A method and device for determining the confidence level of road network data
Through the logistic regression model and dynamic time regularization algorithm, multiple factors are considered comprehensively to evaluate the confidence of road network data, the problem of difficult to determine the confidence of indoor road network data in the existing technology is solved, and the situation of insufficient prior information is realized. Accurate assessment.
Patent Information
- Application Number
- CN202111389722.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-11-19
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2041-11-19
AI Technical Summary
It is difficult for the prior art to accurately determine the confidence of indoor road network data, especially in the scenario where prior estimation information is insufficient, there are errors, missing and errors in indoor data acquisition, making it difficult to evaluate the accuracy of the data.
The road network data confidence confirmation method is adopted based on logistic regression model and dynamic time regularization algorithm. By comprehensively considering factors such as sensor accuracy, driver professionalism, data acquisition time and number of times, a comprehensive confidence calculation method is constructed, and dynamic time is used. The regression algorithm evaluates the matching degree of multiple sets of road network data to determine the accuracy of road network data.
In the scenario where indoor environment and a priori estimation information are insufficient, the accuracy of road network data can be effectively judged, which improves the accuracy of indoor road network data and is suitable for road network data in large areas and small spaces.
Smart Images

Figure CN114091219B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of vehicle technology, and in particular to a method for determining the confidence of road network data and a device for determining the confidence of road network data. Background Art
[0002] With the continuous development of vehicle-road cooperative technology, the importance of road network data continues to increase, becoming the basic information that today's autonomous driving / unmanned driving technology relies on. Road network data refers to the collection of road data, which includes posture information such as road location, direction, obstacles, and artificial information such as speed limits, one-way and two-way roads. The accuracy of the road network skeleton is particularly important for vehicle driving.
[0003] At present, the more mature road network structures include vector map formats developed by Baidu, Google and other map companies, as well as open source formats such as OpenStreetMap. Massive calibrated outdoor road network data can usually be obtained from the aforementioned platforms, but indoor data still needs to be continuously collected. However, for this type of indoor data that needs to be collected, there is often a lack of prior data, and the collected road network data may also have errors, omissions, mistakes, etc., making it difficult to determine the accuracy of the collected road network data. Summary of the invention
[0004] In view of the above problems, embodiments of the present invention are proposed to provide a method for determining the confidence of road network data and a corresponding device for determining the confidence of road network data that overcome the above problems or at least partially solve the above problems.
[0005] The embodiment of the present invention discloses a method for determining the confidence of road network data, the method comprising:
[0006] Obtaining a regression model for determining the confidence of road network data and the collected road network data; the regression model for determining the confidence of road network data is obtained by training using road network influencing factor information of road network sample data and weight values corresponding to each factor, wherein the road network influencing factor information includes a road network data accuracy parameter determined based on a time regression algorithm;
[0007] The confidence of the road network data is obtained based on the collected road network data and the regression model used to determine the confidence of the road network data.
[0008] Optionally, obtaining the confidence of the road network data according to the collected road network data and the regression model for determining the confidence of the road network data includes:
[0009] Determine the road network influencing factor information for the collected road network data;
[0010] The road network influencing factor information is used as an input item of the regression model for determining the confidence of the road network data, and the confidence of the collected road network data is output.
[0011] Optionally, the regression model for determining the confidence of the road network data is generated as follows:
[0012] Acquire the collected road network sample data, determine the road network influencing factor information for the road network sample data, and the weight value corresponding to each factor; the road network influencing factor information includes at least one of the following: data collection times, data collection user difference parameter, data collection time difference parameter, sensor accuracy and road network data accuracy parameter;
[0013] The road network influencing factor information and the weight values corresponding to the various factors are used, and the regression model is trained based on a preset hyperbolic tangent function to obtain a regression model for determining the confidence of the road network data.
[0014] Optionally, the determining of the road network influencing factor information for the road network sample data includes:
[0015] Acquire environmental data from the collected road network sample data, and determine the number of data collection times that affect the road network sample data based on a determination result of whether the collected environmental data is measured;
[0016] and / or, obtaining the role of a collector of the road network sample data, and determining a data collection user difference parameter based on the role of the collector;
[0017] And / or, obtaining the data collection time and the data use time of the road network sample data, and determining the data collection time difference parameter by constructing a decreasing function based on the difference between the data collection time and the data use time;
[0018] and / or, obtaining a sensor measurement error value used to collect the road network sample data, and determining the accuracy of the sensor that affects the road network sample data based on the sensor measurement error value;
[0019] And / or, determining the road network data accuracy parameters that affect the road network sample data through a time warping algorithm.
[0020] Optionally, determining the road network data accuracy parameters affecting the road network sample data by using a time warping algorithm includes:
[0021] Acquire prior data and newly added data from the road network sample data; the prior data includes historical road network data relative to newly collected road network data;
[0022] Extracting trajectory point coordinate sets of the prior data and the newly added data respectively, and constructing a dynamic time warping matrix according to the trajectory point coordinate sets;
[0023] Through the constructed dynamic time warping matrix, the road network data accuracy parameters affecting the road network sample data are obtained.
