Cutin behavior recognition method and device, equipment and storage medium

Abstract

The invention discloses a cutin behavior recognition method and device, equipment and a storage medium, relates to the technical field of behavior recognition of vehicles, and discloses a cutin behavior recognition method which comprises the steps that effective fragments in a graph-free LCC scene are screened from acquired historical driving data; screening a to-be-cut-in vehicle to be cut in the own vehicle lane from the current frame of the effective fragment; based on a preset global map, a bound lane where the to-be-cut-in vehicle is located in the reference frame of the effective fragment is recognized, and the preset global map comprises lanes where the vehicles are located at different frame moments; and based on the bound lanes where the to-be-cut-in vehicle is located in the different reference frames, judging whether the to-be-cut-in vehicle has a cut-in behavior or not. That is, the global map is utilized to accurately determine the lanes where the to-be-cut-in vehicle is located in different frame data, and then through the lanes where the to-be-cut-in vehicle is located at different moments, whether the vehicle in the historical driving data has the cutin behavior or not is accurately identified.

Description

Technical Field

[0001] The present application relates to the technical field of vehicle behavior recognition, and in particular to a cutin behavior recognition method, apparatus, device, and storage medium. Background Art

[0002] Cutin behavior is the behavior of a vehicle cutting into its own lane. Since vehicle cutin behavior can be dangerous, it is necessary to accurately identify cutin behavior from the vehicle's historical driving data to improve vehicle optimization.

[0003] Currently, to accurately identify cutin behavior from historical driving data, high-precision images of surrounding vehicles are typically obtained from the data. These images are then used to extract information such as the position, speed, and acceleration of surrounding vehicles. Algorithms are then used to determine whether surrounding vehicles have engaged in cutin behavior based on this information. However, in map-free LCC scenarios, each frame of map information in the historical driving data is generated in real time by perception. The same lane centerline appears differently in each frame, making it difficult to accurately identify vehicle cutin behavior in map-free LCC scenarios.

[0004] The above content is only used to assist in understanding the technical solution of this application and does not constitute an admission that the above content is prior art. Summary of the Invention

[0005] The main purpose of this application is to provide a cutin behavior recognition method, which aims to solve the technical problem of being unable to recognize the cutin behavior of a vehicle in a non-image LCC scenario.

[0006] To achieve the above objectives, the present application proposes a cutin behavior recognition method, which includes: Filter valid segments in the LCC scenario without image from the acquired historical driving data; Filtering a vehicle to be cut into the own vehicle lane from the current frame of the valid segment; Identifying, based on a preset global map, a bound lane in which the vehicle to be cut into is located in a reference frame of the valid segment, wherein the preset global map includes lanes in which each vehicle is located at different frame times; Based on the bound lanes where the vehicle to be cut in is located in different reference frames, it is determined whether the vehicle to be cut in has a cutin behavior.

[0007] In one embodiment, the step of identifying, based on a preset global map, a bound lane where the vehicle to be cut in is located in a reference frame of the valid segment includes: Searching for the most recent valid frame before the current frame from the valid fragments; Based on the preset global map, obtaining the first bound lane where the vehicle to be cut-in is located in the current frame and the second bound lane where the vehicle is located in the most recent valid frame; determining whether the first bound lane and the second bound lane are the same lane; If the first bound lane and the second bound lane are not the same lane, based on the preset global map, the bound lanes in each reference frame after the current frame in the valid segment are obtained, and the bound lanes in each reference frame before the most recent valid frame in the valid segment are obtained.

[0008] In one embodiment, the step of determining whether the vehicle to be cut-in has cut-in behavior based on the bound lanes in which the vehicle to be cut-in is located in different reference frames includes: Based on the bound lanes of the to-be-cut-in vehicle in different reference frames, counting a first proportion of all frames in which the bound lane in reference frames subsequent to the current frame is the first bound lane; and, a second proportion of all frames in which the bound lane is the second bound lane in the reference frame before the most recent valid frame; Determining whether the first proportion and the second proportion both meet a preset proportion; If the preset proportions are all met, it is determined that the vehicle to be cut in has performed a cutin behavior.

[0009] In one embodiment, the step of screening the vehicle to be cut into the lane of the vehicle from the current frame of the valid segment includes: Extracting cutin candidate vehicles from the current frame of the valid segment; Obtaining the azimuth angle of the ego vehicle in the current frame and the position coordinates of the ego vehicle in the current frame mapped in the global map; Determining the relative lateral displacement and relative longitudinal displacement between the current cutin candidate vehicle and the ego vehicle based on the position information of the current cutin candidate vehicle and the position coordinates, wherein the current cutin candidate vehicle is the cutin candidate vehicle currently located; Determining whether the relative longitudinal displacement is less than the currently recorded minimum longitudinal displacement, and determining whether the relative lateral displacement meets a preset threshold; If the relative lateral displacement is less than the minimum longitudinal displacement and the relative longitudinal displacement meets the preset threshold, the current cutin candidate vehicle is used to update the currently recorded vehicle to be cut in, until all cutin candidate vehicles are judged, and the cutin candidate vehicle with the smallest relative lateral displacement is determined as the vehicle to be cut in.

[0010] In one embodiment, before the step of screening the vehicle to be cut into the lane of the own vehicle from the current frame of the valid segment, the method further includes: Mapping the coordinates of the lane centerline on each frame of data to a global grid to obtain an initial global map, wherein each frame of data constitutes the valid segment; Merging overlapping lanes filtered from the initial global map to obtain a global grid map; Mapping each object in the valid segment to the global grid map, and obtaining the index position of each object in each frame of data in the global grid map; Based on the index position, each object in each frame of data is bound to the global lane centerline in the global grid map to obtain a preset global map.

