A lane feature processing method and device

Abstract

The embodiment of the present invention relates to a lane feature processing method and device, the method comprising: obtaining a high-precision map; constructing a first directed graph; selecting any two first nodes on the first directed graph as the start and end nodes; searching for the connected paths between the two to generate a first path set; counting the total number of node edges of each first path in the first path set to generate the total number of first node edges; and deleting the first node edges whose total number exceeds the first total number threshold as redundant paths to obtain a second path set; counting the total number of paths in the second path set to generate the first total number; and recording each first path in the set as a corresponding second path; encoding the path feature of the second path based on a long short-term memory network; and identifying the lane feature from the start node to the end node; and adding the first lane feature as a supplementary feature of the start node to the first directed graph. The present invention can optimize the search efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method and device for processing lane features. Background Art

[0002] A common prediction method for the trajectory prediction module of the autonomous driving system is to convert the HD map into a lane graph and perform trajectory prediction based on the converted lane graph. The essence of a conventional lane graph is a heterogeneous directed graph consisting of nodes and node edges. Each node corresponds to a lane, and the node input features of each node come from the high-precision map semantic features of the corresponding lane (such as orientation, position, centerline and centerline sampling points, edge line and edge line sampling points, length, width, road curvature, etc.). Each node edge connects two nodes, and the node edge features of each node edge are related to some high-precision map semantic features of the two adjacent lanes and the lane connectivity relationship (such as front-to-back connectivity, left-to-right connectivity, one-way connectivity, etc.). It is not difficult to see from the above description of graph elements that, in a conventional lane graph, there is no lane feature used to characterize the connectivity between every two lanes, or between every two nodes, except for two adjacent nodes with node edges. Therefore, when searching for a path between two points, one can only start from the starting node and search step by step along each adjacent node. The search method is single, the search calculation is large, and the search efficiency is not ideal. Summary of the invention

[0003] The purpose of the present invention is to provide a lane feature processing method, device, electronic device and computer-readable storage medium to address the defects of the prior art; improve the conventional lane map, while retaining all the nodes and node edge features of the conventional lane map, add a set of lane features Y for each node that can reflect the connectivity relationship between the current node and other non-directly connected nodes. Through the present invention, the overall connectivity between every two lanes, i.e., every two nodes, can be confirmed on the lane map based on the lane feature Y. The path search based on the lane map with the node lane feature Y output by the present invention can not only support the traditional node-by-node search method, but also support the jump search method with the node step length, so as to achieve the purpose of diversifying the search method, reducing the search calculation amount, and optimizing the search efficiency.

[0004] To achieve the above object, a first aspect of an embodiment of the present invention provides a method for processing lane features, the method comprising:

[0005] Acquire a high-precision map; the high-precision map includes a plurality of first roads; the first road includes a plurality of first lanes;

[0006] A directed graph is constructed according to the high-precision map to generate a corresponding first directed graph; the first directed graph includes a node set V and a node edge set E; the node set V includes a plurality of first nodes v; the first nodes v correspond to the first lanes one by one; the node edge set E includes a plurality of first node edges e; each of the first node edges e connects two of the first nodes v; each of the first nodes v corresponds to a first lane input feature x v ; Each of the first node edges e corresponds to a first node edge feature x e ;

[0007] Select any two of the first nodes v on the first directed graph as corresponding starting nodes v start and the end node v end ; and from the starting node v start To the end node v end Searching for the connected path between the two in the direction of to generate a corresponding first path set; if the first path set is not empty, counting the total number of node edges of each first path in the first path set to generate a corresponding first node edge total number; and counting the total number of the first node edges in the first path set that exceeds a preset first total number threshold N max The first path is deleted as a redundant path to obtain a corresponding second path set; the first path set includes multiple first paths; the first path includes one or more first node edges e;

[0008] If the second path set is not empty, the total number of the first paths in the second path set is counted to generate a corresponding first total number M; and each of the first paths in the second path set is recorded as a corresponding second path P i , 1≤i≤M; and each of the second paths P i The total number of the first node edges is recorded as the corresponding total number of the first node edges N i , 1≤N i ≤N max ;

[0009] Based on the preset long short-term memory network, according to each of the second paths P i The first node edge feature x of all the first node edges e e For the second path P i The path feature is encoded to generate the corresponding first feature code Φ i ; and according to the first path characteristic Φ of the first total number M obtained i and the end node v end The first lane input feature x v For the starting node v startTo the end node v end The lane features are identified to generate the corresponding first lane features And the first lane feature As the starting node v start A supplementary feature is added to the first directed graph.

