CN115540879A

CN115540879A - Road network matching method and device, computer equipment and readable storage medium

Info

Publication number: CN115540879A
Application number: CN202211142739.7A
Authority: CN
Inventors: 刘健乔; 李隽颖; 李峰
Original assignee: Shenzhen Yishi Huolala Technology Co Ltd
Current assignee: Shenzhen Yishi Huolala Technology Co Ltd
Priority date: 2022-09-19
Filing date: 2022-09-19
Publication date: 2022-12-30

Abstract

The application discloses a road network matching method. The road network matching method comprises the following steps: preprocessing road network data of a map and track data of characteristic vehicles, wherein the road network data comprises a plurality of roads, and the track data comprises a plurality of track points; calculating relative position data between the track point and the road; and obtaining a road matched with the track data according to the relative position data through the path planning model and the path matching model. The application also discloses a road network matching device, computer equipment and a computer readable storage medium. The path can be planned among a plurality of track points with low sampling frequency through the path planning model, and then the path planning model is matched with the path matching model to carry out road network matching, so that the accuracy of the matched road is high.

Description

Road network matching method and device, computer equipment and readable storage medium

Technical Field

The present disclosure relates to the field of road network matching technologies, and in particular, to a road network matching method, a road network matching apparatus, a computer device, and a computer-readable storage medium.

Background

In many intelligent navigation map applications, one of the most important links is to accurately associate a time-ordered GPS position trajectory sequence to a map road network, and this process is called road network matching. Under the condition of low sampling frequency, the GPS signals of adjacent time stamps can have time difference of up to tens of seconds, for example, when a truck travels on a fast lane such as a high speed lane, a national lane, a provincial lane and the like, the truck can travel through a plurality of lanes in a short time and reach a long distance, and these conditions may cause that roads between continuous high-speed track matching points are not communicated, thereby affecting the accuracy of road network matching.

Disclosure of Invention

In order to solve at least one technical problem in the foregoing background, embodiments of the present application provide a road network matching method, a road network matching apparatus, a computer device, and a computer-readable storage medium.

The road network matching method according to the embodiment of the present application includes:

preprocessing road network data of a map and track data of characteristic vehicles, wherein the road network data comprises a plurality of roads, and the track data comprises a plurality of track points;

calculating relative position data between the track point and the road;

and obtaining a road matched with the track data according to the relative position data through a path planning model and a path matching model.

In some embodiments, the road network matching method further comprises:

and updating the path matching model based on the matching result of the track data and the road and the truth value annotation.

In some embodiments, pre-processing road network data of a map and trajectory data of a characteristic vehicle comprises:

acquiring the road network data and the track data;

filtering out non-motor vehicle roads in the road network data; and

and filtering abnormal field data in the track data.

In some embodiments, calculating relative position data between the trajectory point and the road comprises:

indexing a first road segment set matched with each track point;

calculating a first point set matched with the track point in the matched first path set; and

and screening the first road section set and the first point set to obtain a second road section set and a second point set.

In some embodiments, calculating a first set of points that the track points match in the first set of matched segments comprises:

calculating the shortest distance from the track point to each road section in the first road section set; and

and when the shortest distance is obtained, mapping the track points to points on the road section as the points of the first point set.

In some embodiments, screening the first set of road segments and the first set of points to obtain a second set of road segments and a second set of points comprises:

screening out points in the first point set, wherein the shortest distance of the points is greater than a distance threshold value, and obtaining a second point set; and

and screening road sections corresponding to the screened points in the first road section set to obtain a second road section set.

In some embodiments, obtaining a road matched with the trajectory data according to the relative position data by a path planning model and a path matching model includes:

calculating the observation probability of the track points and the initial probability of a track sequence, wherein the track sequence consists of a plurality of track points;

calculating the transition probability from one track point to the next track point; and

and calculating a road matched with the track data according to the initial probability, the observation probability and the transition probability.