[0024] Optionally, the priori data and the newly added data include path number information; and the step of respectively extracting the trajectory point coordinate sets of the priori data and the newly added data includes:
[0025] Acquire, based on the path number information, a trajectory in the newly added data that overlaps with the prior data;
[0026] A set of trajectory point coordinates of the coincident trajectory in the prior data is extracted, and the coincident trajectory in the newly added data is adjusted, and a set of trajectory point coordinates of the adjusted coincident trajectory in the newly added data is extracted.
[0027] Optionally, the trajectory point coordinate set includes the trajectory point coordinates on the overlapping trajectory in the priori data and the trajectory point coordinates on the adjusted overlapping trajectory in the newly added data;
[0028] The constructing a dynamic time warping matrix according to the trajectory point coordinate set includes:
[0029] Determine the matching relationship between the coordinates of the trajectory points on the overlapping trajectory in the prior data and the coordinates of the trajectory points on the adjusted overlapping trajectory in the newly added data, and establish a dynamic time warping matrix for the road network sample data through the determined matching relationship.
[0030] Optionally, the road network data accuracy parameters affecting the road network sample data are obtained through the constructed dynamic time warping matrix, including:
[0031] The shortest path in the constructed dynamic time warping matrix is calculated to obtain the similarity between the curve for the prior data and the curve for the newly added data, so as to obtain the accuracy parameter of the road network data.
[0032] The embodiment of the present invention further discloses a device for determining the confidence of road network data, the device comprising:
[0033] A regression model acquisition module, used to acquire a regression model for determining the confidence of road network data, and the collected road network data; the regression model for determining the confidence of road network data is obtained by training the road network influencing factor information of the road network sample data and the weight values corresponding to each factor, wherein the road network influencing factor information includes a road network data accuracy parameter determined based on a time regression algorithm;
[0034] The confidence determination module is used to obtain the confidence of the road network data based on the collected road network data and the regression model used to determine the confidence of the road network data.
[0035] Optionally, the confidence determination module includes:
[0036] A first road network influencing factor information determination submodule, used to determine the road network influencing factor information for the collected road network data;
[0037] The confidence determination submodule is used to use the road network influencing factor information as an input item of the regression model used to determine the confidence of the road network data, and output the confidence of the collected road network data.
[0038] Optionally, the device further comprises:
[0039] The regression model generation module is used to train the road network influencing factor information of the road network sample data and the corresponding weight value of each factor to obtain a regression model for determining the confidence of the road network data.
[0040] Optionally, the regression model generation module includes:
[0041] The second road network influencing factor information determination submodule is used to obtain the collected road network sample data, determine the road network influencing factor information for the road network sample data, and the weight value corresponding to each factor; the road network influencing factor information includes at least one of the following: data collection times, data collection user difference parameters, data collection time difference parameters, sensor accuracy, and road network data accuracy parameters;
[0042] The regression model training submodule is used to adopt the road network influencing factor information and the weight values corresponding to each factor, and train the regression model based on a preset hyperbolic tangent function to obtain a regression model for determining the confidence of road network data.
[0043] Optionally, the second road network influencing factor information determination submodule includes:
[0044] A data collection frequency determination unit, used to obtain environmental data in the collected road network sample data, and determine the data collection frequency affecting the road network sample data based on a judgment result of whether the collected environmental data is measured;
[0045] A data collection user difference parameter determination unit, used for obtaining the role of a collector of the road network sample data, and determining a data collection user difference parameter based on the role of the collector;
[0046] A data collection time difference parameter determination unit, used for obtaining the data collection time and the data use time of the road network sample data, and determining the data collection time difference parameter by constructing a decreasing function based on the difference between the data collection time and the data use time;
[0047] a sensor accuracy determination unit, configured to obtain a sensor measurement error value used to collect the road network sample data, and determine a sensor accuracy that affects the road network sample data based on the sensor measurement error value;
[0048] The road network data accuracy determination unit is used to determine the road network data accuracy parameters that affect the road network sample data through a time warping algorithm.
[0049] Optionally, the road network data accuracy determination unit includes:
[0050] A data acquisition subunit, used to acquire prior data and newly added data from the road network sample data; the prior data includes historical road network data relative to newly collected road network data;
[0051] A dynamic time warping matrix construction subunit, used to extract the trajectory point coordinate sets of the prior data and the newly added data respectively, and construct a dynamic time warping matrix according to the trajectory point coordinate sets;
[0052] The road network data accuracy parameter determination subunit is used to obtain the road network data accuracy parameters that affect the road network sample data through the constructed dynamic time warping matrix.
[0053] An embodiment of the present invention also discloses a vehicle, comprising: a device for determining the road network data confidence, a processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein the computer program, when executed by the processor, implements the steps of any one of the methods for determining the road network data confidence.
[0054] An embodiment of the present invention further discloses a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of any one of the methods for determining the confidence of road network data are implemented.
