Map construction method and apparatus, terminal device, and storage medium

Abstract

The application discloses a kind of map construction method, device, terminal equipment and storage medium, by obtaining map and each original track, and determine the feature matching relationship of the map and each original track;Based on the feature matching relationship, the joint constraint relationship of the map and each original track is constructed;The joint constraint relationship is iteratively optimized, and an optimization result is obtained;According to the optimization result, the map is updated, and an updated map is obtained.By obtaining map and each original track, and determining the feature matching relationship of map and each original track, then constructing joint constraint relationship and iteratively optimizing, obtaining updated map, avoid individual precision not high track to cause influence on map overall quality, realize reference all track to map is jointly optimized, can improve the overall precision of map.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of map, in particular to a map construction method and device, a terminal equipment and a storage medium. BACKGROUND

[0002] At present, when a map construction platform adopts trajectory information in vehicle crowdsourcing data to construct a map, the trajectory is usually sequentially adopted to update the map, trajectories that are not well aligned with the map are abandoned, trajectories that are well aligned with the map are selected, the trajectory and the map are aligned and stretched, the features of the trajectory are made to enter the map, and an updated map is formed.

[0003] Problems existing in this method include: 1) the order of trajectory processing has a great influence on the accuracy of the map, when a trajectory with low accuracy is inserted in the early stage, the quality of the entire map is reduced, and the accuracy of subsequent trajectory and map alignment is affected; 2) when the map has systematic errors, there is no scheme for jointly optimizing all trajectories, and it is difficult to repair the map problem, thereby resulting in low overall accuracy of the map.

[0004] Therefore, it is necessary to provide a solution for improving the overall accuracy of the map.

[0005] The above content is only used to assist in understanding the technical solutions of the present application, and does not represent the acknowledgement of the above content as prior art. SUMMARY

[0006] The main purpose of the present application is to provide a map construction method and device, a terminal equipment and a storage medium, which aims to improve the overall accuracy of the map.

[0007] To achieve the above purpose, the present application provides a map construction method, which comprises:

[0008] obtaining a map and each original trajectory, and determining a feature matching relationship between the map and the original trajectories;

[0009] based on the feature matching relationship, constructing a joint constraint relationship between the map and the original trajectories;

[0010] iteratively optimizing the joint constraint relationship to obtain an optimization result;

[0011] updating the map according to the optimization result to obtain an updated map.

[0012] Optionally, the step of determining the feature matching relationship between the map and the original trajectories comprises:

[0013] aligning the original trajectories with the map by a preset alignment algorithm to obtain each aligned trajectory;

[0014] determining a feature matching relationship between the map and the original trajectories according to the aligned trajectories.

[0015] Optionally, the step of determining the feature matching relationship between the map and the original trajectories according to the aligned trajectories comprises:

[0016] searching a matching relationship between feature points of the aligned trajectories and feature points of the map by a tree data structure algorithm, wherein the feature points of the aligned trajectories correspond to respective feature serial numbers;

[0017] obtaining the feature matching relationship between the map and the original trajectories according to the matching relationship between the feature points of the aligned trajectories and the feature points of the map, and combining the feature serial numbers.

[0018] Optionally, the feature points of the map comprise respective connection nodes, the joint constraint relationship comprises a trajectory-map constraint relationship, and the step of constructing the joint constraint relationship between the map and the original trajectories based on the feature matching relationship comprises:

[0019] determining respective constraint terms according to the feature matching relationship;

[0020] determining a to-be-optimized term according to the feature points of the map, wherein the to-be-optimized term comprises the connection nodes;

[0021] constructing the trajectory-map constraint relationship based on the to-be-optimized term and the constraint terms.

[0022] Optionally, the joint constraint relationship further comprises a trajectory constraint relationship and / or a map constraint relationship, and the step of constructing the joint constraint relationship between the map and the original trajectories based on the feature matching relationship further comprises:

[0023] constructing the trajectory constraint relationship according to the feature points of the original trajectories; and / or,

[0024] constructing the map constraint relationship according to the feature points of the map.

[0025] Optionally, the step of iteratively optimizing the joint constraint relationship to obtain an optimized result comprises:

[0026] configuring weight parameters of respective constraint relationships in the joint constraint relationship;

[0027] inputting the respective constraint relationships in the joint constraint relationship and corresponding weight parameters into a preset graph optimization iterator to iteratively optimize the connection nodes in the to-be-optimized term by the graph optimization iterator to obtain optimized connection nodes;

[0028] According to the optimized connection nodes, the optimization result is obtained.