[0011] In one embodiment, the step of binding each object in each frame of data to a global lane centerline in the global grid map based on the index position to obtain a preset global map includes: Determining a target lane corresponding to the object in the global grid map based on the index position; Determine the number of objects in the target lane in each frame of data; If the number of objects is not one, determining a projection distance of the object on the global lane centerline of the target lane; Bind the objects in each frame of data to the global lane centerline and projection distance of the corresponding target lane to obtain a preset global map.

[0012] In one embodiment, the step of merging overlapping lanes screened from the initial global map to obtain a global grid map includes: screening overlapping lanes having overlapping areas from the initial global map; Merging the overlapping lanes to obtain a global lane centerline of each lane in the valid segment in the initial global map; Performing at least one thinning process on the centerline coordinates of the global lane centerline to obtain thinned centerline coordinates; Based on the thinned centerline coordinates, the global lane centerlines are gridded to obtain a global grid map.

[0013] In addition, to achieve the above-mentioned purpose, the present application also proposes a cutin behavior recognition device, which includes: The acquisition module is used to filter valid segments in the LCC scenario without image from the acquired historical driving data; A screening module, configured to screen a vehicle to be cut into the own vehicle lane from a current frame of the valid segment; an identification module, configured to identify, based on a preset global map, a bound lane in which the vehicle to be cut into is located in a reference frame of the valid segment, wherein the preset global map includes lanes in which each vehicle is located at different frame times; The judgment module is used to judge whether the vehicle to be cut in has a cutin behavior based on the bound lanes where the vehicle to be cut in is located in different reference frames.

[0014] In addition, to achieve the above-mentioned purpose, the present application also proposes a cutin behavior recognition device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the cutin behavior recognition method described above.

[0015] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and stores a computer program on the storage medium. When the computer program is executed by the processor, the steps of the cutin behavior recognition method described above are implemented.

[0016] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the steps of the cutin behavior recognition method as described above.

[0017] One or more technical solutions proposed in this application have at least the following technical effects: Because historical driving data contains a large number of HD-formatted frame segments, it is necessary to first filter valid segments from the historical driving data for map-free LCC scenarios. Furthermore, since multiple vehicles may be near the ego vehicle, it is necessary to filter the vehicle that is about to cut into the ego vehicle's lane from the current frame of the valid segment. A pre-created global map (preset global map) is then used to determine the bound lanes in which the vehicle is located across different frames. Because the global map maps the lane information for each vehicle at different frame times, the global map can be used to accurately determine the lane in which the vehicle is located across different frames. The lanes in which the vehicle is located at different times can then be used to accurately identify whether a vehicle in the historical driving data has cut in. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0019] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0020] Figure 1 A flowchart of the first embodiment of the cutin behavior recognition method of this application is provided; Figure 2 A flowchart of the second embodiment of the cutin behavior recognition method provided in this application; Figure 3 A flowchart for implementing the third cutin behavior recognition method of this application is provided; Figure 4 This is a schematic diagram of the module structure of the cutin behavior recognition device according to an embodiment of the present application; Figure 5 This is a schematic diagram of the device structure of the hardware operating environment involved in the cutin behavior recognition method in the embodiment of the present application.

[0021] The realization of the objectives, functional features and advantages of this application will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0022] It should be understood that the specific embodiments described herein are merely used to explain the technical solutions of the present application and are not intended to limit the present application.

[0023] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0024] The main solution of the embodiment of the present application is: the data analysis device filters the valid segments in the map-free LCC scenario from the acquired historical driving data; filters the vehicle to be cut into the own vehicle lane from the current frame of the valid segment; based on the preset global map, identifies the bound lane where the vehicle to be cut into is located in the reference frame of the valid segment, and the preset global map includes the lanes where each vehicle is located at different frame times; based on the bound lane where the vehicle to be cut into is located in different reference frames, determines whether the vehicle to be cut into has performed a cutin behavior.

[0025] In this embodiment, for ease of description, the following description is made with the data analysis device as the execution entity.

[0026] Currently, to accurately identify cutin behavior from historical driving data, high-precision images of surrounding vehicles are typically obtained from the data. Information such as the position, speed, and acceleration of surrounding vehicles is extracted from these high-precision images. Algorithms are then used to determine whether surrounding vehicles have engaged in cutin behavior based on this information. However, in map-free LCC scenarios, each frame of map information in the historical driving data is generated in real time by perception. The same lane centerline appears differently in each frame, making it impossible to accurately identify vehicle cutin behavior in map-free LCC scenarios.

[0027] The present application provides a solution. Since there are a large number of HD-format frame segments in the historical driving data, it is necessary to first screen the valid segments in the non-mapped LCC scenario from the historical driving data. And since there may be multiple vehicles near the vehicle, it is necessary to screen the vehicle to be cut into the lane of the vehicle from the current frame of the valid segment, and then use the pre-created global map (preset global map) to determine the bound lane where the vehicle to be cut into is located in different frame data. Since the global map maps the lane information of each vehicle at different frame times, the global map can be used to accurately determine the lane where the vehicle to be cut into is located in different frame data, and then accurately identify whether the vehicle in the historical driving data has cutin behavior by the lane where the vehicle to be cut into is located at different times.

[0028] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, mobile phone, etc., or an electronic device or data analysis device capable of performing the above functions. The following uses a data analysis device as an example to illustrate this embodiment and the following embodiments.