[0010] Preferably, each of the first lanes corresponds to a first lane semantic set.

[0011] Preferably, the step of constructing a directed graph according to the high-precision map to generate a corresponding first directed graph specifically includes:

[0012] Construct the corresponding first node v with each first lane in the high-precision map; and input the first lane feature x of the corresponding first node v according to the first lane semantic set of each first lane. v Set; and form the corresponding node set V by all the first nodes v obtained;

[0013] Every two first lanes in the high-precision map that can be connected longitudinally or transversely are taken as the corresponding first lane group; the two first nodes v corresponding to the first lane group are taken as the corresponding first node pair; and the corresponding first node edge e is constructed for the first node pair; and the connection direction of the corresponding first node edge e is set according to the connection direction of the two first lanes in the first lane group; and the first node edge feature x of the corresponding first node edge e is set according to the connection direction of the two first lanes in the first lane group and the two first lane semantic sets. e ; and all the first node edges e obtained constitute the corresponding node edge set E;

[0014] The obtained node set V and the node edge set E constitute the corresponding first directed graph.

[0015] Preferably, the preset long short-term memory network is based on each of the second paths P i The first node edge feature x of all the first node edges e e For the second path P i The path feature is encoded to generate the corresponding first feature code Φ i , specifically including:

[0016] According to the starting point v start To the end node v end direction, the second path P i The total number of edges of the first node N iThe first node edge feature x of the first node edge e e Sort and generate the corresponding first feature sequence {x e,k}, 1≤k≤N i ; and for the first feature sequence {x e,k} each node edge feature x e,k Convert the encoding format according to the one-hot encoding format;

[0017] The first feature sequence {x e,k} Input the long short-term memory network to calculate and generate the corresponding first path feature Φ i ,

[0018] Φ i =LSTM({x e,k}),

[0019] LSTM() is the model function of the long short-term memory network.

[0020] Preferably, the first path characteristic Φ obtained according to the first total number M i and the end node v end The first lane input feature x v For the starting node v start To the end node v end The lane features are identified to generate the corresponding first lane features Specifically include:

[0021] According to the first path feature Φ obtained from the first total number M i The attention intensity is calculated to generate the corresponding first attention intensity Ψ(v start ,v end ),

[0022]

[0023] The end node v end The first lane input feature x v As the corresponding end lane input feature

[0024] According to the first attention strength Ψ(v start ,v end ) and the end lane input feature Calculate and generate the corresponding first lane feature

[0025]

[0026] Preferably, the method further comprises:

[0027] If the first or second path set is empty, set the first lane feature The preset invalid lane feature Y * , and the first lane feature of the current As the starting node v start A supplementary feature is added to the first directed graph.

[0028] Preferably, all the first lane features of any first node v on the first directed graph After all the additions are completed, each of the first nodes v corresponds to one of the first lane input features x v In addition, it also corresponds to one or more first lane features Y v (v'); v' is another first node, and the number of the first node edges e passing from the current first node v to the first node v' does not exceed the first total threshold value N max .

[0029] A second aspect of an embodiment of the present invention provides a device for implementing the lane feature processing method described in the first aspect, the device comprising: an acquisition module, a first directed graph processing module, a second directed graph processing module, a third directed graph processing module and a lane feature processing module;

[0030] The acquisition module is used to acquire a high-precision map; the high-precision map includes a plurality of first roads; the first road includes a plurality of first lanes;

[0031] The first directed graph processing module is used to construct a directed graph according to the high-precision map to generate a corresponding first directed graph; the first directed graph includes a node set V and a node edge set E; the node set V includes a plurality of first nodes v; the first nodes v correspond to the first lanes one by one; the node edge set E includes a plurality of first node edges e; each of the first node edges e connects two of the first nodes v; each of the first nodes v corresponds to a first lane input feature x v ; Each of the first node edges e corresponds to a first node edge feature x e ;

