer to the actual scenario. This avoids misjudgment of behavioral intention caused by missing curb information or differences in the movement characteristics of different road participants, effectively ensuring the safety of vehicle driving based on this intention, reducing collision risk and improving traffic efficiency.

[0071] To improve the accuracy of determining the behavioral intentions of target road participants, in some embodiments, the behavioral intentions of target road participants are predicted based on the road participant category corresponding to the target road participant and the road edge category corresponding to the target road edge. Before obtaining the target behavioral intentions of the target road participant, the road edge category corresponding to the target road edge is determined first.

[0072] Specifically, the curb category corresponding to the target curb is determined based on whether the target curb is a physical entity and the preset curb height.

[0073] The preset curb height is a fixed height, for example, 15cm.

[0074] Based on the above process for determining the target road edge, it can be determined that when judging whether the target road edge is a physical entity, if the target road edge is determined based on a virtual boundary, then the target road edge is not a physical entity, that is, the road edge category corresponding to the target road edge is a virtual road edge. If the target road edge is directly determined based on the detected lane boundary, then the target road edge is determined to be a physical entity.

[0075] After determining that the target curb is a physical entity, the curb category corresponding to the target curb is further determined by combining the preset curb height.

[0076] If the target curb is not a physical entity, the curb category corresponding to the target curb is determined to be a virtual curb. If the target curb is a physical entity, the curb category corresponding to the target curb is determined to be a high physical curb or a low physical curb (or a high-barrier physical curb or a low-barrier physical curb) based on the relationship between the preset curb height and the height of the target curb.

[0077] Specifically, if the target curb is a physical entity and its height is greater than or equal to the preset curb height, then the curb category corresponding to the target curb is determined to be a high physical curb. If the target curb is a physics question and its height is less than the preset curb height, then the curb category corresponding to the target curb is determined to be a low physical curb.

[0078] In this way, since the obstruction effect of curbs on target road participants varies depending on whether the curb has a physical entity and the curb height, the curb category corresponding to the target curb can be accurately determined based on whether the curb has a physical entity and the preset curb height. This allows for precise prediction of the target road participants' intentions, reducing collision risks and improving traffic efficiency.

[0079] Furthermore, since different types of road participants have varying abilities to cross curbs, curbs of different heights offer varying degrees of traversability for each type. In practical implementation, different preset curb heights can be used to determine the curb type for different road participant types.

[0080] Specifically, the preset curb height for pedestrians is higher than that for non-motorized vehicles, and vice versa. Of course, the preset curb height can be adjusted based on different actual needs.

[0081] In this embodiment, such detailed classification can provide a more accurate basis for predicting the behavioral intentions of target road participants by considering the degree of constraint and influence of different types of curbs on the behavior of road participants. This will enable more accurate prediction of the behavioral intentions of target road participants, ensure the safety of vehicle driving based on the intentions, effectively reduce collision risks and improve traffic efficiency.

[0082] To ensure the accuracy of predicting the behavioral intentions of target road participants, in some embodiments, different methods are used to determine the behavioral intentions of target road participants based on the roadside category corresponding to the target roadside. Specifically, when predicting the target road participant's behavioral intention based on the road participant category and the roadside category corresponding to the target roadside, if the roadside category corresponding to the target roadside is a virtual roadside, the behavioral intention is predicted based on the target image. If the roadside category corresponding to the target roadside is a low physical roadside, the behavioral intention is predicted based on the target image and the road participant type corresponding to the target road participant. If the roadside type corresponding to the target roadside is a high physical roadside, the determined behavioral intention indicates that the target road participant will not cross the target roadside.

[0083] The target image includes the target road participant, which is obtained by acquiring images of the target road participant. Specifically, the number of target images is at least one frame.

[0084] The behavioral intentions of participants on the target road are predicted based on at least one frame of the target image to obtain at least one behavioral intention, wherein there is a one-to-one correspondence between the at least one frame of the target image and the at least one behavioral intention. Based on this at least one behavioral intention, the target behavioral intention is determined.