[0029] Based on this, the present application embodiment provides a cutin behavior recognition method, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the cutin behavior recognition method of this application.

[0030] In this embodiment, the cutin behavior recognition method includes steps S10 to S40: Step S10, filtering valid segments in the non-map LCC scenario from the acquired historical driving data; It should be noted that historical driving data is a fragment of environmental information about the vehicle's surroundings obtained from the vehicle's previously recorded driving data. This historical driving data consists of multiple frames. Because each frame of data is generated in real time by perception in image-free LCC scenarios, even the same lane centerline will appear differently in each frame.

[0031] It is understandable that due to the different formats of HD images and each frame of data in the valid segment, there will be differences in the information obtained for each vehicle. Mapping the image data in both formats to the global grid requires a complex conversion process, which will reduce the efficiency of cutin behavior recognition. Therefore, it is necessary to avoid using HD images when searching for the nearest valid frame.

[0032] Step S20, screening the vehicle to be cut into the own vehicle lane from the current frame of the valid segment; It should be noted that the current frame is the frame data currently being processed within the valid segment. The ego vehicle lane is the lane the ego vehicle is currently traveling in. The pending vehicle is a vehicle that needs to change lanes into the ego vehicle lane and is adjacent to the ego vehicle. The ego vehicle is the vehicle providing historical driving data.

[0033] It is understandable that since there is more than one vehicle that needs to change lanes to the own lane in the current frame, it is necessary to screen out the vehicle that is most likely to cut into the own lane from the vehicles in the current frame to avoid cutting-in confirmation detection for too many vehicles, thereby improving the efficiency of cutin behavior recognition.

[0034] Furthermore, step S20 further includes: Extracting cutin candidate vehicles from the current frame of the valid segment; Obtaining the azimuth angle of the ego vehicle in the current frame and the position coordinates of the ego vehicle in the current frame mapped in the global map; Determining the relative lateral displacement and relative longitudinal displacement between the current cutin candidate vehicle and the ego vehicle based on the position information of the current cutin candidate vehicle and the position coordinates, wherein the current cutin candidate vehicle is the cutin candidate vehicle currently located; Determining whether the relative longitudinal displacement is less than the currently recorded minimum longitudinal displacement, and determining whether the relative lateral displacement meets a preset threshold; If the relative lateral displacement is less than the minimum longitudinal displacement and the relative longitudinal displacement meets the preset threshold, the current cutin candidate vehicle is used to update the currently recorded vehicle to be cut in, until all cutin candidate vehicles are judged, and the cutin candidate vehicle with the smallest relative lateral displacement is determined as the vehicle to be cut in.

[0035] It should be noted that a cutin candidate vehicle is a vehicle that could potentially cut into the ego vehicle's lane from the current lane or another lane in the current frame. The azimuth angle is the orientation angle of the ego vehicle in the current frame. This angle can be expressed relative to a fixed direction (such as true north) based on the ego vehicle's forward direction. The position coordinates are the ego vehicle's current position as mapped in the global map, expressed in two-dimensional or three-dimensional coordinates, reflecting the ego vehicle's specific position in space. The position information is the current position of the current cutin candidate vehicle as mapped in the global map. The relative lateral displacement is the lateral distance between the current cutin candidate vehicle and the ego vehicle, reflecting the candidate vehicle's relative lateral position in the ego vehicle's direction. The relative longitudinal displacement is the longitudinal distance between the current cutin candidate vehicle and the ego vehicle, reflecting the candidate vehicle's relative longitudinal position in the ego vehicle's direction. The minimum longitudinal displacement is the minimum longitudinal displacement relative to the ego vehicle among all cutin candidate vehicles. The preset threshold is the maximum lateral distance between the current cutin candidate vehicle and the ego vehicle when cutting into the ego vehicle's lane. This is the maximum lateral distance used to determine whether the current cutin candidate vehicle has cut into the ego vehicle's lane.

[0036] It can be understood that by accurately calculating the relative lateral and longitudinal displacements between vehicles, the vehicle most likely to cut into the own vehicle's lane can be identified more accurately. Since the relative positions and dynamic changes between vehicles are taken into account, the accuracy of cutin behavior recognition is improved.

[0037] In the specific implementation, the cutin candidate vehicle index cutin_candidate in the current frame is calculated. By comparing the position information of each cutin candidate vehicle, it is determined which cutin candidate vehicle is most likely to complete the cut-in action. Specifically, the temporary minimum value of the longitudinal displacement of the object in the ego vehicle coordinate system (local_x_temp) can be defined and initialized to a large value of 10000; a threshold value (local_y_threshold) for judging whether the object is located inside the lane is defined and set to 2.8 meters; the current cutin candidate vehicle object index (cutin_candidate) that is most likely to cut into the lane is defined and initialized to -1; after completing the above preparations, the ego vehicle (ego) azimuth angle ego_theta of the current frame, the lane index ego_laneid in the global map of the bound scene, and the position coordinates ego_x, ego_y are obtained; the object index list obj_id_list of the current frame is traversed to determine the position of each obj_id_ Is obj_id in the list equal to 0? Since 0 is the index of the ego vehicle, if obj_id is equal to 0, skip the current object. When obj_id is not equal to 0, get the bound lane obj_laneid of the current cutin candidate vehicle obj_id in the global map from the global map. If obj_laneid is not equal to ego_laneid, it is determined that the two are not in the same lane, so the current cutin candidate vehicle needs to be skipped. Get the next cutin candidate vehicle as the current vehicle. If obj_laneid is equal to ego_laneid, get the position coordinates obj_x and obj_y of the current object, and further calculate the position (local_x, local_y) of the current cutin candidate vehicle obj_id relative to the ego vehicle in the current frame, where curr_x is the longitudinal displacement and curr_y is the lateral displacement. The specific calculation process is as follows: x_diff=global_x-ego_x y_diff=global_y-ego_y angle_cos = math.cos(ego_theta) angle_sin=math.sin(ego_theta) local_x=x_diff*angle_cos-y_diff*angle_sin local_y=x_diff*angle_sin+y_diff*angle_cos.

