Road weight determination method, model generation method and route planning method and device

Abstract

The present invention discloses a road weight determination method, a model generation method, and a route planning method and device. The road weight determination method comprises: planning a planned route from the first track point to the last track point according to the first and last track points and the departure time recorded in the historical driving track; matching the track points included in the historical driving track with roads to obtain the roads matched with the track points, and the historical track route corresponding to the historical driving track is formed by the roads; the historical track route and the corresponding planned route are formed into a sample; the machine learning model is trained using the set of samples to output the travel time coefficient and the travel distance coefficient of the road; the road weight is determined according to the regular travel time, travel distance, travel time coefficient and travel distance coefficient of the road. The learning sample can be established by integrating the actual user's travel track and the planned route, so that the learning result is closer to the actual situation of the road.

Description

Technical Field

[0001] The present invention relates to the field of navigation technology, and in particular to a road weight determination method, a model generation method, and a route planning method and device. Background Art

[0002] In the field of navigation technology, when planning a navigation route, the road weight is usually used. The road weight refers to the travel cost of a road (or a road and its corresponding intersection), which integrates factors such as the road's travel time, travel distance, and road conditions. The route planning engine can recall the navigation route based on the road weight and the shortest route algorithm, where the road weight is the key parameter that determines the recalled navigation route. Therefore, how to accurately determine the road weight is a problem that needs to be solved in the field of navigation technology. Summary of the invention

[0003] In view of the above problems, the present invention is proposed to provide a road weight determination method, model generation method, and route planning method and device that overcome the above problems or at least partially solve the above problems.

[0004] In a first aspect, an embodiment of the present invention provides a method for determining a road weight, comprising:

[0005] According to the first and last track points and departure time recorded in the historical driving track, a planned route from the first track point to the last track point is planned;

[0006] Performing road matching on the track points included in the historical driving track to obtain roads matched by the track points, and the roads constitute the historical track route corresponding to the historical driving track;

[0007] The historical trajectory route and the corresponding planned route are combined into a sample;

[0008] Using the set of samples to train a machine learning model, outputting a travel time coefficient and a travel distance coefficient of the road;

[0009] The weight of the road is determined based on the road's regular travel time, travel distance, travel time coefficient and travel distance coefficient.

[0010] In some optional embodiments, planning a route from the first track point to the last track point according to the first and last track points and the departure time recorded in the historical driving track specifically includes:

[0011] The route connecting the first and last track points of the historical driving track record is used as an alternative planning route;

[0012] Determine the sum of the road weights of the roads included in each candidate planned route according to the departure time recorded in the historical driving trajectory and the road weights corresponding to the departure time;

[0013] The planned route is determined according to the sum of the road weights of the alternative planned routes.

[0014] In some optional embodiments, after determining the weight of the road, the method further includes:

[0015] Determine a new planned route according to the first and last track points, the departure time and the weight of the current road of the historical driving track corresponding to the current sample, and add the new planned route to the sample;

[0016] Using the set of new samples to train the currently trained machine learning model, outputting a new travel time coefficient and a new travel distance coefficient for the road;

[0017] Determine the new weight of the road based on the regular travel time, travel distance, new travel time coefficient and new travel distance coefficient of the road;

[0018] Determine whether the preset termination training condition is met. If not, continue to determine the new planned route according to the first and last track points, departure time and current road weight of the historical driving track corresponding to the current sample, and add the new planned route to the sample.

[0019] In some optional embodiments, before adding the new planned route to the sample, the method further includes:

[0020] Determining whether the newly planned route is consistent with the historical trajectory route;

[0021] If not, executing the step of adding the new planned route to the sample;

[0022] If so, delete the sample.

[0023] In some optional embodiments, the determining whether a preset training termination condition is met specifically includes:

[0024] Determine whether the number of training times reaches a preset number threshold; and / or,

[0025] Determine whether the loss function value determined according to the currently trained machine learning model meets the preset conditions; and / or,

[0026] Determine whether the yaw rate determined by the currently trained machine learning model meets the preset conditions.

[0027] In some optional embodiments, the training of the machine learning model using the set of samples specifically includes:

[0028] Obtaining discrete data of set characteristic parameters of each road in the historical trajectory route in the sample, and obtaining discrete data of set characteristic parameters of each road in each planned route in the sample, to obtain discrete samples;

[0029] A selected logistic regression model is trained using the set of discrete samples.

[0030] In some optional embodiments, the training of the machine learning model using the set of samples specifically includes:

[0031] Obtaining continuous feature expression data of set feature parameters of each road in the historical trajectory route in the sample, and obtaining continuous feature expression data of set feature parameters of each road in each planned route in the sample, to obtain a continuous feature expression sample;

[0032] The selected neural network model is trained using the set of continuous feature expression samples.

[0033] In some optional embodiments, before planning a planned route from the first track point to the last track point according to the first and last track points and the departure time recorded in the historical driving track, the method further includes:

[0034] Determine whether the historical driving trajectory meets the preset sample conditions;

[0035] If yes, execute the planning route from the first track point to the last track point according to the first and last track points and the departure time recorded in the historical driving track;

[0036] If not, delete the historical driving track.