The road network matching device according to the embodiment of the present application includes:

the map preprocessing module is used for preprocessing road network data of a map and track data of characteristic vehicles, wherein the road network data comprises a plurality of roads, and the track data comprises a plurality of track points;

the calculation module is used for calculating relative position data between the track point and the road;

and the matching module is used for obtaining a road matched with the track data according to the relative position data through a path planning model and a path matching model.

The computer device of the embodiment of the application comprises: one or more processors; a memory; and one or more computer programs, wherein the one or more computer programs are stored in the memory and configured to be executed by the one or more processors, the one or more computer programs configured to: the road network matching method according to any of the embodiments of the present application is executed.

The non-volatile computer-readable storage medium of the embodiments of the present application stores a computer program, which, when executed by one or more processors, causes the processors to execute the road network matching method according to any of the embodiments of the present application.

In the road network matching method, the road network matching device, the computer equipment and the computer readable storage medium, the path can be planned among a plurality of track points with low sampling frequency through the path planning model, and then the path planning model is matched with the road network matching model to match the road network, so that the accuracy of the matched road is high.

Additional aspects and advantages of embodiments of the present application will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the present application.

Drawings

The above and/or additional aspects and advantages of the present application will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which:

fig. 1 is a schematic flow chart of a road network matching method according to a first embodiment of the present application;

fig. 2 is a schematic flow chart of a road network matching method according to a second embodiment of the present application;

fig. 3 is a schematic flow chart of a road network matching method according to a third embodiment of the present application;

fig. 4 is a schematic flow chart of a road network matching method according to a fourth embodiment of the present application;

fig. 5 is a schematic flowchart of a road network matching method according to a fifth embodiment of the present application;

fig. 6 is a schematic flow chart of a road network matching method according to a sixth embodiment of the present application;

fig. 7 is a schematic view of a scene of a road network matching method according to a seventh embodiment of the present application;

fig. 8 is a block diagram schematically illustrating a road network matching apparatus according to an eighth embodiment of the present application;

fig. 9 is a block diagram schematically illustrating a road network matching apparatus according to a ninth embodiment of the present application;

FIG. 10 is a schematic representation of a computer readable storage medium in communication with a processor according to some embodiments of the present application;

FIG. 11 is a schematic diagram of a computer device according to some embodiments of the present application.

Detailed Description

Reference will now be made in detail to embodiments of the present application, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below by referring to the drawings are exemplary only for the purpose of explaining the embodiments of the present application, and are not to be construed as limiting the embodiments of the present application.

In many intelligent navigation map applications, one of the most important links is to accurately associate a time-ordered GPS position trajectory sequence to a map road network, and this process is called road network matching. Two factors need to be considered for road network matching, wherein the distance between the position of a track point and a road, the included angle between the direction of the track point and the direction of the road, the relation between the speed of the track point and the speed limit of the road, the topological relation between the road and the self attribute of the road are considered.

In road network matching, the following general problems are easy to exist: 1. road network matching of sparse GPS signals caused by low sampling frequency (> 10 s/point); 2. correcting the GPS signals affected by the noise with low confidence coefficient and matching the road network; road network matching under the condition of serious continuous missing of GPS track data; 4. and the map road network is dense, and the map road network is not matched in a good or wrong way under the condition of missing. The general problem described above is solved by the road network matching method according to the embodiment of the present application.

Referring to fig. 1, fig. 1 is a schematic flow chart of a road network matching method according to a first embodiment of the present application, and the road network matching method according to the embodiment of the present application includes the steps of:

01: preprocessing road network data of a map and track data of characteristic vehicles, wherein the road network data comprises a plurality of roads, and the track data comprises a plurality of track points;

02: calculating relative position data between the track point and the road;

03: and obtaining a road matched with the track data according to the relative position data through the path planning model and the path matching model.

The path can be planned among a plurality of track points with low sampling frequency through the path planning model, and then the path planning model is matched with the path matching model to carry out road network matching, so that the accuracy of the matched road is high.