[0085] Specifically, if a preset number of target image frames correspond to behavioral intentions indicating that the target road participant will cross the target curb, then the target behavioral intention is determined to indicate that the target road participant will cross the target curb. Conversely, if the number of behavioral intentions indicating that the target road participant will not cross the target curb in at least one target image frame is less than a preset number, then the target behavioral intention is determined to indicate that the target road participant will not cross the target curb.

[0086] More specifically, if the behavioral intent representation of the target road participant corresponding to a consecutive preset number of target images indicates that the target road participant will cross the target curb, then it is determined that the target behavioral intent representation indicates that the target road participant will cross the target curb.

[0087] The number of participants intending to cross the target road edge can be counted using a counter. Each counter is linked to a specific participant in the target road, with a one-to-one correspondence.

[0088] After detecting a target road participant, an unbound counter is bound to the target road participant. The counter bound to the target road participant is used to count the number of target images in at least one target image containing the target road participant whose corresponding behavioral intent is that the target road participant will cross the target curb.

[0089] Once a target road participant can no longer be detected, the counter is unbound from that target road participant. Simultaneously with unbinding, or when it is necessary to bind the counter to other target road participants, the counter is reset to zero.

[0090] Since the type of high physical curb is determined in combination with the type of target road participant, and based on this, high physical curb is a curb with low traversability, in this case, directly determining the target behavioral intention indicates that the target road participant will not cross the target curb. This can quickly determine the target behavioral intention while ensuring accuracy, so as to meet the autonomous driving needs of the target vehicle.

[0091] In this embodiment, different methods are used to identify the intentions of target road participants based on different curb categories, which can effectively reduce misjudgments, lower the risk of collisions under autonomous driving, and improve traffic speed.

[0092] In order to accurately identify the behavioral intentions of target road participants, the behavioral intentions of target road participants are predicted based on the target image. When the target behavioral intentions are obtained, the target orientation of the target road participants in the target image is identified. Then, the target behavioral intentions are determined based on the size relationship between the target angle and the preset angle.

[0093] The target angle is the angle between the target's orientation and the target's curb.

[0094] Specifically, the angle between the motion orientation vector M and the curb reference vector L, i.e., the target angle θ, is calculated using the vector dot product formula, which is θ = arc cos( ),in, The dot product of the motion orientation vector L and the curb reference vector M, Let θ be the magnitude of the two vectors. The magnitude of θ represents the degree of perpendicularity between the target road participant and the target curb. The larger θ is, the more perpendicular the target road participant's orientation is to the target curb, and the more significant the intention to cross is, meaning the greater the probability that the target road participant will cross the target curb. The smaller θ is, the more parallel the target road participant's orientation is to the target curb, and the more significant the target road participant's intention to move along the curb is, meaning the less likely the target road participant will not cross the target curb.

[0095] Specifically, if the target angle is greater than the preset angle, it is determined that the target behavior intention indicates that the target road participant will cross the target curb; if the target angle is less than or equal to the preset angle, it is determined that the target behavior intention indicates that the target road participant will not cross the target curb.

[0096] For example, an image-based intent recognition process can be as follows: Figure 4 As shown, after detecting a target road participant, a dedicated frame counter is assigned to the target road participant, and the value recorded by the frame counter is initialized to 0, i.e., the count is reset to zero. Then, the orientation of the target road participant is sensed in real time, i.e., orientation information sensing is performed to determine the target angle α. It is then determined whether the target angle α is greater than a preset angle. If the target angle α is greater than the preset angle, it indicates that the target road participant is inclined to cross the target curb under the current circumstances, and the frame counter is updated, i.e., the recorded value of i is updated to i+1. When the count exceeds a threshold, i.e., exceeds a preset number, it is determined that the target road participant has the intention to cross, i.e., the target behavioral intention indicates that the target road participant will cross the target curb. If the count does not exceed the threshold, the orientation information sensing operation is repeated, and the above operation is repeated. Conversely, if the target angle α is less than or equal to the preset angle, the count is reset to zero.

[0097] For example, the preset included angle is 30°.