[0029] Optionally, the feature points of the map further include feature elements corresponding to the connection nodes, and the step of updating the map according to the optimization result to obtain an updated map includes:

[0030] The optimized connection nodes in the optimization result are compared with the connection nodes before optimization to obtain connection node position changes;

[0031] According to the connection node position changes, corresponding connection node neighborhood transformation matrices are obtained respectively;

[0032] The connection node neighborhood transformation matrices are respectively applied to the corresponding feature elements to obtain updated feature elements;

[0033] The updated map is constructed based on the updated feature elements.

[0034] In addition, to achieve the above object, the application further provides a map construction device, which comprises:

[0035] An acquisition module is configured to acquire a map and original trajectories, and determine feature matching relationships between the map and the original trajectories;

[0036] A construction module is configured to construct joint constraint relationships between the map and the original trajectories based on the feature matching relationships;

[0037] An iteration module is configured to iteratively optimize the joint constraint relationships to obtain an optimization result;

[0038] An update module is configured to update the map according to the optimization result to obtain an updated map.

[0039] In addition, to achieve the above object, the application further provides a terminal device, which comprises a memory, a processor, and a map construction program stored in the memory and executable on the processor, and the map construction program realizes the steps of the above map construction method when executed by the processor.

[0040] In addition, to achieve the above object, the application further provides a computer readable storage medium, which stores a map construction program, and the map construction program realizes the steps of the above map construction method when executed by a processor.

[0041] This invention proposes a map construction method, apparatus, terminal device, and storage medium. The method involves acquiring a map and various original trajectories, determining the feature matching relationship between the map and the original trajectories, constructing a joint constraint relationship between the map and the original trajectories based on the feature matching relationship, iteratively optimizing the joint constraint relationship to obtain an optimization result, and updating the map according to the optimization result to obtain an updated map. By acquiring the map and various original trajectories, determining the feature matching relationship between the map and the original trajectories, and then constructing a joint constraint relationship and iteratively optimizing it to obtain an updated map, this method avoids the impact of individual low-precision trajectories on the overall map quality. It achieves joint optimization of the map by referring to all trajectories, thereby improving the overall accuracy of the map. Attached Figure Description

[0042] Figure 1 This is a schematic diagram of the functional modules of the terminal device to which the map building device of this invention belongs;

[0043] Figure 2 This is a flowchart illustrating an exemplary embodiment of the map construction method of the present invention;

[0044] Figure 3 for Figure 2 A schematic diagram of the specific process of step S10 in the embodiment;

[0045] Figure 4 for Figure 2 A schematic diagram of the specific process of step S30 in the embodiment;

[0046] Figure 5 for Figure 2 A detailed flowchart of step S40 in the embodiment;

[0047] Figure 6 This is an exemplary schematic diagram of the map before optimization in an embodiment of the present invention;

[0048] Figure 7 This is an exemplary schematic diagram of the optimized map in an embodiment of the present invention.

[0049] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0050] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0051] The main solution of the embodiment of the present application is: acquiring a map and each original trajectory, determining a feature matching relationship between the map and the each original trajectory, constructing a joint constraint relationship between the map and the each original trajectory based on the feature matching relationship, iteratively optimizing the joint constraint relationship to obtain an optimization result, and updating the map according to the optimization result to obtain an updated map. By acquiring the map and each original trajectory, determining the feature matching relationship between the map and each original trajectory, and then constructing the joint constraint relationship and iteratively optimizing, the updated map is obtained, the influence of the individual trajectory with low precision on the overall quality of the map is avoided, the joint optimization of the map by referring to all trajectories is realized, and the overall precision of the map can be improved.

[0052] The technical terms related to the embodiment of the present application are:

[0053] g2o (general graph optimization): g2o is a general solver and is not limited to some SLAM problems, and can be used to solve most optimization problems expressed by graphs;

[0054] kd-tree (abbreviation of k-dimensional tree) is a tree-shaped data structure for storing instance points in k-dimensional space for fast retrieval, and is mainly applied to search of key data in multidimensional space (such as range search and nearest neighbor search). The K-D tree is a special case of the binary space partitioning tree.

[0055] When artificial mapping is performed by using vehicle crowdsourcing data, the map gradually grows, and the trajectories are sequentially spliced, the trajectories and the existing map are aligned and stretched, the features of the trajectories enter the map and become part of the map, and the trajectories that are not well aligned with the map are abandoned.

[0056] The method has the following problems:

[0057] (1) The order of trajectory processing has a great influence on the map precision, and when a trajectory with low precision is spliced in the early stage, the quality of the entire map is reduced, and the alignment precision of the subsequent trajectory and the map is affected;

[0058] (2) When the map has systematic errors, there is no scheme for jointly optimizing all trajectories to repair the map problem.

[0059] The present application provides a solution, by using a g2o-based multi-trajectory single-map joint optimization method, all trajectory information is read after the map is manually edited, the matching relationship between the trajectory and the map features is found, the systematic problems of the map scale and rotation are repaired, and the overall precision of the map is improved.

