Model training method, navigation trajectory recommendation method, device and vehicle

Abstract

The present disclosure provides a model training method, a navigation trajectory recommendation method, a device and a vehicle, which are related to deep learning of map technology in computer technology, specifically to navigation technology, and can be applied to autonomous driving and intelligent transportation. The specific implementation scheme is: obtaining road network data, and determining the user's navigation trajectory based on the road network data, performing multi-dimensional feature extraction on the user's navigation trajectory, obtaining the path characteristics of the user's navigation trajectory, and constructing a sample trajectory set based on the road network data, the path characteristics, and the user's navigation trajectory, and obtaining a navigation trajectory recommendation model based on the sample trajectory set training, wherein the navigation recommendation model is used to recall and recommend navigation trajectories, so that the navigation trajectory recommendation model has high reliability and effectiveness, thereby improving the accuracy of the recommendation, and further improving the technical effect of the vehicle's driving safety.

Description

Technical Field

[0001] The present disclosure relates to deep learning of map technology in computer technology, specifically to navigation technology, which can be applied to autonomous driving and intelligent transportation, and in particular to a model training method, a navigation trajectory recommendation method, a device and a vehicle. Background Art

[0002] The map navigation service includes a navigation track recommendation service. For example, the map navigation system receives a start point and an end point input by a user, and recalls and outputs a corresponding navigation track based on a pre-built map.

[0003] In the related art, a map navigation system searches for a path with the minimum road cost based on a map as a navigation trajectory, and the minimum road cost is usually determined based on distance or time.

[0004] However, when the above method is used to determine and recommend navigation trajectories to users, there is a problem that the reliability of the recommendation is relatively low. Summary of the invention

[0005] The present disclosure provides a model training method for improving the reliability of navigation trajectory recommendation, a navigation trajectory recommendation method, a device, and a vehicle.

[0006] According to a first aspect of the present disclosure, a model training method is provided, comprising:

[0007] Acquire road network data, and determine the user navigation trajectory according to the road network data;

[0008] Performing multi-dimensional feature extraction on the user navigation trajectory to obtain path features of the user navigation trajectory, and constructing a sample trajectory set based on the road network data, the path features, and the user navigation trajectory;

[0009] A navigation trajectory recommendation model is obtained by training according to the sample trajectory set, wherein the navigation trajectory recommendation model is used to recall and recommend navigation trajectories.

[0010] According to a second aspect of the present disclosure, a method for recommending a navigation trajectory is provided, comprising:

[0011] Obtaining a user's navigation request, wherein the navigation request includes a starting point and an end point;

[0012] Recalling and outputting a navigation trajectory corresponding to the navigation request according to the navigation request and a pre-trained navigation trajectory recommendation model;

[0013] The navigation trajectory recommendation model is obtained based on the method described in the first aspect.

[0014] According to a third aspect of the present disclosure, a model training device is provided, comprising:

[0015] A first acquisition unit, used for acquiring road network data;

[0016] A determination unit, configured to determine a user navigation trajectory according to the road network data;

[0017] An extraction unit, configured to perform multi-dimensional feature extraction on the user navigation trajectory to obtain path features of the user navigation trajectory;

[0018] A construction unit, configured to construct a sample trajectory set based on the road network data, the path features, and the user navigation trajectory;

[0019] A training unit is used to train a navigation trajectory recommendation model according to the sample trajectory set, wherein the navigation trajectory recommendation model is used to recall and recommend navigation trajectories.

[0020] According to a fourth aspect of the present disclosure, a navigation trajectory recommendation device is provided, comprising:

[0021] A second obtaining unit is used to obtain a navigation request of a user, wherein the navigation request includes a starting point and an end point;

[0022] A recall unit, configured to recall a navigation trajectory corresponding to the navigation request according to the navigation request and a pre-trained navigation trajectory recommendation model;

[0023] An output unit, used to output the navigation track corresponding to the navigation request

[0024] The navigation trajectory recommendation model is obtained based on the method described in the first aspect.

[0025] According to a fifth aspect of the present disclosure, there is provided an electronic device, including:

[0026] at least one processor; and

[0027] a memory communicatively connected to the at least one processor; wherein,

[0028] The memory stores instructions that can be executed 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 method described in the first aspect or the second aspect.

[0029] According to a sixth aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause the computer to execute the method according to the first aspect or the second aspect.

[0030] According to a seventh aspect of the present disclosure, a computer program product is provided, comprising: a computer program, wherein the computer program is stored in a readable storage medium, at least one processor of an electronic device can read the computer program from the readable storage medium, and the at least one processor executes the computer program so that the electronic device executes the method described in the first aspect or the second aspect.

[0031] According to an eighth aspect of the present disclosure, a vehicle is provided, comprising: the device as described in the fourth aspect.

[0032] According to the technical solution disclosed in the present invention, the user navigation trajectory is subjected to feature extraction from multiple dimensions to obtain a navigation trajectory recommendation model by combining road network data, path features, and user navigation trajectory training, so that the navigation trajectory recommendation model has high reliability and effectiveness, thereby improving the accuracy of the recommendation, and further improving the technical effect of the driving safety of the vehicle.

[0033] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure.

[0035] Figure 1 is a schematic diagram according to a first embodiment of the present disclosure;

[0036] Figure 2 is a schematic diagram according to a second embodiment of the present disclosure;

[0037] Figure 3 is a schematic diagram of a navigation node according to the present disclosure;

[0038] Figure 4 This is a diagram of the navigation trajectory. Figure 1 ;

[0039] Figure 5 This is a diagram of the navigation trajectory. Figure 2 ;

[0040] Figure 6 This is a diagram of the navigation trajectory. Figure 3 ;

[0041] Figure 7 This is a diagram of the navigation trajectory. Figure 4 ;

[0042] Figure 8 is a schematic diagram according to a third embodiment of the present disclosure;

[0043] Fig. 9 is a schematic diagram of an algorithm framework for the travel cost of a road network according to the present disclosure;

[0044] Fig.10 is a schematic diagram according to a fourth embodiment of the present disclosure;

[0045] Fig.11 is a schematic diagram according to a fifth embodiment of the present disclosure;

[0046] Fig.12 is a schematic diagram according to a sixth embodiment of the present disclosure;

[0047] Fig.13 is a schematic diagram according to a seventh embodiment of the present disclosure;

[0048] Fig.14 It is a block diagram of an electronic device used to implement the model training method and navigation trajectory recommendation method of the embodiments of the present disclosure. DETAILED DESCRIPTION

[0049] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0050] In order to facilitate users' travel, an electronic map can be pre-built, where the electronic map can be a directed graph obtained by abstracting the real road network. The directed graph includes nodes, and the nodes can be intersections in the real road network. The edge formed by two nodes is the navigation track between the two nodes.