[0038] After determining the position of the current cutin candidate relative to the ego vehicle (relative lateral displacement (local_y) and relative longitudinal displacement (local_x)), the cutin candidate's potential for cutting in is further determined. Specifically, local_x is checked to see if it is greater than 0 and less than the currently recorded minimum longitudinal displacement (local_x_temp). Furthermore, the absolute value of local_y is checked to see if it is less than the set threshold (local_y_threshold). If both conditions are met, the current cutin candidate is considered closer to completing the cutin than the previous candidate. Therefore, cutin_candidate is updated to the current cutin candidate's index (obj_id) and local_x_temp is updated to the current value of local_x.

[0039] Step S30, identifying the bound lane where the vehicle to be cut in is located in the reference frame of the valid segment based on a preset global map, wherein the preset global map includes the lanes where each vehicle is located at different frame times; It should be noted that the preset global map is derived by mapping the centerline coordinates of each lane and the positions of each object in the active segment onto the global grid. Therefore, the preset global map includes the lanes occupied by each vehicle at different frame times. The reference frame is used to determine whether the vehicle to be cut-in will initiate a cut-in. Bound lanes are lanes mapped to the global grid and bound to the corresponding vehicle.

[0040] It is understandable that in map-free LCC scenarios, sensors can usually only perceive the current behavior information of surrounding vehicles. The behavior information around the vehicle in the valid frame segment is the behavior information perceived by the vehicle at that moment. Therefore, it is necessary to create a global map that can accurately represent the position and lane of each vehicle. Using the global map and detailed valid segments, the bound lane of the vehicle to be cut in can be accurately identified, thereby improving the accuracy of cutin behavior recognition.

[0041] Step S40: determining whether the vehicle to be cut-in occurs a cut-in behavior based on the bound lanes where the vehicle to be cut-in is located in different reference frames.

[0042] It can be understood that by using the bound lanes of the vehicle to be cut into in different reference frames, it is possible to determine whether the vehicle to be cut into changes lanes to the own vehicle lane in the valid segment, and then accurately determine whether the vehicle to be cut into has performed a cutin behavior, so that the cutin behavior can still be accurately identified in the valid segment in the non-image LCC scenario.

[0043] This embodiment provides a method for identifying cutin behavior. Because historical driving data contains a large number of HD-formatted frame segments, it is necessary to first screen valid segments from the historical driving data for non-mapped LCC scenarios. Furthermore, because multiple vehicles may be near the ego vehicle, it is necessary to screen the vehicle to be cut into the ego vehicle's lane from the current frame of the valid segment. A pre-created global map (preset global map) is then used to determine the bound lanes in which the vehicle to be cut in each frame. Because the global map maps the lane information for each vehicle at each frame moment, the global map can be used to accurately determine the lane in which the vehicle to be cut in each frame moment is located. The lanes in which the vehicle to be cut in each frame moment can then be used to accurately identify whether the vehicle in the historical driving data has engaged in a cutin behavior.

[0044] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above embodiment 1 can be referred to the above introduction and will not be described in detail later. Figure 2 , step S30 further includes steps S31 to S34: Step S31, searching the valid segments for the most recent valid frame before the current frame; Step S32: Based on the preset global map, obtaining the first bound lane where the vehicle to be cut-in is located in the current frame and the second bound lane where the vehicle is located in the most recent valid frame; Step S33: determining whether the first bound lane and the second bound lane are the same lane; In step S34, if the first bound lane and the second bound lane are not the same lane, based on the preset global map, the bound lanes in each reference frame after the current frame in the valid segment are obtained, and the bound lanes in each reference frame before the most recent valid frame in the valid segment are obtained.

[0045] It is understandable that since the vehicle may make collection errors when collecting information about surrounding vehicles, and the vehicle may also give up cutting in during the cutting-in process, a small amount of frame data in the valid segment will be in a lane different from that of the vehicle to be cut in. In order to avoid misidentifying this situation as a cutin behavior of the vehicle to be cut in, when determining that the first bound lane of the vehicle to be cut in the current frame and the second bound lane in the valid frame are different lanes, it is necessary to further determine whether the bound lane of the vehicle to be cut in in each reference frame after the current frame is the first bound lane, and whether the bound lane of the vehicle to be cut in in each reference frame before the most recent valid frame is the second bound lane, so as to improve the accuracy of identifying the cutin behavior.

[0046] Furthermore, step S40 further includes: Based on the bound lanes of the to-be-cut-in vehicle in different reference frames, counting a first proportion of all frames in which the bound lane in reference frames subsequent to the current frame is the first bound lane; and, a second proportion of all frames in which the bound lane is the second bound lane in the reference frame before the most recent valid frame; Determining whether the first proportion and the second proportion both meet a preset proportion; If the preset proportions are all met, it is determined that the vehicle to be cut in has performed a cutin behavior.