[0055] The embodiments of the present invention include the following advantages:
[0056] In the embodiment of the present invention, a regression model for determining the confidence of road network data is obtained, and the obtained regression model is obtained by training the road network influencing factor information of the road network sample data and the weight values corresponding to each factor, wherein the road network influencing factor information includes the road network data accuracy parameter determined based on the time regression algorithm, and then the confidence of the collected road network data is obtained by inputting the collected road network data into the obtained regression model. By comprehensively considering multiple road network influencing factors that affect the accuracy of road network data, a more comprehensive confidence calculation method is constructed using a logistic regression model, and by using a dynamic time regression algorithm to propose an evaluation method for comparing the matching degree of each road in multiple groups of road network data, the confidence of the road network data is determined in an indoor environment or in a scene where a priori estimation information is insufficient, so as to determine the accuracy of the collected road network data. BRIEF DESCRIPTION OF THE DRAWINGS
[0057] Figure 1 It is a flowchart of the steps of an embodiment of a method for determining the confidence of road network data of the present invention;
[0058] Figure 2 is a function graph corresponding to the hyperbolic tangent function provided in an embodiment of the present invention;
[0059] Figure 3 is a flowchart of another method for determining the confidence level of road network data according to an embodiment of the present invention;
[0060] Figure 4 It is a schematic diagram of the process of calculating the accuracy of road network data provided by an embodiment of the present invention;
[0061] Figure 5 is a schematic diagram of constructing a dynamic time warping matrix provided by an embodiment of the present invention;
[0062] Figure 6 It is a structural block diagram of an embodiment of a device for determining the confidence of road network data of the present invention. DETAILED DESCRIPTION
[0063] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.
[0064] The accuracy of the road network skeleton of road network data is particularly important for vehicle driving. At present, the calculation method of the confidence of road network data is often determined by calculating data integrity, road name integrity, geometric information, etc. For example, the confidence of road network data is usually calculated by comparing the newly added road data and remote sensing data. These methods require that the roads can be photographed by remote sensing satellites and can be better applied to the inspection of road network quality in large vector spaces. However, the data confidence judgment for indoor environments such as parking lots and shopping malls or underground spaces lacks consideration of measurement errors and interference; in addition, the existing outdoor road network data confidence calculation method is obtained by comparing satellite remote sensing images with the road network conditions to be measured. For indoor or underground spaces, there is a lack of standardized, complete and accurate samples.
[0065] For indoor data that needs to be collected, there is often a lack of prior data, that is, many terrains are explored for the first time, and the collected road network data may also contain errors, omissions, and mistakes. Taking the indoor parking lot road network data that needs to be collected by autonomous driving vehicles as an example, the accuracy differences of different sensors, whether the driver is professional, whether the parking lot data is collected for the first time, and other factors will lead to different confidence levels in the obtained road network data. For this type of indoor environment, especially in scenarios where prior estimation information is insufficient, it is particularly important to propose a method to determine the confidence level of road network data.
[0066] This proposal mainly starts from human conditions, environmental factors and data accuracy, and designs a road network data confidence confirmation method based on logistic regression model and dynamic time warping algorithm. It is mainly based on determining the parameters corresponding to the factors that can affect the road network data, and then the parameters of the determined related factors are iteratively calculated and solved through network training or parameter fitting method to solve the entire model, that is, training is used to confirm the confidence of the road network data in the future. By comprehensively considering the accuracy differences of sensors, whether the driver is professional, data collection time, number of collections, accuracy and other factors that affect the road network data, a more comprehensive confidence calculation method is constructed using the logistic regression model, and it can be used for road network data in large areas and small spaces. By using the dynamic time regression algorithm, an evaluation method for comparing the matching degree of each road in multiple groups of road network data is proposed, and the confidence of the road network data is determined in indoor environments or scenarios where prior estimation information is insufficient to determine the accuracy of the collected road network data.
[0067] Reference Figure 1 , shows a flowchart of a method for determining the confidence level of road network data according to an embodiment of the present invention, which may specifically include the following steps:
[0068] Step 101, obtaining a regression model for determining the confidence of road network data, and the collected road network data;
[0069] In an embodiment of the present invention, the confidence of road network data can be determined based on a logistic regression model and a dynamic time warping algorithm. The constructed logistic regression model is a comprehensive confidence calculation method, which can be applicable to indoor environments or scenarios with insufficient prior estimation information, and can be used to determine the confidence of road network data in large areas and small spaces.
[0070] Among them, the constructed regression model can realize the function of determining the confidence of comprehensive road network data. This regression model for determining the confidence of road network data is mainly a model obtained by training the road network influencing factor information of road network sample data and the corresponding weight values of each factor. Among them, when constructing the regression model, multiple factors affecting the road network data, such as the accuracy difference of the sensor, whether the driver is professional, the data collection time, the number of collections, and the accuracy, are comprehensively considered, and the confidence can be determined in combination with the time regression algorithm, specifically, the road network data accuracy parameters in the road network influencing factor information are determined based on the time regression algorithm.
[0071] Step 102, obtaining the confidence of the road network data based on the collected road network data and the regression model used to determine the confidence of the road network data.
[0072] When determining the confidence of road network data, a regression model constructed to implement a comprehensive confidence calculation method can be obtained, and the object that needs to be determined in terms of confidence, that is, the collected road network data, can be input into the regression model, and the collected road network data can be determined by using the constructed regression model.