[0060] Specifically, refer toFigure 1 , Figure 1 is a functional module diagram of a terminal device to which the map construction device belongs. The map construction device can be a device capable of map construction independent of the terminal device, which can be carried on the terminal device in the form of hardware or software. The terminal device can be a smart mobile terminal such as a mobile phone or a tablet computer having a data processing function, and can also be a fixed terminal device or a server having a data processing function.

[0061] In this embodiment, the terminal device to which the map construction device belongs at least includes an output module 110, a processor 120, a memory 130, and a communication module 140.

[0062] The memory 130 stores an operating system and a map construction program, and the map construction device can store information such as the obtained map and each original trajectory, each aligned trajectory, the feature matching relationship between the map and each original trajectory, the constructed joint constraint relationship, the optimization result obtained by iterative optimization, and the updated map in the memory 130; the output module 110 can be a display screen or the like. The communication module 140 can include a WIFI module, a mobile communication module, and a Bluetooth module, and the like, and communicates with external devices or servers through the communication module 140.

[0063] The map construction program in the memory 130 is executed by the processor to implement the following steps:

[0064] Obtain the map and each original trajectory, and determine the feature matching relationship between the map and each original trajectory;

[0065] Based on the feature matching relationship, construct the joint constraint relationship between the map and each original trajectory;

[0066] Iteratively optimize the joint constraint relationship to obtain an optimization result;

[0067] Update the map according to the optimization result to obtain an updated map.

[0068] Further, the map construction program in the memory 130 is executed by the processor to implement the following steps:

[0069] Align each original trajectory with the map by a preset alignment algorithm to obtain each aligned trajectory;

[0070] Determine the feature matching relationship between the map and each original trajectory according to each aligned trajectory.

[0071] Further, the map construction program in the memory 130 is executed by the processor to implement the following steps:

[0072] searching, by a tree data structure algorithm, a matching relationship between the feature points of the aligned trajectories and the feature points of the map, wherein the feature points of the aligned trajectories correspond to the feature numbers respectively;

[0073] According to the matching relationship between the feature points of the aligned trajectories and the feature points of the map, and in combination with the feature numbers, a feature matching relationship between the map and the original trajectories is obtained.

[0074] Further, the map construction program in the memory 130, when executed by the processor, also implements the following steps:

[0075] According to the feature matching relationship, determine each constraint term;

[0076] According to the feature points of the map, determine a to-be-optimized term, wherein the to-be-optimized term includes the connection nodes;

[0077] Based on the to-be-optimized term and each constraint term, construct the trajectory map constraint relationship.

[0078] Further, the map construction program in the memory 130, when executed by the processor, also implements the following steps:

[0079] According to the feature points of the original trajectories, construct the trajectory constraint relationship; and / or,

[0080] According to the feature points of the map, construct the map constraint relationship.

[0081] Further, the map construction program in the memory 130, when executed by the processor, also implements the following steps:

[0082] Configure a weight parameter of each constraint relationship in the joint constraint relationship;

[0083] Input each constraint relationship in the joint constraint relationship and the corresponding weight parameter into a preset graph optimization iterator, so as to iteratively optimize each connection node in the to-be-optimized term by the graph optimization iterator, and obtain an optimized each connection node;

[0084] According to the optimized each connection node, obtain the optimization result.

[0085] Further, the map construction program in the memory 130, when executed by the processor, also implements the following steps:

[0086] Compare the optimized each connection node in the optimization result with the each connection node before optimization, and obtain each connection node position change;

[0087] According to the each connection node position change respectively, obtain a corresponding connection node neighborhood transformation matrix;

[0088] The connection node neighborhood transformation matrix is respectively applied to the corresponding feature elements to obtain updated feature elements;

[0089] An updated map is constructed based on the updated feature elements.

[0090] In the embodiment, the map and the original trajectories are obtained, and the feature matching relationship between the map and the original trajectories is determined. Based on the feature matching relationship, a joint constraint relationship between the map and the original trajectories is constructed. The joint constraint relationship is iteratively optimized to obtain an optimization result. The map is updated according to the optimization result to obtain an updated map. The feature matching relationship between the map and the original trajectories is determined, and the joint constraint relationship is constructed and iteratively optimized to obtain the updated map, so that the influence of individual trajectories with low accuracy on the overall quality of the map is avoided, the joint optimization of the map based on all trajectories is realized, and the overall accuracy of the map is improved.

[0091] Based on the terminal device architecture but not limited to the above-mentioned architecture, the method embodiment of the present application is proposed.

[0092] The execution subject of the method embodiment can be a map construction device or a terminal device, and the map construction device is taken as an example in the embodiment.

[0093] Reference Figure 2 , Figure 2 FIG. 1 is a flowchart of an exemplary embodiment of the map construction method of the present application. The map construction method comprises the following steps.