[0098] Of course, depending on the actual needs, the preset angle can be set to other values. Generally, the longer the virtual curb is, the larger the preset angle should be.

[0099] In this embodiment, the behavioral intention of the road participant is identified by combining the orientation of the road participant. The target behavioral intention is determined by whether the angle between the orientation of the target road participant that can be identified in the image and the target road edge is greater than a preset angle. This method has low computational burden and good stability and real-time performance.

[0100] To ensure the accuracy of predicting the behavioral intentions of target road participants, in some embodiments, the behavioral intentions of target road participants are predicted based on the target image and the corresponding road participant type. When the target behavioral intention is obtained, it is determined in different ways based on the corresponding road participant type. Specifically, if the road participant type is pedestrian, the determined behavioral intention indicates that the target road participant will cross the target curb; if the road participant type is motor vehicle, the determined behavioral intention indicates that the target road participant will not cross the target curb; if the road participant type is non-motor vehicle, the behavioral intention is predicted based on the target image to obtain the target behavioral intention.

[0101] The specific implementation of predicting the behavioral intentions of participants on the target road based on the target image to obtain the target behavioral intentions can be found in the above content and will not be repeated here.

[0102] For example, the overall intent recognition process can be as follows: Figure 5 As shown, based on scene data acquired by sensors, a curb information perception operation is performed to obtain curb information. Curb type analysis is then performed based on this curb information to determine the curb type corresponding to the target curb. Specifically, when the curb lacks a physical entity, the curb type corresponding to the target curb is determined to be a virtual curb. In this case, multi-frame motion orientation verification is performed, i.e., the operation described above, which determines the target's behavioral intent based on the target image, is executed. When the curb possesses a physical entity, further curb type analysis is performed. If the curb type corresponding to the target curb is determined to be a high physical curb, then the target road participant is determined to have no intention to cross it, i.e., the target's behavioral intent indicates that the target road participant will not cross the target curb. If the curb type corresponding to the target curb is determined to be a low physical curb, then the target's behavioral intent is determined based on the road participant type corresponding to the target road participant. More specifically, if the road participant type is a non-motorized vehicle, multi-frame motion orientation verification is performed. If the road participant type is a motorized vehicle, then the target road participant has no intention to cross it. If the road participant type is a pedestrian, then the target road participant has an intention to cross it.

[0103] In this embodiment, different methods can be used to determine the target behavioral intent based on the different types of road participants corresponding to the target road participants. This can achieve targeted prediction of the target behavioral intent and effectively ensure the accuracy of the predicted target behavioral intent.

[0104] Furthermore, since pedestrian behavior is less predictable and road edges are more easily crossed by pedestrians, when the road participant type is pedestrian, the target behavior intention is directly determined as the target road participant will cross the target road edge. Motor vehicle behavior is relatively more predictable, and road edges are less easily crossed by motor vehicles. When the road participant type is motor vehicle, the target behavior intention is directly determined as the target road participant will not cross the target road edge. The crossability of road edges for non-motorized vehicles is difficult to determine. In this case, the target behavior intention is judged by combining the target image. In this way, the behavioral intention of the target road participant can be judged more accurately.

[0105] like Figure 6 As shown in the figure, this application embodiment also provides an intent recognition device, including a first determination module 601, a second determination module 602, and a prediction module 603.

[0106] in,

[0107] The first determining module 601 is used to determine the target curb based on the lane where the target vehicle is located;

[0108] The second determining module 602 is used to determine the target road participant based on the location of the target curb, wherein the target road participant includes road participants located outside the target curb;

[0109] The prediction module 603 is used to predict the behavioral intention of the target road participant based on the road participant category corresponding to the target road participant and the road edge category corresponding to the target road edge, so as to obtain the target behavioral intention of the target road participant, wherein the target behavioral intention represents whether the target road participant will cross the target road edge.

[0110] The intent recognition device provided in this embodiment belongs to the same concept as the intent recognition method provided in the above embodiments of this application. It can execute the intent recognition method provided in any of the above embodiments of this application and has the corresponding functional modules and beneficial effects of executing the method. Technical details not described in detail in this embodiment can be found in the specific processing content of the intent recognition method provided in the above embodiments of this application, and will not be repeated here.