Specifically, in step 01, the road network data of the map and the trajectory data of the characteristic vehicles are preprocessed, in one example, the preprocessing includes, but is not limited to, collecting, classifying, screening and the like the road network data and the trajectory data, and the road network data and the trajectory data are preprocessed, so that the road network data and the trajectory data can be conveniently used for calculation. The road network data includes a plurality of roads, which may be represented by a plurality of line segments on a map, for example, and the trajectory data includes a plurality of trajectory points, which may be represented by a plurality of points on the map, without limitation. Wherein the location of the plurality of trace points may be represented by a GPS location.

In step 02, relative position data between the track point and the road is calculated, and the relative position data can reflect the relative position between the track point and the road, such as the distance, angle and other data between the track point and the road, so that the relative position data can be conveniently input into a path planning model and a path matching model for matching.

In step 03, a road matched with the trajectory data is obtained according to the relative position data through the path planning model and the path matching model.

The path planning model can quickly search the shortest communication path and length between the positions of any two continuous road network matching points, and the road network matching correctness under the low acquisition frequency is ensured. Meanwhile, when the method is applied to a freight service scene, the visual analysis finds that partial track points have track loss in different degrees, namely, the situations that the time stamp difference of continuous track points is too long and the coordinate distance of the continuous track points is too far exist in a track sequence. In order to avoid the loss of the road network matching accuracy rate caused by the missing track, when the missing track is detected (for example, the difference between timestamps of adjacent track points is greater than 15 s), the path planning module automatically performs shortest path planning compensation on the track between the missing points, and records the timestamp, the missing time length, the missing linear mileage and other information of the missing point.

Further, in one example, a CH (context technologies) path planning is added to the path planning model, and compared with a multi-purpose Dijkstra (Dijkstra) or a path planning algorithm (both of which have the problems of high time complexity and high resource consumption) in the existing map matching technology, the CH path planning is used for layering and segmenting roads in the whole country, so that the accuracy of long-mileage path planning is efficiently ensured, the influence of low-frequency sampling and trace missing on the road network matching effect is effectively solved, and a data basis is provided for accurate freight positioning, mileage calculation and pricing.

The path matching Model may employ a Hidden Markov Model (HMM) framework. The HMM has the advantages that the information of the road network data and the information of the positioning source can be comprehensively considered, and the accuracy of road network matching is improved.

Referring to fig. 2, fig. 2 is a schematic flow chart of a road network matching method according to a second embodiment of the present application, and in some embodiments, the road network matching method further includes step 04: and updating the path matching model based on the matching result of the track data and the road and the true value label.

Specifically, according to the experience of a real driving scene, the global rule is applied to optimize the matching parameters and the matching effect of the road network matching model, and the global rule comprises the steps of preferentially selecting main road matching, reducing the turning frequency and the like. And inputting the matching result of the track data and the road as training data of each iteration of the machine learning optimizer, and manually marking the matched road with a true value meeting the global rule. And then, the optimizer calculates loss based on a global rule according to actual matching and a true value, calculates gradient updating parameters by applying a stochastic gradient descent Strategy (SGD), and updates the optimized parameters in a path matching model based on hidden Markov (HMM) after multiple rounds of training convergence.

After the road network matching is completed, according to a real freight traffic driving scene and business experience, loss is defined by applying a global rule based on a machine learning method, the method comprises the steps of preferentially selecting main road matching, reducing turning frequency and turning frequency, applying a random gradient descent Strategy (SGD) to internal parameters of a path matching model for updating and optimizing, and the optimized path matching model is more stable in the scenes of complex road network, serious track drift and low-frequency reporting.