[0051] The map navigation system can recall and output a navigation trajectory that meets the user's travel needs based on the electronic point map, so that the vehicle can follow the navigation path trajectory.

[0052] In some embodiments, the map navigation system can receive the starting point and end point input by the user, and determine the node corresponding to the starting point and the node corresponding to the end point on the electronic map. Based on the graph search path planning algorithm, the edge with the smallest travel cost between the two nodes determined on the electronic map (i.e., the starting point and the end point) is searched as the optimal trajectory.

[0053] Exemplarily, the method for determining the optimal trajectory may include the following steps:

[0054] The first step is to initialize the travel cost with preset dimension parameters.

[0055] Among them, the preset dimension parameter can be road grade, travel time, or travel distance.

[0056] For example, taking road grade as an example, if the road is a national highway, the road grade is high, and the toll is 90 per kilometer; if the road is a provincial highway, the road grade is medium, and the toll is 70 per kilometer.

[0057] It should be understood that the above description is only used to exemplify the possible classification of road grades and cannot be understood as a limitation on road grades. For example, in other embodiments, the road grades may also be classified based on whether the road is a highway.

[0058] The second step: Based on the probability optimization algorithm, the initial model is fine-tuned to find the optimal solution, and the edge corresponding to the travel cost that satisfies the optimal solution is determined as the optimal trajectory.

[0059] However, when the above method is used to determine the optimal trajectory, a probability-based optimization algorithm needs to be continuously traversed, resulting in high complexity. The time increases exponentially with the increase in feature difficulty, and it is difficult to introduce features of more dimensions. It is difficult to more accurately characterize road attributes and travel costs, and it is impossible to provide users with navigation trajectories that meet user preferences. The travel cost cannot be updated in time with the road network and navigation trajectory, resulting in changes in the traffic capacity of some road networks, and recommendations for users cannot be updated in time.

[0060] In order to avoid at least one of the above problems, the inventors of the present invention have creatively worked to obtain the inventive concept of the present invention: multi-dimensional feature extraction of user navigation trajectories to obtain path features, and to train a navigation trajectory recommendation model by combining road network data, path features, and user navigation trajectories.

[0061] Based on the above-mentioned inventive concept, the present disclosure provides a model training method, a navigation trajectory recommendation method, a device and a vehicle, which are applied to deep learning of map technology in computer technology, specifically to navigation technology, and can be applied to autonomous driving and intelligent transportation to achieve the reliability of navigation trajectory recommendation.

[0062] Figure 1 is a schematic diagram according to the first embodiment of the present disclosure, such as Figure 1 As shown, the model training method provided by the present disclosure includes:

[0063] S101: Acquire road network data, and determine a user navigation trajectory according to the road network data.

[0064] Exemplarily, the executor of this embodiment may be a model training device, which may be a server (including a local server and a cloud server, which may be a cloud control platform, a vehicle-road management platform, a central subsystem, an edge computing platform, a cloud computing platform, etc.), a roadside device, a terminal device, a processor, a chip, etc., which are not limited in this embodiment.

[0065] Among them, the roadside equipment includes, for example, a roadside sensing device with computing function and a roadside computing device connected to the roadside sensing device. In the intelligent transportation system architecture, the roadside equipment includes a roadside sensing device and a roadside computing device. The roadside sensing device (e.g., a roadside camera) is connected to a roadside computing device (e.g., a roadside computing unit RSCU), and the roadside computing device is connected to a server. The server can communicate with the autonomous driving or assisted driving vehicle in various ways; or, if the roadside sensing device itself includes computing function, the roadside sensing device is directly connected to the server. The above connection can be wired or wireless.

[0066] The road network in this embodiment refers to a road network in the field of transportation, which is used to define the movement trajectory of a moving object. The moving object includes an object that moves on the road, such as a person and / or a vehicle. The movement trajectory includes a movement speed and / or a movement direction.

[0067] Correspondingly, the road network data is the movement trajectory of the mobile object in the road network. It can also be understood that the road network data includes the movement trajectory of mobile objects such as people and vehicles, and the movement trajectory of the vehicle can be called the user navigation trajectory.

[0068] S102: extracting multi-dimensional features of the user's navigation trajectory to obtain path features of the user's navigation trajectory, and constructing a sample trajectory set based on the road network data, the path features, and the user's navigation trajectory.

[0069] In this embodiment, the user navigation trajectory is extracted from multiple dimensions to obtain path features for characterizing the features of the user navigation trajectory in different dimensions, thereby achieving comprehensiveness and diversity in the extraction of user navigation trajectory features.

[0070] In this embodiment, a sample trajectory set is constructed from three aspects (i.e., road network data, path characteristics, and user navigation trajectory), so that the sample trajectory set is comprehensive. Therefore, when a navigation trajectory recommendation model is trained based on the sample trajectory set, the reliability and effectiveness of the navigation trajectory recommendation model can be improved. Furthermore, when a navigation trajectory is recommended to a user based on the navigation trajectory recommendation model, the accuracy of the recommendation is improved, and the safety of the vehicle driving based on the navigation trajectory is improved, thereby satisfying the user's navigation experience.

[0071] S103: Obtain a navigation trajectory recommendation model by training according to the sample trajectory set.

[0072] Among them, the navigation trajectory recommendation model is used to recall and recommend navigation trajectories.

[0073] Based on the above analysis, it can be known that the embodiment of the present disclosure provides a model training method, including: obtaining road network data, and determining the user navigation trajectory based on the road network data, performing multi-dimensional feature extraction on the user navigation trajectory to obtain the path feature of the user navigation trajectory, and constructing a sample trajectory set based on the road network data, the path feature, and the user navigation trajectory, and obtaining a navigation trajectory recommendation model through training according to the sample trajectory set, wherein the navigation recommendation model is used to recall and recommend navigation trajectories. In this embodiment, the user navigation trajectory is subjected to feature extraction from multiple dimensions, and the technical features of the navigation trajectory recommendation model are obtained by combining the road network data, the path feature, and the user navigation trajectory training, so that the navigation trajectory recommendation model has higher reliability and effectiveness, thereby improving the accuracy of the recommendation, and further improving the technical effect of improving the driving safety of the vehicle.