[0111] Another embodiment of this application also provides an electronic device, see [link to relevant documentation] Figure 7 As shown, the electronic device includes a memory 700 and a processor 710.

[0112] The memory 700 is connected to the processor 710 and is used to store programs;

[0113] The processor 710 is configured to implement the intent recognition method disclosed in any of the above embodiments by running the program stored in the memory 700.

[0114] Specifically, the electronic device may also include: a bus, a communication interface 720, an input device 730, and an output device 740.

[0115] The processor 710, memory 700, communication interface 720, input device 730, and output device 740 are interconnected via a bus. Among them:

[0116] A bus can include a pathway for transmitting information between various components of a computer system.

[0117] The processor 710 can be a general-purpose processor, such as a general-purpose central processing unit (CPU), a microprocessor, etc., or an application-specific integrated circuit (ASIC), or one or more integrated circuits used to control the execution of the program of the present application. It can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0118] The processor 710 may include a main processor, as well as a baseband chip, modem, etc.

[0119] The memory 700 stores a program that executes the technical solution of this application, and may also store an operating system and other key business functions. Specifically, the program may include program code, which includes computer operation instructions. More specifically, the memory 700 may include read-only memory (ROM), other types of static storage devices that can store static information and instructions, random access memory (RAM), other types of dynamic storage devices that can store information and instructions, disk storage, flash memory, etc.

[0120] Input device 730 may include a device for receiving data and information input by a user, such as a keyboard, mouse, camera, scanner, light pen, voice input device, touch screen, pedometer, or gravity sensor.

[0121] Output device 740 may include devices that allow information to be output to a user, such as a display screen, printer, speaker, etc.

[0122] The communication interface 720 may include a device that uses any transceiver to communicate with other devices or communication networks, such as Ethernet, Radio Access Network (RAN), Wireless Local Area Network (WLAN), etc.

[0123] The processor 710 executes the program stored in the memory 700 and calls other devices, which can be used to implement the various steps of any of the intent recognition methods provided in the above embodiments of this application.

[0124] Those skilled in the art will understand that Figure 7 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the electronic device to which the present application is applied. The specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0125] This application also proposes a chip including a processor and a data interface. The processor reads and runs a program stored in a memory through the data interface to execute the intent recognition method described in any of the above embodiments. For details of the processing and its beneficial effects, please refer to the above embodiments of the intent recognition method.

[0126] In addition to the methods and apparatus described above, embodiments of this application provide a computer program product comprising computer program instructions that, when executed by a processor, cause the processor to perform the steps of the intent recognition methods according to various embodiments of this application as described in the "Exemplary Methods" section of this specification.

[0127] The computer program product can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of this application. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0128] Furthermore, embodiments of this application also propose a storage medium having a computer program stored thereon, the computer program being executed by a processor of the steps in the intent recognition method according to various embodiments of this application described in the "Exemplary Methods" section above.

[0129] The basic principles of the present invention have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in the present invention are merely examples and not limitations, and should not be considered as essential features of each embodiment of the present invention. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the present invention to the necessity of employing the aforementioned specific details.

[0130] The block diagrams of devices, apparatuses, devices, and systems involved in this invention are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0131] It should also be noted that in the apparatus, device, and method of the present invention, the components or steps can be disassembled and / or recombined. These disassemblies and / or recombinations should be considered as equivalent solutions of the present invention.

[0132] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the invention. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of the invention. Therefore, the invention is not intended to be limited to the aspects shown herein, but rather to be carried out within the widest scope consistent with the principles and novel features disclosed herein.

[0133] It should be understood that the qualifying terms "first", "second", "third", "fourth", "fifth" and "sixth" used in the description of the embodiments of the present invention are only used to more clearly illustrate the technical solutions and are not intended to limit the scope of protection of the present invention.

[0134] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of the invention to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations therein.