The method carries out effect evaluation on the road network matching method of the implementation mode based on real freight service order track data extracted from a production environment, wherein the test data comprises thousands of track data of domestic short distance (0-4000 m), intermediate distance (4000-6000 m), long distance (6000 m-50000 m) and ultra-long distance (= 50000 m), and the road network matching method shows stable and excellent matching effect under the condition that partial test data comprises track loss, noise and road network loss at different degrees, wherein the average point matching accuracy can reach 98%, and the matching is guaranteed to be completed within the time scale of hundred milliseconds. Meanwhile, the relation among the matching accuracy, the time delay of the matching result and the real-time memory occupation amount is explored, and on the premise of ensuring the matching effect of the invention, high timeliness, low memory consumption and high robustness are achieved.

Referring to fig. 3, fig. 3 is a schematic flow chart of a road network matching method according to a third embodiment of the present application, in some embodiments, step 01: the method for preprocessing the road network data of the map and the track data of the characteristic vehicles comprises the following steps:

011: acquiring road network data and track data;

012: filtering non-motor vehicle roads in the road network data; and

013: and filtering abnormal field data in the track data.

Through

steps

011, 012, and 013, the road network data and trajectory data are preprocessed and stored. Firstly, the road network data compiled by binary system is analyzed, the road network data comprises spatial index data and road information data, and the road network data is loaded in a memory in a singleton mode after being analyzed for being called by the subsequent steps. And then filtering the non-motor vehicle lanes and eliminating and filtering the track data abnormal fields in the road network. In one example, the trajectory data includes GPS trajectory points containing longitude, latitude, timestamp, speed, angle, driver id, etc. information. The road network data includes information such as road id, road coordinates, road start and end point id, road grade attributes, and traffic direction.

The

steps

011, 012 and 013 load the national road network data into the memory, which can improve the road network matching operation speed. Meanwhile, the track data is checked and filtered, if abnormal fields (such as longitude and latitude negative numbers, missing, abnormal timestamp and the like) occur, abnormal filtering or track mismatching is prompted, and then the track data is transmitted to the subsequent step of road network matching.

Referring to fig. 4, fig. 4 is a schematic flow chart of a road network matching method according to a fourth embodiment of the present application, in some embodiments, step 02: calculating relative position data between the trajectory point and the road, including:

021: indexing a first path set matched with each track point;

022: calculating a first point set matched with the track point in the matched first path set; and

023: and screening the first road section set and the first point set to obtain a second road section set and a second point set.

Steps

021, 022 and 023 are described in detail below:

in step 021, a first road segment set with each track point matched is indexed, and a track sequence converted according to time sequence is set as { p } ₁ ,p ₂ ,p ₃ ,L,p _n-1 ,p _n N represents the total number of trace points; let p _i And (3) representing the ith track point, wherein i is more than or equal to 1 and less than or equal to n. Generally, a tracing point at least contains information such as longitude coordinates, latitude coordinates, vehicle speed, etc. of the point, i.e. p _i ＝(lon _i ,lat _i ,v _i ) L represents the meaning of an ellipsis (the same applies hereinafter).

In one example, track point p is searched in road network data by using UberH3 hexagonal grid index _i All matching road segments. The specific method comprises the following steps: with p _i Searching p in road network data by taking position point as center _i All road sections around the h3 grid and extending outwards for a circle to be within 7 grid ranges are taken as a matched road section set, and the search range is recorded as

Wherein l _i Representing points of track p _i The total number of the matched road sections, the range r is a parameter which is manually set according to the actual condition and experience of the road sections, and the range of the search grid is 65.9 meters. If l _i If =0, the road network of the road segment is considered to be missing, and no processing is performed. By the method, the first road segment set matched with all the track points can be indexed.

In step 022, a first point set of the track points matching in the first matched segment set is calculated, specifically, please refer to fig. 5, fig. 5 is a flowchart of a road network matching method according to a fifth embodiment of the present application, and in some embodiments, step 022 includes the steps of:

0221: calculating the shortest distance from the track point to each road section in the first road section set; and

0222: and mapping the track points to points on the road section when the shortest distance is obtained, and taking the points as the points of the first point set.

