Information determination model training method and information determination method

Abstract

The invention provides a training method of an information determination model and an information determination method, and relates to the technical field of artificial intelligence, in particular to the technical fields of map navigation, automatic driving, intelligent traffic and the like. The method comprises the steps that based on vehicle track data associated with a target signal lamp, the total signal period of the target signal lamp in different time periods is determined, and the total signal period comprises a green light period and a red light period which are adjacent; merging the vehicle trajectory data with the same total signal period to obtain target trajectory data; and performing model training based on pre-collected period truth value data of the target signal lamp and the target trajectory data to obtain an information determination model for determining period information of the signal lamp. According to the method, the problem of sparse training data during model training is solved, and the accuracy of model training is improved.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, specifically to technical fields such as map navigation, autonomous driving, and intelligent transportation, and in particular to a training method for an information determination model and an information determination method. Background Art

[0002] At present, traffic lights (also known as signal lights) are an important infrastructure for traffic management departments to control vehicle traffic order and adjust road traffic flow. The cycle duration and light changing time of traffic lights are one of the important information that dynamic traffic service providers need to provide users with corresponding traffic services for different road traffic scenarios. Summary of the invention

[0003] The present disclosure provides a training method for an information determination model and an information determination method.

[0004] According to a first aspect of the present disclosure, a method for training an information determination model is provided, comprising: determining the total signal cycle of the target signal light in different time periods based on vehicle trajectory data associated with the target signal light, wherein the total signal cycle includes adjacent green light cycles and red light cycles; merging vehicle trajectory data with the same total signal cycle to obtain target trajectory data; and performing model training based on pre-collected target signal light cycle true value data and target trajectory data to obtain an information determination model for determining the cycle information of the signal light.

[0005] According to a second aspect of the present disclosure, there is provided an information determination method, comprising: acquiring vehicle trajectory data associated with a signal light to be determined; inputting the vehicle trajectory data into an information determination model, and outputting a red light cycle and a green light cycle of the signal light to be determined, wherein the information determination model is trained using the method described in any implementation manner in the first aspect.

[0006] According to a third aspect of the present disclosure, a training device for an information determination model is provided, comprising: a total period determination module, configured to determine the total signal period of a target signal light in different time periods based on vehicle trajectory data associated with the target signal light, wherein the total signal period includes adjacent green light periods and red light periods; a data merging module, configured to merge vehicle trajectory data with the same total signal period to obtain target trajectory data; and a model training module, configured to perform model training based on pre-collected period true value data of the target signal light and the target trajectory data to obtain an information determination model for determining the period information of the signal light.

[0007] According to a fourth aspect of the present disclosure, there is provided an information determination device, comprising: a data acquisition module, configured to acquire vehicle trajectory data associated with a signal light to be determined; a cycle determination module, configured to input the vehicle trajectory data into an information determination model, and output a red light cycle and a green light cycle of the signal light to be determined, wherein the information determination model is trained using the method described in any implementation manner in the first aspect.

[0008] According to a fifth aspect of the present disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method described in any implementation manner of the first aspect or the second aspect.

[0009] According to a sixth aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, where the computer instructions are used to cause a computer to execute the method described in any implementation manner of the first aspect or the second aspect.

[0010] According to a seventh aspect of the present disclosure, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the computer program implements the method described in any implementation manner of the first aspect or the second aspect.

[0011] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure.

[0013] Figure 1 is an exemplary system architecture diagram in which the present disclosure may be applied;

[0014] Figure 2 is a flow chart of an embodiment of a method for training a model determined according to the information disclosed herein;

[0015] Figure 3 is a flowchart of another embodiment of a method for training a model according to the information disclosed herein;

[0016] Figure 4 is a flowchart of another embodiment of a method for training a model determined according to the information disclosed herein;

[0017] Figure 5 is a flow chart of an embodiment of an information determination method according to the present disclosure;

[0018] Figure 6 is a schematic diagram of a structure of an embodiment of a training device for determining a model according to the information disclosed herein;

[0019] Figure 7 is a structural schematic diagram of an embodiment of an information determination device according to the present disclosure;

[0020] Figure 8 It is a block diagram of an electronic device used to implement the training method of the information determination model or the information determination method of the embodiment of the present disclosure. DETAILED DESCRIPTION

[0021] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0022] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present disclosure may be combined with each other. The present disclosure will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0023] Figure 1 An exemplary system architecture 100 is shown to which embodiments of the information determination model training method, information determination method or information determination model training device, and information determination device of the present disclosure can be applied.

[0024] like Figure 1 As shown, the system architecture 100 may include terminal devices 101, 102, 103, 104, a network 105, and a server 106. The network 105 is used to provide a medium for communication links between the terminal devices 101, 102, 103, 104 and the server 106. The network 105 may include various connection types, such as wired, wireless communication links, or optical fiber cables, etc.