[0074] Figure 2 is a schematic diagram according to the second embodiment of the present disclosure, such as Figure 2 As shown, the model training method provided by the present disclosure includes:

[0075] S201: Acquire road network data, and determine the user navigation trajectory according to the road network data.

[0076] It should be understood that in order to avoid cumbersome descriptions, the technical features of this embodiment that are the same as those in the above embodiments will not be described in detail in this embodiment.

[0077] S202: Extract features of the user navigation trajectory based on each dimension in the multiple dimensions to obtain extracted features corresponding to each dimension.

[0078] The multiple dimensions include: at least two of: a road attribute dimension, a real-time traffic condition dimension, an estimated time of arrival (ETA) dimension, and a driving information dimension.

[0079] The road attribute dimension is a dimension that characterizes the characteristics of the road, such as the road grade dimension in the above embodiment, or it can be a road size dimension, such as the width and / or length of the road.

[0080] The real-time traffic condition dimension is a dimension that characterizes the real-time traffic characteristics of the road, such as road congestion information.

[0081] The driving information dimension is a dimension that characterizes driving characteristics, such as driving speed and / or acceleration, and the type of the driving vehicle (a truck or a car, etc.).

[0082] Exemplarily, if the multi-dimensionality includes two dimensions, namely the road attribute dimension and the real-time traffic condition dimension, feature extraction is performed on the user navigation trajectory based on the road attribute dimension to obtain the extracted features corresponding to the road attribute dimension; feature extraction is performed on the user navigation trajectory based on the real-time road condition dimension to obtain the extracted features corresponding to the real-time road condition dimension.

[0083] It should be understood that the above-mentioned multi-dimensional content is only used to exemplify the dimensions that may be included and should not be understood as a limitation on the multi-dimensionality.

[0084] S203: Concatenate the extracted features corresponding to each dimension to obtain path features.

[0085] Based on the above analysis, the extracted features corresponding to the road attribute dimension and the extracted features corresponding to the real-time road condition dimension can be concatenated to obtain the path features.

[0086] In this embodiment, by combining the above dimensions to extract features of the user navigation trajectory and obtaining path features based on splicing processing, the path features can more fully and comprehensively characterize the user navigation trajectory, so that the path features have stronger characterization capabilities and reliability.

[0087] S204: Determine navigation track distribution information of the user's navigation track according to the road network data and the user's navigation track.

[0088] The navigation track distribution information is used to characterize the distribution of user navigation tracks. For example, some users have a large number of navigation tracks and are relatively densely distributed, while some users have a small number of navigation tracks and are relatively sparsely distributed.

[0089] In some embodiments, the road network data includes intersection nodes; S204 may include the following steps:

[0090] The first step: according to the intersection node and the user's navigation trajectory, obtain the navigation node of the user's navigation trajectory.

[0091] The second step: determining navigation track distribution information of the user's navigation track according to the navigation nodes of the user's navigation track.

[0092] For example, Figure 3 As shown in the figure, numbers represent intersection nodes, and link_xx represents directed edges between intersection nodes. Figure 3 In the road section shown, the starting point of the user navigation trajectory is node 4, and the end point of the user navigation trajectory is node 2. Then, according to the intersection nodes, it can be determined that the navigation nodes of the user navigation trajectory are node 4->node 2. Navigation trajectory distribution information.

[0093] like Figure 3As shown, the navigation nodes of the user's navigation trajectory may be determined to be node 4, node 5, and node 2, or may be determined to be node 4, node 1, and node 2, and the navigation trajectory distribution information of the user's navigation trajectory may be determined.

[0094] For example, if there are N user navigation trajectories in total, the distribution information of the user navigation trajectories whose navigation nodes are node 4, node 5, and node 2 (4—>node 5—>node 2) can be determined from the N user navigation trajectories; the distribution information of the user navigation trajectories whose navigation nodes are node 4, node 1, and node 2 (4—>node 1—>node 2) can also be determined from the N user navigation trajectories.

[0095] That is to say, there are multiple user navigation tracks, and for each user navigation track, the navigation track distribution information of the user navigation track is determined according to the navigation nodes of each user navigation track.

[0096] In this embodiment, by combining the intersection nodes to determine the navigation nodes of the user's navigation trajectory, and determining the navigation trajectory distribution information based on the determined navigation nodes, the navigation trajectory distribution information can be made highly consistent with the actual road network to improve the reliability and effectiveness of the navigation trajectory distribution information.

[0097] S205: Construct a sample trajectory set according to the path characteristics and navigation trajectory distribution information.

[0098] In this embodiment, by determining the navigation trajectory distribution information, a sample trajectory set is constructed by combining the navigation trajectory distribution information and the path feature, so that the sample trajectory set can characterize the distribution of the user's navigation trajectory, thereby improving the reliability and effectiveness of the sample trajectory set.

[0099] In some embodiments, S205 may include the following steps:

[0100] The first step: construct sample labels based on navigation trajectory distribution information.

[0101] Among them, the sample label is a positive sample label or a negative sample label.

[0102] In some embodiments, the number of user navigation tracks is multiple; the first step may include the following sub-steps:

[0103] The first sub-step: determining, according to the navigation track distribution information, the user navigation track with the largest number of distributions among the user navigation tracks with the same starting point and end point.

[0104] The second sub-step: construct the sample labels of the user navigation trajectories with the largest distribution as positive sample labels.

[0105] The third sub-step: constructing the sample labels of the user navigation trajectories other than the user navigation trajectories with the largest distribution as negative sample labels.

[0106] Exemplarily, in combination with the above embodiment, the user navigation trajectory 4→5→2 and the user navigation trajectory 4→1→2 have the same starting point and the same end point. If the number of distributions of the user navigation trajectory 4→5→2 is the largest, or is much larger than the number of distributions of the user navigation trajectory 4→1→2, the user navigation trajectory 4→1→2 can be determined as a negative sample, and the user navigation trajectory 4→5→2 can be determined as a positive sample. And both the user navigation trajectory 4→1→2 and the user navigation trajectory 4→5→2 have path features.