[0073] Specifically, to determine the confidence of the collected road network data, it is first necessary to determine the road network influencing factor information for the collected road network data, and then use the road network influencing factor information as an input item of the regression model for determining the confidence of the road network data, so as to output the confidence of the collected road network data.
[0074] Among them, it is assumed that the confidence level of the road network data is P. When P=0, it means that the road network data is unreliable, and when P=1, it means that the road network data is accurate. That is, the closer the confidence level of the road network data is to 1, the higher the accuracy of the road network data. The constructed logistic regression model is constructed by comprehensively considering multiple factors affecting the road network data, such as the accuracy difference of the sensor, whether the driver is professional, the data collection time, the number of collections, and the accuracy. The specific factors affecting the accuracy of the road network data include the number of data collections, abnormal parameters of the data collection user, the data collection time difference parameters, the sensor accuracy, and the data accuracy parameters. These road network influencing parameter information are all positive values.
[0075] Reference Figure 2, shows the function graph corresponding to the hyperbolic tangent function provided in the embodiment of the present invention. In the first quadrant of the function graph, when the independent variable (in the embodiment of the present invention, the road network influencing parameter information) is positive, the interval of the output dependent variable is 0-1, which is consistent with the determination interval for the confidence level of the road network data; and Figure 2 As shown in , the slope of the function rises rapidly at the beginning and tends to be flat later, which is consistent with the expectation that the independent variable changes less at the beginning, but can achieve the expectation of quickly improving its confidence. Based on the above reasons, Figure 2 The image of the hyperbolic tangent function tanh function shown can meet the requirements for determining the confidence of road network data.
[0076] The expression of the regression model constructed for determining the road network data can be expressed by using the hyperbolic tangent function tanh function, specifically:
[0077]
[0078] Z=KX
[0079] Where: X=[a num ,a per ,a time ,a sen ,a acc ] -1
[0080] Wherein, K can be a coefficient matrix, which can be used to represent the weights of various parameters in X, that is, the degree of influence of various parameters in X on the confidence of road network data. For the determination method of the coefficient matrix K, one method can be to evaluate multiple sets of input and output used to train the regression model and set it based on human experience. Another method is that in practical applications, the coefficient matrix K can be defined as the initial value, and different settings of the confidence value P can be iteratively set later; X in the formula is equivalent to the independent variable of the above equation Z = KX, which can be used to represent various influencing factors that can affect the road network data, where a num can be the number of data collection times, a per Can be used for data collection user difference parameters, a time can be the data collection time difference parameter, a sen is the sensor accuracy, a acc It is a data accuracy parameter, which can be determined by a time warping algorithm.
[0081] In the embodiment of the present invention, a regression model for determining the confidence of road network data is obtained, and the obtained regression model is obtained by training the road network influencing factor information of the road network sample data and the weight values corresponding to each factor, wherein the road network influencing factor information includes the road network data accuracy parameter determined based on the time regression algorithm, and then the confidence of the collected road network data is obtained by inputting the collected road network data into the obtained regression model. By comprehensively considering multiple road network influencing factors that affect the accuracy of road network data, a more comprehensive confidence calculation method is constructed using a logistic regression model, and by using a dynamic time regression algorithm to propose an evaluation method for comparing the matching degree of each road in multiple groups of road network data, the confidence of the road network data is determined in an indoor environment or in a scene where a priori estimation information is insufficient, so as to determine the accuracy of the collected road network data.
[0082] Reference Figure 3 , shows a flowchart of another method for determining the confidence of road network data according to an embodiment of the present invention, which may specifically include the following steps:
[0083] Step 301, obtaining the collected road network sample data, determining the road network influencing factor information for the road network sample data, and the weight value corresponding to each factor;
[0084] In the embodiment of the present invention, a more comprehensive confidence calculation method is constructed using a logistic regression model by comprehensively considering multiple factors affecting road network data, such as the accuracy differences of sensors, whether the driver is professional, data collection time, number of collections, and accuracy. The model is mainly trained by using the road network influencing factor information of the road network sample data and the corresponding weight values of each factor.
[0085] The information on road network influencing factors may include at least one of the following: data collection times a num , Data collection user difference parameter a per , data collection time difference parameter a time 、Sensor accuracy a sen And the road network data accuracy parameter a acc .
[0086] For the number of data collection times a num The environmental data in the collected road network sample data can be obtained, and the determination is made based on the judgment result of whether the collected environmental data is measured. Specifically, for the number of data collection times, the more times, the larger the total value of the road network data, and the higher the probability of the road network being complete. num When it is 0, it means that the environmental data has not been measured. The calculation of data accuracy requires at least one set of complete test or incomplete test samples. When the sample set does not exist, it cannot be calculated. acc, at this time, the data accuracy can be directly set to a acc The value of is 0.