[0025] The user can use the terminal devices 101, 102, 103, 104 to interact with the server 106 via the network 105 to receive or send information, etc. Various client applications can be installed on the terminal devices 101, 102, 103, 104.

[0026] Terminal devices 101, 102, 103, 104 may be hardware or software. When terminal devices 101, 102, 103, 104 are hardware, they may be various electronic devices, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, etc. When terminal devices 101, 102, 103, 104 are software, they may be installed in the above electronic devices. They may be implemented as multiple software or software modules, or as a single software or software module. No specific limitation is made here.

[0027] The server 106 can provide various services. For example, the server 106 can analyze and process the vehicle trajectory data obtained from the terminal devices 101, 102, 103, and 104, and generate a processing result (eg, an information determination model).

[0028] It should be noted that the server 106 can be hardware or software. When the server 106 is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or it can be implemented as a single server. When the server 106 is software, it can be implemented as multiple software or software modules (for example, for providing distributed services), or it can be implemented as a single software or software module. No specific limitation is made here.

[0029] It should be noted that the information determination model training method and information determination method provided in the embodiments of the present disclosure are generally executed by the server 106 , and accordingly, the information determination model training device and information determination device are generally arranged in the server 106 .

[0030] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0031] Continue to refer Figure 2 , which shows a process 200 of an embodiment of a training method for an information determination model according to the present disclosure. The training method for the information determination model comprises the following steps:

[0032] Step 201 : determining the total signal cycle of the target signal light in different time periods based on the vehicle trajectory data associated with the target signal light.

[0033] In this embodiment, the execution subject of the training method of the information determination model (for example Figure 1The server 105 shown in the figure will determine the total signal cycle of the target signal light in different time periods based on the vehicle trajectory data associated with the target signal light, wherein the total signal cycle includes adjacent green light cycles and red light cycles. Here, the target signal light is the signal light for which cycle information needs to be predicted. The vehicle trajectory data associated with the target signal light is the trajectory data of the vehicle in front of the target signal light, which generally includes information such as longitude and latitude, timestamp, and vehicle speed. The vehicle trajectory data can be offline type trajectory data or online type trajectory data, as long as it is trajectory data in front of the target signal light. This embodiment does not limit the trajectory type and acquisition method of the vehicle trajectory data. At the same time, the vehicle trajectory data can be data in front of the target signal light within a period of time, or it can be data collected by real-time statistics at the current moment. When the vehicle trajectory data is offline data, the offline data can be locally stored data, or it can be reliable true value data provided by a third party, etc. After obtaining the vehicle trajectory data, it will also be cleaned to filter abnormal data.

[0034] The total signal cycle includes adjacent green light cycles and red light cycles. The green light cycle includes the green light duration and the yellow light duration, and the red light cycle is the red light duration. The total signal cycle can be the total duration from the red light switching to the green light state to the next red light switching to the green light state, or the total duration from the green light switching to the yellow light state to the next green light switching to the yellow light state, or the total duration from the yellow light switching to the red light state to the next yellow light switching to the red light state.

[0035] Since the total signal cycle in different time periods of a day is different, for example, the regular information and total signal cycle of traffic lights in the morning and evening peak time periods (such as 17:00-18:00) and other time periods (such as 21:00-22:00) are generally different. For another example, the regular information and total signal cycle of traffic lights in the late night and early morning time periods (such as 00:00-01:00) and the noon time period (such as 11:00-12:00) are generally different. Therefore, in this embodiment, a day will be divided into multiple time periods based on the statistical analysis of historical data, and the total signal cycle of the traffic lights in each time period is the same.

[0036] The above-mentioned execution entity will analyze and count the vehicle start and stop characteristics, timestamps and other information contained in the vehicle trajectory data, so as to determine the total signal cycle of the target traffic light in different time periods.

[0037] Step 202: merge the vehicle trajectory data with the same total signal period to obtain target trajectory data.

[0038] In this embodiment, since the total signal cycle of the target traffic light in different time periods of each day has been calculated, and the total signal cycle of different time periods may be the same, for example, the total signal cycle of the time period 11:00-12:00 and the time period 21:00-22:00 may be the same, the above-mentioned execution subject will merge the vehicle trajectory data with the same total signal cycle in multiple days to obtain the merged target trajectory data. Since there are multiple values ​​of the total signal cycle, for each total signal cycle, the vehicle trajectory data corresponding to the total signal cycle will be merged. Therefore, the target trajectory data includes multiple merged vehicle trajectory data, and each merged vehicle trajectory data corresponds to a total signal cycle value.

