Traffic element information processing method and device, equipment and medium

Abstract

The invention relates to the field of intelligent driving, in particular to a traffic element information processing method and device, equipment and a medium, and the method comprises the steps: obtaining road element information and lane element information at a current moment; extracting a traffic element feature set corresponding to the traffic element information set at the current moment; according to a target traffic element feature and a target neighborhood feature at the current moment, a mapping relation between the target traffic element feature and the target neighborhood feature is obtained by using a target adaptive convolution model, the target traffic element feature is any feature in a traffic element feature set, and the target neighborhood feature is any feature in the traffic element feature set; the target neighborhood feature is the traffic element feature of any neighborhood in the target traffic element features, and the target adaptive convolution model has strong feature learning and mapping capabilities, and can establish a mapping relationship between traffic elements according to the feature difference at different moments and the relationship between the neighborhood features.

Description

Technical Field

[0001] The present application relates to the field of intelligent driving, and in particular to a method, device, equipment and medium for processing traffic element information. Background Art

[0002] In densely populated urban environments, smart cars must navigate more complex traffic rules and dynamic scenarios than at highway speeds. Their perception systems must simultaneously detect multiple traffic signs and signals in each lane. Furthermore, lanes are complex, encompassing not only the car's own lane but also multiple other lanes at intersections. When a smart car's perception equipment perceives traffic elements on urban roads, it may misidentify them. For example, if two traffic lights are close together, a more distant one might be mistaken for the one in the current lane. This can affect the driver's driving judgment or the autonomous driving system's driving planning.

[0003] Therefore, how to determine the relationship between traffic element information to effectively reduce violations of traffic rules and the occurrence of no braking is a technical problem that needs to be solved urgently in this field. Summary of the Invention

[0004] The purpose of this application is to provide a traffic element information processing method, device, equipment and medium that can determine the relationship between traffic element information.

[0005] In a first aspect, a method for processing traffic element information is provided, comprising: obtaining a traffic element information set at a current moment; the traffic element information set comprising: ground arrow information, zebra crossing information, stop line information, traffic light information, and lane information; extracting a traffic element feature set corresponding to the traffic element information set at the current moment; and obtaining a mapping relationship between the target traffic element feature and the target neighborhood feature using a target adaptive convolution model based on the target traffic element feature and the target neighborhood feature at the current moment, wherein the target traffic element feature is any feature in the traffic element feature set, and the target neighborhood feature is the traffic element feature of any neighborhood in the target traffic element feature.

[0006] In a preferred example, the present application can be further configured as follows: based on the target traffic element characteristics and target neighborhood characteristics at the current moment, using the target adaptive convolution model, a mapping relationship between the target traffic element characteristics and the target neighborhood characteristics is obtained, including: determining multiple neighborhood characteristics corresponding to the target traffic element characteristics at the current moment; subtracting the target traffic element characteristics at the current moment from each corresponding neighborhood characteristic to obtain a feature difference; based on the feature difference between the target traffic element characteristics and the target neighborhood characteristics at the current moment, using the target adaptive convolution model, a mapping relationship between the target traffic element characteristics and the target neighborhood characteristics is obtained.

[0007] In a preferred example, the present application can be further configured as follows: based on the feature difference between the target traffic element feature and the target neighborhood feature at the current moment, a target adaptive convolution model is used to obtain a mapping relationship between the target traffic element feature and the target neighborhood feature, including: if the current moment is the first moment, the feature difference between the target traffic element feature and the target neighborhood feature at the first moment is input into the target adaptive convolution model to obtain a mapping relationship between the target traffic element feature and the target neighborhood feature; if the current moment is not the first moment, the feature difference between the target traffic element feature and the target neighborhood feature at the current moment and the feature difference at the previous moment are input into the target adaptive convolution model to obtain a mapping relationship between the target traffic element feature and the target neighborhood feature.

[0008] In a preferred example, the present application can be further configured as follows: it also includes: obtaining training data, the training data including: a traffic element information training feature set; inputting the training data into the adaptive convolution model to obtain a training mapping relationship; using a loss function according to the training mapping relationship to train the adaptive convolution model to obtain a target adaptive convolution model.

[0009] In a preferred example, the present application can be further configured as follows: obtaining the traffic element information set at the current moment, including: obtaining the initial traffic element information set at the current moment; the initial traffic element information set includes: initial road element information and initial lane element information; each initial element information includes: coordinates and confidence; removing the initial element information with a confidence less than a preset confidence threshold from the initial traffic element information set at the current moment to obtain the traffic element information set at the current moment.