[0037] In some optional embodiments, the determining whether the historical driving trajectory meets the preset sample condition specifically includes:

[0038] Determine whether the historical driving trajectory contains violation record information; and / or,

[0039] Determine whether the time interval between any two adjacent trajectory points in the historical driving trajectory is less than a preset time threshold.

[0040] In some optional embodiments, after the historical trajectory route corresponding to the historical driving trajectory is formed by the road, the method further includes:

[0041] Determine whether the ratio of the travel distance of the historical trajectory route to the travel distance of each corresponding planned route is less than a preset ratio threshold;

[0042] If so, the historical trajectory route and the corresponding planned route are combined into a sample.

[0043] In a second aspect, an embodiment of the present invention provides a route planning method, including:

[0044] The navigation route is determined according to the user's navigation starting point, end point, request time and the road weight determined according to the above road weight determination method.

[0045] In a third aspect, an embodiment of the present invention provides a method for generating a road right model, including:

[0046] According to the first and last track points and the departure time recorded in the historical driving track, a planned route from the first track point to the last track point is planned;

[0047] Performing road matching on the track points included in the historical driving track to obtain roads matched by the track points, and the roads constitute the historical track route corresponding to the historical driving track;

[0048] The historical trajectory route and the corresponding planned route are combined into a sample;

[0049] The set of samples is used to train a machine learning model, and the machine learning model is used to output a travel time coefficient and a travel distance coefficient of the road.

[0050] In a fourth aspect, an embodiment of the present invention provides a road weight determination device, comprising:

[0051] A planning module, used to plan a route from the first track point to the last track point according to the first and last track points and the departure time recorded in the historical driving track;

[0052] A matching module, used for performing road matching on the track points included in the historical driving track to obtain roads matched by the track points, and the roads constitute a historical track route corresponding to the historical driving track;

[0053] A combining module, used for combining the historical trajectory route planned by the planning module and the corresponding planned route matched by the matching module into a sample;

[0054] A training module, used to train a machine learning model using the set of samples combined by the combination module, and output a travel time coefficient and a travel distance coefficient of a road;

[0055] The determination module is used to determine the weight of the road according to the regular travel time, travel distance, travel time coefficient and travel distance coefficient of the road.

[0056] In a fifth aspect, an embodiment of the present invention provides a road right model generation device, including:

[0057] A planning module, used to plan a route from the first track point to the last track point according to the first and last track points and the departure time recorded in the historical driving track;

[0058] A matching module, used for performing road matching on the track points included in the historical driving track to obtain roads matched by the track points, and the roads constitute a historical track route corresponding to the historical driving track;

[0059] A combining module, used for combining the historical trajectory route planned by the planning module and the corresponding planned route matched by the matching module into a sample;

[0060] A training module is used to train a machine learning model using a set of samples combined by the combination module, and the machine learning model is used to output a travel time coefficient and a travel distance coefficient of a road.

[0061] In a sixth aspect, an embodiment of the present invention provides a navigation device, wherein the navigation device is provided with the above-mentioned road weight determination device;

[0062] The navigation device is used to determine a navigation route according to a user's navigation starting point, end point, request time and the road weight determined by the road weight determination device.

[0063] In a seventh aspect, an embodiment of the present invention provides a computer-readable storage medium having computer instructions stored thereon, which, when executed by a processor, implement the above-mentioned road weight determination method, or implement the above-mentioned route planning method, or implement the above-mentioned road right model generation method.

[0064] The beneficial effects of the above technical solution provided by the embodiment of the present invention include at least:

[0065] The road weight determination method provided by the embodiment of the present invention plans a planned route from the first track point to the last track point according to the first and last track points and departure time recorded in the historical driving track; matches the track points included in the historical driving track with roads to obtain the roads matched by the track points, and the historical track route corresponding to the historical driving track is formed by the roads; the historical track route and the corresponding planned route form a sample; the machine learning model is trained using the sample set to output the travel time coefficient and the travel distance coefficient of the road; the road weight is determined according to the regular travel time, travel distance, travel time coefficient and travel distance coefficient of the road. Each sample includes the historical track route corresponding to the real historical driving track, and the characteristics related to user behavior are very important information, so the learning of the real historical track route can improve the training effect; and the actual business is that even if the various situations are the same, different users may choose different routes, which is a personalized behavior, so for each driving history track, one or more planned routes are determined to be added to the sample data set, enriching the sample data, so that the learning result is closer to the real situation of the road.

[0066] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings.

[0067] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0069] Figure 1 This is a flow chart of a method for determining a road weight in Embodiment 1 of the present invention;

[0070] Figure 2 for Figure 1 Specific implementation flow chart of step S11;

[0071] Figure 3 This is a specific implementation flow chart of the road weight determination method in the second embodiment of the present invention;

[0072] Figure 4 This is another specific implementation flow chart of the method for determining road weight in Embodiment 3 of the present invention;

[0073] Figure 5 Schematic diagram of the structure of a device for determining road weight in an embodiment of the present invention;

[0074] Figure 6 Schematic diagram of the structure of the road right model generating device in an embodiment of the present invention. DETAILED DESCRIPTION

[0075] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0076] In order to solve the problem of low accuracy in determining road weights in the prior art, the embodiments of the present invention provide a road weight determination method, a model generation method, and a route planning method and device, which can comprehensively establish learning samples based on the user's actual trajectory and planned route, so that the learning results are closer to the actual situation of the road.