[0039] Step 203 , model training is performed based on the pre-collected target traffic light cycle true value data and target trajectory data to obtain an information determination model for determining the cycle information of the traffic light.

[0040] In this embodiment, the above-mentioned execution entity will collect the cycle true value data corresponding to the target signal light, that is, the real signal cycle information of the target signal light in different time periods, including the real red light cycle, the real green light cycle (including green light cycle and yellow light cycle), so as to obtain the red light cycle true value and green light cycle true value of the target signal light.

[0041] Then, the initial model is trained according to the collected true value of the red light cycle, the true value of the green light cycle and the target trajectory data, so as to obtain a trained information determination model, which is used to predict the cycle information of the traffic light. Here, the initial model can be a DNN (Deep Neural Network) regression model, that is, the collected true value data of the cycle and the target trajectory data are input into the DNN regression model, so that the DNN regression model learns the position of the red-green cycle dividing point, and then obtains the corresponding red light cycle and green light cycle based on the position information of the dividing point and the total cycle information, so that the trained information determination model can be used to predict the red light cycle and green light cycle of the traffic light. For example, assuming that the total signal cycle is 180 seconds, the position of the red-green cycle dividing point is at the 80th second of a complete signal cycle, so that it can be determined that the red light cycle is 80 seconds and the green light cycle is 100 seconds.

[0042] Optionally, technologies such as Attention (attention mechanism) and Dropout (referring to ignoring units in the training phase for a group of randomly selected neurons) can be used to improve the expressiveness and generalization ability of the information determination model. Since Attention and Dropout are existing technologies, they will not be described in detail here.

[0043] The training method of the information determination model provided in the embodiment of the present disclosure first determines the total signal period of the target signal light in different time periods based on the vehicle trajectory data associated with the target signal light; then, the vehicle trajectory data with the same total signal period are merged to obtain the target trajectory data; finally, the model is trained based on the pre-collected target signal light cycle true value data and the target trajectory data to obtain an information determination model for determining the cycle information of the signal light. The training method of the information determination model in this embodiment superimposes multiple days of data by merging trajectory data with the same total signal period, thereby training the model under large-scale data, thereby solving the problem of sparse training data during model training and improving the accuracy of model training.

[0044] In addition, in the technical solutions involved in this disclosure, the acquisition, storage, use, processing, transportation, provision and disclosure of user personal information (such as the vehicle trajectory data involved in the subsequent disclosure) are in compliance with the relevant laws and regulations and do not violate public order and good morals.

[0045] Continue to refer Figure 3 , Figure 3 A process 300 of another embodiment of a method for training an information determination model according to the present disclosure is shown. The method for training an information determination model comprises the following steps:

[0046] Step 301: extracting vehicle driving characteristic information from vehicle trajectory data associated with a target traffic light.

[0047] In this embodiment, the execution subject of the training method of the information determination model (for example Figure 1 The server 105 shown in the figure can extract the driving characteristic information of the vehicle from the vehicle trajectory data. The driving characteristic information includes the driving state information and timestamp of the vehicle. The driving state information includes stop or start.

[0048] In some optional implementations of this embodiment, step 301 includes: determining a target lane corresponding to a target signal light; determining lane trajectory data corresponding to the target lane from the full trajectory data of the intersection where the target signal light is located; determining vehicle trajectory data associated with the target signal light from the lane trajectory data, and extracting driving characteristic information of the vehicle from the vehicle trajectory data.

[0049] In this implementation, the above-mentioned execution subject will first determine the lane corresponding to the target signal light, that is, the target lane. Generally speaking, the signal light corresponds to the lane, for example, the left-turn lane corresponds to the left-turn signal light, the straight lane corresponds to the straight-through signal light, and the right-turn lane corresponds to the right-turn signal light (some right-turn lanes do not have signal lights), so the above-mentioned execution subject will first determine the lane corresponding to the target signal light. Then, the lane trajectory data corresponding to the target lane is determined from the full trajectory data of the intersection where the target signal light is located. After that, the vehicle trajectory data associated with the target signal light is determined from the lane trajectory data, for example, the data with a distance from the target signal light within a preset range is determined as the vehicle trajectory data associated with the target signal light. Finally, the driving characteristic information of the vehicle is extracted from the vehicle trajectory data. Thus, the vehicle trajectory data associated with the target signal light is determined based on the lane information, and the correlation between the vehicle trajectory data and the target signal light is ensured.

[0050] In some optional implementations of this embodiment, the vehicle trajectory data associated with the target signal light includes: vehicle trajectory data whose distance to the target signal light is within a preset range and whose driving direction matches the target signal light.