[0087] For data collection user difference parameter a per By determining , we can obtain the role of the collector of the road network sample data, and determine the difference parameters of the data collection users based on the role of the collector. Among them, data collection users can be divided into professional testers and users. When collecting data for mapping, professionals generally drive at a suitable speed, traverse all important locations of the indoor scene, and may use some test tools to assist in the operation to ensure the integrity and reliability of the road network data. However, when different non-professional users collect data, there are certain differences, which cannot be calculated quantitatively. Specifically, the function of the classifier can be used, which is expressed as follows:
[0088]
[0089] Among them, based on the function of the classifier, different personnel roles can be distinguished by different values, and the k value can be any value, which can be set according to the actual application situation, for example, 1, and the embodiment of the present invention is not limited.
[0090] For the data acquisition time difference parameter a time By determining the data collection time and data usage time of the road network sample data, the data collection time difference parameter can be determined by constructing a decreasing function based on the difference between the data collection time and the data usage time. Specifically, the data collection time also has a great impact on the confidence of the road network data. As time passes, the map may change to a certain extent. For example, the shopping mall road network data collected a year ago cannot be fully used because of its secondary renovation or expansion. At this time, in order to quantify the impact of this data collection time on the collected road network data, a method based on the data collection time t can be constructed. g and data usage time t u The decreasing function of the difference value indicates the degree of influence on the data acquisition time:
[0091]
[0092] That is, by showing that the impact of a road network data can be determined based on the data collection time and data usage time, the accuracy of the road network data may decrease as the collection time and usage time increase.
[0093] For sensor accuracy a senThe determination of can obtain the sensor measurement error value used to collect road network sample data, and make the determination based on the sensor measurement error value. In practical applications, there are many types of sensors at present. For the collection vehicle, monocular / binocular / trinocular cameras, ultrasonic radars, lidars and other sensors may be used at the same time. In addition, considering the cost issue, the accuracy of sensors arranged in different directions of some vehicles will also be different. The differences in these sensors used to collect road network data will cause a certain error value in the collected data point cloud. At this time, based on the measurement error of the sensor, the degree of influence of sensor accuracy on road network data can be expressed as:
[0094]
[0095] Among them, the higher the sensor accuracy, the smaller the measurement error value e sen The smaller it is, the greater the confidence level will be.
[0096] For the road network data accuracy parameter a acc The determination can be made through the time warping algorithm (Dynamic TimeWarping, referred to as DTW, is a method to measure the similarity of two time series of different lengths).
[0097] Reference Figure 4 , showing a schematic diagram of the process of calculating the accuracy of road network data provided by an embodiment of the present invention. The process of calculating the accuracy of road network data based on the dynamic time warping algorithm can be mainly divided into a trajectory point coordinate extraction stage, a dynamic time warping matrix construction stage, and an accuracy calculation stage.
[0098] like Figure 5 As shown in the figure, the descriptions of various road network structures for the location, connection and other information of roads generally include the following parts: location information Location, which mainly involves the three-dimensional coordinate information of points on the road, number information ID, which mainly classifies and identifies each road, and link information Link, which is the connection between each road or between the road and other surrounding information.
[0099] In the embodiment of the present invention, the calculation of the determinism of the road network data first needs to obtain the prior data and the newly added data in the road network sample data. The acquired prior data includes the historical road network data relative to the newly collected road network data, that is, the indoor road network data can be collected multiple times in the same indoor scene. Generally, the road network data collected for the first n times is called the prior data for the next collection of road network data, and the route data collected for the next time can be called the newly added data. For example, suppose there is the following scenario: After a parking garage is newly built, the tester can follow the following steps: Figure 5The trajectory shown in the upper left figure is used to move to obtain information about the three roads. At this time, this information can be stored as prior data. Since this information cannot fully reproduce the road network data of the parking garage, another tester can follow the same method as shown in the following figure. Figure 5 The trajectory in the upper right figure shown in the figure is collected again to store the new data. In theory, a more complete scene map can be obtained by fusing the two types of data, namely the prior data and the new data. However, due to the influence of factors such as the starting position, collection speed, and collection time of the second collection, the collected road network data will be different. At this time, the accuracy of the collected road network data needs to be measured through a time warping algorithm.
[0100] In the trajectory point coordinate extraction stage, the trajectory point coordinate sets of the prior data and the newly added data can be extracted respectively, wherein the prior data and the newly added data include path number information, and the extracted trajectory point coordinate set is a coordinate point set for the overlapping trajectory in different data. First, based on the path number information, the overlapping trajectory in the newly added data and the prior data can be obtained, and then the trajectory point coordinate set of the overlapping trajectory in the prior data can be extracted, and the overlapping trajectory in the newly added data can be adjusted, and the trajectory point coordinate set of the adjusted overlapping trajectory in the newly added data can be extracted.
[0101] If there is no overlap between the prior data and the newly added data, that is, there is no road section with the same road number information, then the accuracy parameter a of the road network data can be directly used. acc The value is 0.
[0102] In practical applications, the coordinate set (Location) of the trajectory points of the road section (i.e., the coincident trajectory) in the prior data and the newly added information is extracted, where the prior data point data set can be recorded as When marking the information points of the newly added data, the overlapping parts of the trajectory can be obtained based on the road ID information, that is, the road number information, for example Figure 4 Road 1 in the middle is a coincident trajectory with the same ID. After changing the position and posture of the coincident trajectory in the newly added data so that the newly added trajectory approaches the prior data, the newly added information point data set can be recorded as Among them, i can represent the i-th measurement, r is the path number, and the lower right subscript is the number of the point data. Represents the data of the nth point on the rth path in the i-th measurement.