[0077] Embodiment 1

[0078] Embodiment 1 of the present invention provides a method for determining a road weight, the process of which is as follows: Figure 1 As shown, the following steps are included:

[0079] Step S11: planning a route from the first track point to the last track point according to the first and last track points and the departure time recorded in the historical driving track.

[0080] In one embodiment, referring to Figure 2 As shown, the following steps may be included:

[0081] Step S21: taking a route connecting the first and last track points of the historical driving track record as an alternative planning route.

[0082] For example, a large number of users' historical driving trajectories can be collected through navigation software. According to the connectivity relationship of the roads, one or more routes connecting the first and last track points of each historical driving trajectory record are determined as alternative planning routes.

[0083] Step S22: Determine the sum of the road weights of the roads included in each candidate planned route according to the departure time recorded in the historical driving trajectory and the road weights corresponding to the departure time.

[0084] The weight of the road corresponding to the above departure time may be the road right of the corresponding time determined by other road right model learning methods, or may be an empirical value obtained according to statistical data. The specific determination method is not limited in this embodiment.

[0085] Step S23: Determine the planned route according to the sum of the road weights of the candidate planned routes.

[0086] The determined planned route can be one, that is, the alternative planned route with the smallest sum of road weights is determined as the planned route. Optionally, there can be multiple planned routes, which can be the number of specific planned routes corresponding to each historical driving trajectory; or a ratio threshold of the sum of the weights of the planned routes to the sum of the weights of the historical driving trajectories can be set, and one or more alternative planned routes whose ratio of the sum of the weights to the sum of the weights of the historical driving trajectories is less than the ratio threshold are determined as the planned routes.

[0087] Step S12: performing road matching on the track points included in the historical driving track to obtain roads matched with the track points, and the roads constitute a historical track route corresponding to the historical driving track.

[0088] The historical driving trajectory is matched with the road, so that the obtained historical trajectory route includes the roads and intersections that the historical driving trajectory has passed; the subsequent planned route also includes roads and intersections, so the travel time coefficient and travel distance coefficient of the road can be obtained through the machine learning model.

[0089] The specific method of matching the trajectory points to the road is not limited in this embodiment.

[0090] Step S13: The historical trajectory route and the corresponding planned route are combined into a sample.

[0091] Specifically, the historical trajectory route includes not only the trajectory points and the roads matching the trajectory points, but also one or more of the following information:

[0092] The actual travel time of each track point, where the actual travel time of the first track point is the departure time;

[0093] The location of each trajectory point;

[0094] The historical trajectory route includes the road grade, road capacity, number of lanes, road connectivity and dynamic traffic conditions of each road;

[0095] The historical trajectory route includes the actual travel time of each road.

[0096] Specifically, the planned route may include one or more of the following information:

[0097] The planned route includes the road grade, road capacity, number of lanes, road connectivity and dynamic traffic conditions of each road;

[0098] The planned travel time of each road included in the planned route.

[0099] Specifically, the actual travel time or planned travel time of the above-mentioned road does not only refer to the actual travel time or planned travel time of the road, but also includes the travel time of the intersection corresponding to the road. For example, if road A passes through intersection B to reach road C, the actual travel time of road A can be the actual travel time of road A and intersection B.

[0100] Step S14: Use the sample set to train the machine learning model and output the travel time coefficient and travel distance coefficient of the road.

[0101] In one embodiment, for each sample in the sample set, discrete data of set feature parameters of each road (including intersections) in the historical trajectory route in the sample is obtained, and discrete data of set feature parameters of each road (including intersections) in each planned route in the sample is obtained to obtain discrete samples; and the set of discrete samples is used to train the selected logistic regression model.

[0102] Specifically, a machine learning model is trained using a set of samples. In one embodiment, for each sample in the sample set, continuous feature expression data of set feature parameters of each road (including intersections) in the historical trajectory route in the sample is obtained, and continuous feature expression data of set feature parameters of each road (including intersections) in each planned route in the sample is obtained to obtain continuous feature expression samples; and the set of continuous feature expression samples is used to train the selected neural network model.

[0103] That is, when the logistic regression model is selected for training and learning, the set characteristic parameters of each historical trajectory route and the road included in the planned route in the sample set need to be converted into an enumeration format.

[0104] When a neural network model is selected for training and learning, the set feature parameters of the roads included in each historical trajectory route and planned route in the sample set need to be converted into a vector form of continuous feature expression. Based on Road2Vec (Road to Vector, R2V), roads and intersections are mapped to vector space for representation. Similar roads or intersections will have similar vector representations. Since neurons are shared, similar but not identical scenes can use each other's training data, thereby solving the problem of inaccurate road weight coefficients in low-frequency scenes, making the final training and learning results more accurate. The above-mentioned neural network model can be a deep neural network (Deep Neural Networks, DNN) neural network model or other neural network models.

[0105] Optionally, other machine learning models may also be used, which is not specifically limited in this embodiment.

[0106] Step S15: Determine the weight of the road according to the regular travel time, travel distance, travel time coefficient and travel distance coefficient of the road.