[0094] In step S10, the map and the original trajectories are obtained, and the feature matching relationship between the map and the original trajectories is determined.

[0095] Specifically, in the process of vehicle driving, the sensors such as laser radar and camera of the vehicle can collect trajectory information, and each vehicle uploads the trajectory information to the cloud to form crowd-sourced trajectories. The map construction platform obtains the original trajectory information in the crowd-sourced trajectories from the cloud. Each original trajectory and the map are essentially composed of point clouds composed of a plurality of discrete points. The types of the discrete points can be determined by classifying each discrete point, such as lane lines or parking spaces. The points that can be classified and matched in the discrete points are taken as feature points in the trajectories or the map. The types of the feature points include special points of connection nodes. The connection nodes have a binding relationship with the surrounding feature points. With the update of the position of the connection nodes, the positions of the surrounding feature points can be updated synchronously. In the initial stage of mapping, the map can be initially constructed according to the trajectories, and then the existing map can be updated and optimized by using the newly added original trajectories.

[0096] When the crowdsourcing trajectory data of vehicles is used for map construction and updating, the original trajectories and the map are obtained from the cloud, and the original trajectories and the map can be aligned by manual editing or by using an automatic alignment algorithm.

[0097] As one of the embodiments, after the map and the original trajectories are obtained, the original trajectories are aligned with the map by using a preset alignment algorithm to obtain aligned trajectories, and then the feature matching relationship between the map and the original trajectories is determined according to the aligned trajectories.

[0098] In the embodiment, the map is constructed by using a map construction platform, and the original trajectories collected and uploaded by vehicles are obtained from the cloud. The original trajectories and the map are essentially composed of point clouds composed of a plurality of discrete points. The types of the discrete points, such as lane lines or parking spaces, are determined by classifying the discrete points. The points that can be classified and matched are used as feature points in the trajectories or the map. The special points of the types of connection nodes in the feature points have a binding relationship with the surrounding feature points. The updated positions of the connection nodes (link points) are obtained by iterative optimization, and the positions of the feature points (i.e., feature elements) corresponding to the connection nodes are updated synchronously, thereby realizing the updating of the map.

[0099] The original trajectories are sequentially aligned with the map by using a preset alignment algorithm, thereby obtaining aligned trajectories. In this process, if there is a trajectory that fails to be aligned, it means that the trajectory has a large difference in form from the map, and the trajectory is not used, thereby avoiding the influence of trajectories with low precision on the overall quality of the map.

[0100] As one of the embodiments, the feature matching relationship between the aligned trajectories and the map is searched by using a tree data structure algorithm, and then the feature matching relationship between the map and the original trajectories is obtained by combining the feature serial numbers corresponding to the aligned trajectories and the original trajectories.

[0101] Specifically, in the embodiment, a tree data structure algorithm (kd-tree) is used to quickly search the feature matching relationship between the aligned trajectories and the map. Since the feature serial numbers corresponding to the features in the trajectories before and after alignment are unchanged, the feature matching relationship between the original trajectories and the map can be obtained by combining the feature serial numbers corresponding to the features before and after alignment according to the feature matching relationship between the aligned trajectories and the map. Compared with the aligned trajectories, the original trajectories have not been aligned and stretched into the map, and therefore the feature matching relationship between the original trajectories and the map can be used to avoid systematic errors in the map, and then used to repair systematic problems in the map.

[0102] Step S20, based on the feature matching relationship, constructing the joint constraint relationship of the map and the original trajectories;

[0103] Further, in the process of constructing the joint constraint relationship, the embodiment of the present application not only considers the feature matching relationship between the map and the original trajectories, but also includes the constraint relationship inside the map and the constraint relationship inside the original trajectories, ensuring the stability of the internal form of the trajectories and the map. In this process, since the parameterized trajectories and the map include a large number of features, such as lane lines, parking spaces, speed bumps, and positions of trajectory points at each time, in order to make the overall calculation scale controllable, the embodiment of the present application selects key features in the trajectories and the map to be included in the calculation.

[0104] As one of the implementation manners, the joint constraint relationship includes a trajectory-map constraint relationship, and the step of constructing the joint constraint relationship of the map and the original trajectories based on the feature matching relationship includes:

[0105] According to the feature matching relationship, determining each constraint term;

[0106] According to the feature points of the map, determining a to-be-optimized term, wherein the to-be-optimized term includes the connection nodes;

[0107] Based on the to-be-optimized term and each constraint term, constructing the trajectory-map constraint relationship.

[0108] As another implementation manner, the joint constraint relationship further includes a trajectory constraint relationship and / or a map constraint relationship, and the step of constructing the joint constraint relationship of the map and the original trajectories based on the feature matching relationship further includes:

[0109] According to the feature points of the original trajectories, constructing the trajectory constraint relationship; and / or,

[0110] According to the feature points of the map, constructing the map constraint relationship.