[0103] It should be noted that the position and posture of the overlapping trajectories are changed, specifically, the same features of the two trajectories can be found (for example, reference to the garage entrance and exit, the shopping mall elevator position and other features), and the overlapping trajectories in the newly added data with the prior information are aligned based on the relatively fixed entrance and exit or elevator position features. The alignment processing performed may include up and down translation, left and right translation, rotation, etc., that is, adjusting the non-fixed points based on the fixed points in the trajectory, and regularizing the characteristic points on the same baseline. This is not limited in the embodiments of the present invention.
[0104] In the dynamic time warping matrix construction stage, the dynamic time warping matrix can be constructed according to the trajectory point coordinate set. Specifically, the trajectory point coordinate set includes the trajectory point coordinates on the overlapping trajectory in the prior data and the trajectory point coordinates on the adjusted overlapping trajectory in the newly added data. At this time, the matching relationship between the trajectory point coordinates on the overlapping trajectory in the prior data and the trajectory point coordinates on the adjusted overlapping trajectory in the newly added data can be determined. The dynamic time warping matrix for the road network sample data is established through the determined matching relationship. The constructed dynamic time warping matrix can be as follows Figure 5 shown.
[0105] The matching relationship between the coordinates of the trajectory points contained in the two trajectories can be determined by calculating the Euclidean metric of all points on the two trajectories. For example, the determination of the Euclidean metric can be expressed as:
[0106]
[0107] In the formula, Represents the trajectory on the rth path i S r The nth point in i+1 S r The Euclidean metric of the mth point on the .
[0108] In the accuracy calculation stage, the road network data accuracy parameters that affect the road network sample data can be obtained through the constructed dynamic time warping matrix. Specifically, the shortest path in the constructed dynamic time warping matrix can be calculated to obtain the similarity between the curve for the prior data and the curve for the newly added data to obtain the road network data accuracy parameters.
[0109] Among them, based on the dynamic time warping algorithm, it can be known that the shortest path in the dynamic time warping matrix can be used to represent the similarity between two curves. At this time, the shortest path to the coordinate point (x, y) can be g(x, y), and the specific coordinate point (x, y) can be a relatively fixed feature point, such as the entrance and exit of a garage, the location of a shopping mall elevator, etc., and its expression can be:
[0110]
[0111] Through continuous iteration, the similarity g(m,n) of the two curves can be obtained to obtain the accuracy parameter for the collected road network data, that is, a acc =g(m,n). The trajectories under the same path are matched by curves and the similarity value is calculated. If the similarity is less than a given threshold, it can be considered that the confidence of the new data is insufficient.
[0112] It should be noted that the threshold given for the similarity is somewhat similar to the concept of confidence. For example, when the similarity is 0, it can indicate complete dissimilarity, and when the similarity is 1, it can indicate complete identity. The result of whether the two trajectories are similar can be mainly obtained through the quantitative index of the similarity. This threshold can usually be determined by engineering requirements, and the embodiments of the present invention are not limited to this.
[0113] Step 302: using the information of the factors influencing the road network and the weight values corresponding to the factors, and training the regression model based on a preset hyperbolic tangent function, to obtain a regression model for determining the confidence of the road network data.
[0114] In one embodiment of the present invention, the constructed regression model can realize the function of determining the confidence of comprehensive road network data. This regression model for determining the confidence of road network data is mainly a model obtained by training the road network influencing factor information of road network sample data and the corresponding weight values of each factor. When constructing the regression model, multiple factors affecting the road network data, such as the accuracy difference of the sensor, whether the driver is professional, the data collection time, the number of collections, and the accuracy, are comprehensively considered, and the confidence can be determined in combination with the time regression algorithm, specifically, the road network data accuracy parameters in the road network influencing factor information are determined based on the time regression algorithm.
[0115] Specifically, the information of the road network influencing factors determined for the road network sample data, including the number of data collection times a num , Data collection user difference parameter a per , data collection time difference parameter a time 、Sensor accuracy a sen And the road network data accuracy parameter a acc , combined with the K value of the influence of the factor, the hyperbolic tangent function tanh function is substituted to build a regression model for determining the confidence of the road network data. The specific regression model construction expression can refer to the above embodiment part, which will not be repeated here.
[0116] In practical applications, the entire regression model can be iteratively calculated and solved through network training or parameter fitting methods, and then applied to the subsequent confirmation of the confidence of road network data. It is worth noting that when calculating the confidence, the various influencing factors and corresponding weights are mainly combined in a linear weighted manner, and the value range of each independent variable (i.e., each road network data influencing factor) will have an important impact on the result. For example, the data collection time difference a time The value range of is 0-1, while the value range of sensor accuracy can be far beyond this interval, possibly reaching 0-100. Before iterative calculation of the model, each part can be normalized as follows:
[0117]
[0118] The maximum value and the minimum value of the impact factor a may vary according to the actual application scenario, and the embodiment of the present invention does not limit this.
