Device and method for predicting a passage time and method for predicting a travel time

Abstract

The present embodiment relates to an apparatus and method of predicting a passing time and a method of predicting a travel time. More specifically, the present disclosure relates to a technology for predicting a passing time of a vehicle in a waiting line section set behind a cross section by inputting information on a speed of the vehicle to a trained prediction model and more accurately predicting a time taken for a user to arrive at his or her destination.

Description

[0001] Cross-references to related applications

[0002] This application claims priority to Korean Patent Application No. 10-2021-0186117, filed on December 23, 2021, the entire contents of which are incorporated herein by reference. Technical Field

[0003] This disclosure relates to a method for predicting intersection passage time by taking into account irregular rotation and signal waiting, and to apparatus for using the method. Background Technology

[0004] Traffic flow refers to the number of vehicles passing through a point within a unit of time. Uninterrupted flow refers to traffic flow that does not involve external influences controlling the traffic flow. Interrupted flow refers to traffic flow that is discontinuous and involves traffic lights or other external influences.

[0005] Examples of disrupted flow can include signalized intersections, unsignalized intersections, and roundabouts. At such intersections, waiting lines can form, potentially leading to congestion due to factors such as vehicles turning around and waiting for signals.

[0006] Even though waiting lines can significantly impact vehicle speed and predicted arrival times, waiting lines formed within interrupted flow also suffer from low traffic information collection rates. Therefore, it remains difficult to predict the accurate travel time to the destination.

[0007] With the increasing attention given to artificial intelligence (AI) technology, research into AI is actively underway. If artificial neural networks are used in such AI technologies, it is believed that reliable guidance on transit times and destination arrival times can be provided to users by relatively accurately predicting traffic conditions that are difficult to predict due to disruptions. Summary of the Invention

[0008] In this context, the purpose of this embodiment is to calculate the transit time in the waiting line section based on the direction of the vehicle's travel at the intersection, calculate the route that the vehicle can take to reach its destination in the shortest possible time by using a prediction model to derive the predicted transit time in each future direction of the vehicle's travel, and provide the route to the user.

[0009] To achieve this objective, one embodiment can provide a method for predicting transit times, comprising: collecting traffic information including the speed of vehicles traveling in each direction through a waiting line section located behind an intersection; calculating the transit time of vehicles traveling in each direction through the waiting line section based on the traffic information and the length of the waiting line section, and calculating the average transit time for each direction based on the transit time; inputting the average transit time into a prediction model trained to predict future transit times; and deriving the predicted transit time for each direction from the trained prediction model.

[0010] In methods for predicting transit time, the average transit time can be derived by calculating the transit time for each direction of travel corresponding to the 10 minutes before and after the derived timing of vehicles that have already passed the waiting line section using an arithmetic averaging method. The derived timing can indicate the derived average transit time in 5-minute increments.

[0011] The speed data collected in 5-minute intervals may have missing values. However, these missing values ​​can be corrected by reprocessing the vehicle transit times before and after the derived timing using an arithmetic averaging method. Furthermore, since recent past and recent future traffic information are integrated equally, irregularities in vehicle transit times attributable to turnaround waiting or signal waiting at intersections can be smoothed out.

[0012] In methods for predicting transit time, a predictive model can be trained by receiving average transit times accumulated from the past based on a time series. The average transit time accumulated from the past can be derived by calculating the transit time for each direction of travel corresponding to the 10 minutes before and after the derived timing of vehicles that have already passed the waiting line section, using an arithmetic averaging method. The derived timing can indicate the derived average transit time in 5-minute increments. The predicted transit time can be derived for each direction of travel by dividing the current time into 5-minute increments for the next hour.

[0013] In the method for predicting transit time, waiting line segments are set differently for each direction of vehicle travel.

[0014] In methods for predicting transit time, the trained prediction model can be a model trained using deep learning based on artificial neural networks.

[0015] Methods for predicting transit time can further include generating waiting line segment information that includes the length of the waiting line segment in each direction of travel, road gradient, road type and number of lanes, signal type at intersections, and environmental information including day of the week, time, and administrative district. Average transit time, waiting line segment information, and environmental information can be input into a trained prediction model.

[0016] To achieve this objective, another embodiment may provide an apparatus for predicting transit times, comprising: a collection circuit configured to collect traffic information including the speed of vehicles traveling in each direction through a waiting line section located behind an intersection; an arithmetic circuit configured to calculate the transit time of vehicles traveling in each direction through the waiting line section based on the traffic information and the length of the waiting line section, and to calculate the average transit time for each direction based on the transit time; and a prediction circuit including a prediction model trained to predict future transit times, and configured to input the average transit time into the trained prediction model and derive the predicted transit time for each direction.