[0111] Specifically, in the embodiment of the present application, g2o is used to optimize the map. In this process, first, the vertices and edges in the optimization relationship are defined, the connection nodes in the map are taken as to-be-optimized terms, i.e., defined as vertices, the feature points of the original trajectories are key feature points selected from the original trajectories, the feature points of the map are key feature points selected from the map, which include discrete connection nodes, the surrounding feature points having a binding relationship with each connection node are called corresponding feature elements, the joint constraint relationship between the map and the trajectories and inside the map and the trajectories is defined as edges, i.e., as constraint terms, wherein the constraint terms can further include distances and angles between the connection nodes, pose changes of the feature points, and feature matching relationships, etc., each constraint relationship is jointly included in the equation of the iterator to be solved, which can achieve the purpose of adjustment, so as to minimize the average error between the map and the original trajectories.

[0112] Step S30, the joint constraint relationship is iteratively optimized to obtain an optimization result;

[0113] Further, after the joint constraint relationship is constructed, a preset map optimization iterator can be input, and the belief degree of the map or the trajectory is controlled by configuring the weight parameter of each constraint relationship, and then each connection node after optimization is obtained. Through iteration, each connection node after optimization is obtained. According to the position change of each connection node before and after optimization, the update of each feature element can be realized, so that the update of the feature points around the connection node is realized. The optimization item selected in the optimization is mainly the key feature points in the map, such as each connection point. The corresponding connection node neighborhood transformation matrix is obtained through the optimized connection points, and then acts on the surrounding feature points. Therefore, it is not necessary to perform iterative calculation on all feature points, and the overall calculation scale is controlled while the overall update of the map is realized.

[0114] Step S40, the map is updated according to the optimization result to obtain an updated map.

[0115] Further, according to the position change of each connection node before and after optimization, the transformation matrix in the neighborhood of each connection node can be solved, and then acts on the corresponding feature elements to obtain an updated map. In addition, with the continuous addition of crowd-sourced trajectories, more trajectories can be further obtained for iterative solving to realize the update and optimization of the map.

[0116] In the embodiment, the map and each original trajectory are obtained, and the feature matching relationship between the map and the original trajectories is determined. Based on the feature matching relationship, the joint constraint relationship between the map and the original trajectories is constructed. The joint constraint relationship is iteratively optimized to obtain an optimization result. The map is updated according to the optimization result to obtain an updated map. By obtaining the map and each original trajectory and aligning, the aligned trajectory is obtained to further determine the feature matching relationship between each original trajectory and the map, and then the joint constraint relationship is constructed and iteratively optimized to obtain an updated map. The influence of the individual trajectory with low accuracy on the overall quality of the map is avoided, the joint optimization of the map is realized by referring to all trajectories, the systematic problems of the map can be repaired, and the overall accuracy of the map is improved.

[0117] Reference Figure 3 , Figure 3 To Figure 2 The specific flowchart of determining the feature matching relationship between the map and the original trajectories in step S10 in the embodiment is shown. The embodiment is based on the above Figure 2 The embodiment shown in the above

[0118] In step S101, the original trajectories are aligned with the map by using a preset alignment algorithm to obtain aligned trajectories.

[0119] Specifically, after the map and the original trajectories are obtained, the feature points in the original trajectories are preliminarily matched with the feature points in the map by using a preset alignment algorithm, and then the original trajectories are stretched and translated according to the matching result, so that the feature points in the trajectories are aligned with the feature points in the map, and the trajectories after alignment are taken as the aligned trajectories.

[0120] In step S102, the feature matching relationship between the map and the original trajectories is determined according to the aligned trajectories.

[0121] Further, after the original trajectories are stretched and translated to align the feature points in the trajectories with the feature points in the map to obtain the aligned trajectories, since the feature points in the trajectories all have corresponding feature serial numbers, the positions of the feature points in the trajectories may change in the alignment process, but the feature serial numbers corresponding to the feature points remain unchanged, so the feature matching relationship between the original trajectories and the map can be determined according to the feature matching relationship between the aligned features and the map.

[0122] As one of the embodiments, the step of determining the feature matching relationship between the original trajectories and the map according to the feature matching relationship between the aligned features and the map includes:

[0123] The feature matching relationship between the feature points of the aligned trajectories and the feature points of the map is searched by using a tree data structure algorithm, wherein the feature points of the aligned trajectories correspond to the feature serial numbers respectively.

[0124] According to the feature matching relationship between the feature points of the aligned trajectories and the feature points of the map, and in combination with the feature serial numbers, the feature matching relationship between the map and the original trajectories is obtained.