Calculating the trace point p _i To each section of the first route set

And the point p is located below the shortest distance _i To a mapping point on the road section

As p _i A point mapped on the link, the point being a point in the first set of points. If p is _i To road section

If there is vertical distance, selecting the vertical mapping point under the vertical distance as the point of the first point set, if there is no vertical distance, selecting the point p on the road segment to the track point _i The closest end point, i.e. the starting point or the end point, is the point in the first set of points. All points forming a locus point p _i Is recorded as a first set of points

In step 023, the first road segment set and the first point set are filtered to obtain a second road segment set and a second point set, specifically referring to fig. 5, in some embodiments, step 023 includes the steps of:

0231: screening out points in the first point set, wherein the shortest distance is greater than a distance threshold value, and obtaining a second point set; and

0232: and screening road sections corresponding to the screened points in the first road section set to obtain a second road section set.

After calculating to obtain a first road section set and a first point set of all track points in the track sequence, roughly screening the first road section set and the first point set, specifically, setting a threshold value D, and sequentially comparing each point in the first point set to a track point p _i Of (2) is

And the size of D, if

Deleting the points in the corresponding first point set and the road section in the first road section, and otherwise, keeping the points. Finally, a new second road section set is obtained

And a second set of points

Wherein k is _i Indicating the number of links and points remaining after screening. The parameter D is related to the actual conditions of the road network surroundings, and in one example, may be 80 meters.

Referring to fig. 6, fig. 6 is a schematic flow chart of a road network matching method according to a sixth embodiment of the present application, in some embodiments, step 03: obtaining a road matched with the track data according to the relative position data through a path planning model and a path matching model, and the method comprises the following steps:

031: calculating the observation probability of the track points and the initial probability of a track sequence, wherein the track sequence consists of a plurality of track points;

032: calculating the transition probability from one track point to the next track point; and

033: and calculating a road matched with the track data according to the initial probability, the observation probability and the transition probability.

Specifically, in step 031, the trajectory point p is calculated _i And the initial probability of the trajectory sequence. Wherein the initial probability is p ₁ The probability of observation of (2). The observation probability is influenced by two characteristics, i.e., the shortest distance between the track point and the road and the angle difference, and is the product of the distance observation probability and the angle observation probability in this embodiment. According to practical conditions and experience, supposing a track point p _i The distance observation probability obeys the Gaussian distribution of the Euclidean distance between the track point and the matching point, the angle observation probability obeys the Gaussian distribution of the angle difference, and then when the track point p is subjected to the Gaussian distribution _i The time matching point is

Probability of (2)

Satisfies the following formula:

wherein, mu and sigma ² The expected variance and the variance are respectively regarded as matching parameters and are optimized and updated after being initialized manually. When the observation probability is lower than a threshold, we make the observation probability equal to the threshold in order to make the probability still exist, the threshold is set to 10 in the present invention ^-7 。

In step 032, calculate the trace point p _i-1 To p _i The transition probability. The transition probability is the product of the distance transition probability and the angle transition probability.

Calculating the trace point p _i-1 To p _i The manner of transferring the angle probability of (1) is: calculating p assuming that the angle probability obeys the Gaussian distribution of the angle _i-1 And p _i The included angle between the formed vector angle and the road on which the second concentrated points of the two track points are positioned

And normalization processing is carried out to ensure that the angles all meet

Obtaining corresponding transition angle probability P in the same way as the formula (1) _q 。

Calculating to obtain a track point p _i-1 To p _i The transition distance probability of (2) is: suppose a locus point p _i The distances to the points in each second point set satisfy the Gaussian distribution of the distances, and due to the low-frequency sampling, a CH-based path planning module is applied to obtain the point p _i-1 To p _i Then calculating the difference between the shortest path distance and the two-point straight line distance and projecting the difference to Gaussian distribution, and obtaining the corresponding transition distance probability P in the same way as the formula (1) _d 。

Calculating to obtain a track point p _i-1 To p _i The transition probability of (c) is as follows: transition probability P _T Is the product of the angle probability and the distance probability, i.e.