[0017] In a device for predicting transit time, an arithmetic circuit can derive an average transit time by calculating the transit time corresponding to the 10 minutes before and after the derived timing of vehicles that have already passed the waiting line section in each direction of travel using an arithmetic averaging method. The derived timing can indicate the derived average transit time in 5-minute increments. Speeds collected in 5-minute increments may have missing values. However, these missing values ​​can be corrected by reprocessing the transit times of vehicles before and after the derived timing using an arithmetic averaging method. Furthermore, since recent past traffic information and recent future traffic information are integrated equally, irregular movements in the transit times of vehicles attributed to turning waits or signal waits at intersections can be smoothed out.

[0018] The device for predicting transit time may further include: management circuitry configured to generate waiting line segment information including the length of the waiting line segment in each direction of travel, road gradient, road type and number of lanes, signal type of the intersection, and environmental information including information about the day of the week, time, and administrative district.

[0019] The device for predicting transit time may further include: a transmission circuit configured to transmit the predicted transit time to a vehicle navigation management server.

[0020] The apparatus for predicting transit time may further include: a learning circuit configured to train the prediction model by inputting an average transit time accumulated from the past based on a time series. The average transit time accumulated from the past may be derived by calculating, using an arithmetic averaging method, the transit time corresponding to 10 minutes before and after the derived timing of vehicles that have already passed the waiting line section for each direction of travel, prior to and after the derived timing. The derived timing may indicate a timing in 5-minute increments for deriving the average transit time.

[0021] To achieve this objective, another embodiment may provide a method for predicting travel time, comprising: collecting traffic information including the speed of vehicles traveling in each direction through a waiting line segment located behind an intersection; calculating the transit time of vehicles traveling through each waiting line segment and in each direction based on the traffic information and the length of the waiting line segment, and calculating the average transit time in each direction based on the transit time; inputting the average transit time into a prediction model trained to predict future transit times; deriving the predicted transit time in each direction from the trained prediction model; and deriving a route with the minimum sum of predicted transit times from the waiting line segments through which vehicles travel to reach their destinations.

[0022] In methods for predicting travel time, the average transit time can be derived by calculating the transit time for each direction of travel corresponding to the 10 minutes before and after the derived timing of vehicles that have already passed the waiting line segment using an arithmetic averaging method. The derived timing can indicate the average transit time in 5-minute increments. Speeds collected in 5-minute increments may have missing values. However, these missing values ​​can be corrected by reprocessing the transit times of vehicles before and after the derived timing using an arithmetic averaging method. Furthermore, since recent past and recent future traffic information are integrated equally, irregular movements in vehicle transit times attributable to turning waits or signal waits at intersections can be smoothed out.

[0023] The method for predicting travel time may further include generating waiting line segment information including the length of the waiting line segment in each direction of travel, road gradient, road type and number of lanes, signal type at intersections, and environmental information including the day of the week, time, and administrative district. The trained prediction model can be a model trained using deep learning based on artificial neural networks. Average travel time, waiting line segment information, and environmental information can be input into the trained prediction model.

[0024] In methods for predicting travel time, a predictive model can be trained by receiving average transit times accumulated from the past based on a time series. The average transit time accumulated from the past can be derived by calculating the transit time for each direction of travel corresponding to the 10 minutes before and after the derived timing of vehicles that have already passed the waiting line section, using an arithmetic averaging method. The derived timing can indicate the derived average transit time in 5-minute increments. The predicted transit time can be derived for each direction of travel by dividing the actual time into 5-minute increments for the next hour.

[0025] As described above, the advantage of this embodiment is that by accurately predicting the time it takes for a vehicle to pass through an intersection in each direction of its travel, the user's arrival time at their destination can be accurately predicted. Furthermore, it addresses the problem that when collecting speeds based on the speeds of vehicles that have completely passed through the intersection in each direction of their travel, the collection rate may be low, and the reliability of predicting the future based on information with a low collection rate may be low. Additionally, it achieves the effect of predicting stable passage times in the waiting lines for each direction of travel, which may be progressively more reliable for the vehicle compared to passage times attributed to irregular rotations within the intersection section or inconsistent signals for each vehicle waiting. Attached Figure Description

[0026] Figure 1 This is a diagram used to schematically illustrate this embodiment in relation to intersections and waiting line sections.

[0027] Figure 2 This is a diagram illustrating a device for predicting transit time according to this embodiment.

[0028] Figure 3 This is a diagram illustrating the method for predicting transit time according to this embodiment.