[0051] In this implementation, the matching of the vehicle's driving direction and the target signal light refers to the target signal light being used to indicate the vehicle's driving direction. For example, if the vehicle is traveling south, the target signal light is located to the south of the vehicle, that is, the target signal light is used to indicate the vehicle's southward traffic signal. The preset range of the distance from the target signal light within the preset range can be set according to actual conditions, for example, it is set to 100 meters, that is, the vehicle trajectory data within a distance of less than 100 meters from the target signal light is obtained. Therefore, based on the above conditions, the correlation between the obtained vehicle trajectory data and the target signal light is further guaranteed.

[0052] Step 302: Determine the total signal cycle of the target signal light in different time periods based on the driving characteristic information.

[0053] In this embodiment, for a certain time period, the above-mentioned execution subject will determine the total signal cycle of the target signal light based on the driving characteristic information of the vehicle extracted within the time period, wherein the total signal cycle includes adjacent green light cycles and red light cycles, and the green light cycle includes the green light duration and the yellow light duration. That is, by statistically analyzing the characteristic information such as the start-stop characteristics and timestamps of all vehicles within a certain time period, the total signal cycle of the target signal light can be obtained. Since the vehicle waiting in front of the signal light will generally start at the first time when the signal light switches from red to green, the red-to-green timestamp obtained based on the vehicle trajectory data is relatively accurate. Therefore, the above-mentioned execution subject can determine the total signal cycle based on the timestamp of the moment when two adjacent red lights switch to green. Thereby, the total cycle information of the signal light is determined based on the driving characteristics of the vehicle, and the accuracy of the total cycle information is improved.

[0054] In some optional implementations of this embodiment, the driving characteristic information includes time information of the starting trajectory point; and step 302 includes: inputting the time information of the starting trajectory point collected on the same day into a pre-trained cycle mining model, and outputting the total signal cycle of the target traffic light in different time periods.

[0055] In this implementation, the execution subject will use the regularity of the vehicle startup information to first obtain the time information of the vehicle's startup trajectory point, and then input the time information of all the startup trajectory points collected on the same day into the pre-trained cycle mining model, so as to output the total signal cycle of the target signal light in different time periods. The cycle mining model is used to mine the total cycle information of the signal light, and its specific training process can be implemented using existing technology, which will not be repeated here. Therefore, the total cycle information of the signal light is determined based on the vehicle's startup time information and the cycle mining model, further improving the accuracy of the total cycle information.

[0056] Step 303 , mapping the vehicle trajectory data with the same signal total period within the same day into a total period interval to obtain mapped trajectory data.

[0057] In this embodiment, the execution subject maps the vehicle trajectory data with the same total cycle in a single day into an interval, which is an interval with 0 and the total cycle value as endpoints, thereby obtaining the mapped trajectory data. That is, in this example, the trajectory data of different time periods with the same total cycle in a single day are first mapped into the same interval.

[0058] Step 304 , merging the mapped trajectory data of different days to obtain target trajectory data.

[0059] In this embodiment, the mapped trajectory data obtained through the above steps contains all trajectory data with the same total period in a single day, and then the above execution entity will merge the mapped trajectory data of multiple days to obtain the merged target trajectory data. For each signal total period, the vehicle trajectory data corresponding to the signal total period will be merged. Therefore, the target trajectory data includes multiple merged vehicle trajectory data, and each merged vehicle trajectory data corresponds to a signal total period value. This solves the problem of sparse trajectories on a certain day or a certain time period.

[0060] Step 305 , taking the target trajectory data as input and the true value of the red light cycle and the true value of the green light cycle of the target traffic light as output, training the initial model to obtain an information determination model for determining the cycle information of the traffic light.

[0061] In this embodiment, the target signal light cycle true value data includes the red light cycle true value and the green light cycle true value. The cycle true value data can be collected by a true value vehicle or obtained by other means. This embodiment does not specifically limit this. The above-mentioned execution subject will take the target trajectory data as input and the red light cycle true value and green light cycle true value of the target signal light as output, and perform supervised training on the initial model to obtain an information determination model for predicting the cycle information of the signal light. The initial model here can be a DNN regression model. Therefore, model training is performed based on the collected true value data and the trajectory data obtained by the above steps to obtain a trained information determination model.

[0062] from Figure 3 It can be seen that Figure 2 Compared with the corresponding embodiment, the information determination model training method in this embodiment highlights the steps of mapping and merging trajectory data with the same total period. By first mapping the trajectory data with the same total period in a single day, and then merging the mapped trajectory data of multiple days, the problem of sparse trajectories on a certain day or in a certain time period is solved.

[0063] Continue to refer Figure 4 , Figure 4 A process 400 of another embodiment of a method for training an information determination model according to the present disclosure is shown. The method for training an information determination model comprises the following steps:

[0064] Step 401 : determining the total signal cycle of the target signal light in different time periods based on the vehicle trajectory data associated with the target signal light.