The transition probability comprehensively considers the actual condition of road network sections and track point data, a bottom probability threshold value is set for the transition probability in the embodiment, when the transition probability is smaller than the threshold value, the transition probability is equal to the threshold value, and the threshold value is set to be 10 in the invention ^-7 。

In step 033, according to the initial probability, the observation probability and the transition probability, calculating a road matched with the track sequence by using a viterbi algorithm module:

starting from the track sequence p ₁ Starting a point, taking a point in the n second point sets with the highest observation probability (hereinafter referred to as a matching point) as p ₁ The candidate link binding point of (2) is selected to be 10 according to the experience n and recorded.

Calculating p ₁ Each matching point to p ₂ Cumulative probability P of each matching point, where P = P _O gP _T Selecting 10 p that maximizes the cumulative probability ₂ Matching point as p ₂ Binding the matching points to the candidate road sections. And the cumulative probability of each path is normalized and recorded.

Analogizing in turn, for the locus point p _i-1 To p _i Cumulative probability of, p to be recorded _i-1 Cumulative probability of and p _i-1 To p _i Multiplying the transition probabilities of the matching points, chosen such that p _i The 10 matching points with the maximum point accumulation probability are taken as p _i Binding the matching points to the candidate road sections. The cumulative probabilities are normalized and recorded.

After the whole track sequence is processed, the obtained probability matrix is traced back, the matching point with the highest end point probability is selected to trace back the index record value in the Viterbi algorithm, as shown in FIG. 7, the matching point with the highest probability shown by the solid black point at the time of t +1 is traced back to the time of t-1, the final road section binding matching point of each track point can be obtained, the traced back path is the best path matched with the road network, and the road network matching is completed after the path is decompressed and restored according to the time sequence.

Further, in the implementation of step 04, the global rule for optimization is as follows:

a) High main road weight

According to the experience of the freight scene, the truck is more inclined to drive on the main road with high speed limit. Therefore, different weights are set for different road grades, higher weights are set for high speed limit roads, and the weights are reduced along with the reduction of the speed limit. Here, the weight is defined in the following manner

l is the rate limit, in m/s. Then calculating the current matching profit r by the weighted calculation negative weighted projection distance _m I.e. r _m And = w · d, where d is the projected distance of the trajectory to the road. According to the matching yield, calculating the difference value between the truth value yield and the matching yield as the loss l _m I.e. l _m ＝Δr _m And the method is used for iteratively optimizing the matching model parameters.

b) Low frequency of turning

To avoid unnecessary turns, the difference between the number of turns between the true value and the matching result is measured, i.e. the lower the difference, the lower the loss, which is defined as follows

Here T _x True number of turns, T _y To match the number of turns.

c) Low frequency of turning around

To avoid unnecessary turnaround, the difference between the number of turnaround between the step measurement truth value and the matching result is measured, i.e. the lower the difference, the lower the loss. Loss l _U Is defined as formula (2) wherein T _x Substituted by true value of tap number U _x ，T _y Substituted by a matching number of taps U _y 。

In combination with the above three global rules, the final match optimization total loss is defined in the following way

l＝l _m ·λ _m +l _T ·λ _T +l _U ·λ _U (4)

Wherein λ _m ，λ _T ，λ _U Is a constant. And then calculating a loss gradient, and applying a random gradient descent strategy to update the parameters of the matching model in a plurality of rounds of iteration until the loss is converged. And finally, transplanting the parameters with stable multiple training to a matching model to obtain a stable and optimal matching effect.

In addition, the subject hidden markov model in the present application mainly includes the following 5 elements:

observation state O: states that can be obtained by direct observation, i.e. the appearance of implicit states. The observation state in the present invention refers to track data information (such as latitude and longitude, travel speed, angle, road information, etc.).