[0047] It should be noted that the first proportion is the proportion of frame data corresponding to the first bound lane in the reference frames after the current frame. For example, if the reference frame is 15 frames after the current frame, and the number of frames in which the lane to be cut into is the first bound lane in these 15 frames is 14, then the first proportion is 94%. The second proportion is the proportion of frame data corresponding to the second bound lane in the reference frame before the most recent valid frame. For example, if the reference frame is 15 frames after the current frame, and the number of frames in which the lane to be cut into is the first bound lane in these 15 frames is 100%. The preset proportion is the threshold for determining whether the lane to be cut into in the reference frame is the first bound lane or the second bound lane when the vehicle to be cut into performs a cutin behavior.

[0048] It is understandable that in order to accurately identify that the vehicle to be cut in will perform the cutin behavior and will not exit the cutin behavior when preparing to cut in or during the cutin process, therefore, after obtaining the first proportion of all frames in the reference frame after the current frame in which the bound lane is the first bound lane, and the second proportion of all frames in the reference frame before the most recent valid frame in which the bound lane is the second bound lane, it is necessary to further determine whether the first proportion and the second proportion both meet the preset proportion. When the first proportion and the second proportion both meet the preset proportion, it is determined that the vehicle to be cut in will perform the cutin behavior, so as to improve the accuracy of cutin behavior recognition.

[0049] In the specific implementation, it is determined whether the most likely cutin candidate object cutin_candidate in the most recent valid frame is true and a cutin behavior occurs, that is, it is determined whether the current cutin candidate vehicle index obj_id is in the current cutin candidate vehicle index list obj_id_list_pre of the most recent valid previous frame. If not, this loop is skipped; the information obstacle_item of the current cutin candidate vehicle index obj_id in the most recent current frame is obtained, and based on the global map, the bound lane obj_laneid of the vehicle to be cut in obj_id in the global map is obtained; and the obj_laneid of the lane obj_id in the global map is obtained. Obtain the object information (obstacle_item_pre) for the object obj_id in the most recent valid previous frame, and obtain the lane (obj_laneid_pre) bound to the current object obj_id in the global map at that frame. If obj_laneid == -1 or obj_laneid == obj_laneid_pre, skip this loop. If obj_laneid and obj_laneid_pre differ, calculate the number of frames (count_pre) in which the object was bound to the lane index obj_laneid_pre in the previous frame and the frames before it (usually 15 frames). Calculate the number of frames (count_next) in which the object was bound to the lane index obj_laneid in the current frame and the frames after it (usually 15 frames). If both count_pre and count_next are greater than 9 (a preset threshold (90%)), a lane change is considered to have occurred. If utin_candidate == obj_id, the vehicle to be cut in has performed a cutin.

[0050] Based on the first and second embodiments of the present application, in the third embodiment of the present application, the same or similar contents as those in the above embodiments can be referred to the above introduction and will not be described in detail later. Figure 3 Before step S20, the cutin behavior recognition method further includes steps S01 to S04: Step S01, mapping the coordinates of the lane centerline on each frame data to a global grid to obtain an initial global map, wherein each frame data constitutes the valid segment; Step S02: merging overlapping lanes screened from the initial global map to obtain a global grid map; Step S03, mapping each object in the valid segment to the global grid map, and obtaining the index position of each object in each frame of data in the global grid map; Step S04: Based on the index position, each object in each frame of data is bound to the global lane centerline in the global grid map to obtain a preset global map.

[0051] It should be noted that the lane centerline is a virtual line on the road used to indicate the direction of vehicle travel, typically located in the center of the lane. A global grid is a two- or three-dimensional coordinate system used to map the actual road environment into a unified, standardized space. This grid system typically has a fixed resolution and size and covers the entire area where an autonomous vehicle may travel. The initial global map is a preliminary map obtained by mapping the lane centerline coordinates in each frame of data onto the global grid. This map contains the basic shape and direction of the road but may contain overlapping or inconsistent lane information. Overlapping lanes occur in the initial global map due to differences or errors between different frames of data, resulting in multiple lane centerlines at the same or similar locations. The global grid map is obtained by merging overlapping lanes in the initial global map and performing further processing and optimization. The index position is the coordinate or position information corresponding to each object in the global grid map.

[0052] It can be understood that the coordinates of the lane centerlines on each frame of data are mapped to a unified global grid to form an initial global map, which ensures the spatial consistency of lane information. In the initial global map, due to overlap or errors between valid segments, multiple lane centerlines may be in the same or similar positions. By screening and merging these overlapping lanes, redundancy and contradictions can be eliminated, and a more accurate and consistent global grid map can be obtained. By mapping each object in the valid segment (such as vehicles, pedestrians, traffic signs, etc.) to the global grid map, the index position of each object in each frame of data in the global grid map can be obtained, and the lanes in the global grid map are corrected to ensure the correctness of each lane in the global grid map. Finally, based on the index position of the object in the global grid map, the object in each frame of data is bound to the global lane centerline.

[0053] It can be understood that by integrating lane centerline information from multiple frames of data and filtering and merging it, redundancy and contradictions between data are effectively eliminated. At the same time, by binding objects to lane centerlines, the accuracy and reliability of the map are further improved.

[0054] Furthermore, step S04 further includes: Determining a target lane corresponding to the object in the global grid map based on the index position; Determine the number of objects in the target lane in each frame of data; If the number of objects is not one, determining a projection distance of the object on the global lane centerline of the target lane; Bind the objects in each frame of data to the global lane centerline and projection distance of the corresponding target lane to obtain a preset global map.