[0010] In a preferred example, the present application can be further configured as follows: it also includes: constructing a mapping chain based on the mapping relationship between each traffic element feature and the target neighborhood feature based on the logical transitivity rule, wherein the mapping chain includes multiple associated traffic element features; and mounting the mapping chain in the lane geometry space for display.

[0011] In a preferred example, the present application can be further configured to: mount the mapping chain of lane information in the lane geometric space for display, including: if multiple lane information is included, determining the first target lane information corresponding to the vehicle based on the vehicle's position information and each lane information, and mounting the mapping chain of the first target lane information in the lane geometric space for display; and / or, the method also includes: estimating the second target lane information of the lane that the vehicle is about to enter based on the first target lane information corresponding to the vehicle and the corresponding ground arrow information, and mounting the mapping chain of the second target lane information in the lane geometric space for display.

[0012] In the second aspect, a traffic element information processing device is provided, including: an acquisition module for acquiring a traffic element information set at a current moment; the traffic element information set includes: ground arrow information, zebra crossing information, stop line information, traffic light information and lane information; an extraction module for extracting a traffic element feature set corresponding to the traffic element information set at the current moment; a processing module for obtaining a mapping relationship between target traffic element features and target neighborhood features based on the target traffic element features and target neighborhood features at the current moment, using a target adaptive convolution model, wherein the target traffic element feature is any feature in the traffic element feature set, and the target neighborhood feature is the traffic element feature of any neighborhood in the target traffic element feature.

[0013] According to a third aspect, an electronic device is provided, comprising a memory and a processor, wherein a computer program is stored in the memory, and the processor executes any one of the methods described in the first aspect when running the computer program.

[0014] In a fourth aspect, a computer-readable storage medium is provided, wherein at least one program code is stored in the computer-readable storage medium, and the program code is loaded and executed by a processor to implement any method as described in the first aspect.

[0015] In a fifth aspect, a computer program product is provided, comprising a computer program or instructions, which, when executed by a processor, implements any method described in the first aspect.

[0016] In summary, the traffic element information processing method provided by this application has the following beneficial technical effects:

[0017] Obtain the traffic element information set at the current moment; the traffic element information set includes: ground arrow information, zebra crossing information, stop line information, traffic light information and lane information; extract the traffic element feature set corresponding to the traffic element information set at the current moment; according to the target traffic element features and target neighborhood features at the current moment, use the target adaptive convolution model to obtain the mapping relationship between the target traffic element features and the target neighborhood features, wherein the target traffic element feature is any feature in the traffic element feature set, and the target neighborhood feature is the traffic element feature of any neighborhood in the target traffic element feature. The target adaptive convolution model has powerful feature learning and mapping capabilities, and can establish a mapping relationship between each traffic element based on the feature differences at different times and the relationship between neighborhood features.

[0018] In addition, the present application also provides a traffic element information processing device, equipment and medium, all of which have the above-mentioned beneficial technical effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the embodiments of the present application or the technical solutions of the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0020] Figure 1 This is a flow chart of a method for processing traffic element information provided by an embodiment of the present application;

[0021] Figure 2 This is a schematic diagram of a traffic scenario provided by an embodiment of the present application;

[0022] Figure 3 This is a data processing diagram of an adaptive convolution model provided in an embodiment of the present application;

[0023] Figure 4 This is a schematic diagram of the structure of a traffic element information processing device provided in an embodiment of the present application;

[0024] Figure 5 This is a structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0025] This specific embodiment is merely an explanation of the present application and is not a limitation of the present application. After reading this specification, those skilled in the art may make non-creative modifications to the present embodiment as needed, but as long as they are within the scope of the present application, they are protected by patent law.

[0026] It should be noted that in the optional embodiments of the present application, the object information and other related data involved, when the embodiments in the present application are applied to specific products or technologies, need to obtain the permission or consent of the object, and the collection, use and processing of the relevant data need to comply with the relevant laws, regulations and standards of the relevant countries and regions. In other words, if the embodiments of the present application involve data related to the object, it needs to be obtained with the permission of the object, the permission of the relevant department, and in compliance with the relevant laws, regulations and standards of the country and region. If personal information is involved in the embodiments, the acquisition of all personal information requires the consent of the individual. If sensitive information is involved, the separate consent of the information subject needs to be obtained. The embodiments also need to be implemented with the permission of the object.

[0027] To make the purpose, technical solutions, and advantages of the embodiments of this application more clear, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0028] In this document, the term "and / or" simply describes a relationship between related objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, the character " / " in this document, unless otherwise specified, generally indicates an "or" relationship between the related objects.