[0032] The second directed graph processing module is used to select any two of the first nodes v on the first directed graph as corresponding starting nodes v start and end node v end ; and from the starting node v start To the end node v endSearching for the connected path between the two in the direction of to generate a corresponding first path set; if the first path set is not empty, counting the total number of node edges of each first path in the first path set to generate a corresponding first node edge total number; and counting the total number of the first node edges in the first path set that exceeds a preset first total number threshold N max The first path is deleted as a redundant path to obtain a corresponding second path set; the first path set includes multiple first paths; the first path includes one or more first node edges e;

[0033] The third directed graph processing module is used to, when the second path set is not, count the total number of the first paths in the second path set to generate a corresponding first total number M; and record each of the first paths in the second path set as a corresponding second path P i , 1≤i≤M; and each of the second paths P i The total number of the first node edges is recorded as the corresponding total number of the first node edges N i , 1≤N i ≤N max ;

[0034] The lane feature processing module is used to process the lane feature according to each of the second paths P based on a preset long short-term memory network. i The first node edge feature x of all the first node edges e e For the second path P i The path feature is encoded to generate the corresponding first feature code Φ i ; and according to the first path characteristic Φ of the first total number M obtained i and the end node v end The first lane input feature x v For the starting node v start To the end node v end The lane features are identified to generate the corresponding first lane features And the first lane feature As the starting node v start A supplementary feature is added to the first directed graph.

[0035] A third aspect of an embodiment of the present invention provides an electronic device, including: a memory, a processor, and a transceiver;

[0036] The processor is used to be coupled to the memory, read and execute instructions in the memory, so as to implement the method steps described in the first aspect above;

[0037] The transceiver is coupled to the processor, and the processor controls the transceiver to send and receive messages.

[0038] A fourth aspect of an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions. When the computer instructions are executed by a computer, the computer executes the instructions of the method described in the first aspect above.

[0039] The embodiment of the present invention provides a lane feature processing method, device, electronic device and computer-readable storage medium; improves the conventional lane map, while retaining all nodes and node edge features of the conventional lane map, adds a set of lane features Y for each node that can reflect the connectivity relationship between the current node and other non-directly connected nodes. Through the present invention, the overall connectivity between every two lanes, that is, every two nodes, can be confirmed on the lane map based on the lane feature Y. The path search based on the lane map with the node lane feature Y output by the present invention can not only support the traditional node-by-node search method, but also support the jump search method with the node step length, so that the search method becomes diversified, the search calculation amount is reduced, and the search efficiency is optimized. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 A schematic diagram of a lane feature processing method provided in Embodiment 1 of the present invention;

[0041] Figure 2 A module structure diagram of a lane feature processing device provided in Embodiment 2 of the present invention;

[0042] Figure 3 A schematic diagram of the structure of an electronic device provided in Embodiment 3 of the present invention. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] Embodiment 1 of the present invention provides a method for processing lane features, such as Figure 1 As shown in the schematic diagram of a lane feature processing method provided in the first embodiment of the present invention, the method mainly includes the following steps:

[0045] Step 1: Get a high-precision map;

[0046] Among them, the high-precision map includes multiple first roads; the first road includes multiple first lanes; each first lane corresponds to a first lane semantic set.

[0047] Here, the first lane semantic set corresponding to each first lane is the high-precision map semantic features of the lane in the high-precision map. The first lane semantic set should at least include the following semantic features: orientation, position, center line and center line sampling points, edge line and edge line sampling points, length, width, road curvature, etc.

[0048] Step 2: construct a directed graph according to the high-precision map to generate a corresponding first directed graph;

[0049] The first directed graph includes a node set V and a node edge set E; the node set V includes a plurality of first nodes v; the first nodes v correspond one-to-one to the first lanes; the node edge set E includes a plurality of first node edges e; each first node edge e connects two first nodes v; each first node v corresponds to a first lane input feature x v ; Each first node edge e corresponds to a first node edge feature x e ;

[0050] Specifically, it includes: step 21, constructing a corresponding first node v with each first lane in the high-precision map; and inputting a first lane feature x of the corresponding first node v according to the first lane semantic set of each first lane. v Set up; and form a corresponding node set V with all the first nodes v obtained;

[0051] Here, according to the first lane semantic set of each first lane, the first lane input feature x of the corresponding first node v v When setting, a feature configuration rule can be formulated in advance according to the implementation requirements, and the first lane input feature x corresponding to the first node v of some or all semantic feature pairs can be extracted from the first lane semantic set according to the feature configuration rule. v Set up;