[0029] Figure 4 This is a diagram used to describe the predicted transit time derived by using a prediction model according to this embodiment.

[0030] Figure 5 This is an exemplary diagram of a route guide using a method for predicting travel time according to this embodiment. Detailed Implementation

[0031] In the following, some embodiments are described in detail with reference to the exemplary accompanying drawings. When adding reference numerals to the elements in each drawing, it should be noted that even if the same element is shown in different drawings, it should have the same reference numerals as much as possible. Furthermore, in describing embodiments of this disclosure, detailed descriptions of related well-known configurations or functions will be omitted if it is determined that such detailed descriptions interfere with the understanding of the embodiments of this disclosure.

[0032] Furthermore, in describing the elements of embodiments of this disclosure, terms such as first, second, A, B, (a), and (b) may be used. Such terms are used only to distinguish one component from another, and the nature, order, or arrangement of the corresponding components is not limited by the terms. When it is said that a component is “connected,” “combined,” or “coupled” to another component, a component may be directly connected or coupled to another component; however, it should also be understood that a third component may be “connected,” “combined,” or “coupled” between two components.

[0033] Figure 1 This is a diagram used to schematically illustrate this embodiment in relation to intersections and waiting line sections.

[0034] Based on the time-prediction method according to this embodiment, the waiting line section 120 can be pre-set after the intersection 110. Vehicles passing through the intersection 110 may slow down or stop for various reasons, such as stopping while waiting for a traffic light signal at the intersection or stopping while waiting for a left or right turn by slowing down. The waiting line section 120 can refer to a section at the intersection 110 where vehicles can stop or proceed for various reasons.

[0035] Waiting line segments 120 can be set differently for each direction a vehicle travels in at an intersection (e.g., left turn, right turn, or straight ahead). Therefore, when predicting the transit time of vehicles in the waiting line segments using a predictive model, the transit time can be predicted more accurately because the waiting line segments 120 are determined individually based on the direction the vehicle is traveling.

[0036] The length of the waiting line section 120, road gradient, road type and number of lanes, signal type of intersection 110, information about the day of the week, time and administrative district, etc., can be pre-stored in the management circuit of the device for predicting passage time according to this embodiment.

[0037] Speed ​​information of waiting line section 120 can be collected by detection equipment that has entered and completely passed through waiting line section 120 in each direction, or by detection equipment installed in the road.

[0038] like Figure 1 As shown, waiting lines at intersection 110 can be generated for various reasons, such as waiting lines generated by traffic lights indicating straight or left turns and waiting lines generated for vehicles turning.

[0039] Furthermore, the predicted passage time can be calculated by using the speed of vehicles at the waiting line section and the calculated passage time smoothed and used as an average to calculate the passage time in the waiting line section. Therefore, the predicted passage time can be the passage time that has already incorporated the turning times of all waiting and waiting vehicles attributable to the traffic lights set at intersection 110.

[0040] Apart from Figure 1 In addition to the intersection 110 shown, the intersection according to this embodiment may include an intersection with five or more directions, at which vehicles can even travel in the straight direction, left-turn direction, right-turn direction and diagonal direction.

[0041] Figure 2 This is a diagram illustrating a device for predicting transit time according to this embodiment.

[0042] The apparatus 200 for predicting transit time according to this embodiment may include a collection circuit 210, an arithmetic circuit 220 and a prediction circuit 230, and may further include a management circuit 240.

[0043] The collection circuit 210 can collect traffic information, including the speed of vehicles traveling in each direction within the waiting line section located behind the intersection. The traffic information collected by the collection circuit 210 can include the speed of vehicles traveling in each direction at the intersection. The waiting line sections can be set differently for each direction of vehicle travel.

[0044] The traffic information collected by the collection circuit 210 can be collected by detection equipment installed in each vehicle, or by detection equipment installed on the road.

[0045] Arithmetic circuit 220 can calculate the transit time of vehicles passing through the waiting line section for each direction of vehicle movement based on traffic information and the length of the waiting line section, and can calculate the average transit time of vehicles in each direction of movement based on the transit time. The length of the waiting line section can be pre-stored in the device 200 for predicting transit time. Specifically, the length of the waiting line section can be pre-stored in the arithmetic circuit 220 of the device 200 for predicting transit time. If the device 200 for predicting transit time further includes management circuit 240, the length of the waiting line section can be pre-stored in the management circuit 240.