[0029] In related technologies, the sensing equipment of smart cars perceives traffic element information on urban roads, and there may be cases of misidentification. For example, if two traffic lights are very close, the farther traffic light may be mistaken for the traffic light in the current lane, thereby affecting the driver's driving judgment or the driving planning of the autonomous driving equipment.

[0030] Therefore, this application can process traffic element information such as ground arrows, zebra crossings, stop lines, traffic lights and lanes emitted by sensing devices, and construct associated relationships so that subsequent planning modules can use this information.

[0031] Specifically, the embodiment of the present application provides a method for processing traffic element information, such as Figure 1 As shown, the method provided in the embodiment of the present application can be executed by an electronic device, which is a server. The server can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services. The terminal device can be a smartphone, a tablet computer, a laptop computer, a desktop computer, a vehicle-mounted device, etc., but is not limited to these. The terminal device and the electronic device can be directly or indirectly connected via wired or wireless communication, which is not limited in the embodiment of the present application. The method includes:

[0032] S101. Obtain a current traffic element information set; the traffic element information set includes: ground arrow information, zebra crossing information, stop line information, traffic light information, and lane information.

[0033] In an embodiment of the present application, the sensing device continuously collects traffic element information and sends it to the electronic device, so that the electronic device obtains the traffic element information set S(t)={sp|p=1, 2,…, N} at the current moment, where t represents the t-th frame, t=1, 2,…T, T represents the number of system operation frames; sp represents the traffic element information of the p-th point at time t, and N is the total number of traffic element information.

[0034] In one possible scenario, the road element information includes: ground arrow information, zebra crossing information, stop line information, traffic light information, and lane information. It is understandable that the information is not limited to the above-mentioned element information, and may also include information such as signboards.

[0035] Specifically, ground arrow information includes: longitudinal coordinate Ax, transverse coordinate Ay, vertical coordinate Az, arrow type Atype, confidence Ac, and arrow direction Aheading. The coordinates refer to the center point of the ground arrow. It is understood that vertical coordinates exist due to ground flatness and the height of the sensing device. Arrow types include, but are not limited to, straight arrows, turn arrows, and U-turn arrows. The confidence level is the confidence level of the ground arrow information output by the sensing device.

[0036] The zebra crossing information includes: the coordinates of the four corner points of the zebra crossing (Cix, Ciy, Ciz) (i ranges from 1 to 4), and the confidence level Cc of the zebra crossing. For example, see Figure 2 The zebra crossing information is shown in Figure 1, with the coordinates of the lower right corner being (S4x, S4y, S4z).

[0037] The stop line information includes: the coordinates of the four corner points of the stop line (Six, Siy, Siz) (i ranges from 1 to 4) and the confidence level Sc of the stop line.

[0038] Traffic light information includes: traffic light coordinates (Lx, Ly, Lz), traffic light confidence Lc, traffic light type Ltype, traffic light color Lcolor, and whether the traffic light is flashing Lblink. The traffic light coordinates refer to the center coordinates of the lit light in the traffic light.

[0039] Lane information includes: lane ID, coordinates of lane boundary and center line points (Rx, Ry, Rz) (sampling interval of each point is 1 meter), and lane line confidence Rc.

[0040] S102: Extracting a traffic element feature set corresponding to the traffic element information set at the current moment.

[0041] Among them, all the extracted feature sets are recorded as Fs={ft|t=1, 2, ..., P}, where ft is the traffic element feature set corresponding to time t.

[0042] S103. According to the target traffic element characteristics and target neighborhood characteristics at the current moment, a target adaptive convolution model is used to obtain a mapping relationship between the target traffic element characteristics and the target neighborhood characteristics, wherein the target traffic element characteristics are any characteristics in the traffic element feature set, and the target neighborhood characteristics are traffic element characteristics of any neighborhood in the target traffic element characteristics.

[0043] In the embodiment of the present application, the feature map is constructed by using the information of each point (traffic element features), and the adaptive convolution layer of the target adaptive convolution model is designed to generate a set of N-dimensional features to reflect the spatiotemporal characteristics between different element features. i The center point of the graph convolution is the target traffic element feature, N(i) = {j: (i, j) ∈ E} is a set of point indices in its neighborhood, and the whole represents a set of features. It can be understood that in the feature graph structure, the neighborhood N(i) of the center point i is the set of all center points directly connected to the center point i.

[0044] In the embodiment of the present application, the target adaptive convolution model is used to determine the mapping relationship between the target traffic element characteristics and its multiple domain characteristics. The structure of the adaptive convolution model is not limited in the embodiment of the present application. For example, see Figure 2 The obtained mapping relationships include: the mapping relationship between zebra crossing 1 and stop line 1, the mapping relationship between zebra crossing 1 and traffic light 1, the mapping relationship between zebra crossing 1 and lane 1, the mapping relationship between stop line 1 and lane 1, the mapping relationship between stop line 1 and traffic light 1, and the mapping relationship between lane 1 and traffic light 1.