[0052] Step 22, take every two first lanes that can be connected longitudinally or transversely in the high-precision map as the corresponding first lane group; and take the two first nodes v corresponding to the first lane group as the corresponding first node pair; and construct the corresponding first node edge e for the first node pair; and set the connection direction of the corresponding first node edge e according to the connection direction of the two first lanes in the first lane group; and set the first node edge feature x of the corresponding first node edge e according to the connection direction of the two first lanes in the first lane group and the two first lane semantic sets. e ; And all the first node edges e obtained constitute the corresponding node edge set E;

[0053] Here, the first node edge feature x of the corresponding first node edge e is set according to the connection direction of the two first lanes in the first lane group and the two first lane semantic sets. e When , a feature configuration rule can be formulated in advance according to the implementation requirements, and based on the feature configuration rule, the first node edge feature x of the corresponding first node edge e is determined according to the connection direction of the two first lanes and part or all of the semantic features of the semantic sets of the two first lanes. e Set up;

[0054] Step 23, constructing a corresponding first directed graph using the obtained node set V and node edge set E.

[0055] Step 3: Select any two first nodes v in the first directed graph as the corresponding starting nodes v start and end node v end ; and press from the starting node v start To the end node v end The connected paths between the two are searched in the direction to generate the corresponding first path set; if the first path set is not empty, the total number of node edges of each first path in the first path set is counted to generate the corresponding first node edge total number; and the total number of first node edges in the first path set exceeds the preset first total number threshold N max The first path is deleted as a redundant path to obtain a corresponding second path set;

[0056] The first path set includes multiple first paths; the first path includes one or more first node edges e.

[0057] Here, the first total threshold N max It is usually set to an integer greater than or equal to 2; if the first path set is not empty, it indicates the starting node v start and end node v end are connected, then the first total threshold N max The paths in the first path set are filtered for the maximum path length threshold; otherwise, if the first path set is empty, it means that the starting node v start and end node v end The two lanes are not connected. In this case, the embodiment of the present invention stops executing the subsequent steps and directly sets the first lane feature. The preset invalid lane feature Y * , and the current first lane feature As the starting point v start Add a complementary feature of to the first directed graph, and then go to the first step of step 3 to reselect the next pair of starting nodes v start and end node v end .

[0058] Step 4: If the second path set is not empty, the total number of first paths in the second path set is counted to generate a corresponding first total number M; and each first path in the second path set is recorded as a corresponding second path P. i , 1≤i≤M; and each second path P i The total number of first node edges is recorded as the corresponding total number of first node edges N i , 1≤N i ≤N max .

[0059] Here, the first paths in the second path set all start from the starting node v start and the end node v end The path length, i.e. the number of nodes and edges passed by the path, does not exceed the first total threshold N. max ; If the second path set is empty, it means that the starting node v start and the end node v end The number of node edges between them does not exceed the first total threshold N max The connected path does not exist. In this case, the embodiment of the present invention stops executing the subsequent steps and directly sets the first lane feature. The preset invalid lane feature Y * , and the current first lane feature As the starting point v start Add a complementary feature of to the first directed graph, and then go to the first step of step 3 to reselect the next pair of starting nodes v start and the end node v end .

[0060] Step 5: Based on the preset long short-term memory network, the second paths P i The first node edge feature x of all first node edges e e For the second path P i The path feature is encoded to generate the corresponding first feature code Φ i ; and according to the first path characteristic Φ of the first total number M obtained i and the end node v end The first lane input feature x v For the starting node v start To the end node v end The lane features are identified to generate the corresponding first lane features And the first lane feature As the starting point v start A complementary feature of is added to the first directed graph;

[0061] Specifically, step 51 includes: based on a preset long short-term memory network, according to each second path Pi The first node edge feature x of all first node edges e e For the second path P i The path feature is encoded to generate the corresponding first feature code Φ i ;

[0062] Specifically include: step 511, according to the starting node v start To the end node v end direction, for the current second path P i The total number of first node edges N i The first node edge feature x of the first node edge e e Sort and generate the corresponding first feature sequence {x e,k}, 1≤k≤N i ; and for the first feature sequence {x e,k} each node edge feature x e,k Convert the encoding format to One Hot Encoding format;