[0046] Arithmetic circuit 220 calculates the transit time for each direction of vehicle movement. Therefore, arithmetic circuit 220 can predict the future transit time of vehicles in the waiting line segment for each direction of vehicle movement based on the direction of vehicle movement. Since the smoothed transit time of vehicles in the waiting line segment is used, it can be assumed that all information (such as signal waiting at the intersection, information on turning waiting, etc.) has been incorporated into the average transit time. Furthermore, the average transit time of vehicles passing through the waiting line segment can be calculated and derived using an arithmetic averaging method. Specifically, the average transit time is derived by calculating the transit time of vehicles in the waiting line segment using an arithmetic averaging method, which corresponds to 10 minutes before and after the derived timing of vehicles passing through the waiting line segment in each direction of movement. The derived timing can indicate the derived average transit time in 5-minute increments. Therefore, by calculating and smoothing the transit time of vehicles in the waiting line segment using the arithmetic averaging method described above, a more accurate prediction of vehicle transit time can be made relative to irregular signals at the intersection and the turning of waiting vehicles. That is, speeds collected in 5-minute increments may have missing values. However, missing values ​​can be corrected by reprocessing the vehicle transit times before and after the derived timing using an arithmetic averaging method. Furthermore, since recent past and recent future traffic information are integrated equally, irregularities in vehicle transit times attributable to turnaround waiting or signal waiting at intersections can be smoothed out.

[0047] Prediction circuit 230 includes a prediction model that has been trained to predict future transit times. Prediction circuit 230 can input the average transit time of a vehicle into the trained prediction model, and can also input information stored in management circuit 240, including the length of the waiting line segment, road gradient, road type and number of lanes, signal type, day of the week, time, and administrative district (including average transit time), into the trained prediction model. Prediction circuit 230 can derive the predicted transit time of the vehicle for each direction of its travel.

[0048] A predicted passage time derived by prediction circuit 230 can be generated for each direction of vehicle movement at the intersection, and this predicted passage time can be generated by dividing the actual time into 5-minute increments for the next hour.

[0049] The trained prediction model included in the prediction circuit 230 can be a model that has been trained by deep learning based on artificial neural networks.

[0050] That is, the prediction model can learn the average transit time of vehicles in each forward direction generated in the past. The device 200 for predicting transit time may further include learning circuitry (not shown) for training the prediction model.

[0051] The learning circuit can be trained by inputting the prediction model with the average vehicle transit time accumulated from the past. The average transit time accumulated from the past can be derived by calculating the vehicle transit time corresponding to the 10 minutes before and after the derived timing of vehicles that have already passed through the waiting line segment in each direction of vehicle movement using an arithmetic averaging method. The derived timing can indicate the derived average transit time in 5-minute increments. That is, smoothing the transit time using an arithmetic averaging method allows for more accurate prediction of vehicle transit times in waiting line segments that incorporate irregular signal waiting and rotating waiting vehicles. Specifically, speeds collected in 5-minute increments may have missing values. However, these missing values ​​can be corrected by reprocessing the vehicle transit times before and after the derived timing using an arithmetic averaging method. Furthermore, since recent past traffic information and recent future traffic information are integrated equally, irregular movements in vehicle transit times attributable to rotating waiting or signal waiting at intersections can be smoothed.

[0052] If the device 200 for predicting passage time further includes a learning circuit, the learning circuit can train a prediction model using past average passage times. The prediction circuit 230 can derive the predicted passage time using the trained prediction model. Furthermore, if the device 200 for predicting passage time includes a management circuit 240, the learning circuit can learn information obtained by operating on the average passage time calculated by the arithmetic circuit 220, as well as waiting line segment information and environmental information generated by the management circuit 240. In this case, the past average passage time can be derived by calculating the passage time of vehicles corresponding to the 10 minutes before and after the derivation timing of vehicles passing through the waiting line segment using an arithmetic averaging method for each direction of vehicle movement. By learning the past average passage time, the learning circuit can derive a more accurate passage time in the waiting line segment when the average passage time is received. The management circuit 240 can generate waiting line segment information including the length of the waiting line segment in each direction of vehicle movement, road gradient, road type and number of lanes, and signal type of the intersection, as well as environmental information including information about the day of the week, time, and administrative district.

[0053] Basic information generated by management circuit 240 for waiting line segment information and environmental information can be pre-stored in management circuit 240. This basic information can be collected by detection devices installed in each vehicle operating on the road, or by detection devices installed along the road.

[0054] Furthermore, the prediction circuit 230 can receive average transit time from the arithmetic circuit 220, and waiting line segment information and environmental information from the management circuit 240. It can derive the predicted transit time by inputting the average transit time, waiting line segment information, and environmental information into a trained prediction model. The average transit time, waiting line segment information, and environmental information can interact with each other before being input into the trained prediction model. The interacting information can be input into the trained prediction model. Therefore, a predicted transit time can be derived, which includes information on each waiting line segment and waiting line segments in each direction of vehicle travel, and includes information on the time of vehicle operation, day of the week, and administrative district.