[0045] It can be seen that in the embodiment of the present application, the traffic element information set at the current moment is obtained; the traffic element information set includes: ground arrow information, zebra crossing information, stop line information, traffic light information and lane information; the traffic element feature set corresponding to the traffic element information set at the current moment is extracted; according to the target traffic element features and target neighborhood features at the current moment, the target adaptive convolution model is used to obtain the mapping relationship between the target traffic element features and the target neighborhood features, wherein the target traffic element feature is any feature in the traffic element feature set, and the target neighborhood feature is the traffic element feature of any neighborhood in the target traffic element feature. The target adaptive convolution model has powerful feature learning and mapping capabilities, and can establish a mapping relationship between each traffic element based on the feature differences at different times and the relationship between the neighborhood features.

[0046] A possible implementation method of the embodiment of the present application is to obtain a mapping relationship between the target traffic element characteristics and the target neighborhood characteristics based on the target traffic element characteristics and the target neighborhood characteristics at the current moment by using a target adaptive convolution model, including: determining multiple neighborhood characteristics corresponding to the target traffic element characteristics at the current moment; subtracting the target traffic element characteristics at the current moment from each corresponding neighborhood characteristic to obtain a feature difference; and obtaining a mapping relationship between the target traffic element characteristics and the target neighborhood characteristics based on the feature difference between the target traffic element characteristics and the target neighborhood characteristics at the current moment by using a target adaptive convolution model.

[0047] For each traffic element feature, multiple neighborhood features corresponding to the traffic element feature are determined. The specific method for determining the neighborhood features can be divided by spatial distance, which is not limited in the embodiments of the present application. After obtaining the neighborhood feature, the traffic element feature is subtracted from the corresponding neighborhood feature to obtain the feature difference. For each traffic element feature, based on the feature difference between the target traffic element feature and the target neighborhood feature, a target adaptive convolution model is used to obtain the mapping relationship between the target traffic element feature and the target neighborhood feature.

[0048] It can be seen that in the embodiment of the present application, the feature difference reflects the degree of difference between the target feature and the neighborhood feature. By calculating the feature difference, the changes between the features can be quantified, and then the target adaptive convolution model can be used to more accurately obtain the mapping relationship between the various traffic elements, thereby improving the accuracy and reliability of traffic element information processing.

[0049] Specifically, a possible implementation method of the embodiment of the present application is to use a target adaptive convolution model to obtain a mapping relationship between the target traffic element characteristics and the target neighborhood characteristics based on the feature difference between the target traffic element characteristics and the target neighborhood characteristics at the current moment, including: if the current moment is the first moment, then the feature difference between the target traffic element characteristics and the target neighborhood characteristics at the first moment is input into the target adaptive convolution model to obtain a mapping relationship between the target traffic element characteristics and the target neighborhood characteristics; if the current moment is not the first moment, then the feature difference between the target traffic element characteristics and the target neighborhood characteristics at the current moment and the feature difference at the previous moment are input into the target adaptive convolution model to obtain a mapping relationship between the target traffic element characteristics and the target neighborhood characteristics.

[0050] Generally speaking, target-adaptive convolution kernels build local graph structures and exploit the spatial relationship between the center point and its neighbors. In BEV data, adaptive convolution can be extended to the spatiotemporal dimension. By introducing feature differences in the temporal dimension, the convolution kernel simultaneously captures the relationship between spatial neighborhoods and temporal continuity, and then outputs based on these relationships.

[0051] In the examples of this application, see Figure 3, an adaptive kernel is designed to capture the mapping relationship between each set of points. For each channel in the output N-dimensional feature, the adaptive kernel uses point features The function above dynamically generates a kernel, specifically, , where m=1, 2…N, represents the N output dimensions corresponding to a single filter defined in the adaptive convolution layer. This kernel is able to extract feature differences in global data and local neighborhood data. It can be understood that global data features usually represent the overall properties of the target; local difference features Reflects the offset of neighborhood point j relative to center point i, i.e., the characteristic difference. As the input features of the adaptive kernel, [·,·] is a cascade operation used to avoid information loss while preserving the original features and the difference features. W is the weight matrix. It is the adaptive kernel generated at the previous moment, and the kernel at the current moment is iterated based on the previous kernel.

[0052] Specifically, if the current moment is the first moment, the feature difference between the target traffic element feature and the target neighborhood feature at the first moment is input into the target adaptive convolution model to obtain the mapping relationship between the target traffic element feature and the target neighborhood feature. , and get the mapping relationship.