[0107] Correspondingly, after constructing a sample trajectory set, the sample trajectory set includes positive samples (f(4→1→2)=1) and negative samples (f(4→5→2)=0), where f is a path feature.

[0108] It is worth noting that the network model has model parameters. Correspondingly, the positive sample can be expressed as (w*f(4→1→2)=1) and the negative sample can be expressed as (w*f(4→5→2)=0), where w represents the model parameters.

[0109] In this embodiment, positive samples and negative samples are determined by combining the distribution quantity, and both the positive samples and negative samples have path features, so that the positive samples and negative samples can have higher accuracy and reliability.

[0110] The second step: determining the user navigation trajectory of the positive sample with path features according to the positive sample label, and determining the user navigation trajectory of the negative sample with path features according to the negative sample label.

[0111] The sample trajectory set includes user navigation trajectories of positive samples and user navigation trajectories of negative samples, and both the user navigation trajectories of positive samples and the user navigation trajectories of negative samples have path features.

[0112] Combined with the above analysis, it can be seen that the path feature can characterize the characteristics of the user's navigation trajectory from multiple dimensions, and the navigation trajectory distribution information can characterize the quantity distribution of the user's navigation trajectory. Therefore, in this embodiment, by combining the navigation trajectory distribution information and the path feature to construct a sample label, the sample label can have higher accuracy and reliability, so that when the corresponding positive samples and negative samples are determined in combination with the sample label, the determined positive samples and negative samples have higher accuracy, thereby avoiding interference between positive samples and negative samples and improving the effectiveness and reliability of training.

[0113] S206: Training a preset linear regression model according to the sample trajectory set to obtain a navigation trajectory recommendation model.

[0114] Among them, the navigation trajectory recommendation model is used to recall and recommend navigation trajectories.

[0115] It is worth noting that, in this field, a graph search algorithm is usually used to determine the shortest travel distance or the fastest travel speed (i.e., the shortest travel time) to recommend a navigation trajectory. However, in this embodiment, a linear regression model is used for model training, and a random gradient descent method can be used to solve a navigation trajectory recommendation model that can achieve the best user preference classification effect under a set of sample trajectories, so that the navigation trajectory recommendation model can recall and recommend navigation trajectories based on user preferences, thereby improving the flexibility and reliability of the recommendation.

[0116] For example, Figure 4 As shown in the figure, the starting point is X Building and the end point is Y Subway Station. If we take the shortest travel time as an example and use the graph search algorithm, the recalled navigation trajectory is as follows Figure 4 The navigation trajectory shown in FIG. 1 includes the Z highway.

[0117] However, relatively speaking, users prefer the G auxiliary road in this section. Although the road grade of the G auxiliary road is relatively low and the travel time is slightly longer, it can avoid getting on and off the highway, occasional vehicle inspections, and tolls. The navigation trajectory recommendation model obtained by training the linear regression model according to the sample trajectory set adopted in this embodiment can be recalled and recommended for users. Figure 5 The navigation trajectory shown includes the G auxiliary road.

[0118] Combined with the above analysis, it can be seen that when using a graph search algorithm, navigation trajectories are usually recommended based on features of a single dimension. However, the training of the linear regression model based on the sample trajectory set adopted in this embodiment takes relatively less time and can incorporate features of more dimensions, such as the road attribute dimension, real-time road condition dimension, estimated arrival time dimension, driving information dimension, user preference dimension, etc. in the above embodiment, so as to recall a navigation trajectory with shorter time, better road conditions, and closer to user preferences.

[0119] For example, Figure 6 As shown, the starting point is Building A and the end point is South Railway Station. If the shortest travel time is taken as an example, the recalled navigation trajectory includes the navigation trajectory of Z Expressway and the Second Ring Road using the graph search algorithm. The schematic diagram of the navigation trajectory can be found in Figure 6 Although the travel time of this navigation track is the shortest, it passes through many congested sections, and the user experience is poor.

[0120] The navigation trajectory recommendation model obtained by training the linear regression model according to the sample trajectory set adopted in this embodiment can be used to recommend relatively smoother, more convenient, and more preferred routes for users. Figure 7 The navigation track shown.

[0121] In some embodiments, S206 may include the following steps:

[0122] The first step: perform multiple rounds of model training on the linear regression model according to the sample trajectory set to obtain the network model corresponding to each round of model training.

[0123] The number of training rounds for the linear regression model may be set based on demand, historical records, and experiments, and is not limited in this embodiment.

[0124] The second step: verify each network model based on the preset verification set to obtain the ratio between the accuracy and recall rate of each network model.

[0125] The validation set can be understood as a sample navigation trajectory of the user to verify the accuracy and / or reliability of the network model.

[0126] In this embodiment, for each network model, the accuracy and recall of the network model can be determined using a validation set to calculate the quotient between the accuracy and recall of the network model.

[0127] The third step: from each ratio, obtain the ratio closest to 1, and determine the network model corresponding to the obtained ratio as the navigation trajectory recommendation model.

[0128] Relatively speaking, the closer the ratio is to 1, the stronger the balance between accuracy and recall is, that is, the higher the reliability of the corresponding network model, and the higher the accuracy and recall effectiveness are, the more likely this network model is to be used as the navigation trajectory recommendation model, thereby improving the accuracy and reliability of the navigation trajectory recommendation model.

[0129] Figure 8 is a schematic diagram according to the third embodiment of the present disclosure, Figure 8 As shown, the navigation trajectory recommendation method provided by the present disclosure includes:

[0130] S801: Obtain a navigation request from a user, wherein the navigation request includes a starting point and an end point.

[0131] Exemplarily, the executor of this embodiment is a navigation trajectory recommendation device (hereinafter referred to as the recommendation device), which can be the same device as the model training device or a different device, which is not limited in this embodiment.

[0132] In some embodiments, the user may input a navigation request based on a user device, and the user device may be a mobile user device such as a mobile phone, or may be an in-vehicle terminal installed in a vehicle.