[0055] The apparatus 200 for predicting transit time according to this embodiment may further include a transmission circuit (not shown). The transmission circuit can transmit the information about the predicted transit time generated by the prediction circuit 230 to a management server that manages vehicle navigation. Furthermore, the management server can calculate the route taken by the user in the shortest time to reach their destination using the predicted transit time, and can provide the user with information about that route.

[0056] Figure 3 This is a diagram illustrating the method for predicting transit time according to this embodiment.

[0057] In the method for predicting transit time according to this embodiment, step S310 can be performed, which involves collecting traffic information, including the speed of vehicles passing through the waiting line section set behind the intersection in each direction of their forward movement within the waiting line section.

[0058] Traffic information may include measured speeds of vehicles traveling in each direction at the intersection. Waiting line sections may be set differently for each direction of vehicle travel. Traffic information may be collected by detection devices installed in each vehicle, or by detection devices installed in the road. Specifically, information about the speed of vehicles in the waiting line sections may be collected by detection devices that have entered and completely passed through the waiting line sections 120 in each direction, or by detection devices installed in the road.

[0059] Step S320 can be performed, which is to calculate the passage time of vehicles in each direction of travel through the waiting line section based on traffic information and the length of the waiting line section, and to calculate the average passage time in each direction of travel based on the passage time.

[0060] In step S320, the passage time for each direction of vehicle travel is calculated. Therefore, based on the direction of vehicle travel, the future passage time of vehicles in the waiting line segment can be predicted for each direction of vehicle travel. Since the smoothed passage time of vehicles in the waiting line segment is used, it can be assumed that all information (such as signal waiting and turning waiting information at the intersection) has been incorporated into the average passage time. Furthermore, the average passage time of vehicles passing through the waiting line segment can be calculated and derived using an arithmetic averaging method. Specifically, the average passage time can be derived by calculating the passage time of vehicles in the waiting line segment using an arithmetic averaging method, which corresponds to 10 minutes before and after the derived timing of vehicles passing through the waiting line segment in each direction of travel. The derived timing can indicate the timing in 5-minute increments for deriving the average passage time. That is, speeds collected in 5-minute increments may have missing values. However, since the passage times of vehicles before and after the derived timing are processed again using an arithmetic averaging method, missing values ​​can be corrected. Furthermore, since recent past traffic information and recent future traffic information are integrated equally, irregular movement in the transit time of vehicles attributable to turnaround waiting or signal waiting at intersections can be smoothed out.

[0061] A further step (not shown) can be performed to train the predictive model to learn past average transit times. Specifically, the predictive model can be trained by receiving average transit times accumulated from the past. The accumulated average transit times can be derived by calculating the transit times of vehicles in each direction of vehicle movement using an arithmetic averaging method, corresponding to the transit times of vehicles 10 minutes before and after the derived timing of vehicles that have already passed through the waiting line segment. The derived timing can indicate the derived average transit times in 5-minute increments. That is, by smoothing the transit times using an arithmetic averaging method, the transit times in waiting line segments, which already take into account irregular signals and the rotation of waiting vehicles, can be predicted more accurately.

[0062] Furthermore, the predictive model can learn from information obtained through operations such as average transit time, waiting line segment information, and environmental information. Therefore, when the real-time average transit time is input into the predictive model, it can be trained to predict future average transit times. In this case, the past average transit time can be derived by calculating the transit time of vehicles in each direction of vehicle movement using an arithmetic averaging method, corresponding to the transit time of vehicles 10 minutes before and after the derived timing of vehicles passing through the waiting line segment. The derived timing can indicate the timing in 5-minute increments for deriving the average transit time. By learning the derived timing, when the received average transit time is input, the learning circuit can derive a more accurate transit time within the waiting line segment.

[0063] Step S330 can be performed, which involves inputting the average transit time into a prediction model that has been trained to predict future transit times, and deriving the predicted transit time for each direction of the vehicle's movement from the trained prediction model.

[0064] The trained prediction model can be a model that has been trained using deep learning based on artificial neural networks. The predicted transit time can be derived by dividing the actual time into 5-minute increments and extending it to the next hour.