[0107] Specifically, the road weight e can be determined according to the following formula: i :

[0108] e i =W t ×t+W d ×d

[0109] In the above formula, i is the road number, w t is the travel time coefficient of road i output by the model, t is the regular travel time of road i, and w d is the travel distance coefficient of road i output by the model, and d is the travel distance of road i.

[0110] The regular travel time and distance of the above-mentioned roads are predetermined. The regular travel time of a road, i.e., the time required to travel the road and the corresponding intersection under normal circumstances, can be determined based on factors such as the comprehensive road grade, road speed limit, and the normal driving speed of the vehicle.

[0111] In the above method of this embodiment, according to the first and last track points and departure time recorded in the historical driving track, a planned route from the first track point to the last track point is planned; the track points included in the historical driving track are matched with roads to obtain the roads matched with the track points, and the historical track route corresponding to the historical driving track is formed by the road; the historical track route and the corresponding planned route are formed into a sample; the machine learning model is trained using the sample set to output the travel time coefficient and travel distance coefficient of the road; the weight of the road is determined according to the regular travel time, travel distance, travel time coefficient and travel distance coefficient of the road. Each sample includes the historical track route corresponding to the real historical driving track, and the characteristics related to user behavior are very important information, so the learning of the real historical track route can improve the training effect; and the actual business is that even if the various situations are the same, different users may choose different routes, which is a personalized behavior, so for each driving historical track, one or more planned routes are determined to be added to the sample data set, enriching the sample data, so that the learning result is closer to the real situation of the road.

[0112] Embodiment 2

[0113] Embodiment 2 of the present invention provides a specific implementation process of a method for determining a road weight, and the process is as follows: Figure 3 As shown, the following steps are included:

[0114] Step S301: Filter the historical driving trajectory.

[0115] In one embodiment, judging whether the historical driving trajectory meets the preset sample conditions may include judging whether the historical driving trajectory contains violation record information; and / or judging whether the time interval between any two adjacent trajectory points in the historical driving trajectory is less than a preset time threshold, that is, judging whether the historical driving trajectory has a stop along the way. The historical driving trajectory that does not meet the preset sample conditions is deleted.

[0116] Specifically, historical driving tracks with other abnormal driving conditions may be deleted.

[0117] Step S302: planning a route from the first track point to the last track point according to the first and last track points and the departure time recorded in the historical driving track.

[0118] Step S303: performing road matching on the track points included in the historical driving track to obtain roads matched with the track points, and the roads constitute a historical track route corresponding to the historical driving track.

[0119] Step S304: determine whether the ratio of the travel distance of the historical trajectory route to the travel distance of each corresponding planned route is less than a preset ratio threshold.

[0120] It is determined that the ratio of the travel distance of the historical trajectory route to the travel distance of each corresponding planned route is less than the preset ratio threshold, verifying that the user trajectory has not taken a detour, and executing step S306; if the judgment is no, it is determined that the user trajectory has taken a detour, and executing step S305.

[0121] Step S305: Delete the historical track route and the corresponding historical driving track.

[0122] Steps S301 to S305 determine the planned route based on the historical driving trajectory, screen the historical driving trajectory, and delete the historical driving trajectory with violations, detours or stops, so that the established sample set only contains normal and reasonable historical trajectory routes and their corresponding planned routes, thereby increasing the accuracy of the training results of the machine learning model finally trained using the sample set.

[0123] Step S306: The historical trajectory route and the corresponding planned route are combined into a sample to obtain a sample set.

[0124] Step S307: Use the sample set to train the machine learning model and output the travel time coefficient and travel distance coefficient of the road.

[0125] Step S308: Determine the weight of the road according to the regular travel time, travel distance, travel time coefficient and travel distance coefficient of each road.

[0126] Step S309: for each sample in the current sample set, a new planned route is determined according to the first and last track points, the departure time and the weight of the current road of the historical driving track corresponding to the current sample, and the new planned route is added to the sample.

[0127] Step S310: Use the set of new samples to train the currently trained machine learning model and output the new travel time coefficient and new travel distance coefficient of the road.

[0128] Step S311: Determine a new weight of the road according to the normal travel time, travel distance, new travel time coefficient and new travel distance coefficient of the road.

[0129] Step S312: Determine whether a preset training termination condition is met.

[0130] In one embodiment, it may be to determine whether the number of training times reaches a preset threshold number; and / or to determine whether the loss function value determined according to the currently trained machine learning model meets a preset condition; and / or to determine whether the yaw rate determined according to the currently trained machine learning model meets a preset condition.

[0131] If the judgment result of step S312 is no, continue to execute step S309 until the judgment result of step S312 is yes, indicating that the learning of the model has reached the optimum, and execute step S313.

[0132] Step S313: Terminate the model training.

[0133] The above steps are the same as those in Example 1, and are not described in detail here. For details, please refer to the corresponding steps in Example 1.

[0134] In the second embodiment, each time the model is trained and the new weight of the road is obtained, the new planned route corresponding to each historical trajectory route in the sample set is determined according to the new weight of the road, and an adversarial network is generated. The new planned route is added to the sample set, and the new sample set is used to train the model again. The training is iterated back and forth until the training result meets the preset requirements, so that the final training result is more accurate and closer to the actual situation of the road.