[0133] For example, taking the vehicle-mounted terminal as an example, the vehicle-mounted terminal includes a display device, and the user can input the navigation request by touching the screen. Alternatively, the vehicle-mounted terminal includes a sound pickup device, such as a microphone, and the user can input the navigation request by voice.

[0134] S802: According to the navigation request and the pre-trained navigation trajectory recommendation model, recall and output the navigation trajectory corresponding to the navigation request.

[0135] The navigation trajectory recommendation model is obtained based on the model training method described in any of the above embodiments.

[0136] In some embodiments, S802 may include the following steps:

[0137] The first step: determine the route to be navigated based on the starting point and the end point.

[0138] The road section to be navigated refers to the road section that needs to be navigated, that is, the road section between the starting point and the end point.

[0139] The second step: extracting the path features of the road section to be navigated, and determining at least one suspected navigation track according to the road section to be navigated.

[0140] Similarly, in order to avoid cumbersome statements, the present embodiment will no longer limit the technical features that are the same as those of the above-mentioned embodiments.

[0141] The suspected navigation track is a possible navigation track from the starting point to the end point. There are generally multiple suspected navigation tracks, that is, the vehicle can use different navigation tracks to travel from the starting point to the end point.

[0142] Step 3: Recall and output the navigation trajectory corresponding to the navigation request according to the navigation trajectory recommendation model, each suspected navigation trajectory, and the path characteristics of the to-be-navigated road section.

[0143] In some embodiments, the third step may include the following sub-steps:

[0144] The first sub-step: obtaining model parameters of the navigation trajectory recommendation model.

[0145] The second sub-step: determining the travel cost of each suspected navigation trajectory according to the model parameters and the path characteristics of the road section to be navigated.

[0146] Among them, the passage cost can be understood as the passage cost of the suspected navigation trajectory. The higher the passage cost of the suspected navigation trajectory, the higher the passage cost of the suspected navigation trajectory. Conversely, the lower the passage cost of the suspected navigation trajectory, the smaller the passage cost of the suspected navigation trajectory.

[0147] For example, combined with the above analysis, if the path feature of the navigation section is f, one suspected navigation trajectory is (4→1→2), and the other suspected navigation trajectory is (4→5→2), and the model parameter is w, then the linear regression method can be used to calculate the passage cost w*f(4→1→2) of the suspected navigation trajectory (4→1→2), and the passage cost w*f(4→5→2) of the suspected navigation trajectory (4→1→2).

[0148] The third sub-step: determining the minimum travel cost from the travel costs, and determining the suspected navigation trajectory corresponding to the minimum travel cost as the navigation trajectory corresponding to the navigation request.

[0149] Fourth sub-step: outputting the navigation track corresponding to the navigation request.

[0150] Exemplarily, if w*f(4→1→2)>w*f(4→5→2), the navigation track (4→5→2) is recalled, that is, the navigation track (4→5→2) is the navigation track corresponding to the navigation request.

[0151] Combined with the above analysis, in this embodiment, the algorithm framework of the road network travel cost includes two parts, namely: Fig. 9 The offline module and the online module shown, wherein the offline module is used for training the navigation trajectory recommendation model, and the online module is used for applying the navigation trajectory recommendation model.

[0152] like Fig. 9 As shown, the offline module may include:

[0153] The user navigation trajectory processing submodule is used to perform feature extraction and determine the distribution information of the navigation trajectory as in the above embodiment.

[0154] The sample trajectory collection submodule is used to execute the steps of constructing sample labels and constructing sample trajectory collections as described in the above embodiment.

[0155] The training submodule is used to perform multiple rounds of training as in the above embodiment.

[0156] The parameter selection submodule is used to execute the navigation trajectory recommendation model including model parameters determined based on multiple rounds of training as in the above embodiment.

[0157] like Fig. 9 As shown, online modules may include:

[0158] The feature construction submodule is used to execute the extraction of path features of the to-be-navigated road section in the above embodiment, etc.

[0159] The travel cost calculation submodule is used to perform the determination of the travel cost of each suspected navigation trajectory as in the above embodiment.

[0160] The path planning submodule is used to execute the recall based on each travel cost and output the navigation trajectory corresponding to the navigation request as in the above embodiment.

[0161] Fig.10 is a schematic diagram according to a fourth embodiment of the present disclosure, Fig.10 As shown, the model training device 1000 provided by the present disclosure includes:

[0162] The first acquisition unit 1001 is used to acquire road network data.

[0163] The determination unit 1002 is used to determine the user navigation trajectory according to the road network data.

[0164] The extraction unit 1003 is used to perform multi-dimensional feature extraction on the user navigation trajectory to obtain the path feature of the user navigation trajectory.

[0165] The construction unit 1004 is used to construct a sample trajectory set based on the road network data, path characteristics, and user navigation trajectory.

[0166] The training unit 1005 is used to train a navigation trajectory recommendation model according to the sample trajectory set, wherein the navigation trajectory recommendation model is used to recall and recommend navigation trajectories.

[0167] Fig.11 is a schematic diagram according to a fifth embodiment of the present disclosure, Fig.11 As shown, the model training device 1100 provided by the present disclosure includes:

[0168] The first acquisition unit 1101 is used to acquire road network data.

[0169] The determination unit 1102 is used to determine the user navigation trajectory according to the road network data.

[0170] The extraction unit 1103 is used to perform multi-dimensional feature extraction on the user navigation trajectory to obtain the path features of the user navigation trajectory.

[0171] In some embodiments, the multiple dimensions include: at least two of the road attribute dimension, the real-time traffic condition dimension, the estimated arrival time dimension, and the driving information dimension; Fig.11 It can be seen that the extraction unit 1103 includes:

[0172] The extraction subunit 11031 is used to extract features of the user navigation trajectory based on each dimension in the multiple dimensions to obtain the extracted features corresponding to each dimension.

[0173] The concatenation subunit 11032 is used to concatenate the extracted features corresponding to each dimension to obtain path features.

[0174] The construction unit 1104 is used to construct a sample trajectory set based on the road network data, path characteristics, and user navigation trajectory.

[0175] Combination Fig.11 It can be seen that, in some embodiments, the construction unit 1104 includes:

[0176] The first determining subunit 11041 is used to determine navigation track distribution information of the user's navigation track according to the road network data and the user's navigation track.