[0065] Before step S330, which derives the predicted transit time, further steps can be performed to generate waiting line segment information for each direction of vehicle travel, including the length of the waiting line segment, road gradient, road type and number of lanes, and signal type at intersections, as well as environmental information (not shown) including information about the day of the week, time, and administrative district. The basic information for the waiting line segment information and environmental information can be pre-stored in the management circuitry of the device used to predict transit time, which performs the method for predicting transit time. This basic information can be collected by detection devices installed in each vehicle operating on the road, or by detection devices installed along the road.

[0066] Furthermore, average transit time, waiting line segment information, and environmental information can be input into the trained prediction model. Therefore, a predicted transit time can be derived, which includes information on each waiting line segment and the waiting line segments in each direction of vehicle travel, as well as information on the time of travel, day of the week, and administrative district.

[0067] A further step (not shown) can be performed: after deriving the predicted transit time, transmit the information about the predicted transit time to the vehicle navigation management server. The management server can use the predicted transit time to calculate the route taken by the user in the shortest time to reach their destination and can provide the user with information about that route.

[0068] Figure 4 This is a diagram used to describe the predicted transit time derived by using a prediction model according to this embodiment.

[0069] To calculate the predicted passage time 450 for passing through the waiting line section set behind the intersection, the average passage time 410 of vehicles passing through the waiting line section can be input into the prediction model 440. For example... Figure 4 As shown, the waiting line segment information 420 and environmental information 430, along with the average transit time 410, can be input into the prediction model 440.

[0070] The average transit time 410 can be data in matrix form, where the average transit time for vehicles has been scheduled for each waiting line segment and each time period. The waiting line segment information 420 and the environmental information 430 can be data in matrix or vector form.

[0071] An average passage time 410 can be generated by a device for predicting passage time based on traffic information, including the speed of vehicles passing through the waiting line section. Traffic information can be collected by detection devices installed in each vehicle, or by detection devices installed in the road. Specifically, information about the speed of vehicles in the waiting line section can be collected by detection devices that have entered and completely passed through the waiting line section 120 in each direction, or by detection devices installed in the road.

[0072] Waiting line section information 420 may include the length of the waiting line section for each direction of vehicle travel, road gradient, road type and number of lanes, and intersection signal type. Environmental information 430 may include information about the day of the week, time of day, and administrative district.

[0073] Waiting line segment information 420 and environmental information 430 can be generated based on basic information pre-stored in the management circuit of the device used to predict transit time.

[0074] Figure 5 This is an exemplary diagram of a route guide using a method for predicting travel time according to this embodiment.

[0075] like Figure 5 As shown, when a vehicle is in motion, it typically passes through multiple intersections. Therefore, if the transit time of vehicles in the waiting line sections set behind multiple intersections can be predicted, users of the vehicle can more accurately check the time it will take to reach their destination and can be provided with guidance on the route requiring the shortest travel time.

[0076] Therefore, in the method for predicting travel time according to this embodiment, a step of collecting traffic information can be performed, which includes the speed of vehicles traveling in each direction within the waiting line sections set behind each intersection. The waiting line sections can be set differently for each intersection based on the direction of vehicle travel (e.g., straight, left turn, and right turn). Traffic information can be collected by detection devices installed in each vehicle or by detection devices installed in the road. Specifically, information about the speed of vehicles in the waiting line sections can be collected by detection devices that have entered and completely passed through the waiting line sections 120 in each direction, or by detection devices installed in the road. Furthermore, the traffic information may include information about vehicle speed.

[0077] The process can perform steps based on traffic information and the length of waiting line segments to calculate the transit time for each waiting line segment and for each direction of vehicle travel, and to calculate the average transit time based on the transit time. The average transit time can be derived by calculating the transit times of vehicles in the waiting line segments using an arithmetic averaging method, which corresponds to 10 minutes before and after the derived timing for vehicles traveling through the waiting line segment in each direction of travel. The derived timing can indicate the derived average transit time in 5-minute increments. That is, speeds collected in 5-minute increments may have missing values. However, since the transit times of vehicles before and after the derived timing are processed again using an arithmetic averaging method, missing values ​​can be corrected. Furthermore, since recent past traffic information and recent future traffic information are integrated equally, irregular movements in the transit times of vehicles attributed to turning waits or signal waits at intersections can be smoothed out.

[0078] Therefore, since the average transit time is calculated by smoothing and using the transit time of vehicles in the waiting line section, information such as signal waiting of vehicles at the intersection and turning waiting of vehicles may have been incorporated into the average transit time.

[0079] The steps of generating waiting line segment information and environmental information can be further performed. The waiting line segment information includes the length of the waiting line segment for each direction of vehicle travel, road gradient, road type and number of lanes, and signal type of the intersection. The environmental information includes information about the day of the week, time and administrative district.