[0135] The method in the first embodiment above only trains the machine learning model once, and does not introduce new planned routes to generate adversarial samples. The sample data used for training only consists of historical trajectory routes and one planned route. This generates fewer constraints, and the scope of application of the learned right of way is limited. It is impossible to give a reasonable route planning when the sample is not covered; but the amount of calculation is small and the learning efficiency is high. The method in the second embodiment introduces new planned routes to generate adversarial sample data. The wrong route planning caused by insufficient coverage of the previous sample data will be used as a negative sample to enter the model training again, thereby improving the effect of route planning; but the amount of calculation is also relatively increased.

[0136] Embodiment 3

[0137] Embodiment 3 of the present invention provides another specific implementation process of the road weight determination method, and its process is as follows: Figure 3 As shown, the following steps are included:

[0138] Step S401: Filter the historical driving trajectory.

[0139] Step S402: planning a route from the first track point to the last track point according to the first and last track points and the departure time recorded in the historical driving track.

[0140] Step S403: performing road matching on the track points included in the historical driving track to obtain roads matched with the track points, and the roads constitute a historical track route corresponding to the historical driving track.

[0141] Step S404: determine whether the ratio of the travel distance of the historical trajectory route to the travel distance of each corresponding planned route is less than a preset ratio threshold.

[0142] It is determined that the ratio of the travel distance of the historical trajectory route to the travel distance of each corresponding planned route is less than the preset ratio threshold, verifying that the user trajectory has not taken a detour, and executing step S406; if the judgment is no, it is determined that the user trajectory has taken a detour, and executing step S405.

[0143] Step S405: Delete the historical track route and the corresponding historical driving track.

[0144] Step S406: The historical trajectory route and the corresponding planned route are combined into a sample to obtain a sample set.

[0145] Step S407: Use the sample set to train the machine learning model and output the travel time coefficient and travel distance coefficient of the road.

[0146] Step S408: Determine the weight of the road according to the regular travel time, travel distance, travel time coefficient and travel distance coefficient of each road.

[0147] Step S409: for each sample in the current sample set, a new planned route is determined according to the first and last track points, the departure time and the weight of the current road of the historical driving track corresponding to the current sample.

[0148] Step S410: Determine whether the new planned route is consistent with the historical trajectory route.

[0149] Determine whether any of the newly planned routes is consistent with the historical trajectory route in the sample. If not, execute step S412; if so, execute step S411.

[0150] Step S411: Delete the sample corresponding to the planned route in the new sample set.

[0151] Step S412: adding the new planned route to the samples to obtain a new sample set.

[0152] Step S413: Use the set of new samples to train the currently trained machine learning model and output the new travel time coefficient and new travel distance coefficient of the road.

[0153] Step S414: Determine a new weight of the road according to the normal travel time, travel distance, new travel time coefficient and new travel distance coefficient of the road.

[0154] Step S415: Determine whether a preset training termination condition is met.

[0155] If the judgment in step S415 is no, continue to execute step S409 until it is judged to be yes, indicating that the learning of the model has reached the optimum, and execute step S416.

[0156] Step S416: Terminate the model training.

[0157] The above steps are the same as those in Embodiment 1 or Embodiment 2, and are not described in detail here. For details, please refer to the corresponding steps in Embodiment 1 or Embodiment 2.

[0158] In the above-mentioned third embodiment, during the iterative training of the machine learning model, for each sample in the current sample set, after determining the new planned route according to the new weights of the first and last track points, departure time and roads corresponding to the historical track route, it is determined whether the new planned route is consistent with the historical track route. If any new planned route is consistent with the historical track route, it is determined that the sample has been learned, and the sample is deleted; if the new planned routes are inconsistent with the historical track routes, the new planned routes are added to the samples to obtain a new sample set, and the machine learning model is retrained using the new sample set. In the process of iterative training of the machine learning model, the samples that have been learned are continuously deleted, which greatly reduces the amount of calculation and improves the learning efficiency.

[0159] The method for determining the road weight in the present invention is not limited to the method in the above embodiment, and any combination of the steps in the above embodiment can be understood as the method for determining the road weight in the present invention.

[0160] Based on the inventive concept of the present invention, an embodiment of the present invention further provides a route planning method, including:

[0161] The navigation route is determined according to the user's navigation starting point, end point, request time and the road weight determined according to the above road weight determination method.

[0162] Based on the inventive concept of the present invention, an embodiment of the present invention further provides a method for generating a road right model, comprising:

[0163] According to the first and last track points and the departure time recorded in the historical driving track, a planned route from the first track point to the last track point is planned;

[0164] Performing road matching on the track points included in the historical driving track to obtain roads matched by the track points, and the roads constitute the historical track route corresponding to the historical driving track;

[0165] The historical trajectory route and the corresponding planned route are combined into a sample;

[0166] The set of samples is used to train a machine learning model, and the machine learning model is used to output a travel time coefficient and a travel distance coefficient of the road.

[0167] Optionally, the road right model generation method may also be generated by using any of the road right value determination methods in the above embodiments.