[0053] If the current moment is not the first moment, the feature difference between the target traffic element feature and the target neighborhood feature at the current moment and the feature difference at the previous moment are input into the target adaptive convolution model to obtain the mapping relationship between the target traffic element feature and the target neighborhood feature.

[0054] In one achievable manner, the , get the mapping relationship, Represents the feature difference at time t-1, and W is used as a weight matrix to concatenate the feature differences at the previous and next moments.

[0055] Of course, you can also , the embodiments of this application are no longer limited.

[0056] In the embodiment of the present application, the adaptive convolution layer is based on the individual features of each pair of points. An adaptive kernel is generated for each pair of points. Then, the adaptive kernel is applied to pairs of points in order to output their mapping relationship, , where K(t) represents the mapping relationship between the traffic light, zebra crossing, stop line and lane information finally extracted from frame t.

[0057] It can be seen that in an embodiment of the present application, if the current moment is the first moment, the feature difference between the target traffic element feature and the target neighborhood feature at the first moment is input into the target adaptive convolution model to obtain a mapping relationship between the target traffic element feature and the target neighborhood feature; if the current moment is not the first moment, the feature difference between the target traffic element feature and the target neighborhood feature at the current moment and the feature difference at the previous moment are input into the target adaptive convolution model to obtain a mapping relationship between the target traffic element feature and the target neighborhood feature, taking into account the feature difference information between the current moment and the previous moment. By introducing the correlation information on the time series, the dynamic change trend of the traffic element information can be better captured, so that the mapping relationship more accurately reflects the relationship between the target traffic element feature and the neighborhood feature at the current moment.

[0058] A possible implementation method of an embodiment of the present application also includes: obtaining training data, the training data including: a training feature set of traffic element information; inputting an adaptive convolution model according to the training data to obtain a training mapping relationship; and using a loss function according to the training mapping relationship to train the adaptive convolution model to obtain a target adaptive convolution model.

[0059] The loss function of the present invention is: ;Wherein, λ is the balance coefficient and MLP is the multi-layer perceptron.

[0060] A possible implementation method of an embodiment of the present application is to obtain a traffic element information set at the current moment, including: obtaining an initial traffic element information set at the current moment; the initial traffic element information set includes: initial road element information and initial lane element information; each initial element information includes: coordinates and confidence; removing the initial element information with a confidence less than a preset confidence threshold from the initial traffic element information set at the current moment to obtain the traffic element information set at the current moment.

[0061] Among them, the preset confidence threshold can be customized by the user, and the same or different preset confidence thresholds can be set for different traffic elements. The embodiment of the present application is no longer limited, and the user can set it according to actual needs.

[0062] It can be seen that in the embodiment of the present application, by screening the obtained perception data according to the confidence level, the perception data with low confidence level can be removed, and reliable data can be obtained, making the mapping analysis more accurate and faster.

[0063] Furthermore, generally, if there is a stop line on a lane, there is usually a zebra crossing in front of it and a ground arrow on the bottom. Therefore, it is possible that several elements exist at the same time.

[0064] Based on this, after obtaining the traffic element information sets of multiple consecutive moments at the current moment, it is also possible to determine several first traffic elements that exist at the same time. If the several first traffic elements are not completely present in the traffic element information set, it may mean that the perception device did not perceive the traffic element due to unclear markings on the road or other reasons, or the confidence after perception is low. Therefore, the preset confidence threshold can be lowered from the initial traffic element information set, and the traffic element information corresponding to the specified traffic element can be selected, and the specified traffic element can be designated as the non-existent first traffic element; or according to the preset distance between the non-existent first traffic element and the existing first traffic element, the non-existent first traffic element can be simulated and displayed to the user.

[0065] A possible implementation of the embodiment of the present application further includes: performing driving planning based on a mapping relationship between target traffic element characteristics and target neighborhood characteristics.

[0066] Specifically, according to the mapping relationship between each traffic element feature and the target neighborhood feature, a mapping chain is constructed based on the logical transitivity rule. The mapping chain includes multiple associated traffic element features; the mapping chain is mounted in the lane geometry space for display.

[0067] Among them, the logical transitivity rule represents a logical relationship, that is, if there is a mapping relationship from traffic element feature 1 to traffic element feature 2, and a mapping relationship from traffic element feature 2 to traffic element feature 3, then according to the logical transitivity rule, it can be inferred that there is an indirect mapping relationship from traffic element feature 1 to traffic element feature 3.