[0055] It should be noted that the target lane is the lane where the object is located, determined in the global grid map based on the index position of the object. The number of objects is the total number of objects that exist in a specific target lane in a certain frame of data. Since the number of objects in the target lane is one, it means that there is only one car in the target lane, so the coordinates of the target lane can be assigned to the object. The projection distance is the vertical projection distance of the object on the global lane centerline of the target lane, that is, the shortest distance from the object position to the lane centerline. The local lane centerline is a virtual line in the global grid map used to represent the lane direction and position, located in the center of the lane.

[0056] It can be understood that by constructing a preset global map based on the index position and target lane method, there is no need to globally scan and reconstruct the entire road environment, which reduces the complexity and computational complexity of data processing. By calculating the projection distance of the object on the global lane centerline of the target lane, the relative position of the object in the lane can be accurately described, and the driving trajectory and possible behavior of the object can be more accurately judged, thereby improving the accuracy of identifying cutin behavior.

[0057] It is understandable that since it can handle the situation where there are multiple objects in the target lane, by judging the number of objects and calculating the projection distance, it provides the autonomous driving system with rich environmental information, which can better adapt to complex traffic scenarios such as congestion and intersections, thereby improving the adaptability and robustness of the system.

[0058] Furthermore, step S02 further includes: screening overlapping lanes having overlapping areas from the initial global map; Merging the overlapping lanes to obtain a global lane centerline of each lane in the valid segment in the initial global map; Performing at least one thinning process on the centerline coordinates of the global lane centerline to obtain thinned centerline coordinates; Based on the thinned centerline coordinates, the global lane centerlines are gridded to obtain a global grid map.

[0059] It should be noted that thinning refers to simplifying the centerline coordinates of the global lane centerline to reduce the number of data points while maintaining the overall shape and direction of the lane centerline. Thinned centerline coordinates refer to the coordinates of the remaining data points on the global lane centerline after thinning. Gridding refers to the process of dividing the global lane centerline into multiple grid cells according to specific rules.

[0060] It can be understood that by screening overlapping lanes and merging them, the overlapping phenomenon in the initial global map can be eliminated, making the lane centerlines clearer and more accurate, which is conducive to improving the accuracy and readability of the global map. The thinning processing step can significantly reduce the number of data points of the global lane centerlines, thereby reducing the complexity and computational complexity of data processing, simplifying the map construction process, and improving the efficiency of global map generation. When using the global map, it can also avoid wasting too much time in searching for the lane centerline of the vehicle to be cut into in the global map due to too many centerline coordinates, thereby improving the efficiency of identifying cutin behavior.

[0061] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the cutin behavior recognition method of the present application. More simple transformations based on this technical concept are all within the scope of protection of the present application.

[0062] This application also provides a cutin behavior recognition device, please refer to Figure 4 , the cutin behavior recognition device includes: An acquisition module 10 is used to filter valid segments in a non-map LCC scenario from the acquired historical driving data; A screening module 20 is configured to screen a vehicle to be cut into the own vehicle lane from the current frame of the valid segment; An identification module 30 is configured to identify the bound lane where the vehicle to be cut in is located in the reference frame of the valid segment based on a preset global map, wherein the preset global map includes the lanes where each vehicle is located at different frame times; The judgment module 40 is configured to judge whether the vehicle to be cut-in has cut-in behavior based on the bound lanes where the vehicle to be cut-in is located in different reference frames.

[0063] Optionally, the identification module 30 is further used to search the most recent valid frame before the current frame from the valid segment; based on the preset global map, obtain the first bound lane where the vehicle to be cut in is located in the current frame, and the second bound lane where the vehicle is located in the most recent valid frame; determine whether the first bound lane and the second bound lane are the same lane; if the first bound lane and the second bound lane are not the same lane, then based on the preset global map, obtain the bound lanes in each reference frame after the current frame in the valid segment, and obtain the bound lanes in each reference frame before the most recent valid frame in the valid segment.

[0064] Optionally, the identification module 30 is further used to count a first proportion of all frames in which the bound lane is the first bound lane in the reference frames after the current frame, based on the bound lanes where the vehicle to be cut in different reference frames is located; and a second proportion of all frames in which the bound lane is the second bound lane in the reference frames before the most recent valid frame; determine whether the first proportion and the second proportion both meet the preset proportions; if both meet the preset proportions, determine that the vehicle to be cut in has performed a cutin behavior.

[0065] Optionally, the screening module 20 is further configured to extract a cutin candidate vehicle from the current frame of the valid segment; obtain the azimuth angle of the ego vehicle in the current frame and the position coordinates of the ego vehicle in the current frame mapped in the global map; determine the relative lateral displacement and relative longitudinal displacement between the current cutin candidate vehicle and the ego vehicle based on the position information and the position coordinates of the current cutin candidate vehicle, wherein the current cutin candidate vehicle is the cutin candidate vehicle currently located; determine whether the relative longitudinal displacement is less than a currently recorded minimum longitudinal displacement, and determine whether the relative lateral displacement meets a preset threshold; if the relative lateral displacement is less than the minimum longitudinal displacement and the relative longitudinal displacement meets the preset threshold, use the current cutin candidate vehicle to update the currently recorded vehicle to be cut in, until all cutin candidate vehicles are determined, and determine the cutin candidate vehicle with the smallest relative lateral displacement as the vehicle to be cut in.

[0066] Optionally, the screening module 20 is further used to map the coordinates of the lane centerline on each frame data to a global grid to obtain an initial global map, wherein each frame data constitutes the valid segment; merge the overlapping lanes screened from the initial global map to obtain a global grid map; map each object in the valid segment to the global grid map to obtain an index position of each object in each frame data in the global grid map; and based on the index position, bind each object in each frame data to the global lane centerline in the global grid map to obtain a preset global map.