Probability of observation P _O : the probability that an implicit state is in an observed state at a certain moment. The probability that the observed track data is mapped to the candidate road sections in the map road network is referred to in the invention, generally speaking, the observation probability is related to the distance and angle difference between the track data point and each road in the road network, and the larger the distance and angle difference is, the smaller the observation probability is.

Initial probability P ₁ : the probability that an implicit state is in a certain state at the initial moment. In the present invention, the probability that the trajectory data is located on a certain road segment in the road network at the initial time is generally expressed by the observation probability at the initial time.

Implicit State S: the actual hidden state is usually not available by direct observation. The Markov property is satisfied between the states, i.e., the present state is conditionally independent from the past state. Implicit in the present invention is the track of the location where an object (e.g., truck, person, etc.) with a GPS device is actually located.

Transition probability P _T : the probability that an implicit state is another implicit state at the next moment in time. In the present invention, the next time the vehicle is transferred toThe probability of another position point, generally speaking, the closer the distance of the communication path between the front and rear matching points on the real road section of the road network is to the distance between the front and rear track points, and the closer the angle difference is, the greater the transition probability is.

In summary, the road network matching method in the embodiment of the present application can solve the problem of high difficulty road network matching under the conditions of low sampling frequency, poor signal quality, missing track and extremely dense road network.

Referring to fig. 8, fig. 8 is a block diagram illustrating a road network matching device 10 according to an eighth embodiment of the present application, in which in some embodiments, the road network matching device 10 includes a preprocessing module 11, a calculating module 12, and a matching module 13. The preprocessing module 11 may be configured to perform step 01, that is, the preprocessing module 11 may be configured to preprocess road network data of a map and trajectory data of a characteristic vehicle, where the road network data includes a plurality of roads, and the trajectory data includes a plurality of trajectory points; the calculation module 12 may be configured to implement step 02, that is, the calculation module 12 may be configured to calculate relative position data between the trajectory point and the road; the matching module 13 may be configured to perform step 03, that is, the matching module 13 may be configured to obtain a road matched with the trajectory data according to the relative position data through the path planning model and the path matching model.

Referring to fig. 9, fig. 9 is a block diagram illustrating a road network matching device 10 according to a ninth embodiment of the present invention, in some embodiments, the road network matching device 10 further includes an updating module 14, and the updating module 14 is configured to implement step 04, that is, the updating module 14 is configured to update the path matching model based on the matching result of the trajectory data and the road and the truth label.

Referring again to fig. 8, in some embodiments, the preprocessing module 11 can be used to implement

steps

011, 012, and 013, that is, the preprocessing module 11 can be used to obtain road network data and trajectory data; filtering non-motor vehicle roads in the road network data; and filtering abnormal field data in the track data.

Referring again to fig. 8, in some embodiments, calculation module 12 may be configured to perform

steps

021, 022, and 023, i.e., calculation module 12 may be configured to index the first set of segments for which each trace point matches; calculating a first point set matched with the track points in the matched first path set; and screening the first road section set and the first point set to obtain a second road section set and a second point set.

Referring back to fig. 8, in some embodiments, calculation module 12 may be configured to perform

steps

0221 and 0222, i.e., calculation module 12 may be configured to calculate the shortest distance from the point of the track to each segment in the first set of segments; and when the shortest distance is obtained, mapping the track points to points on the road section as the points of the first point set.

Referring again to fig. 8, in some embodiments, the calculation module 12 may be configured to perform

steps

0231 and 0232, i.e., the calculation module 12 may be configured to filter out the points in the first set of points whose shortest distance is greater than the distance threshold to obtain a second set of points; and screening the road sections corresponding to the screened points in the first road section set to obtain a second road section set.

Referring back to fig. 8, in some embodiments, the matching module 13 may be configured to implement

steps

031, 032, and 033, that is, the matching module 13 may be configured to calculate an observation probability of a track point and an initial probability of a track sequence, where the track sequence is composed of a plurality of track points; calculating the transition probability from one track point to the next track point; and calculating a road matched with the track data according to the initial probability, the observation probability and the transition probability.