Claims

1. An intent recognition method, characterized in that, The method includes: Determine the target curb based on the lane where the target vehicle is located; Target road participants are determined based on the location of the target curb, including road participants located outside the target curb. Based on the road participant category corresponding to the target road participant and the curb category corresponding to the target curb, the behavioral intention of the target road participant is predicted to obtain the target behavioral intention of the target road participant. The road participant category corresponding to the target road participant represents the motion characteristics of the target road participant, the curb category of the target curb represents the traversability of the target curb, and the target behavioral intention represents whether the target road participant will cross the target curb.

2. The intent recognition method according to claim 1, characterized in that, Before predicting the target road participant's behavioral intention based on the road participant category corresponding to the target road participant and the road edge category corresponding to the target road edge, and obtaining the target behavioral intention of the target road participant, the method further includes: Based on whether the target curb is a physical entity and the preset curb height, the curb category corresponding to the target curb is determined.

3. The intent recognition method according to claim 2, characterized in that, The step of determining the curb category corresponding to the target curb based on whether the target curb is a physical entity and a preset curb height includes: If the target curb is not a physical entity, then the curb category corresponding to the target curb is determined to be a virtual curb; If the target curb is a physical entity, then based on the relationship between the preset curb height and the height of the target curb, the curb category corresponding to the target curb is determined to be either a high physical curb or a low physical curb.

4. The intent recognition method according to claim 3, characterized in that, The step of predicting the behavioral intention of the target road participant based on the road participant category corresponding to the target road participant and the road edge category corresponding to the target road edge, to obtain the target behavioral intention of the target road participant, includes: If the curb category corresponding to the target curb is a virtual curb, then the behavioral intention of the target road participant is predicted based on the target image to obtain the target behavioral intention, wherein the target image includes the target road participant; If the curb category corresponding to the target curb is a low physical curb, then based on the target image and the road participant type corresponding to the target road participant, the behavioral intention of the target road participant is predicted to obtain the target behavioral intention; If the curb category corresponding to the target curb is a high physical curb, then the target behavioral intent is determined to represent that the target road participant will not cross the target curb.

5. The intent recognition method according to claim 4, characterized in that, The step of predicting the behavioral intentions of the target road participants based on the target image to obtain the target behavioral intentions includes: Identify the target orientation of the target road participant in the target image; Based on the relationship between the target angle and the preset angle, the target's behavioral intention is determined, where the target angle is the angle between the target's orientation and the target curb.

6. The intent recognition method according to claim 4, characterized in that, The step of predicting the behavioral intent of the target road participant based on the target image and the road participant type corresponding to the target road participant, to obtain the target behavioral intent, includes: If the road participant type corresponding to the target road participant is a pedestrian, then the target behavioral intent is determined to indicate that the target road participant will cross the target curb; If the road participant type corresponding to the target road participant is a motor vehicle, then the target behavioral intent is determined to indicate that the target road participant will not cross the target curb. If the target road participant is a non-motorized vehicle, then the behavioral intention of the target road participant is predicted based on the target image to obtain the target behavioral intention.

7. An intent recognition device, characterized in that, The device includes: The first determining module is used to determine the target curb based on the lane where the target vehicle is located; The second determining module is used to determine the target road participant based on the location of the target curb, wherein the target road participant includes the road participant located outside the target curb; The prediction module is used to predict the behavioral intention of the target road participant based on the road participant category corresponding to the target road participant and the curb category corresponding to the target curb, thereby obtaining the target behavioral intention of the target road participant. The road participant category corresponding to the target road participant represents the motion characteristics of the target road participant, the curb category of the target curb represents the traversability of the target curb, and the target behavioral intention represents whether the target road participant will cross the target curb.

8. An electronic device, characterized in that, Including memory and processor; The memory is connected to the processor and is used to store programs; The processor is configured to implement the intent recognition method as described in any one of claims 1 to 6 by running a program in the memory.

9. A storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the intent recognition method as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, It includes computer program instructions that, when executed by a processor, cause the processor to implement the intent recognition method as described in any one of claims 1-6.

Images