[0067] Optionally, the screening module 20 is further used to determine the target lane corresponding to the object in the global grid map based on the index position; determine the number of objects existing in the target lane in each frame data; if the number of objects is not one, determine the projection distance of the object on the global lane centerline of the target lane; bind the object in each frame data with the global lane centerline and projection distance of the corresponding target lane to obtain a preset global map.

[0068] Optionally, the screening module 20 is further used to screen overlapping lanes with overlapping areas from the initial global map; merge the overlapping lanes to obtain the global lane center lines of each lane in the valid segment in the initial global map; perform at least one thinning process on the center line coordinates of the global lane center lines to obtain thinned center line coordinates; and grid the global lane center lines based on the thinned center line coordinates to obtain a global grid map.

[0069] The cutin behavior recognition device provided in this application, employing the cutin behavior recognition method described in the aforementioned embodiments, can address the technical issue of being unable to identify a vehicle's cutin behavior in non-image LCC scenarios. Compared to the prior art, the cutin behavior recognition device provided in this application achieves the same beneficial effects as the cutin behavior recognition method described in the aforementioned embodiments. Other technical features of the cutin behavior recognition device are the same as those disclosed in the aforementioned embodiments and are not further elaborated here.

[0070] The present application provides a cutin behavior recognition device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the cutin behavior recognition method of the above-mentioned embodiment 1.

[0071] Reference below Figure 5 , which shows a schematic diagram of the structure of a cutin behavior recognition device suitable for implementing the embodiments of the present application. The cutin behavior recognition device in the embodiments of the present application may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), in-vehicle terminals (such as in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 5 The cutin behavior recognition device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0072] like Figure 5As shown, the cutin behavior recognition device may include a processing device 1001 (e.g., a central processing unit, graphics processing unit, etc.), which can perform various appropriate actions and processes based on programs stored in read-only memory (ROM) 1002 or programs loaded from storage device 1003 into random access memory (RAM) 1004. RAM 1004 also stores various programs and data required for the operation of the cutin behavior recognition device. Processing device 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007, such as a touchscreen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008, such as a liquid crystal display (LCD), speaker, vibrator, etc.; storage device 1003, such as a magnetic tape or hard disk; and communication device 1009. The communication device 1009 can allow the cutin behavior recognition device to communicate wirelessly or wired with other devices to exchange data. Although the figure shows a cutin behavior recognition device with various systems, it should be understood that it is not required to implement or have all of the systems shown. More or fewer systems can be implemented or provided instead.

[0073] In particular, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed in the present application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program comprising program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0074] The cutin behavior recognition device provided in this application, employing the cutin behavior recognition method described in the aforementioned embodiment, can resolve the technical issue of being unable to identify a vehicle's cutin behavior in non-image LCC scenarios. Compared to the prior art, the cutin behavior recognition device provided in this application achieves the same beneficial effects as the cutin behavior recognition method described in the aforementioned embodiment. Other technical features of the cutin behavior recognition device are the same as those disclosed in the aforementioned embodiment and are not further elaborated here.

[0075] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0076] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0077] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, a computer program) stored thereon, the computer-readable program instructions being used to execute the cutin behavior recognition method in the above-mentioned embodiment.

[0078] The computer-readable storage medium provided herein may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems, or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including, but not limited to, wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0079] The computer-readable storage medium may be included in the cutin behavior recognition device; or may exist independently without being assembled into the cutin behavior recognition device.

[0080] The above-mentioned computer-readable storage medium carries one or more programs. When the above-mentioned one or more programs are executed by the cutin behavior recognition device, the cutin behavior recognition device: filters out valid segments in the map-free LCC scenario from the acquired historical driving data; filters out the vehicle to be cut into the own vehicle lane from the current frame of the valid segment; based on a preset global map, identifies the bound lane where the vehicle to be cut into is located in the reference frame of the valid segment, and the preset global map includes the lanes where each vehicle is located at different frame times; based on the bound lanes where the vehicle to be cut into is located in different reference frames, determines whether the vehicle to be cut into has performed a cutin behavior.

[0081] Computer program code for performing the operations of the present application may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages such as Java, Smalltalk, C++, and conventional procedural programming languages such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0082] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, program segment or a part of code, and the module, program segment or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented by a dedicated hardware-based system that performs the specified function or operation, or can be implemented by a combination of dedicated hardware and computer instructions.

[0083] The modules described in the embodiments of the present application may be implemented in software or hardware, wherein the name of a module does not necessarily limit the unit itself.

[0084] The computer-readable storage medium provided in this application stores computer-readable program instructions (i.e., a computer program) for executing the aforementioned cutin behavior recognition method. This computer-readable storage medium can address the technical issue of being unable to recognize a vehicle's cutin behavior in non-image-based LCC scenarios. Compared to the prior art, the computer-readable storage medium provided in this application offers the same beneficial effects as the cutin behavior recognition method provided in the aforementioned embodiments, and therefore is not further elaborated here.

[0085] The present application also provides a computer program product, comprising a computer program, which implements the steps of the above-mentioned cutin behavior recognition method when executed by a processor.

[0086] The computer program product provided in this application can solve the technical problem of being unable to identify a vehicle's cutin behavior in non-image LCC scenarios. Compared to the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the cutin behavior identification method provided in the above embodiment, and will not be elaborated here.