[0063] Step 512: convert the first feature sequence {x e,k} Input the long short-term memory network to calculate and generate the corresponding first path feature Φ i ,

[0064] Φ i =LSTM({x e,k}),

[0065] LSTM() is the model function of the long short-term memory network;

[0066] Here, the embodiment of the present invention first encodes the second path P in a one-hot encoding format. i All first node edge features x e Convert the encoding format, and then convert the first node edge feature x in the one-hot encoding format e The first characteristic sequence {x e,k} Input into the Long Short-Term Memory (LSTM) network for feature encoding to obtain the corresponding first path feature Φ i ;

[0067] Step 52: based on the obtained first total number M of first path characteristics Φ i and the end node v end The first lane input feature x v For the starting node v start To the end node v end The lane features are identified to generate the corresponding first lane features

[0068] Specifically, step 521 includes: i Perform attention intensity calculation to generate the corresponding first attention intensity Ψ(v start ,v end ),

[0069]

[0070] Step 522, end node v end The first lane input features x v As the corresponding end lane input feature

[0071] Step 523, according to the first attention strength Ψ(v start ,v end ) and end lane input features Calculate and generate the corresponding first lane features

[0072]

[0073] Step 53: The first lane feature As the starting point v start A supplementary feature is added to the first directed graph.

[0074] In summary, by repeating the above steps 3-5, all the first lane features of any first node v on the first directed graph After all the additions are completed, each first node v on the first directed graph finally obtained corresponds to a first lane input feature x v In addition, it also corresponds to one or more first lane features Y v (v'); v' is another first node, and the number of first node edges e from the current first node v to the first node v' does not exceed the first total threshold N max .

[0075] Based on the first directed graph output by the embodiment of the present invention, whether any two first nodes v have connectivity can be quickly determined, specifically:

[0076] Identify whether there is a corresponding first node edge e between the current two first nodes v;

[0077] If so, it is confirmed that there is connectivity between the two;

[0078] If it does not exist, one of the two first nodes v is used as the first starting point v s1 , and another as the first end point v e1 ; and for the first starting point vs1 The first lane features Is it the preset invalid lane feature Y? * Confirm; if not, confirm that there is connectivity between the two; if so, set the first end point v e1 As the corresponding second starting point v s2 , the first starting point v s1 As the corresponding second end point v e2 , and for the second starting point v s2 The first lane features Is it the preset invalid lane feature Y? * Confirm; if not, confirm that there is connectivity between the two; if so, confirm that there is no connectivity between the two.

[0079] Based on the first directed graph output by the embodiment of the present invention, when searching for a path between any two first nodes v having connectivity, a node step size λ greater than 1 can be used for fast search to reduce the amount of search calculation and optimize search efficiency.

[0080] Figure 2 This is a module structure diagram of a lane feature processing device provided in the second embodiment of the present invention. The device is a terminal device or server that implements the aforementioned method embodiment, and can also be a device that enables the aforementioned terminal device or server to implement the aforementioned method embodiment. For example, the device can be a device or chip system of the aforementioned terminal device or server. Figure 2 As shown, the device includes: an acquisition module 201, a first directed graph processing module 202, a second directed graph processing module 203, a third directed graph processing module 204 and a lane feature processing module 205.

[0081] The acquisition module 201 is used to acquire a high-precision map; the high-precision map includes multiple first roads; the first road includes multiple first lanes.

[0082] The first directed graph processing module 202 is used to construct a directed graph according to the high-precision map to generate a corresponding first directed graph; the first directed graph includes a node set V and a node edge set E; the node set V includes a plurality of first nodes v; the first nodes v correspond one-to-one to the first lanes; the node edge set E includes a plurality of first node edges e; each first node edge e connects two first nodes v; each first node v corresponds to a first lane input feature x v ; Each first node edge e corresponds to a first node edge feature x e .

[0083] The second directed graph processing module 203 is used to select any two first nodes v on the first directed graph as corresponding starting nodes v start and end node v end; and press from the starting node v start To the end node v end The connected paths between the two are searched in the direction to generate the corresponding first path set; if the first path set is not empty, the total number of node edges of each first path in the first path set is counted to generate the corresponding first node edge total number; and the total number of first node edges in the first path set exceeds the preset first total number threshold N max The first path is deleted as a redundant path to obtain a corresponding second path set; the first path set includes multiple first paths; the first path includes one or more first node edges e.