[0177] In some embodiments, the road network data includes intersection nodes; the first determining subunit 11041 includes:

[0178] The first acquisition module is used to acquire the navigation node of the user navigation trajectory according to the intersection node and the user navigation trajectory.

[0179] The first determining module is used to determine navigation track distribution information of the user's navigation track according to the navigation nodes of the user's navigation track.

[0180] In some embodiments, there are multiple user navigation trajectories; the first determination module is used to determine, for each user navigation trajectory, navigation trajectory distribution information of the user navigation trajectory according to the navigation nodes of each user navigation trajectory.

[0181] The construction subunit 11042 is used to construct a sample trajectory set according to the path characteristics and navigation trajectory distribution information.

[0182] In some embodiments, construct subunit 11042 includes:

[0183] The construction module is used to construct a sample label according to the navigation trajectory distribution information, wherein the sample label is a positive sample label or a negative sample label.

[0184] In some embodiments, the number of user navigation tracks is multiple; the building module includes:

[0185] The determination submodule is used to determine the user navigation track with the largest number of distributions among the user navigation tracks with the same starting point and end point according to the navigation track distribution information.

[0186] The first construction submodule is used to construct the sample labels of the user navigation trajectories with the largest distribution number as positive sample labels.

[0187] The second construction submodule is used to construct the sample labels of the user navigation trajectories other than the user navigation trajectories with the largest distribution number as negative sample labels.

[0188] The second determination module is used to determine the user navigation trajectory of the positive sample with path characteristics according to the positive sample label, and determine the user navigation trajectory of the negative sample with path characteristics according to the negative sample label, wherein the sample trajectory set includes the user navigation trajectory of the positive sample and the user navigation trajectory of the negative sample.

[0189] The training unit 1105 is used to train a navigation trajectory recommendation model according to the sample trajectory set, wherein the navigation trajectory recommendation model is used to recall and recommend navigation trajectories.

[0190] In some embodiments, the training unit 1105 is used to train a preset linear regression model according to the sample trajectory set to obtain a navigation trajectory recommendation model.

[0191] Combination Fig.11 It can be seen that, in some embodiments, the training unit 1105 includes:

[0192] The training subunit 11051 is used to perform multiple rounds of model training on the linear regression model according to the sample trajectory set to obtain the network model corresponding to each round of model training.

[0193] The verification subunit 11052 is used to verify each network model based on a preset verification set to obtain the ratio between the accuracy and recall of each network model.

[0194] The acquisition subunit 11053 is used to acquire the ratio closest to 1 from among the ratios.

[0195] The second determining subunit 11054 is configured to determine the network model corresponding to the acquired ratio as the navigation trajectory recommendation model.

[0196] Fig.12 is a schematic diagram according to a sixth embodiment of the present disclosure, Fig.12 As shown, the navigation trajectory recommendation device 1200 provided by the present disclosure includes:

[0197] The second obtaining unit 1201 is used to obtain a navigation request from a user, wherein the navigation request includes a starting point and an end point.

[0198] The recall unit 1202 is used to recall the navigation trajectory corresponding to the navigation request according to the navigation request and the pre-trained navigation trajectory recommendation model.

[0199] Combination Fig.12 It can be seen that, in some embodiments, the recall unit 1202 includes:

[0200] The third determining subunit 12021 is used to determine the road section to be navigated according to the starting point and the end point.

[0201] The extraction subunit 12022 is used to extract the path features of the road section to be navigated.

[0202] The fourth determining subunit 12023 is used to determine at least one suspected navigation trajectory according to the road section to be navigated.

[0203] The recall subunit 12024 is used to recall the navigation trajectory corresponding to the navigation request according to the navigation trajectory recommendation model, each suspected navigation trajectory, and the path characteristics of the to-be-navigated road section.

[0204] In some embodiments, the recall subunit 12024 includes:

[0205] The second acquisition module is used to acquire model parameters of the navigation trajectory recommendation model.

[0206] The third determination module is used to determine the travel cost of each suspected navigation trajectory according to the model parameters and the path characteristics of the to-be-navigated road section.

[0207] The fourth determination module is used to determine the minimum travel cost from the travel costs, and determine the suspected navigation trajectory corresponding to the minimum travel cost as the navigation trajectory corresponding to the navigation request.

[0208] The output unit 1203 is used to output the navigation track corresponding to the navigation request.

[0209] The navigation trajectory recommendation model is obtained based on the model training method described in any of the above embodiments.

[0210] According to another aspect of the embodiments of the present disclosure, the embodiments of the present disclosure further provide a vehicle, comprising: a navigation trajectory recommendation device as described in any of the above embodiments.

[0211] Fig.13 is a schematic diagram according to the seventh embodiment of the present disclosure, Fig.13 As shown, the electronic device 1300 in the present disclosure may include: a processor 1301 and a memory 1302 .

[0212] The memory 1302 is used to store programs. The memory 1302 may include volatile memory, such as random-access memory (RAM), such as static random-access memory (SRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), etc. The memory may also include non-volatile memory, such as flash memory. The memory 1302 is used to store computer programs (such as applications, functional modules, etc. that implement the above methods), computer instructions, etc. The above computer programs, computer instructions, etc. may be partitioned and stored in one or more memories 1302. And the above computer programs, computer instructions, data, etc. may be called by the processor 1301.

[0213] The above-mentioned computer programs, computer instructions, etc. may be stored in partitions in one or more memories 1302 . And the above-mentioned computer programs, computer instructions, etc. may be called by the processor 1301 .

[0214] The processor 1301 is used to execute the computer program stored in the memory 1302 to implement each step of the method involved in the above embodiment.

[0215] For details, please refer to the relevant description in the previous method embodiment.

[0216] The processor 1301 and the memory 1302 may be independent structures or integrated structures. When the processor 1301 and the memory 1302 are independent structures, the memory 1302 and the processor 1301 may be coupled and connected via a bus 1303 .

[0217] The electronic device of this embodiment can execute the technical solution in the above method, and its specific implementation process and technical principle are the same, which will not be repeated here.

[0218] In the technical solution of the present disclosure, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0219] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

[0220] According to an embodiment of the present disclosure, the present disclosure also provides a computer program product, which includes: a computer program, the computer program is stored in a readable storage medium, at least one processor of an electronic device can read the computer program from the readable storage medium, and at least one processor executes the computer program so that the electronic device executes the solution provided by any of the above embodiments.