[0065] In this embodiment, the execution subject of the training method of the information determination model (for example Figure 1 The server 105 shown in the figure will determine the total signal cycle of the target signal light in different time periods based on the vehicle trajectory data associated with the target signal light, wherein the total signal cycle includes adjacent green light cycles and red light cycles. Step 401 is basically the same as step 201 of the above embodiment, and the specific implementation method can refer to the above description of step 201, which will not be repeated here.

[0066] Step 402 : generating relative time information of the trajectory data to be processed based on the timestamps and time periods of the trajectory data to be processed with the same total signal period in the same day.

[0067] In this embodiment, the trajectory data to be processed will include all vehicle trajectory data with the same total signal cycle in the same day. For each trajectory data to be processed, the above-mentioned execution subject will first obtain the time information of the trajectory point on the day based on the timestamp of the trajectory data to be processed. Since the timestamp refers to the total number of seconds from 00:00:00 Greenwich time on January 1, 1960 (08:00:00 Beijing time on January 1, 1960) to the present, the timestamp is first converted to the time of the day when the trajectory data to be processed is collected to obtain the first time information, which is the total number of seconds from 00:00:00 on the day to the collection time of the trajectory data to be processed. Then, the time period of the trajectory data to be processed is determined, so as to further convert the obtained first time information into relative time information with 00:00:00 on the day of collection as the reference point. For example, assuming that the time period of the trajectory data to be processed is 02:00-03:00, the time period endpoint 02:00 is 2 hours away from 00:00, and there are 3600 seconds per hour, then the relative time information of the trajectory data to be processed can be obtained by subtracting 2*3600 from the first time information.

[0068] Step 403, performing a modulo operation on the total period of the signal using the relative time information.

[0069] In this embodiment, the execution subject will perform a modulo operation on the total period of the signal through the relative time information. The modulo operation is a mathematical calculation method used to find the remainder of a number divided by another number. Here, the remainder of the relative time information divided by the total period is found.

[0070] Step 404 , using the result of the remainder operation as the interval time information of the trajectory data to be processed within the total period interval, to obtain the mapped trajectory data.

[0071] In this embodiment, the calculation result of the remainder operation is used as the interval time information of the trajectory data to be processed within the total cycle interval, so as to obtain the mapped trajectory data.

[0072] For example, assuming that the total cycle of the current signal is 180 seconds, then the total cycle interval is [0,180). After calculation, the relative time information of the trajectory data to be processed is 2001. The relative time information (2001) modulo the total cycle of the signal (180) is used to obtain the modulo result of 21, which is the interval time information of the trajectory data to be processed within the total cycle interval. According to the above steps, each trajectory data to be processed can be mapped to the total cycle interval.

[0073] This enables the trajectory data with the same total period within a single day to be mapped into the same interval.

[0074] Step 405 , performing debiasing processing on the mapped trajectory data to obtain debiased trajectory data.

[0075] In this embodiment, the above steps have mapped the trajectory data with the same total signal period within the same day into a total period interval. The mapped data may be relatively concentrated in a certain segment and relatively sparse in another segment within the total period interval. Therefore, the above execution entity will use the clustering algorithm to calculate the earliest starting point offset in the segment of data, and debias the mapped trajectory data according to the offset.

[0076] In some optional implementations of this embodiment, step 405 includes: determining the first started target trajectory data from the mapped trajectory data; using the interval time information of the target trajectory data as an offset; and debiasing the mapped trajectory data based on the offset to obtain debiased trajectory data.

[0077] In this implementation, the execution subject will determine the first started target trajectory data from the mapped trajectory data, and use the interval time information of the target trajectory data in the total cycle interval as the offset, and debias the mapped trajectory data based on the offset, that is, subtract the offset from the interval time information of the mapped trajectory data in the total cycle, and use the obtained value as the new interval time information of the mapped trajectory data, thereby obtaining the debiased trajectory data. By debiasing the trajectory data mapped for a single day, trajectory data with the same total cycle on different days can be merged later.

[0078] Step 406 , merging the debiased trajectory data of different days to obtain target trajectory data.

[0079] In this embodiment, the trajectory data mapped within a single day has been debiased through the above steps, and then the above execution body will merge the debiased trajectory data of multiple days to obtain merged target trajectory data.

[0080] Since there are multiple values ​​of the total signal period, for each total signal period, the vehicle trajectory data of a single day corresponding to the total signal period will be mapped and debiased, and then the debiased trajectory data of multiple days will be merged. That is, the trajectory data corresponding to each total signal period will be processed in the manner described in steps 402-406. Therefore, the target trajectory data obtained includes multiple merged vehicle trajectory data, and each merged vehicle trajectory data corresponds to a total signal period value. Thus, through the above steps, the data with the same total period of multiple days are superimposed, the number of trajectories with the same total period is enriched, and the problem of sparse training data is solved.