[0080] The step of inputting the average transit time into a predictive model trained to predict future transit times can be performed. The trained predictive model can be a model trained using deep learning based on artificial neural networks. If both waiting line segment information and environmental information are used in the method for predicting travel time, the average transit time, waiting line segment information, and environmental information can be input into the training model. The average transit time, waiting line segment information, and environmental information can interact with each other before being input into the trained model.

[0081] Furthermore, to generate a trained prediction model, a learning step can be performed by inputting past average transit times into the prediction model. If both waiting line segment information and environmental information are used in the method for predicting travel time, the prediction model can be trained by inputting information obtained through interoperation of the average transit time, waiting line segment information, and environmental information into the prediction model. In this case, the past average transit time can be derived by calculating the transit time corresponding to 10 minutes before and after the derived timing of a vehicle passing through a waiting line segment in each direction of vehicle movement using an arithmetic averaging method. In this case, the derived timing can indicate the derived average transit time in 5-minute increments.

[0082] A predictive model can be trained by receiving average transit times accumulated from the past. These accumulated average transit times can be derived by calculating the transit times of vehicles in each direction of vehicle movement using an arithmetic averaging method, corresponding to the transit times of vehicles that have already passed the waiting line segment 10 minutes before and after the derived timing. The derived timing can be a 5-minute interval indicating the derived average transit time. That is, smoothing transit times using an arithmetic averaging method allows for more accurate predictions of transit times in waiting line segments that already account for irregular signal waiting and vehicle rotation. Specifically, speeds collected in 5-minute intervals may have missing values. However, these missing values ​​can be corrected by reprocessing the vehicle transit times before and after the derived timing using an arithmetic averaging method. Furthermore, since recent past and recent future traffic information are integrated equally, irregular movements in vehicle transit times attributable to vehicle rotation waiting or signal waiting at intersections can be smoothed.

[0083] The steps can be performed to derive the predicted passage time for each waiting line segment and each direction of travel from a trained prediction model. Predicted passage times can be generated by dividing the actual time into 5-minute increments within the next hour. Therefore, the passage time within the waiting line segments of each intersection can be predicted.

[0084] The steps can be performed to derive a route with the minimum sum of predicted passage times from the waiting line segments that vehicles can pass through in order to reach their destination. The travel route with the minimum time can be derived by the sum of the predicted passage times from the waiting line segments at each intersection that the user must travel through to reach their destination.

[0085] Therefore, it can solve the problem of interrupted flow waiting lines with low information collection rates, and can provide users with accurate destination and scheduled arrival time guidance.

[0086] Unless otherwise specifically mentioned, the terms "comprising," "constituting," or "having" above mean that they may include the corresponding components. Therefore, it should be interpreted that the term may further include another component, but does not exclude another component. Unless otherwise defined in this disclosure, all terms used herein (including technical or scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains. Commonly used terms, such as those defined in common dictionaries, should be interpreted as having the same meaning as in the context of the prior art, and should not be construed as having an ideal or overly formal meaning unless expressly defined elsewhere in this specification.

[0087] The above description is merely a description of the technical spirit of this disclosure, and those skilled in the art can change and modify this disclosure in various ways without departing from its essential characteristics. Therefore, the embodiments described in this disclosure should not be construed as limiting the technical spirit of this disclosure, but rather as describing it. The technical spirit of this disclosure is not limited to the embodiments. The scope of protection of this disclosure should be interpreted based on the appended claims, and all technical spirit within the equivalent scope of this disclosure should be interpreted as included within the scope of the rights of this disclosure.

Claims

1. A method for predicting transit time, comprising the following steps: Collect traffic information, including the speed of vehicles passing through a waiting line section set behind an intersection in each of a plurality of directions of travel of the vehicles in the waiting line section; Based on the traffic information and the length of the waiting line section, the transit time of vehicles passing through the waiting line section in each direction of travel is calculated, and the average transit time in each direction of travel is calculated based on the transit time; The average transit time is input into a prediction model trained to predict future transit times; The predicted transit time for each direction of travel is derived from the trained prediction model; as well as Based on the predicted transit time for each direction of travel, the route is displayed on a specific geographic area of ​​the map.

2. The method according to claim 1, wherein: The average passage time is derived by calculating the passage time for each direction of travel using an arithmetic averaging method, corresponding to 10 minutes before and after the derived timing of the vehicle that has already passed the waiting line section. The export timing indicator exports the average transit time in 5-minute increments.