[0221] Fig.14 A schematic block diagram of an example electronic device 1400 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0222] like Fig.14 As shown, the device 1400 includes a computing unit 1401, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 1402 or a computer program loaded from a storage unit 1408 into a random access memory (RAM) 1403. In the RAM 1403, various programs and data required for the operation of the device 1400 can also be stored. The computing unit 1401, the ROM 1402, and the RAM 1403 are connected to each other via a bus 1404. An input / output (I / O) interface 1405 is also connected to the bus 1404.

[0223] A number of components in the device 1400 are connected to the I / O interface 1405, including: an input unit 1406, such as a keyboard, a mouse, etc.; an output unit 1407, such as various types of displays, speakers, etc.; a storage unit 1408, such as a disk, an optical disk, etc.; and a communication unit 1409, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 1409 allows the device 1400 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0224] The computing unit 1401 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 1401 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 1401 performs the various methods and processes described above, such as a model training method and a method for recommending a navigation trajectory. For example, in some embodiments, the model training method and the method for recommending a navigation trajectory may be implemented as a computer software program, which is tangibly included in a machine-readable medium, such as a storage unit 1408. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 1400 via the ROM 1402 and / or the communication unit 1409. When the computer program is loaded into the RAM 1403 and executed by the computing unit 1401, one or more steps of the model training method and the method for recommending a navigation trajectory described above may be executed. Alternatively, in other embodiments, the computing unit 1401 may be configured to execute the model training method and the navigation trajectory recommendation method in any other appropriate manner (eg, by means of firmware).

[0225] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0226] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0227] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0228] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0229] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0230] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship between the client and the server is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services ("Virtual Private Server", or "VPS" for short). The server may also be a server of a distributed system, or a server combined with a blockchain.

[0231] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0232] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A model training method, comprising: Acquire road network data, and determine the user navigation trajectory according to the road network data; Performing multi-dimensional feature extraction on the user navigation trajectory to obtain path features of the user navigation trajectory, and constructing a sample trajectory set based on the road network data, the path features, and the user navigation trajectory; training a navigation trajectory recommendation model according to the sample trajectory set, wherein the navigation trajectory recommendation model is used to recall and recommend navigation trajectories; The number of the user navigation trajectories is multiple, and before constructing the sample trajectory set, the method further includes: Determine, according to the navigation track distribution information, the user navigation track with the largest number of distributions among the user navigation tracks having the same starting point and end point, wherein the navigation track distribution information is determined according to the road network data and the user navigation tracks; The sample labels of the user navigation trajectories with the largest distribution number are constructed as positive sample labels.

2. The method according to claim 1, wherein: Constructing a sample trajectory set based on the road network data, the path features, and the user navigation trajectory, including: Determining navigation track distribution information of the user navigation track according to the road network data and the user navigation track; The sample trajectory set is constructed according to the path characteristics and the navigation trajectory distribution information.

3. The method according to claim 2, wherein: The road network data includes intersection nodes; determining navigation track distribution information of the user navigation track according to the road network data and the user navigation track, including: According to the intersection node and the user navigation trajectory, obtaining the navigation node of the user navigation trajectory; Navigation track distribution information of the user navigation track is determined according to the navigation nodes of the user navigation track.

4. The method according to claim 3, wherein: The number of the user navigation tracks is multiple; and determining the navigation track distribution information of the user navigation track according to the navigation nodes of the user navigation track includes: For each user navigation track, navigation track distribution information of the user navigation track is determined according to the navigation nodes of each user navigation track.

5. The method according to any one of claims 2 to 4, wherein: The sample label is a positive sample label or a negative sample label, and the sample trajectory set is constructed according to the path feature and the navigation trajectory distribution information, including: A user navigation trajectory of a positive sample having the path feature is determined according to the positive sample label, and a user navigation trajectory of a negative sample having the path feature is determined according to the negative sample label, wherein the sample trajectory set includes the user navigation trajectory of the positive sample and the user navigation trajectory of the negative sample.

6. The method according to claim 1, after determining, according to the navigation track distribution information, the user navigation track with the largest number of distributions among the user navigation tracks with the same starting point and end point, the method further comprises: The sample labels of the user navigation trajectories other than the user navigation trajectories with the largest distribution number are constructed as negative sample labels.

7. The method according to any one of claims 1 to 4, wherein: The multiple dimensions include: at least two of a road attribute dimension, a real-time traffic condition dimension, an estimated arrival time dimension, and a driving information dimension; performing multi-dimensional feature extraction on the user navigation trajectory to obtain path features of the user navigation trajectory includes: Extracting features of the user navigation trajectory based on each dimension of the multiple dimensions to obtain extracted features corresponding to each dimension; The extracted features corresponding to each dimension are concatenated to obtain the path feature.

8. The method according to any one of claims 1 to 4, wherein: The navigation trajectory recommendation model is obtained by training the sample trajectory set, including: A preset linear regression model is trained according to the sample trajectory set to obtain the navigation trajectory recommendation model.

9. The method according to claim 8, wherein: The preset linear regression model is trained according to the sample trajectory set to obtain the navigation trajectory recommendation model, including: Performing multiple rounds of model training on the linear regression model according to the sample trajectory set to obtain a network model corresponding to each round of model training; Each network model is verified based on the preset verification set to obtain the ratio between the accuracy and recall of each network model; From the ratios, a ratio closest to 1 is obtained, and a network model corresponding to the obtained ratio is determined as the navigation trajectory recommendation model.

10. A method for recommending a navigation trajectory, comprising: Obtaining a user's navigation request, wherein the navigation request includes a starting point and an end point; Recalling and outputting a navigation trajectory corresponding to the navigation request according to the navigation request and a pre-trained navigation trajectory recommendation model; Wherein, the navigation trajectory recommendation model is obtained based on the method according to any one of claims 1-9.

11. The method according to claim 10, wherein: Recalling and outputting a navigation trajectory corresponding to the navigation request according to the navigation request and a pre-trained navigation trajectory recommendation model, including: Determine the road section to be navigated according to the starting point and the end point; Extracting the path features of the road section to be navigated, and determining at least one suspected navigation track according to the road section to be navigated; The navigation trajectory corresponding to the navigation request is recalled and outputted according to the navigation trajectory recommendation model, each suspected navigation trajectory, and the path characteristics of the to-be-navigated road section.