[0084] The third directed graph processing module 204 is used to count the total number of first paths in the second path set to generate a corresponding first total number M when the second path set is not ; and record each first path in the second path set as a corresponding second path P i , 1≤i≤M; and each second path P i The total number of first node edges is recorded as the corresponding total number of first node edges N i , 1≤N i ≤N max .

[0085] The lane feature processing module 205 is used to process the lane feature according to each second path P based on a preset long short-term memory network. i The first node edge feature x of all first node edges e e For the second path P i The path feature is encoded to generate the corresponding first feature code Φ i ; and according to the first path characteristic Φ of the first total number M obtained i and end node v end The first lane input features x v For the starting node v start To the end node v end The lane features are identified to generate the corresponding first lane features And the first lane feature As the starting point v start A supplementary feature is added to the first directed graph.

[0086] A lane feature processing device provided in an embodiment of the present invention can execute the method steps in the above method embodiment. Its implementation principle and technical effects are similar and will not be repeated here.

[0087] It should be noted that it should be understood that the division of the various modules of the above device is only a division of logical functions. In actual implementation, they can be fully or partially integrated into one physical entity, or they can be physically separated. And these modules can all be implemented in the form of software called by processing elements; they can also be all implemented in the form of hardware; some modules can also be implemented in the form of software called by processing elements, and some modules can be implemented in the form of hardware. For example, the acquisition module can be a separately established processing element, or it can be integrated in a chip of the above device. In addition, it can also be stored in the memory of the above device in the form of program code, and called and executed by a processing element of the above device. The function of the above-mentioned module is determined. The implementation of other modules is similar. In addition, these modules can be fully or partially integrated together, or they can be implemented independently. The processing element described here can be an integrated circuit with signal processing capabilities. In the implementation process, each step of the above method or each module above can be completed by an integrated logic circuit of hardware in the processor element or instructions in the form of software.

[0088] For example, the above modules may be one or more integrated circuits configured to implement the above methods, such as one or more application specific integrated circuits (ASIC), or one or more digital signal processors (DSP), or one or more field programmable gate arrays (FPGA). For another example, when a module above is implemented in the form of a processing element scheduling program code, the processing element may be a general-purpose processor, such as a central processing unit (CPU) or other processor that can call program code. For another example, these modules may be integrated together and implemented in the form of a system-on-a-chip (SOC).

[0089] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the above method embodiments are generated. The above computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The above-mentioned computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the above-mentioned computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, Bluetooth, microwave, etc.) methods. The above-mentioned computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media integrated. The above-mentioned available medium can be a magnetic medium (such as a floppy disk, a hard disk, a tape), an optical medium (such as a DVD), or a semiconductor medium (such as a solid state disk (SSD)), etc.

[0090] Figure 3 This is a schematic diagram of the structure of an electronic device provided in Embodiment 3 of the present invention. The electronic device may be the aforementioned terminal device or server, or may be a terminal device or server connected to the aforementioned terminal device or server to implement the method of the embodiment of the present invention. Figure 3 As shown, the electronic device may include: a processor 301 (such as a CPU), a memory 302, and a transceiver 303; the transceiver 303 is coupled to the processor 301, and the processor 301 controls the transceiver 303. Various instructions may be stored in the memory 302 to complete various processing functions and implement the processing steps described in the aforementioned method embodiment. Preferably, the electronic device involved in the embodiment of the present invention also includes: a power supply 304, a system bus 305 and a communication port 306. The system bus 305 is used to realize the communication connection between components. The above-mentioned communication port 306 is used for connecting and communicating between the electronic device and other peripherals.

[0091] exist Figure 3The system bus 305 mentioned in the figure can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The system bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus. The communication interface is used to realize the communication between the database access device and other devices (such as clients, read-write libraries, and read-only libraries). The memory may include random access memory (RAM), and may also include non-volatile memory (Non-Volatile Memory), such as at least one disk storage.

[0092] The above-mentioned processor can be a general-purpose processor, including a central processing unit CPU, a network processor (Network Processor, NP), a graphics processing unit (Graphics Processing Unit, GPU), etc.; it can also be a digital signal processor DSP, an application-specific integrated circuit ASIC, a field programmable gate array FPGA or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0093] It should be noted that an embodiment of the present invention further provides a computer-readable storage medium, in which instructions are stored. When the computer-readable storage medium is run on a computer, the computer executes the method and processing process provided in the above embodiments.