[0068] In one achievable method, the inference rule may be: if a mapping relationship exists between the first traffic element feature and the second traffic element feature, then all other traffic element features of the second traffic element feature that exist in a mapping relationship may be used as traffic element features that exist in a mapping relationship with the first traffic element feature, so as to obtain a mapping chain of the first traffic element feature. For example, a mapping chain of the second traffic element feature is obtained. Furthermore, if a mapping relationship exists between the first traffic element feature and the second traffic element feature, then it is determined whether the difference between the number of other traffic element features that exist in a mapping relationship with the first traffic element feature and the number of other traffic element features that exist in a mapping relationship with the second traffic element feature is within a preset difference range. If so, the other traffic element features that exist in a mapping relationship with the first traffic element feature and the other traffic element features that exist in a mapping relationship with the second traffic element feature are combined to form a mapping chain as the mapping chain of the first traffic element feature and the second traffic element feature. The preset difference range can be customized by the user, such as 1, 2, etc.

[0069] The mapping chain is composed of multiple associated traffic element features to express the logical relationship between the traffic element features. Continuing with the above example, the mapping chain is: traffic element feature 1, traffic element feature 2 and traffic element feature 3.

[0070] Among them, after obtaining the mapping chain, the mapping chain is mounted on the lane geometry space for display, realizing the association between the mapping chain and the lane geometry space, so that the various traffic element features in the mapping chain can be located at the corresponding feature position in the lane geometry space, realizing the purpose of combining the mapping relationship with the actual environment, and providing more intuitive information for the driver / algorithm.

[0071] After mounting, visualization technology can be used to mark the traffic element features in the mapping chain with graphics, colors, etc., and display them in the lane geometry model, so that the traffic element features belonging to the mapping chain can be directly determined through the displayed content, allowing drivers to pay attention to traffic information in advance, so that they can reasonably control the speed of the vehicle according to the lane and the distance from the stop line, and ensure that they stop or pass the intersection in a safe location. Alternatively, after the autonomous vehicle identifies the traffic light, stop line and lane information through perception equipment and algorithms, it makes decisions based on the mapping relationship, such as accurately controlling the vehicle's parking and passage according to the traffic light status and stop line position; accurately controlling the vehicle's parking and passage according to the traffic light status and stop line position; and avoiding accidents through safety redundancy design and dynamic adjustment.

[0072] Of course, the driver or the autonomous driving system is mainly concerned with the traffic element characteristics of the own lane. Therefore, only the traffic element characteristics that have a mapping relationship with the own lane characteristics and the traffic element characteristics that have a mapping relationship with the traffic element characteristics can be displayed.

[0073] Specifically, a possible implementation method of an embodiment of the present application is to mount the mapping chain of lane information in the lane geometric space for display, including: if multiple lane information is included, then based on the position information of the vehicle and each lane information, determine the first target lane information corresponding to the vehicle, and mount the mapping chain of the first target lane information in the lane geometric space for display, so as to focus on the traffic element information of the lane.

[0074] A possible implementation method of an embodiment of the present application further includes: estimating the second target lane information of the lane that the vehicle is about to enter based on the first target lane information corresponding to the vehicle and the corresponding ground arrow information, and mounting the mapping chain of the second target lane information in the lane geometric space for display.

[0075] Based on the arrows on the ground, the current lane's direction of travel is determined, allowing the vehicle to directly estimate the second target lane information for the lane it is about to enter based on this direction. Furthermore, the second target lane information for the lane the vehicle is about to enter can be estimated based on the direction of travel in the preceding lane, the distribution of surrounding lanes, the indications of arrows on the ground, and common traffic regulations. Once the second target lane information is obtained, its mapping chain is mapped and associated with the lane geometry space for display on the vehicle's display interface or related systems, allowing the driver or the autonomous driving system to intuitively understand the lane the vehicle is about to enter.

[0076] In summary, this application can help smart cars obtain the information about traffic lights, stop lines, and zebra crossings that should be paid attention to when driving in the current lane when driving in urban areas, avoid accidental deceleration of the autonomous driving function in urban areas, and improve the safety of the entire system. By accurately mounting traffic lights, zebra crossings, and stop lines to the lane geometry, the dynamic mapping problem of traffic rule elements and road structure is solved; it ensures that autonomous vehicles perceive and comply with traffic rules in real time, avoiding risks such as running red lights and crossing lines due to positioning deviations; supports high-precision path planning, reduces sudden braking or congestion caused by misjudgment, and improves road traffic efficiency; adapts to heterogeneous scenarios such as complex intersections and variable lanes, and enhances the robustness and scalability of intelligent transportation systems.