It should be noted that details of implementation and effects achieved when the road network matching device 10 implements the road network matching method according to any embodiment of the present application may refer to the description of the road network matching method, and are not described herein again.

Referring to fig. 10, an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the road network matching method according to any of the above embodiments is implemented. The computer-readable storage medium includes, but is not limited to, any type of disk including floppy disks, hard disks, optical disks, CD-ROMs, and magnetic-optical disks, ROMs (Read-Only memories), RAMs (Random Access memories), EPROMs (Erasable Programmable Read-Only memories), EEPROMs (Electrically Erasable Programmable Read-Only memories), flash memories, magnetic cards, or optical cards. That is, a storage device includes any medium that stores or transmits information in a form readable by a device (e.g., a computer, a cellular phone), and may be a read-only memory, a magnetic or optical disk, or the like.

The contents of the method embodiments of the present application are all applicable to the storage medium embodiments, the functions specifically implemented by the storage medium embodiments are the same as those of the method embodiments, and the beneficial effects achieved by the storage medium embodiments are also the same as those achieved by the method described above, and for details, refer to the description of the method embodiments, and are not described herein again.

In addition, referring to fig. 11, an embodiment of the present application further provides a computer device, where the computer device described in this embodiment may be a server, a personal computer, a network device, and other devices. The computer device includes one or more processors, memory, and one or more computer programs. Wherein the one or more computer programs are stored in the memory and configured to be executed by the one or more processors. One or more computer programs are configured to perform the road network matching method according to any of the above embodiments.

In the description herein, reference to the description of the terms "one embodiment," "some embodiments," "an illustrative embodiment," "an example," "a specific example" or "some examples" or the like means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the application. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, various embodiments or examples and features of different embodiments or examples described in this specification can be combined and combined by one skilled in the art without contradiction.

Any process or method descriptions in flow charts or otherwise described herein may be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing specific logical functions or steps of the process, and the scope of the preferred embodiments of the present application includes other implementations in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order, depending on the functionality involved, as would be understood by those reasonably skilled in the art of the present application.

Although embodiments of the present application have been shown and described above, it is to be understood that the above embodiments are exemplary and not to be construed as limiting the present application, and that changes, modifications, substitutions and alterations can be made to the above embodiments by those of ordinary skill in the art within the scope of the present application.

Claims

1. A road network matching method is characterized by comprising the following steps:

calculating relative position data between the track point and the road; and

2. The road network matching method according to claim 1, further comprising:

and updating the path matching model based on the matching result of the track data and the road and the true value label.

3. The road network matching method according to claim 1, wherein preprocessing road network data of a map and trajectory data of a characteristic vehicle comprises:

acquiring the road network data and the track data;

filtering out non-motor vehicle roads in the road network data; and

and filtering abnormal field data in the track data.

4. The road network matching method according to claim 1, wherein calculating relative position data between said trajectory point and said road comprises:

indexing a first road section set matched with each track point;

calculating a first point set matched with the track points in the matched first path set; and

5. The road network matching method according to claim 4, wherein calculating a first set of points of the track points matching in the first set of matched road segments comprises:

6. The road network matching method according to claim 5, wherein the step of screening the first road segment set and the first point set to obtain a second road segment set and a second point set comprises:

screening out points in the first point set, wherein the shortest distance is greater than a distance threshold value, and obtaining a second point set; and

7. The road network matching method according to claim 1, wherein obtaining the road matched with the trajectory data according to the relative position data by a path planning model and a path matching model comprises:

8. A road network matching device, comprising:

9. A computer device, comprising:

one or more processors;

a memory; and

one or more computer programs, wherein the one or more computer programs are stored in the memory and configured to be executed by the one or more processors, the one or more computer programs configured to: performing the road network matching method according to any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium storing a computer program, wherein the computer program, when executed by one or more processors, causes the processors to perform the road network matching method according to any one of claims 1 to 7.