[0168] Based on the inventive concept of the present invention, an embodiment of the present invention further provides a road weight determination device, which can be arranged in a navigation device. The structure of the device is as follows: Figure 5 As shown, including:

[0169] A planning module 51, for planning a route from the first track point to the last track point according to the first and last track points and the departure time recorded in the historical driving track;

[0170] A matching module 52, configured to perform road matching on the track points included in the historical driving track to obtain roads matched by the track points, and the roads constitute a historical track route corresponding to the historical driving track;

[0171] A combining module 53, used to combine the historical trajectory route planned by the planning module 51 and the corresponding planned route matched by the matching module 52 into a sample;

[0172] A training module 54, used to train a machine learning model using the set of samples combined by the combination module 54, and output a travel time coefficient and a travel distance coefficient of the road;

[0173] The determination module 55 is used to determine the weight of the road according to the normal travel time, travel distance, travel time coefficient and travel distance coefficient of the road.

[0174] In some embodiments, the planning module 51 is specifically configured to:

[0175] The route connecting the first and last track points of the historical driving trajectory record is used as an alternative planned route; the sum of the road weights of each road included in each alternative planned route is determined according to the departure time of the historical driving trajectory record and the weight of the road corresponding to the departure time; and the planned route is determined according to the sum of the road weights of the alternative planned routes.

[0176] In some embodiments, the road weight determination device further includes a first judgment module 56;

[0177] After the determination module 55 determines the weight of the road, the planning module 51 is also used to determine a new planned route according to the first and last trajectory points, the departure time and the weight of the current road of the historical driving trajectory corresponding to the current sample, and add the new planned route to the sample; the training module 54 is also used to train the currently trained machine learning model using the set of new samples, and output the new travel time coefficient and new travel distance coefficient of the road; the determination module 55 is also used to determine the new weight of the road according to the regular travel time, travel distance, new travel time coefficient and new travel distance coefficient of the road; the first judgment module 56 is used to judge whether the preset termination training conditions are met; if the first judgment module 56 judges as yes, the planning module 51, the training module 54 and the determination module 55 continue to perform the above functions in sequence.

[0178] In some embodiments, the above-mentioned road weight determination device also includes a second judgment module 57, which is used to judge whether the new planned route is consistent with the historical trajectory route before the planning module 51 adds the new planned route to the sample; if the second judgment module 57 judges as yes, the planning module 51 executes to add the new planned route to the sample; if the second judgment module 57 judges as no, the planning module 51 is used to delete the sample.

[0179] In some embodiments, the first determination module 56 is specifically configured to:

[0180] Determine whether the number of training times reaches a preset threshold; and / or, determine whether the loss function value determined according to the currently trained machine learning model meets the preset conditions; and / or, determine whether the yaw rate determined according to the currently trained machine learning model meets the preset conditions.

[0181] In some embodiments, the training module 54 is specifically configured to:

[0182] Obtain discrete data of set characteristic parameters of each road in the historical trajectory route in the sample, obtain discrete data of set characteristic parameters of each road in each planned route in the sample, and obtain discrete samples; use the set of discrete samples to train a selected logistic regression model.

[0183] In some embodiments, the training module 54 is specifically configured to:

[0184] Obtain continuous feature expression data of set feature parameters of each road in the historical trajectory route in the sample, obtain continuous feature expression data of set feature parameters of each road in each planned route in the sample, and obtain continuous feature expression samples; use the set of continuous feature expression samples to train the selected neural network model.

[0185] In some embodiments, the above-mentioned road weight determination device also includes a third judgment module 58, which is used for the planning module 51 to judge whether the historical driving trajectory meets the preset sample conditions before planning the planned route from the first trajectory point to the last trajectory point according to the first and last trajectory points and the departure time recorded in the historical driving trajectory; if the third judgment module 58 judges as yes, the planning module 51 executes the planning of the planned route from the first trajectory point to the last trajectory point according to the first and last trajectory points and the departure time recorded in the historical driving trajectory; if the third judgment module 58 judges as no; the planning module 51 is used to delete the historical driving trajectory.

[0186] In some embodiments, the third determination module 58 is specifically configured to:

[0187] Determine whether the historical driving trajectory contains violation record information; and / or determine whether the time interval between any two adjacent trajectory points in the historical driving trajectory is less than a preset time threshold.

[0188] In some embodiments, the above-mentioned road weight determination device also includes a fourth judgment module 59, which is used to, after the planning module 51 forms the historical trajectory route corresponding to the historical driving trajectory from the road, determine whether the ratio of the travel distance of the historical trajectory route to the travel distance of the corresponding planned routes is less than a preset ratio threshold; if the fourth judgment module 59 judges to be yes, the combination module 53 executes the historical trajectory route and the corresponding planned route into a sample.

[0189] Based on the inventive concept of the present invention, an embodiment of the present invention further provides a road right model generation device, which can be arranged in a navigation device. The structure of the device is as follows: Figure 6 As shown, including:

[0190] A planning module 61, for planning a route from the first track point to the last track point according to the first and last track points and the departure time recorded in the historical driving track;

[0191] A matching module 62, configured to perform road matching on the track points included in the historical driving track to obtain roads matched by the track points, and the roads constitute a historical track route corresponding to the historical driving track;

[0192] A combining module 63, used to combine the historical trajectory route planned by the planning module 61 and the corresponding planned route matched by the matching module 62 into a sample;

[0193] The training module 64 is used to train a machine learning model using the set of samples combined by the combination module 63, and the machine learning model is used to output the travel time coefficient and the travel distance coefficient of the road.