[0077] The following is an introduction to a traffic element information processing device provided by an embodiment of the present application. The traffic element information processing device described below and the traffic element information processing method described above can be referenced to each other. The traffic element information processing device of this embodiment is set in an electronic device, and reference Figure 4 , Figure 4 This is a structural block diagram of a device according to one embodiment of the present application, including:

[0078] The acquisition module 210 is used to obtain the current traffic element information set; the traffic element information set includes: ground arrow information, zebra crossing information, stop line information, traffic light information and lane information;

[0079] Extraction module 220, for extracting a traffic element feature set corresponding to the traffic element information set at the current moment;

[0080] The processing module 230 is used to obtain a mapping relationship between the target traffic element feature and the target neighborhood feature based on the target traffic element feature and the target neighborhood feature at the current moment using a target adaptive convolution model, wherein the target traffic element feature is any feature in the traffic element feature set, and the target neighborhood feature is the traffic element feature of any neighborhood in the target traffic element feature.

[0081] In one feasible manner, the processing module 230 is further used to: determine multiple neighborhood features corresponding to the target traffic element feature at the current moment; subtract the target traffic element feature at the current moment from each corresponding neighborhood feature to obtain a feature difference; and obtain a mapping relationship between the target traffic element feature and the target neighborhood feature based on the feature difference between the target traffic element feature at the current moment and the target neighborhood feature using a target adaptive convolution model.

[0082] In one feasible manner, the processing module 230 is further used to: if the current moment is the first moment, then the feature difference between the target traffic element feature and the target neighborhood feature at the first moment is input into the target adaptive convolution model to obtain the mapping relationship between the target traffic element feature and the target neighborhood feature; if the current moment is not the first moment, then the feature difference between the target traffic element feature and the target neighborhood feature at the current moment and the feature difference at the previous moment are input into the target adaptive convolution model to obtain the mapping relationship between the target traffic element feature and the target neighborhood feature.

[0083] In one feasible manner, it also includes: a training module, which is used to: obtain training data, the training data including: a training feature set of traffic element information; input the training data into the adaptive convolution model to obtain a training mapping relationship; and train the adaptive convolution model using a loss function according to the training mapping relationship to obtain a target adaptive convolution model.

[0084] In one feasible manner, the acquisition module 210 is further used to: obtain the initial traffic element information set at the current moment; the initial traffic element information set includes: initial road element information and initial lane element information; each initial element information includes: coordinates and confidence; remove the initial element information with a confidence less than a preset confidence threshold from the initial traffic element information set at the current moment to obtain the traffic element information set at the current moment.

[0085] In one feasible manner, it further includes: a planning module, which is used to: construct a mapping chain based on the mapping relationship between each traffic element feature and the target neighborhood feature and based on the logical transitivity rule, the mapping chain including multiple associated traffic element features; and mount the mapping chain in the lane geometry space for display.

[0086] In one feasible manner, the planning module is specifically used to: mount the mapping chain of lane information in the lane geometric space for display, including: if multiple lane information is included, determining the first target lane information corresponding to the ego vehicle based on the position information of the ego vehicle and each lane information, and mounting the mapping chain of the first target lane information in the lane geometric space for display; the planning module is also used to: estimate the second target lane information of the lane that the ego vehicle is about to enter based on the first target lane information corresponding to the ego vehicle and the corresponding ground arrow information, and mounting the mapping chain of the second target lane information in the lane geometric space for display.

[0087] An electronic device is provided in an embodiment of the present application, as shown in Figure 5. Figure 5 The electronic device 300 shown includes: at least one processor 301 ( Figure 5 301 and memory 303. The processor 301 and memory 303 are connected, for example, via a bus 302. Optionally, the electronic device 300 may further include a transceiver 304. It should be noted that in practical applications, the number of transceivers 304 is not limited to one, and the structure of the electronic device 300 does not constitute a limitation on the embodiments of the present application.

[0088] Processor 301 can be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic device, transistor logic device, hardware component, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 301 can also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0089] Bus 302 may include a path for transmitting information between the above components. Bus 302 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. Bus 302 may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, Figure 5 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0090] The memory 303 may be a ROM (Read Only Memory) or other type of static storage device that can store static information and instructions, a RAM (Random Access Memory) or other type of dynamic storage device that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto.

[0091] The memory 303 is used to store application code for executing the solution of the present application, and the execution is controlled by the processor 301. The processor 301 is used to execute the application code stored in the memory 303 to implement the content shown in the above method embodiment.

[0092] Figure 5 The electronic device shown is merely an example and should not limit the functions and scope of use of the embodiments of the present application.

[0093] An embodiment of the present application provides a computer-readable storage medium, which stores at least one program code. When the computer-readable storage medium is run on a computer, the computer can execute the corresponding content of the aforementioned method embodiment.