[0094] An embodiment of the present invention further provides a chip for executing instructions, wherein the chip is used to execute the processing steps described in the aforementioned method embodiment.

[0095] The embodiment of the present invention provides a lane feature processing method, device, electronic device and computer-readable storage medium; improves the conventional lane map, while retaining all nodes and node edge features of the conventional lane map, adds a set of lane features Y for each node that can reflect the connectivity relationship between the current node and other non-directly connected nodes. Through the present invention, the overall connectivity between every two lanes, that is, every two nodes, can be confirmed on the lane map based on the lane feature Y. The path search based on the lane map with the node lane feature Y output by the present invention can not only support the traditional node-by-node search method, but also support the jump search method with the node step length, so that the search method becomes diversified, the search calculation amount is reduced, and the search efficiency is optimized.

[0096] The professionals should further realize that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to the function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0097] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0098] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for processing lane features, It is characterized in that The method comprises: Acquire a high-precision map; the high-precision map includes a plurality of first roads; the first road includes a plurality of first lanes; A directed graph is constructed according to the high-precision map to generate a corresponding first directed graph; the first directed graph includes a node set V and a node edge set E; the node set V includes a plurality of first nodes v; the first nodes v correspond to the first lanes one by one; the node edge set E includes a plurality of first node edges e; each of the first node edges e connects two of the first nodes v; each of the first nodes v corresponds to a first lane input feature x v ; Each of the first node edges e corresponds to a first node edge feature x e ; Select any two of the first nodes v on the first directed graph as corresponding starting nodes v start and the end node v end ; and from the starting node v start To the end node v end Searching for the connected path between the two in the direction of to generate a corresponding first path set; if the first path set is not empty, counting the total number of node edges of each first path in the first path set to generate a corresponding first node edge total number; and counting the total number of the first node edges in the first path set that exceeds a preset first total number threshold N max The first path is deleted as a redundant path to obtain a corresponding second path set; the first path set includes multiple first paths; the first path includes one or more first node edges e; If the second path set is not empty, the total number of the first paths in the second path set is counted to generate a corresponding first total number M; and each of the first paths in the second path set is recorded as a corresponding second path P i , 1≤i≤M; and each of the second paths P i The total number of the first node edges is recorded as the corresponding total number of the first node edges N i , 1≤N i ≤N max ; Based on the preset long short-term memory network, according to each of the second paths P i The first node edge feature x of all the first node edges e e For the second path P i The path feature is encoded to generate the corresponding first feature code Φ i ; and according to the first path characteristic Φ of the first total number M obtained i and the end node v end The first lane input feature x v For the starting node v start To the end node v end The lane features are identified to generate the corresponding first lane features And the first lane feature As the starting node v start A supplementary feature is added to the first directed graph.

2. The method for processing lane features according to claim 1, It is characterized in that Each of the first lanes corresponds to a first lane semantic set.

3. The method for processing lane features according to claim 2, It is characterized in that The step of constructing a directed graph according to the high-precision map to generate a corresponding first directed graph specifically includes: Construct the corresponding first node v with each first lane in the high-precision map; and input the first lane feature x of the corresponding first node v according to the first lane semantic set of each first lane. v Set; and form the corresponding node set V by all the first nodes v obtained; Every two first lanes in the high-precision map that can be connected longitudinally or transversely are taken as the corresponding first lane group; the two first nodes v corresponding to the first lane group are taken as the corresponding first node pair; and the corresponding first node edge e is constructed for the first node pair; and the connection direction of the corresponding first node edge e is set according to the connection direction of the two first lanes in the first lane group; and the first node edge feature x of the corresponding first node edge e is set according to the connection direction of the two first lanes in the first lane group and the two first lane semantic sets. e ; and all the first node edges e obtained constitute the corresponding node edge set E; The obtained node set V and the node edge set E constitute the corresponding first directed graph.