[0119] In the embodiment of the present invention, by comprehensively considering multiple factors affecting road network data such as the accuracy difference of sensors, whether the driver is professional, data collection time, number of collections, accuracy, etc., a more comprehensive confidence calculation method is constructed using a logistic regression model, and it can be applied to road network data in large areas and small spaces. By using a dynamic time regression algorithm, an evaluation method for comparing the matching degree of each road in multiple groups of road network data is proposed. The confidence of the road network data is determined in indoor environments or scenarios where prior estimation information is insufficient to determine the accuracy of the collected road network data.
[0120] It should be noted that, for the sake of simplicity, the method embodiments are described as a series of action combinations, but those skilled in the art should be aware that the embodiments of the present invention are not limited by the order of the actions described, because according to the embodiments of the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily required by the embodiments of the present invention.
[0121] Reference Figure 6 , shows a structural block diagram of an embodiment of a device for determining the confidence level of road network data of the present invention, which may specifically include the following modules:
[0122] The regression model acquisition module 601 is used to acquire a regression model for determining the confidence of the road network data and the collected road network data; the regression model for determining the confidence of the road network data is obtained by training the road network influencing factor information of the road network sample data and the weight values corresponding to each factor, wherein the road network influencing factor information includes the road network data accuracy parameter determined based on the time regression algorithm;
[0123] The confidence determination module 602 is used to obtain the confidence of the road network data based on the collected road network data and the regression model used to determine the confidence of the road network data.
[0124] In one embodiment of the present invention, the confidence determination module 602 may include the following submodules:
[0125] A first road network influencing factor information determination submodule, used to determine the road network influencing factor information for the collected road network data;
[0126] The confidence determination submodule is used to use the road network influencing factor information as an input item of the regression model used to determine the confidence of the road network data, and output the confidence of the collected road network data.
[0127] In one embodiment of the present invention, the device may further include the following modules:
[0128] The regression model generation module is used to train the road network influencing factor information of the road network sample data and the corresponding weight value of each factor to obtain a regression model for determining the confidence of the road network data.
[0129] In one embodiment of the present invention, the regression model generation module may include the following submodules:
[0130] The second road network influencing factor information determination submodule is used to obtain the collected road network sample data, determine the road network influencing factor information for the road network sample data, and the weight value corresponding to each factor; the road network influencing factor information includes at least one of the following: data collection times, data collection user difference parameters, data collection time difference parameters, sensor accuracy, and road network data accuracy parameters;
[0131] The regression model training submodule is used to adopt the road network influencing factor information and the weight values corresponding to each factor, and train the regression model based on a preset hyperbolic tangent function to obtain a regression model for determining the confidence of road network data.
[0132] In one embodiment of the present invention, the second road network influencing factor information determination submodule may include the following units:
[0133] A data collection frequency determination unit, used to obtain environmental data in the collected road network sample data, and determine the data collection frequency affecting the road network sample data based on a judgment result of whether the collected environmental data is measured;
[0134] A data collection user difference parameter determination unit, used for obtaining the role of a collector of the road network sample data, and determining a data collection user difference parameter based on the role of the collector;
[0135] A data collection time difference parameter determination unit, used for obtaining the data collection time and the data use time of the road network sample data, and determining the data collection time difference parameter by constructing a decreasing function based on the difference between the data collection time and the data use time;
[0136] a sensor accuracy determination unit, configured to obtain a sensor measurement error value used to collect the road network sample data, and determine a sensor accuracy that affects the road network sample data based on the sensor measurement error value;
[0137] The road network data accuracy determination unit is used to determine the road network data accuracy parameters that affect the road network sample data through a time warping algorithm.
[0138] In one embodiment of the present invention, the road network data accuracy determination unit may include the following subunits:
[0139] A data acquisition subunit, used to acquire prior data and newly added data from the road network sample data; the prior data includes historical road network data relative to newly collected road network data;
[0140] A dynamic time warping matrix construction subunit, used to extract the trajectory point coordinate sets of the prior data and the newly added data respectively, and construct a dynamic time warping matrix according to the trajectory point coordinate sets;
[0141] The road network data accuracy parameter determination subunit is used to obtain the road network data accuracy parameters that affect the road network sample data through the constructed dynamic time warping matrix.
[0142] As for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.
[0143] An embodiment of the present invention further provides a vehicle, comprising:
[0144] It includes the above-mentioned road network data confidence determination device, a processor, a memory and a computer program stored in the memory and capable of running on the processor. When the computer program is executed by the processor, it implements the various processes of the above-mentioned road network data confidence determination method embodiment and can achieve the same technical effect. To avoid repetition, it will not be repeated here.
[0145] An embodiment of the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the various processes of the above-mentioned method for determining the confidence level of road network data can be implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.
[0146] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.
[0147] It will be appreciated by those skilled in the art that the embodiments of the present invention may be provided as methods, devices, or computer program products. Therefore, the embodiments of the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the embodiments of the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.
[0148] The embodiments of the present invention are described with reference to the flowcharts and / or block diagrams of the methods, terminal devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0149] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0150] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device so that a series of operating steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable terminal device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0151] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present invention.