[0125] Specifically, the tree data structure algorithm is used to quickly search the feature matching relationship between the aligned trajectories and the map in the embodiment of the present application. Since the feature serial numbers corresponding to the features in the trajectories remain unchanged before and after alignment, the feature matching relationship between the original trajectories and the map can be obtained according to the feature matching relationship between the aligned trajectories and the map in combination with the feature serial numbers corresponding to the features before and after alignment, so as to form a joint constraint relationship and iteratively optimize to update the map. Compared with the aligned trajectories, the original trajectories have not been aligned and stretched into the map, so that the feature matching relationship between the original trajectories and the map can be used to avoid systematic errors of the map, such as problems in the scale or rotation of the map, repair the systematic problems of the map, and thus improve the overall accuracy of the map.

[0126] The embodiment aligns each original trajectory with the map through a preset alignment algorithm to obtain each aligned trajectory, and determines the feature matching relationship between the map and each original trajectory according to the aligned trajectories. The aligned trajectories are obtained by aligning the obtained map with each original trajectory, and the feature matching relationship between the aligned trajectories and the map is searched, and then the feature matching relationship between each original trajectory and the map is determined according to the characteristic that the feature serial number is unchanged before and after trajectory alignment, so as to form a joint constraint relationship and iterative optimization, realize the update of the map, realize the repair of the systematic problem of the map, and improve the overall accuracy of the map.

[0127] With reference to Figure 4 , Figure 4 For Figure 2 The specific flowchart of step S30 in the embodiment is shown. The embodiment is based on the above Figure 2 The embodiment shown in the above embodiment, in the embodiment, the step S30 comprises:

[0128] Step S301, configuring the weight parameters of each constraint relationship in the joint constraint relationship;

[0129] Specifically, after the joint constraint relationship including the trajectory constraint relationship, the map constraint relationship and the trajectory map constraint relationship is constructed, the weight parameters of each constraint relationship can be configured, and the specific weight parameters can be set according to the credibility of the map and the trajectory in the actual situation, which is not limited in the embodiment.

[0130] It should be noted that in order to fully consider the stability of the internal form of the trajectory and the internal form of the map, the joint constraint relationship constructed in the embodiment includes the trajectory constraint relationship and the map constraint relationship. In other embodiments, only the trajectory map constraint relationship can be used, or one of the trajectory constraint relationship and the map constraint relationship can be added as the joint constraint relationship for iterative optimization, and the embodiment is not limited in detail.

[0131] Step S302, inputting each constraint relationship in the joint constraint relationship and the corresponding weight parameter into a preset graph optimization iterator to iteratively optimize each connection node in the to-be-optimized item through the graph optimization iterator to obtain each optimized connection node;

[0132] Further, the trajectory constraint relationship, the map constraint relationship, the trajectory map constraint relationship and the weight parameters corresponding to each constraint relationship are input into the preset graph optimization iterator, that is, each constraint of the trajectory and the map is unified into an equation for iterative solution to obtain a solution that minimizes the equation, and the equation type is as follows:

[0133] Wherein, x, y are trajectory and map respectively, f function represents self constraint, w function represents constraint between x and y, and λ and θ are weight parameters for controlling proportion between constraint terms. The equation is solved iteratively to obtain a set of x, y to minimize the value of the equation. The process is optimization. The number of iterations can be set according to actual situation.

[0134] Step S303, obtaining the optimization result according to the optimized connection nodes.

[0135] Further, the graph optimization iterator is used to iteratively optimize each connection node in the to-be-optimized item, and y in the obtained result is the optimized connection node in the map. The obtained each optimized connection node is taken as the optimization result. The position of the optimized connection node is used to synchronously update the positions of the remaining feature points having a corresponding relationship with the connection node, thereby realizing the update of the map.

[0136] In the embodiment, the weight parameters of each constraint relationship in the joint constraint relationship are configured. Each constraint relationship in the joint constraint relationship and the corresponding weight parameter are input into a preset graph optimization iterator. Each connection node in the to-be-optimized item is iteratively optimized by the graph optimization iterator to obtain each optimized connection node. The optimization result is obtained according to the optimized connection nodes. When joint optimization is performed, the constraints of all trajectories and maps are added to minimize the average error between the map and the trajectory. The constraint relationship of each original trajectory is referred to to realize the joint optimization of the map and repair the systematic problem of the map, thereby improving the overall accuracy of the map.

[0137] Reference Figure 5 , Figure 5 To Figure 2 The specific flowchart of step S40 in the embodiment is shown. The embodiment is based on the above Figure 2 The embodiment shown in the above embodiment. In the embodiment, step S40 includes:

[0138] Step S401, comparing each optimized connection node in the optimization result with each connection node before optimization to obtain the position change of each connection node.