12. The method according to claim 11, wherein: Recalling and outputting a navigation trajectory corresponding to the navigation request according to the navigation trajectory recommendation model, each suspected navigation trajectory, and the path characteristics of the to-be-navigated road section, including: Obtaining model parameters of the navigation trajectory recommendation model; Determining the travel cost of each suspected navigation trajectory according to the model parameters and the path characteristics of the to-be-navigated road section; Determining a minimum travel cost from among the travel costs, and determining a suspected navigation trajectory corresponding to the minimum travel cost as the navigation trajectory corresponding to the navigation request; A navigation track corresponding to the navigation request is output.

13. A model training device, comprising: A first acquisition unit, used for acquiring road network data; A determination unit, configured to determine a user navigation trajectory according to the road network data; An extraction unit, configured to perform multi-dimensional feature extraction on the user navigation trajectory to obtain path features of the user navigation trajectory; A construction unit, configured to construct a sample trajectory set based on the road network data, the path features, and the user navigation trajectory; A training unit, configured to obtain a navigation trajectory recommendation model by training the sample trajectory set, wherein the navigation trajectory recommendation model is used to recall and recommend navigation trajectories; The number of the user navigation tracks is multiple, the construction unit includes a construction subunit, the construction subunit includes a construction module for constructing a sample label, and the construction module includes: a determination submodule, configured to determine, according to the navigation track distribution information, the user navigation track with the largest number of distributions among the user navigation tracks having the same starting point and end point, wherein the navigation track distribution information is determined according to the road network data and the user navigation tracks; The first construction submodule is used to construct the sample labels of the user navigation trajectories with the largest distribution number as positive sample labels.

14. The device according to claim 13, wherein: The construction unit further includes a first determining subunit: The first determining subunit is used to determine the navigation track distribution information of the user navigation track according to the road network data and the user navigation track; The construction subunit is used to construct the sample trajectory set according to the path characteristics and the navigation trajectory distribution information.

15. The device according to claim 14, wherein: The road network data includes intersection nodes; the first determining subunit includes: A first acquisition module, used for acquiring a navigation node of the user navigation trajectory according to the intersection node and the user navigation trajectory; The first determining module is used to determine the navigation track distribution information of the user navigation track according to the navigation nodes of the user navigation track.

16. The device according to claim 15, wherein: There are multiple user navigation tracks; the first determining module is used to determine the navigation track distribution information of each user navigation track according to the navigation nodes of each user navigation track.

17. The device according to any one of claims 14 to 16, wherein: The sample label is a positive sample label or a negative sample label, and the construction subunit further includes a second determination module: The second determination module is used to determine the user navigation trajectory of the positive sample with the path feature according to the positive sample label, and determine the user navigation trajectory of the negative sample with the path feature according to the negative sample label, wherein the sample trajectory set includes the user navigation trajectory of the positive sample and the user navigation trajectory of the negative sample.

18. The apparatus according to claim 13, wherein the building module further comprises: The second construction submodule is used to construct the sample labels of the user navigation trajectories other than the user navigation trajectories with the largest distribution number as negative sample labels.

19. The device according to any one of claims 13 to 16 and 18, wherein: The multiple dimensions include: at least two of: a road attribute dimension, a real-time traffic condition dimension, an estimated arrival time dimension, and a driving information dimension; the extraction unit includes: An extraction subunit, configured to extract features of the user navigation trajectory based on each dimension of the multiple dimensions to obtain extraction features corresponding to each dimension; The splicing subunit is used to splice the extracted features corresponding to each dimension to obtain the path feature.

20. The device according to any one of claims 13 to 16 and 18, wherein: The training unit is used to train a preset linear regression model according to the sample trajectory set to obtain the navigation trajectory recommendation model.

21. The device according to claim 20, wherein: The training unit comprises: A training subunit, used to perform multiple rounds of model training on the linear regression model according to the sample trajectory set to obtain a network model corresponding to each round of model training; A verification subunit, used to verify each network model based on a preset verification set to obtain the ratio between the accuracy and recall of each network model; An acquisition subunit, used for acquiring a ratio closest to 1 from among the ratios; The second determining subunit is configured to determine the network model corresponding to the acquired ratio as the navigation trajectory recommendation model.

22. A navigation trajectory recommendation device, comprising: A second obtaining unit is used to obtain a navigation request of a user, wherein the navigation request includes a starting point and an end point; A recall unit, configured to recall a navigation trajectory corresponding to the navigation request according to the navigation request and a pre-trained navigation trajectory recommendation model; An output unit, configured to output a navigation track corresponding to the navigation request; Wherein, the navigation trajectory recommendation model is obtained based on the method according to any one of claims 1-9.

23. The device according to claim 22, wherein: The recall unit comprises: A third determining subunit, configured to determine a road section to be navigated according to the starting point and the end point; An extraction subunit, used for extracting the path features of the road section to be navigated; A fourth determining subunit, configured to determine at least one suspected navigation trajectory according to the to-be-navigated road section; The recall subunit is used to recall the navigation trajectory corresponding to the navigation request according to the navigation trajectory recommendation model, each suspected navigation trajectory, and the path characteristics of the to-be-navigated road section.

24. The device according to claim 23, wherein: The recall subunit comprises: A second acquisition module is used to acquire model parameters of the navigation trajectory recommendation model; A third determination module is used to determine the travel cost of each suspected navigation trajectory according to the model parameters and the path characteristics of the to-be-navigated road section; The fourth determination module is used to determine the minimum travel cost from the various travel costs, and determine the suspected navigation trajectory corresponding to the minimum travel cost as the navigation trajectory corresponding to the navigation request.

25. An electronic device comprising: at least one processor; as well as 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 so that the at least one processor can execute the method described in any one of claims 1 to 9; or so that the at least one processor can execute the method described in any one of claims 10 to 12.

26. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-9; or, the computer instructions are used to cause the computer to execute the method according to any one of claims 10-12.

27. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements the steps of the method according to any one of claims 1 to 9; or, when the computer program is executed by a processor, the computer program implements the steps of the method according to any one of claims 10 to 12.

28. A vehicle comprising: A device as claimed in any one of claims 22 to 24.

Images