3. The method according to claim 1, wherein: The prediction model is trained by receiving the average elapsed time accumulated from the past based on the time series. The average passage time accumulated from the past is derived by calculating the passage time for each direction of travel using an arithmetic averaging method, corresponding to the 10 minutes before and after the departure time of the vehicle that has already passed through the waiting line section. The export timing indicator exports the average transit time in 5-minute increments, and The predicted transit time is derived for each direction of travel by dividing the actual time into 5-minute increments for the next hour.

4. The method according to claim 1, wherein, The waiting line segments are set differently for each direction of the vehicle's movement.

5. The method according to claim 1, wherein, The trained prediction model is a model trained using deep learning based on artificial neural networks.

6. The method of claim 1, further comprising the following steps: Generate waiting line segment information including the length of the waiting line segment in each direction of travel, road gradient, road type and number of lanes, signal type of the intersection, and environmental information including the day of the week, time, and administrative district. The average transit time, the waiting line segment information, and the environmental information are input into the trained prediction model.

7. An apparatus for predicting transit time, comprising: A collection circuit is configured to collect traffic information, including the speed of vehicles passing through a waiting line section located behind an intersection in each of a plurality of directions of travel of the vehicles in the waiting line section. An arithmetic circuit is configured to calculate, based on the traffic information and the length of the waiting line section, the transit time of vehicles passing through the waiting line section in each direction of travel, and to calculate, based on the transit time, the average transit time in each direction of travel; The prediction circuit includes a prediction model trained to predict future transit times and is configured to input the average transit time into the trained prediction model and derive the predicted transit time for each direction of travel. as well as The display screen shows the route on a specific geographic area of ​​the map based on the predicted transit time for each direction of travel.

8. The apparatus according to claim 7, wherein: The arithmetic circuit is configured to derive the average passage time by calculating the passage time corresponding to 10 minutes before and after the derived timing of the vehicle that has passed through the waiting line section in each direction of advance using an arithmetic averaging method. The export timing indicator exports the average transit time in 5-minute increments.

9. The apparatus according to claim 7, further comprising: The management circuit is configured to generate waiting line segment information including the length of the waiting line segment in each direction of travel, road gradient, road type and number of lanes, signal type of the intersection, and environmental information including information about the day of the week, time and administrative district.

10. The apparatus according to claim 7, further comprising: The transmission circuit is configured to transmit the prediction over time to the vehicle navigation management server.

11. The apparatus of claim 7, further comprising: The learning circuit is configured to train the prediction model by inputting the prediction model with the average elapsed time accumulated from the past based on the time series. The average passage time accumulated from the past is derived by calculating the passage time for each direction of travel using an arithmetic averaging method, corresponding to the 10 minutes before and after the derived timing of the vehicle that has already passed the waiting line section. The export timing indicator exports the average transit time in 5-minute increments.

12. A method for predicting travel time, comprising the following steps: Collect traffic information, including the speed of vehicles passing through a waiting line section set behind an intersection in each of a plurality of directions of travel of the vehicles in the waiting line section; Based on the traffic information and the length of the waiting line segment, the transit time of vehicles passing through the waiting line segment in each direction of travel is calculated for each waiting line segment, and the average transit time in each direction of travel is calculated based on the transit time. The average transit time is input into a prediction model trained to predict future transit times; The predicted transit time for each direction of travel is derived from the trained prediction model; Derive a route with the minimum sum of predicted passage times from the waiting line segments that the vehicle passes through in order to reach its destination. as well as The route is displayed on a specific geographic area of ​​the map based on the route having the minimum sum of predicted transit times.

13. The method according to claim 12, wherein: The average passage time is derived by calculating the passage time for each direction of travel using an arithmetic averaging method, corresponding to 10 minutes before and after the derived timing of the vehicle that has already passed the waiting line section. The export timing indicator exports the average transit time in 5-minute increments.

14. The method of claim 12, further comprising the following steps: Generate waiting line segment information including the length of the waiting line segment in each direction of travel, road gradient, road type and number of lanes, signal type of the intersection, and environmental information including day of the week, time, and administrative district. The trained prediction model is a model trained using deep learning based on artificial neural networks, and The average transit time, the waiting line segment information, and the environmental information are input into the trained prediction model.

15. The method according to claim 12, wherein: The prediction model is trained by receiving the average elapsed time accumulated from the past based on the time series. The average passage time accumulated from the past is derived by calculating the passage time for each direction of travel using an arithmetic averaging method, corresponding to the 10 minutes before and after the departure time of the vehicle that has already passed through the waiting line section. The export timing indicator exports the average transit time in 5-minute increments, and The predicted transit time is derived for each direction of travel by dividing the actual time into 5-minute increments for the next hour.

Images