[0152] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or terminal device including the elements.
[0153] The above is a detailed introduction to a method for determining the confidence of road network data and a device for determining the confidence of road network data provided by the present invention. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.
Claims
1. A method for determining the confidence of road network data, characterized in that: The method comprises: Obtaining a regression model for determining the confidence of road network data and the collected road network data; the regression model for determining the confidence of road network data is obtained by using road network influencing factor information of road network sample data and the weight values corresponding to each factor in the road network influencing factor information, and training the regression model based on a preset hyperbolic tangent function, wherein the road network influencing factor information includes a road network data accuracy parameter determined based on a time regression algorithm, a data collection number of times, a data collection user difference parameter, a data collection time difference parameter, and a sensor accuracy; The confidence of the road network data is obtained based on the collected road network data and the regression model used to determine the confidence of the road network data.
2. The method according to claim 1, characterized in that: The step of obtaining the confidence of the road network data based on the collected road network data and the regression model for determining the confidence of the road network data comprises: Determine the road network influencing factor information for the collected road network data; The road network influencing factor information is used as an input item of the regression model for determining the confidence of the road network data, and the confidence of the collected road network data is output.
3. The method according to claim 1, characterized in that The method further comprises: Acquire environmental data from the collected road network sample data, and determine the number of data collection times that affect the road network sample data based on a determination result of whether the collected environmental data is measured; and / or, obtaining the role of a collector of the road network sample data, and determining a data collection user difference parameter based on the role of the collector; And / or, obtaining the data collection time and the data use time of the road network sample data, and determining the data collection time difference parameter by constructing a decreasing function based on the difference between the data collection time and the data use time; and / or, obtaining a sensor measurement error value used to collect the road network sample data, and determining the accuracy of the sensor that affects the road network sample data based on the sensor measurement error value; And / or, determining the road network data accuracy parameters that affect the road network sample data through a time warping algorithm.
4. The method according to claim 3, characterized in that The determining of the road network data accuracy parameters affecting the road network sample data by a time warping algorithm includes: Acquire prior data and newly added data from the road network sample data; the prior data includes historical road network data relative to newly collected road network data; Extracting trajectory point coordinate sets of the prior data and the newly added data respectively, and constructing a dynamic time warping matrix according to the trajectory point coordinate sets; Through the constructed dynamic time warping matrix, the road network data accuracy parameters affecting the road network sample data are obtained.
5. The method according to claim 4, characterized in that The priori data and the newly added data include path number information; and the step of respectively extracting the trajectory point coordinate sets of the priori data and the newly added data includes: Acquire, based on the path number information, a trajectory in the newly added data that overlaps with the prior data; A set of trajectory point coordinates of the coincident trajectory in the prior data is extracted, and the coincident trajectory in the newly added data is adjusted, and a set of trajectory point coordinates of the adjusted coincident trajectory in the newly added data is extracted.
6. The method according to claim 4 or 5, characterized in that: The trajectory point coordinate set includes the trajectory point coordinates on the overlapping trajectory in the priori data and the trajectory point coordinates on the adjusted overlapping trajectory in the newly added data; The step of constructing a dynamic time warping matrix according to the trajectory point coordinate set includes: Determine the matching relationship between the coordinates of the trajectory points on the overlapping trajectory in the prior data and the coordinates of the trajectory points on the adjusted overlapping trajectory in the newly added data, and establish a dynamic time warping matrix for the road network sample data through the determined matching relationship.
7. The method according to claim 4 or 5, characterized in that: The road network data accuracy parameters affecting the road network sample data are obtained through the constructed dynamic time warping matrix, including: The shortest path in the constructed dynamic time warping matrix is calculated to obtain the similarity between the curve for the prior data and the curve for the newly added data, so as to obtain the accuracy parameter of the road network data.
8. A device for determining the confidence of road network data, characterized in that: The device comprises: A regression model acquisition module, used to acquire a regression model for determining the confidence of road network data, and the collected road network data; the regression model for determining the confidence of road network data is obtained by using the road network influencing factor information of the road network sample data and the weight values corresponding to each factor in the road network influencing factor information, and training the regression model based on a preset hyperbolic tangent function, wherein the road network influencing factor information includes a road network data accuracy parameter determined based on a time regression algorithm, a data collection number, a data collection user difference parameter, a data collection time difference parameter, and a sensor accuracy; The confidence determination module is used to obtain the confidence of the road network data based on the collected road network data and the regression model used to determine the confidence of the road network data.
9. A vehicle, characterized in that: include: The device for determining the road network data confidence as described in claim 8, the processor, the memory and the computer program stored in the memory and capable of running on the processor, wherein when the computer program is executed by the processor, the steps of the method for determining the road network data confidence as described in any one of claims 1 to 7 are implemented.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for determining the confidence of road network data as described in any one of claims 1-7 are implemented.
Citation Information
Patent Citations
Blood oxygen saturation measurement confidence degree calculating method and system and storage medium
CN109512392A
Energy consumption saving method for unmanned vehicles based on open-pit mining area paths
CN110486017A