[0094] An embodiment of the present application provides a computer program product, including a computer program or instructions, which implements the corresponding contents of the aforementioned method embodiment when the computer program or instructions are executed by a processor.

[0095] It should be understood that although the steps in the flowcharts of the accompanying drawings are shown in sequence as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some of the steps in the flowcharts of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.

[0096] The above are only some of the implementation methods of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for processing traffic element information, characterized in that: include: Get the traffic element information set at the current moment; The traffic element information set includes: ground arrow information, zebra crossing information, stop line information, traffic light information and lane information; Extracting the traffic element feature set corresponding to the traffic element information set at the current moment; According to the target traffic element features and target neighborhood features at the current moment, a target adaptive convolution model is used to obtain a mapping relationship between the target traffic element features and the target neighborhood features, wherein the target traffic element feature is any feature in the traffic element feature set, and the target neighborhood feature is the traffic element feature of any neighborhood in the target traffic element feature.

2. The method according to claim 1, characterized in that According to the target traffic element characteristics and target neighborhood characteristics at the current moment, the target adaptive convolution model is used to obtain the mapping relationship between the target traffic element characteristics and the target neighborhood characteristics, including: Determine multiple neighborhood features corresponding to the target traffic element feature at the current moment; Subtract the target traffic element feature at the current moment from the corresponding features of each neighborhood to obtain the feature difference; According to the feature difference between the target traffic element feature and the target neighborhood feature at the current moment, the mapping relationship between the target traffic element feature and the target neighborhood feature is obtained using the target adaptive convolution model.

3. The method according to claim 2, characterized in that According to the feature difference between the target traffic element feature and the target neighborhood feature at the current moment, the target adaptive convolution model is used to obtain the mapping relationship between the target traffic element feature and the target neighborhood feature, including: If the current moment is the first moment, the feature difference between the target traffic element feature and the target neighborhood feature at the first moment is input into the target adaptive convolution model to obtain the mapping relationship between the target traffic element feature and the target neighborhood feature; If the current moment is not the first moment, the feature difference between the target traffic element feature and the target neighborhood feature at the current moment and the feature difference at the previous moment are input into the target adaptive convolution model to obtain the mapping relationship between the target traffic element feature and the target neighborhood feature.

4. The method according to claim 1, wherein Also includes: Acquiring training data, the training data including: a traffic element information training feature set; Input the training data into the adaptive convolution model to obtain the training mapping relationship; The adaptive convolution model is trained using the loss function according to the training mapping relationship to obtain the target adaptive convolution model.

5. The method according to claim 1, wherein Get the traffic element information set at the current moment, including: Obtaining an initial traffic element information set at the current moment; the initial traffic element information set includes: initial road element information and initial lane element information; each initial element information includes: coordinates and confidence level; Initial element information with a confidence level less than a preset confidence threshold is removed from the initial traffic element information set at the current moment to obtain the traffic element information set at the current moment.

6. The method according to claim 1, characterized in that Also includes: According to the mapping relationship between each traffic element feature and the target neighborhood feature, a lane information mapping chain is constructed based on the logical transitive rule. The mapping chain includes multiple associated traffic element features. The mapping chain of lane information is mounted in the lane geometry space for display.

7. The method according to claim 6, characterized in that The mapping chain of lane information is mounted in the lane geometry space for display, including: If multiple lane information is included, the first target lane information corresponding to the vehicle is determined based on the vehicle's position information and each lane information, and the mapping chain of the first target lane information is mounted in the lane geometry space for display; And / or, the method further comprises: Based on the first target lane information corresponding to the ego vehicle and the corresponding ground arrow information, the second target lane information of the lane that the ego vehicle is about to enter is estimated, and the mapping chain of the second target lane information is mounted in the lane geometry space for display.

8. A traffic element information processing device, characterized in that: include: The acquisition module is used to obtain the traffic element information set at the current moment; The traffic element information set includes: ground arrow information, zebra crossing information, stop line information, traffic light information and lane information; An extraction module, used to extract a traffic element feature set corresponding to the traffic element information set at the current moment; The processing module is used to obtain a mapping relationship between the target traffic element feature and the target neighborhood feature based on the target traffic element feature and the target neighborhood feature at the current moment using a target adaptive convolution model, wherein the target traffic element feature is any feature in the traffic element feature set, and the target neighborhood feature is the traffic element feature of any neighborhood in the target traffic element feature.

9. An electronic device, characterized in that: The electronic device includes a memory and a processor, wherein a computer program is stored in the memory, and the processor executes the method according to any one of claims 1 to 7 when running the computer program.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores at least one program code, and the program code is loaded and executed by a processor to implement the method according to any one of claims 1 to 7.

Images