[0081] Step 407 , taking the target trajectory data as input and the true value of the red light cycle and the true value of the green light cycle of the target traffic light as output, training the initial model to obtain an information determination model for determining the cycle information of the traffic light.

[0082] Step 407 is basically the same as step 305 of the aforementioned embodiment. The specific implementation method can refer to the aforementioned description of step 305, which will not be repeated here.

[0083] from Figure 4 It can be seen that Figure 3 Compared with the corresponding embodiments, the information determination model training method in this embodiment highlights the steps of mapping and debiasing the trajectory data with the same total period of a single day, and merges the processed data of different days, thereby solving the problem that the single-day data is sparse and the trajectory data with the same total period of multiple days cannot be directly merged.

[0084] Continue to refer Figure 5 , which shows a process 500 of an embodiment of an information determination method according to the present disclosure. The information determination method comprises the following steps:

[0085] Step 501, obtaining vehicle trajectory data associated with the signal light to be determined.

[0086] In this embodiment, the execution subject of the information determination method (for example Figure 1 The server 105 shown in the figure will obtain the vehicle trajectory data associated with the signal light to be determined. Here, the signal light to be determined is the signal light whose periodic information needs to be determined, and the vehicle trajectory data associated with the signal light to be determined is the trajectory data of the vehicle before the signal light to be determined, which can be offline trajectory data or online trajectory data. At the same time, the vehicle trajectory data can be data before the target signal light within a period of time, or data collected by real-time statistics at the current moment.

[0087] Step 502: input the vehicle trajectory data into the information determination model, and output the red light cycle and green light cycle of the signal light to be determined.

[0088] In this embodiment, the above-mentioned execution subject will input the acquired vehicle trajectory data into the information determination model, thereby outputting the red light cycle and green light cycle of the signal light to be determined, where the red light cycle includes the red light duration, and the green light cycle includes the green light duration and the yellow light duration. The information determination model can be trained using the method described in the above embodiment. For all intersections, the trained information determination model is used to predict the red light cycle and green light cycle of the signal light, and is transmitted to the online use in the form of offline data.

[0089] The information determination method provided in the embodiment of the present disclosure first obtains vehicle trajectory data associated with the signal light to be determined; then, the vehicle trajectory data is input into the information determination model, and the red light cycle and green light cycle of the signal light to be determined are output. The information determination method in this embodiment determines the cycle information of the signal light through a pre-trained information determination model, thereby improving the efficiency of cycle information prediction and the accuracy of prediction.

[0090] Further references Figure 6 As an implementation of the methods shown in the above figures, the present disclosure provides an embodiment of a training device for an information determination model. Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0091] like Figure 6 As shown, the training device 600 of the information determination model of this embodiment includes: a total cycle determination module 601, a data merging module 602 and a model training module 603. Among them, the total cycle determination module 601 is configured to determine the total signal cycle of the target signal light in different time periods based on the vehicle trajectory data associated with the target signal light, wherein the total signal cycle includes adjacent green light cycles and red light cycles; the data merging module 602 is configured to merge the vehicle trajectory data with the same total signal cycle to obtain the target trajectory data; the model training module 603 is configured to perform model training based on the pre-collected target signal light cycle true value data and the target trajectory data to obtain the information determination model for determining the cycle information of the signal light.

[0092] In the training device 600 for determining the information model, the specific processing of the total cycle determination module 601, the data merging module 602 and the model training module 603 and the technical effects thereof can be referred to in Figure 2 The relevant descriptions of steps 201 - 203 in the corresponding embodiment are not repeated here.

[0093] In some optional implementations of this embodiment, the data merging module includes: a mapping submodule, configured to map vehicle trajectory data with the same signal total period within the same day into a total period interval to obtain mapped trajectory data; and a merging submodule, configured to merge the mapped trajectory data of different days to obtain target trajectory data.

[0094] In some optional implementations of this embodiment, the merging submodule includes: a debiasing unit, configured to debias the mapped trajectory data to obtain debiased trajectory data; and a merging unit, configured to merge the debiased trajectory data of different days to obtain target trajectory data.

[0095] In some optional implementations of this embodiment, the mapping submodule is further configured to: generate relative time information of the trajectory data to be processed based on the timestamp and time period of the trajectory data to be processed with the same total signal period within the same day; perform a modulo operation on the total signal period through the relative time information; and use the result of the modulo operation as the interval time information of the trajectory data to be processed within the total period interval to obtain the mapped trajectory data.