[0139] Step S402, obtaining the corresponding connection node neighborhood transformation matrix according to the position change of each connection node.

[0140] Step S403, applying the connection node neighborhood transformation matrix to the corresponding feature element to obtain each updated feature element.

[0141] Step S404, constructing the updated map based on each updated feature element.

[0142] Specifically, by iteration, the optimized connection nodes are obtained, and according to the position changes of the connection nodes before and after optimization, the transformation matrix in the neighborhood of each connection node can be obtained, and the transformation matrix of the connection node neighborhood is applied to the corresponding feature elements, so that the update of the feature elements can be realized, thereby realizing the update of the feature points around the connection nodes, and the map composed of the updated feature points is the updated map, which is described with reference to Figure 6 and Figure 7 , Figure 6 is an exemplary schematic diagram of the map before optimization in the embodiment of the present application, Figure 7 is an exemplary schematic diagram of the map after optimization, as shown in the figure, in the optimized map, the feature points are clearer, and the positions of the two lane lines tend to be more parallel, thereby repairing the systematic problems of the scale and rotation of the map, and the overall accuracy of the map is improved. In addition, as the crowd trajectory is continuously added, more trajectories can be further obtained for iterative solving to realize the update and optimization of the map. The optimization items selected during optimization are mainly the key feature points in the map, such as the connection points, the connection node neighborhood transformation matrix corresponding to the optimized connection points is obtained, and then it is applied to the surrounding feature points, so that the iterative calculation of all feature points is not required, the overall update of the map is realized, and the overall calculation scale is controllable.

[0143] In the embodiment, the optimized connection nodes in the optimization result are compared with the connection nodes before optimization to obtain the position changes of the connection nodes, the connection node neighborhood transformation matrix corresponding to each connection node position change is obtained, the connection node neighborhood transformation matrix is applied to the corresponding feature elements to obtain the updated feature elements, and the updated map is constructed based on the updated feature elements. The optimization items selected during optimization are mainly the key feature points in the map, such as the connection points, the connection node neighborhood transformation matrix corresponding to the optimized connection points is obtained, and then it is applied to the surrounding feature points, so that the iterative calculation of all feature points is not required, the overall update of the map is realized, and the overall calculation scale is controllable.

[0144] In addition, the embodiment of the present application also provides a map construction device, which comprises:

[0145] The acquisition module is configured to acquire a map and each original trajectory, and determine the feature matching relationship between the map and the original trajectories.

[0146] The construction module is configured to construct the joint constraint relationship between the map and the original trajectories based on the feature matching relationship.

[0147] an iteration module configured to iteratively optimize the joint constraint relationship to obtain an optimization result;

[0148] an updating module configured to update the map according to the optimization result to obtain an updated map.

[0149] The principle and implementation process of the map construction are implemented in the embodiment, and please refer to the above embodiments, which will not be repeated here.

[0150] In addition, the embodiment of the present application also proposes a terminal device, which comprises a memory, a processor, and a map construction program stored in the memory and executable on the processor. When the map construction program is executed by the processor, the steps of the map construction method described above are implemented.

[0151] Since the map construction program is executed by the processor, all the technical solutions of the above-mentioned embodiments are adopted, and at least all the beneficial effects brought by all the technical solutions of the above-mentioned embodiments are achieved, which will not be repeated here.

[0152] In addition, the embodiment of the present application also proposes a computer readable storage medium, which stores a map construction program. When the map construction program is executed by the processor, the steps of the map construction method described above are implemented.

[0153] Since the map construction program is executed by the processor, all the technical solutions of the above-mentioned embodiments are adopted, and at least all the beneficial effects brought by all the technical solutions of the above-mentioned embodiments are achieved, which will not be repeated here.

[0154] Compared with the prior art, the map construction method, device, terminal device and storage medium proposed by the embodiment of the present application, by acquiring a map and each original trajectory, and determining the feature matching relationship between the map and each original trajectory; based on the feature matching relationship, constructing the joint constraint relationship between the map and each original trajectory; iteratively optimizing the joint constraint relationship to obtain an optimization result; updating the map according to the optimization result to obtain an updated map. By acquiring the map and each original trajectory and aligning, the aligned trajectory is further determined to determine the feature matching relationship between each original trajectory and the map, and then the joint constraint relationship is constructed and iteratively optimized to obtain an updated map, which avoids the influence of individual low-precision trajectory on the overall quality of the map, realizes the joint optimization of the map with reference to all trajectories, can repair the systematic problems of the map, and improves the overall precision of the map.

[0155] It should be noted that, in this document, the terms "comprises", "comprising", or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element preceded by "comprises... a" does not, without more constraints, foreclose the existence of additional identical elements in the process, method, article, or apparatus that comprises the recited element.