[0194] Based on the inventive concept of the present invention, an embodiment of the present invention also provides a navigation device, which is provided with the above-mentioned road weight determination device; the navigation device is used to determine the navigation route according to the user's navigation starting point, end point, request time and the road weight determined by the road weight determination device.

[0195] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0196] Based on the inventive concept of the present invention, an embodiment of the present invention provides a computer-readable storage medium on which computer instructions are stored. When the instructions are executed by a processor, the above-mentioned road weight determination method, or the above-mentioned route planning method, or the above-mentioned road right model generation method is implemented.

[0197] Unless otherwise specifically stated, terms such as processing, computing, calculating, determining, displaying, etc. may refer to the actions and / or processes of one or more processing or computing systems, or similar devices, which operate and convert data represented as physical (e.g., electronic) quantities within registers or memories of a processing system into other data similarly represented as physical quantities within memories, registers, or other such information storage, transmission, or display devices of the processing system. Information and signals may be represented using any of a variety of different techniques and methods. For example, data, instructions, commands, information, signals, bits, symbols, and chips mentioned throughout the above description may be represented by voltages, currents, electromagnetic waves, magnetic fields or particles, light fields or particles, or any combination thereof.

[0198] It should be understood that the specific order or hierarchy of steps in the disclosed process is an example of an exemplary method. Based on design preferences, it should be understood that the specific order or hierarchy of steps in the process can be rearranged without departing from the scope of protection of the present disclosure. The attached method claims present the elements of the various steps in an exemplary order and are not intended to be limited to the specific order or hierarchy described.

[0199] In the above detailed description, various features are grouped together in a single embodiment to simplify the disclosure. This method of disclosure should not be interpreted as reflecting an intention that the embodiments of the claimed subject matter require more features than are clearly stated in each claim. On the contrary, as reflected in the appended claims, the invention is in a state of having less than all the features of the disclosed individual embodiments. Therefore, the appended claims are hereby expressly incorporated into the detailed description, with each claim standing on its own as a separate preferred embodiment of the invention.

[0200] Those skilled in the art will also appreciate that the various illustrative logic blocks, modules, circuits, and algorithmic steps described in conjunction with the embodiments herein can all be implemented as electronic hardware, computer software, or a combination thereof. In order to clearly illustrate the interchangeability between hardware and software, various illustrative components, blocks, modules, circuits, and steps are generally described above around their functions. Whether such functions are implemented as hardware or software depends on specific applications and the design constraints imposed on the entire system. A skilled person can implement the described functions in an alternative manner for each specific application, but such implementation decisions should not be interpreted as departing from the scope of protection of the present disclosure.

[0201] The steps of the method or algorithm described in conjunction with the embodiments herein may be directly embodied as hardware, a software module executed by a processor, or a combination thereof. The software module may be located in a RAM memory, a flash memory, a ROM memory, an EPROM memory, an EEPROM memory, a register, a hard disk, a mobile disk, a CD-ROM, or any other form of storage medium well known in the art. An exemplary storage medium is connected to the processor so that the processor can read information from the storage medium and can write information to the storage medium. Of course, the storage medium may also be an integral part of the processor. The processor and the storage medium may be located in an ASIC. The ASIC may be located in a user terminal. Of course, the processor and the storage medium may also be present in a user terminal as discrete components.

[0202] For software implementation, the techniques described in this application can be implemented with modules (e.g., procedures, functions, etc.) that perform the functions described in this application. These software codes can be stored in a memory unit and executed by a processor. The memory unit can be implemented within the processor or outside the processor. In the latter case, it is coupled to the processor in a communication manner via various means, which are well known in the art.

[0203] The above description includes examples of one or more embodiments. Of course, it is impossible to describe all possible combinations of components or methods for the purpose of describing the above embodiments, but it should be recognized by those skilled in the art that the various embodiments may be further combined and arranged. Therefore, the embodiments described herein are intended to cover all such changes, modifications and variations that fall within the scope of protection of the appended claims. In addition, with respect to the term "comprising" used in the specification or claims, the word is covered in a manner similar to the term "including", just as "including," is explained as a transitional word in the claims. In addition, any term "or" used in the specification of the claims is intended to mean "non-exclusive or".

Claims

1. A method for determining road weight, include: According to the first and last track points and departure time recorded in the historical driving track, one or more planned routes are planned from the first track point to the last track point; Performing road matching on the track points included in the historical driving track to obtain roads matched by the track points, and the roads constitute the historical track route corresponding to the historical driving track; The historical trajectory route and the corresponding planned route are combined into a sample; Using the set of samples to train a machine learning model, outputting a travel time coefficient and a travel distance coefficient of the road; The weight of the road is determined based on the road's regular travel time, travel distance, travel time coefficient and travel distance coefficient.

2. The method according to claim 1, wherein the planning route from the first track point to the last track point is planned according to the first and last track points and the departure time recorded in the historical driving track, specifically include: The route connecting the first and last track points of the historical driving track record is used as an alternative planning route; Determine the sum of the road weights of the roads included in each candidate planned route according to the departure time recorded in the historical driving trajectory and the road weights corresponding to the departure time; The planned route is determined according to the sum of the road weights of the alternative planned routes.