[0096] In some optional implementations of this embodiment, the debiasing unit is further configured to: determine the first started target trajectory data from the mapped trajectory data; use the interval time information of the target trajectory data as an offset; and debias the mapped trajectory data based on the offset to obtain the debiased trajectory data.

[0097] In some optional implementations of this embodiment, the cycle true value data includes the red light cycle true value and the green light cycle true value; and the model training module is further configured to: take the target trajectory data as input, and the red light cycle true value and the green light cycle true value as output, train the initial model, and obtain an information determination model for determining the cycle information of the traffic light.

[0098] In some optional implementations of this embodiment, the total period determination module includes: an extraction submodule, configured to extract the vehicle's driving characteristic information from the vehicle trajectory data associated with the target signal light; and a determination submodule, configured to determine the total signal period of the target signal light in different time periods based on the driving characteristic information.

[0099] In some optional implementations of this embodiment, the extraction submodule is further configured to: determine the target lane corresponding to the target signal light; determine the lane trajectory data corresponding to the target lane from the full trajectory data of the intersection where the target signal light is located; determine the vehicle trajectory data associated with the target signal light from the lane trajectory data, and extract the vehicle's driving characteristic information from the vehicle trajectory data.

[0100] In some optional implementations of this embodiment, the vehicle trajectory data associated with the target signal light includes: vehicle trajectory data whose distance to the target signal light is within a preset range and whose driving direction matches the target signal light.

[0101] In some optional implementations of this embodiment, the driving characteristic information includes the time information of the starting trajectory point; and the determination submodule is further configured to: input the time information of the starting trajectory point collected on the same day into a pre-trained cycle mining model, and output the total signal cycle of the target traffic light in different time periods.

[0102] Further references Figure 7As an implementation of the methods shown in the above figures, the present disclosure provides an embodiment of an information determination device, and the device embodiment is Figure 5 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0103] like Figure 7 As shown, the information determination device 700 of this embodiment includes: a data acquisition module 701 and a cycle determination module 702. The data acquisition module 701 is configured to acquire vehicle trajectory data associated with the signal light to be determined; the cycle determination module 702 is configured to input the vehicle trajectory data into the information determination model, and output the red light cycle and green light cycle of the signal light to be determined, wherein the information determination model is trained by the method described in the above embodiment.

[0104] In this embodiment, in the information determination device 700, the specific processing of the data acquisition module 701 and the period determination module 702 and the technical effects thereof can be referred to in Figure 5 The relevant descriptions of steps 501-502 in the corresponding embodiment are not repeated here.

[0105] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

[0106] Figure 8 A schematic block diagram of an example electronic device 800 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0107] like Figure 8 As shown, the device 800 includes a computing unit 801, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 802 or a computer program loaded from a storage unit 808 into a random access memory (RAM) 803. In the RAM 803, various programs and data required for the operation of the device 800 can also be stored. The computing unit 801, the ROM 802, and the RAM 803 are connected to each other via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0108] Multiple components in the device 800 are connected to the I / O interface 805, including: an input unit 806, such as a keyboard, a mouse, etc.; an output unit 807, such as various types of displays, speakers, etc.; a storage unit 808, such as a disk, an optical disk, etc.; and a communication unit 809, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 809 allows the device 800 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0109] The computing unit 801 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 801 performs the various methods and processes described above, such as the training method of the information determination model or the information determination method. For example, in some embodiments, the training method of the information determination model or the information determination method may be implemented as a computer software program, which is tangibly included in a machine-readable medium, such as a storage unit 808. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 800 via the ROM 802 and / or the communication unit 809. When the computer program is loaded into the RAM 803 and executed by the computing unit 801, one or more steps of the training method of the information determination model or the information determination method described above may be executed. Alternatively, in other embodiments, the computing unit 801 may be configured to execute the information determination model training method or the information determination method in any other suitable manner (eg, by means of firmware).

[0110] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0111] Program codes for implementing the methods of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general purpose computer, a special purpose computer, or other programmable information determination model training device, so that when the program codes are executed by the processor or controller, the functions / operations specified in the flow charts and / or block diagrams are implemented. The program codes may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0112] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0113] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0114] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0115] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0116] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0117] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A training method for an information determination model, comprising: Determine, based on the vehicle trajectory data associated with the target signal light, a total signal cycle of the target signal light in different time periods, wherein the total signal cycle includes adjacent green light cycles and red light cycles; Merging the vehicle trajectory data with the same total signal period to obtain target trajectory data; Model training is performed based on the pre-collected true value data of the period of the target signal light and the target trajectory data to obtain an information determination model for determining the period information of the signal light.

2. The method according to claim 1, wherein: The vehicle trajectory data with the same total signal period are merged to obtain target trajectory data, including: Mapping the vehicle trajectory data with the same signal total period in the same day into a total period interval to obtain mapped trajectory data; The mapped trajectory data of different days are combined to obtain the target trajectory data.