[0156] The above-mentioned sequence numbers of the embodiments of the present application are only for description, and do not represent the advantages or disadvantages of the embodiments.

[0157] From the above description of the embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be realized by means of software and necessary general hardware platforms, and of course can also be realized by hardware, but in many cases the former is a better embodiment. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which is stored in a storage medium (such as a ROM / RAM, a magnetic disk, or an optical disk) and includes a plurality of instructions for causing a terminal device (which can be a mobile phone, a computer, a server, a controlled terminal, or a network device) to execute the method of each embodiment of the present application.

[0158] The above is only the preferred embodiment of the present application, and does not limit the patent scope of the present application, and any equivalent structure or equivalent flow transformation made by using the content of the specification and drawings, or direct or indirect application in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A map construction method characterized by comprising: The map construction method comprises the following steps: acquiring a map and each original trajectory, and determining a feature matching relationship between the map and the original trajectories; based on the feature matching relationship, constructing a joint constraint relationship between the map and the original trajectories, the joint constraint relationship comprising a trajectory-map constraint relationship, a trajectory constraint relationship and / or a map constraint relationship; iteratively optimizing the joint constraint relationship to obtain an optimization result; updating the map according to the optimization result to obtain an updated map; the step of iteratively optimizing the joint constraint relationship to obtain an optimization result comprises: configuring a weight parameter of each constraint relationship in the joint constraint relationship; inputting each constraint relationship in the joint constraint relationship and the corresponding weight parameter into a preset graph optimization iterator to iteratively optimize each connection node of the map through the graph optimization iterator to obtain an optimized each connection node; obtaining the optimization result according to the optimized each connection node.

2. The map construction method according to claim 1, wherein The step of determining the feature matching relationship between the map and the original trajectories comprises: aligning the original trajectories with the map through a preset alignment algorithm to obtain each aligned trajectory; determining the feature matching relationship between the map and the original trajectories according to the each aligned trajectory.

3. The map construction method according to claim 2, wherein The step of determining the feature matching relationship between the map and the original trajectories according to the each aligned trajectory comprises: searching for a matching relationship between feature points of the each aligned trajectory and feature points of the map through a tree data structure algorithm, wherein the feature points of the each aligned trajectory correspond to each feature serial number respectively; obtaining the feature matching relationship between the map and the original trajectories according to the matching relationship between the feature points of the each aligned trajectory and the feature points of the map in combination with the each feature serial number.

4. The map construction method according to claim 3, wherein The feature points of the map comprise each connection node, and the step of constructing the joint constraint relationship between the map and the original trajectories based on the feature matching relationship comprises: determining each constraint term according to the feature matching relationship; determining a to-be-optimized term according to the feature points of the map, wherein the to-be-optimized term comprises the each connection node; constructing the trajectory-map constraint relationship based on the to-be-optimized term and each constraint term.

5. The map construction method according to claim 4, wherein The step of constructing the joint constraint relationship between the map and the original trajectories based on the feature matching relationship further comprises: constructing the trajectory constraint relationship according to the feature points of the original trajectories; and / or constructing the map constraint relationship according to the feature points of the map.

6. The map construction method according to claim 5, wherein The feature points of the map further comprise feature elements corresponding to the each connection node, and the step of updating the map according to the optimization result to obtain an updated map comprises: comparing the optimized each connection node in the optimization result with the each connection node before optimization to obtain each connection node position change; obtaining a corresponding connection node neighborhood transformation matrix according to the each connection node position change respectively; applying the connection node neighborhood transformation matrix to the corresponding feature element respectively to obtain an updated each feature element; An updated map is constructed based on the updated feature elements.

7. A map construction apparatus characterized by comprising: The map construction device comprises: an acquisition module configured to acquire a map and original trajectories, and determine a feature matching relationship between the map and the original trajectories; a construction module configured to construct a joint constraint relationship between the map and the original trajectories based on the feature matching relationship, the joint constraint relationship comprising a trajectory-map constraint relationship, a trajectory constraint relationship, and / or a map constraint relationship; an iteration module configured to iteratively optimize the joint constraint relationship to obtain an optimization result; an update module configured to update the map according to the optimization result to obtain an updated map; the iteration module is further configured to configure weight parameters of each constraint relationship in the joint constraint relationship; input each constraint relationship in the joint constraint relationship and the corresponding weight parameters into a preset graph optimization iterator to iteratively optimize each connection node of the map by using the graph optimization iterator to obtain optimized connection nodes; and obtain the optimization result according to the optimized connection nodes.

8. A terminal device, comprising: The terminal device comprises a memory, a processor, and a map construction program stored on the memory and executable on the processor, and the map construction program, when executed by the processor, implements the steps of the map construction method according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a map construction program, and the map construction program, when executed by the processor, implements the steps of the map construction method according to any one of claims 1-6.

Images