3. The method according to claim 1, wherein after determining the weight of the road, include: Determine a new planned route according to the first and last track points, the departure time and the weight of the current road of the historical driving track corresponding to the current sample, and add the new planned route to the sample; Using the set of new samples to train the currently trained machine learning model, outputting a new travel time coefficient and a new travel distance coefficient for the road; Determine the new weight of the road based on the regular travel time, travel distance, new travel time coefficient and new travel distance coefficient of the road; Determine whether the preset termination training condition is met. If not, continue to determine the new planned route according to the first and last track points, departure time and current road weight of the historical driving track corresponding to the current sample, and add the new planned route to the sample.

4. The method according to claim 3, wherein before adding the new planned route to the sample, include: Determining whether the newly planned route is consistent with the historical trajectory route; If not, executing the step of adding the new planned route to the sample; If so, delete the sample.

5. The method according to claim 3, wherein the determining whether the preset termination condition of training is met comprises: include: Determine whether the number of training times reaches a preset number threshold; and / or, Determine whether the loss function value determined by the currently trained machine learning model meets the preset conditions; and / or, Determine whether the yaw rate determined by the currently trained machine learning model meets the preset conditions.

6. The method according to claim 1, wherein the machine learning model is trained using the set of samples, specifically: include: Obtaining discrete data of set characteristic parameters of each road in the historical trajectory route in the sample, and obtaining discrete data of set characteristic parameters of each road in each planned route in the sample, to obtain discrete samples; A selected logistic regression model is trained using the set of discrete samples.

7. The method of claim 1, wherein the machine learning model is trained using the set of samples, specifically: include: Obtaining continuous feature expression data of set feature parameters of each road in the historical trajectory route in the sample, and obtaining continuous feature expression data of set feature parameters of each road in each planned route in the sample, to obtain a continuous feature expression sample; The selected neural network model is trained using the set of continuous feature expression samples.

8. The method according to any one of claims 1 to 7, wherein before planning a route from the first track point to the last track point based on the first and last track points and the departure time recorded in the historical driving track, include: Determine whether the historical driving trajectory meets the preset sample conditions; If yes, execute the planning route from the first track point to the last track point according to the first and last track points and the departure time recorded in the historical driving track; If not, delete the historical driving track.

9. The method according to claim 8, wherein the step of determining whether the historical driving trajectory meets the preset sample condition comprises: include: Determine whether the historical driving trajectory contains violation record information; and / or, Determine whether the time interval between any two adjacent trajectory points in the historical driving trajectory is less than a preset time threshold.

10. The method according to claim 8, wherein after the road forms the historical trajectory route corresponding to the historical driving trajectory, include: Determine whether the ratio of the travel distance of the historical trajectory route to the travel distance of each corresponding planned route is less than a preset ratio threshold; If so, the historical trajectory route and the corresponding planned route are combined into a sample.

11. A route planning method, include: The navigation route is determined according to the user's navigation starting point, end point, request time and the road weight determined according to the road weight determination method according to any one of claims 1-10.

12. A method for generating a road right model. include: According to the first and last track points and the departure time recorded in the historical driving track, one or more planned routes are planned from the first track point to the last track point; Performing road matching on the track points included in the historical driving track to obtain roads matched by the track points, and the roads constitute the historical track route corresponding to the historical driving track; The historical trajectory route and the corresponding planned route are combined into a sample; The set of samples is used to train a machine learning model, and the machine learning model is used to output a travel time coefficient and a travel distance coefficient of the road.

13. A road weight determination device, include: A planning module, used to plan one or more planned routes from the first track point to the last track point according to the first and last track points and departure time recorded in the historical driving track; A matching module is used to match the track points included in the historical driving track with roads to obtain roads matched by the track points, and the roads constitute the historical track routes corresponding to the historical driving track; a combining module is used to combine the historical track routes planned by the planning module and the corresponding planned routes matched by the matching module into a sample; A training module, used to train a machine learning model using the set of samples combined by the combination module, and output a travel time coefficient and a travel distance coefficient of a road; The determination module is used to determine the weight of the road according to the regular travel time, travel distance, travel time coefficient and travel distance coefficient of the road.

14. A road right model generation device, include: A planning module, used to plan one or more planned routes from the first track point to the last track point according to the first and last track points and departure time recorded in the historical driving track; A matching module is used to match the track points included in the historical driving track with roads to obtain roads matched by the track points, and the roads constitute the historical track routes corresponding to the historical driving track; a combining module is used to combine the historical track routes planned by the planning module and the corresponding planned routes matched by the matching module into a sample; A training module is used to train a machine learning model using a set of samples combined by the combination module, and the machine learning model is used to output a travel time coefficient and a travel distance coefficient of a road.

15. A navigation device, the navigation device being provided with the road weight determination device according to claim 13; The navigation device is used to determine a navigation route according to a user's navigation starting point, end point, request time and the road weight determined by the road weight determination device.

16. A computer-readable storage medium having computer instructions stored thereon, which, when executed by a processor, implement the road weight determination method described in any one of claims 1 to 10, or implement the route planning method described in claim 11, or implement the road right model generation method described in claim 12.

Images