4. The method for processing lane features according to claim 1, It is characterized in that The preset long short-term memory network is based on each of the second paths P i The first node edge feature x of all the first node edges e e For the second path P i The path feature is encoded to generate the corresponding first feature code Φ i , specifically including: According to the starting point v start To the end node v end direction, the second path P i The total number of edges of the first node N i The first node edge feature x of the first node edge e e Sort and generate the corresponding first feature sequence {x e,k }, 1≤k≤N i ; and for the first feature sequence {x e,k } each node edge feature x e,k Convert the encoding format according to the one-hot encoding format; The first feature sequence {x e,k } Input the long short-term memory network to calculate and generate the corresponding first path feature Φ i , Φ i =LSTM({x e,k }), LSTM() is the model function of the long short-term memory network.

5. The method for processing lane features according to claim 1, It is characterized in that The first path characteristic Φ obtained according to the first total number M i and the end node v end The first lane input feature x v For the starting node v start To the end node v end The lane features are identified to generate the corresponding first lane features Specifically include: According to the first path feature Φ obtained from the first total number M i Perform attention intensity calculation to generate the corresponding first attention intensity Ψ(v start ,v end ), The end node v end The first lane input feature x v As the corresponding end lane input feature According to the first attention strength Ψ(v start ,v end ) and the end lane input feature Calculate and generate the corresponding first lane feature 6. The method for processing lane features according to claim 1, It is characterized in that The method further comprises: If the first or second path set is empty, set the first lane feature The preset invalid lane feature Y * , and the first lane feature of the current As the starting node v start A supplementary feature is added to the first directed graph.

7. The method for processing lane features according to any one of claims 1 and 6, It is characterized in that All the first lane features of any first node v on the first directed graph After all the additions are completed, each of the first nodes v corresponds to one of the first lane input features x v In addition, it also corresponds to one or more first lane features Y v (v'); v' is another first node, and the number of the first node edges e passing from the current first node v to the first node v' does not exceed the first total threshold value N max .

8. A device for executing the lane feature processing method according to any one of claims 1 to 7, It is characterized in that The device comprises: an acquisition module, a first directed graph processing module, a second directed graph processing module, a third directed graph processing module and a lane feature processing module; The acquisition module is used to acquire a high-precision map; the high-precision map includes a plurality of first roads; the first road includes a plurality of first lanes; The first directed graph processing module is used to construct a directed graph according to the high-precision map to generate a corresponding first directed graph; the first directed graph includes a node set V and a node edge set E; the node set V includes a plurality of first nodes v; the first nodes v correspond to the first lanes one by one; the node edge set E includes a plurality of first node edges e; each of the first node edges e connects two of the first nodes v; each of the first nodes v corresponds to a first lane input feature x v ; Each of the first node edges e corresponds to a first node edge feature x e ; The second directed graph processing module is used to select any two of the first nodes v on the first directed graph as corresponding starting nodes v start and the end node v end ; and from the starting node v start To the end node v end Searching for the connected path between the two in the direction of to generate a corresponding first path set; if the first path set is not empty, counting the total number of node edges of each first path in the first path set to generate a corresponding first node edge total number; and counting the total number of the first node edges in the first path set that exceeds a preset first total number threshold N max The first path is deleted as a redundant path to obtain a corresponding second path set; the first path set includes multiple first paths; the first path includes one or more first node edges e; The third directed graph processing module is used to, when the second path set is not, count the total number of the first paths in the second path set to generate a corresponding first total number M; and record each of the first paths in the second path set as a corresponding second path P i , 1≤i≤M; and each of the second paths P i The total number of the first node edges is recorded as the corresponding total number of the first node edges N i , 1≤N i ≤N max ; The lane feature processing module is used to process the lane feature according to each of the second paths P based on a preset long short-term memory network. i The first node edge feature x of all the first node edges e e For the second path P i The path feature is encoded to generate the corresponding first feature code Φ i ; and according to the first path characteristic Φ of the first total number M obtained i and the end node v end The first lane input feature x v For the starting node v start To the end node v end The lane features are identified to generate the corresponding first lane features And the first lane feature As the starting node v start A supplementary feature is added to the first directed graph.

9. An electronic device, It is characterized in that include: memory, processors, and transceivers; The processor is used to couple with the memory, read and execute instructions in the memory, so as to implement the method steps described in any one of claims 1 to 7; The transceiver is coupled to the processor, and the processor controls the transceiver to send and receive messages.

10. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores computer instructions, which, when executed by a computer, enable the computer to execute the method according to any one of claims 1 to 7.

Images