[0087] The above description is only part of the embodiments of the present application and does not limit the patent scope of the present application. All equivalent structural transformations made by using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect application in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A cutin behavior recognition method, characterized in that: The cutin behavior recognition method includes: Filter valid segments in the LCC scenario without image from the acquired historical driving data; Filtering a vehicle to be cut into the own vehicle lane from the current frame of the valid segment; Identifying, based on a preset global map, a bound lane in which the vehicle to be cut into is located in a reference frame of the valid segment, wherein the preset global map includes lanes in which each vehicle is located at different frame times; Based on the bound lanes where the vehicle to be cut in is located in different reference frames, it is determined whether the vehicle to be cut in has performed a cutin behavior.

2. The cutin behavior recognition method according to claim 1, characterized in that: The step of identifying the bound lane where the vehicle to be cut in is located in the reference frame of the valid segment based on the preset global map includes: Searching for the most recent valid frame before the current frame from the valid fragments; Based on the preset global map, obtaining the first bound lane where the vehicle to be cut-in is located in the current frame and the second bound lane where the vehicle is located in the most recent valid frame; determining whether the first bound lane and the second bound lane are the same lane; If the first bound lane and the second bound lane are not the same lane, based on the preset global map, the bound lanes in each reference frame after the current frame in the valid segment are obtained, and the bound lanes in each reference frame before the most recent valid frame in the valid segment are obtained.

3. The cutin behavior recognition method according to claim 2, characterized in that: The step of determining whether the vehicle to be cut-in has performed a cut-in behavior based on the bound lanes where the vehicle to be cut-in is located in different reference frames includes: Based on the bound lanes of the to-be-cut-in vehicle in different reference frames, counting a first proportion of all frames in which the bound lane in reference frames subsequent to the current frame is the first bound lane; and, a second proportion of all frames in which the bound lane is the second bound lane in the reference frame before the most recent valid frame; Determining whether the first proportion and the second proportion both meet a preset proportion; If the preset proportions are all met, it is determined that the vehicle to be cut in has performed a cutin behavior.

4. The cutin behavior recognition method according to claim 1, wherein: The step of screening the vehicle to be cut into the own vehicle lane from the current frame of the valid segment includes: Extracting cutin candidate vehicles from the current frame of the valid segment; Obtaining the azimuth angle of the ego vehicle in the current frame and the position coordinates of the ego vehicle in the current frame mapped in the global map; Determining the relative lateral displacement and relative longitudinal displacement between the current cutin candidate vehicle and the ego vehicle based on the position information of the current cutin candidate vehicle and the position coordinates, wherein the current cutin candidate vehicle is the cutin candidate vehicle currently located; Determining whether the relative longitudinal displacement is less than the currently recorded minimum longitudinal displacement, and determining whether the relative lateral displacement meets a preset threshold; If the relative lateral displacement is less than the minimum longitudinal displacement and the relative longitudinal displacement meets the preset threshold, the current cutin candidate vehicle is used to update the currently recorded vehicle to be cut in, until all cutin candidate vehicles are judged, and the cutin candidate vehicle with the smallest relative lateral displacement is determined as the vehicle to be cut in.

5. The cutin behavior recognition method according to claim 1, wherein: Before the step of screening the vehicle to be cut into the own vehicle lane from the current frame of the valid segment, the method further includes: Mapping the coordinates of the lane centerline on each frame of data to a global grid to obtain an initial global map, wherein each frame of data constitutes the valid segment; Merging overlapping lanes filtered from the initial global map to obtain a global grid map; Mapping each object in the valid segment to the global grid map, and obtaining the index position of each object in each frame of data in the global grid map; Based on the index position, each object in each frame of data is bound to the global lane centerline in the global grid map to obtain a preset global map.

6. The cutin behavior recognition method according to claim 5, characterized in that: The step of binding each object in each frame of data to a global lane centerline in the global grid map based on the index position to obtain a preset global map includes: Determining a target lane corresponding to the object in the global grid map based on the index position; Determine the number of objects in the target lane in each frame of data; If the number of objects is not one, determining a projection distance of the object on the global lane centerline of the target lane; Bind the objects in each frame of data to the global lane centerline and projection distance of the corresponding target lane to obtain a preset global map.

7. The cutin behavior recognition method according to claim 5, characterized in that: The step of merging overlapping lanes screened from the initial global map to obtain a global grid map comprises: screening overlapping lanes having overlapping areas from the initial global map; Merging the overlapping lanes to obtain a global lane centerline of each lane in the valid segment in the initial global map; Performing at least one thinning process on the centerline coordinates of the global lane centerline to obtain thinned centerline coordinates; Based on the thinned centerline coordinates, the global lane centerlines are gridded to obtain a global grid map.

8. A cutin behavior recognition device, characterized in that: The cutin behavior recognition device includes: The acquisition module is used to filter valid segments in the LCC scenario without image from the acquired historical driving data; A screening module, configured to screen a vehicle to be cut into the own vehicle lane from a current frame of the valid segment; an identification module, configured to identify, based on a preset global map, a bound lane in which the vehicle to be cut into is located in a reference frame of the valid segment, wherein the preset global map includes lanes in which each vehicle is located at different frame times; The judgment module is used to judge whether the vehicle to be cut in has a cutin behavior based on the bound lanes where the vehicle to be cut in is located in different reference frames.

9. A cutin behavior recognition device, characterized in that: The cutin behavior recognition device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the cutin behavior recognition method according to any one of claims 1 to 7.

10. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the cutin behavior recognition method according to any one of claims 1 to 7 are implemented.

Images