3. The method according to claim 2, wherein: The step of merging the mapped trajectory data of different days to obtain the target trajectory data includes: Performing debiasing processing on the mapped trajectory data to obtain debiased trajectory data; The debiased trajectory data of different days are combined to obtain the target trajectory data.

4. The method according to claim 2 or 3, wherein: Mapping the vehicle trajectory data with the same signal total cycle in the same day into a total cycle interval to obtain the mapped trajectory data includes: Generate relative time information of the trajectory data to be processed based on the timestamp and time period of the trajectory data to be processed with the same signal total period in the same day; Performing a modulo operation on the total period of the signal using the relative time information; The calculation result of the remainder operation is used as the interval time information of the trajectory data to be processed within the total cycle interval to obtain the mapped trajectory data.

5. The method according to claim 3, wherein: The debiasing process is performed on the mapped trajectory data to obtain the debiased trajectory data, including: Determining the first started target trajectory data from the mapped trajectory data; Using the interval time information of the target trajectory data as an offset; Debiasing is performed on the mapped trajectory data based on the offset to obtain the debiased trajectory data.

6. The method according to any one of claims 1 to 5, wherein: The cycle true value data includes a red light cycle true value and a green light cycle true value; as well as The model training is performed based on the pre-collected period true value data of the target signal light and the target trajectory data to obtain an information determination model for determining the period information of the signal light, including: The target trajectory data is used as input, and the red light cycle true value and the green light cycle true value are used as output, and an initial model is trained to obtain an information determination model for determining the cycle information of a traffic light.

7. The method according to any one of claims 1 to 5, wherein: The determining, based on the vehicle trajectory data associated with the target signal light, the total signal cycle of the target signal light in different time periods comprises: Extracting driving characteristic information of the vehicle from the vehicle trajectory data associated with the target signal light; The total signal cycle of the target signal lamp in different time periods is determined based on the driving characteristic information.

8. The method according to claim 7, wherein: The extracting the driving characteristic information of the vehicle from the vehicle trajectory data associated with the target signal light comprises: Determining a target lane corresponding to the target signal light; Determine lane trajectory data corresponding to the target lane from the full trajectory data of the intersection where the target traffic light is located; Vehicle trajectory data associated with the target signal light is determined from the lane trajectory data, and driving characteristic information of the vehicle is extracted from the vehicle trajectory data.

9. The method according to claim 8, wherein: The vehicle trajectory data associated with the target signal light includes: The vehicle trajectory data includes a distance from the target signal light within a preset range and a driving direction of the vehicle matching the target signal light.

10. The method according to claim 7, wherein: The driving characteristic information includes time information of the starting trajectory point; as well as The determining the total signal cycle of the target signal light in different time periods based on the driving characteristic information includes: The time information of the start trajectory points collected on the same day is input into a pre-trained cycle mining model, and the total signal cycle of the target signal light in different time periods is output.

11. A method for determining information, comprising: Obtaining vehicle trajectory data associated with the signal light to be determined; The vehicle trajectory data is input into an information determination model, and the red light cycle and green light cycle of the signal light to be determined are output, wherein the information determination model is trained using the method described in any one of claims 1-10.

12. A training device for an information determination model, comprising: A total cycle determination module is configured to determine the total signal cycle of the target signal light in different time periods based on the vehicle trajectory data associated with the target signal light, wherein the total signal cycle includes adjacent green light cycles and red light cycles; A data merging module is configured to merge the vehicle trajectory data with the same signal total period to obtain target trajectory data; The model training module is configured to perform model training based on the pre-collected period true value data of the target signal light and the target trajectory data to obtain an information determination model for determining the period information of the signal light.

13. The device according to claim 12, wherein: The data merging module comprises: A mapping submodule is configured to map the vehicle trajectory data with the same signal total period in the same day into a total period interval to obtain mapped trajectory data; The merging submodule is configured to merge the mapped trajectory data of different days to obtain the target trajectory data.

14. The device according to claim 13, wherein: The merging submodule comprises: a debiasing unit, configured to perform debiasing processing on the mapped trajectory data to obtain debiased trajectory data; The merging unit is configured to merge the debiased trajectory data of different days to obtain the target trajectory data.

15. The device according to claim 13 or 14, wherein: The mapping submodule is further configured to: Generate relative time information of the trajectory data to be processed based on the timestamp and time period of the trajectory data to be processed with the same signal total period in the same day; Performing a modulo operation on the total period of the signal using the relative time information; The calculation result of the remainder operation is used as the interval time information of the trajectory data to be processed within the total cycle interval to obtain the mapped trajectory data.

16. The device according to claim 14, wherein: The depolarization unit is further configured to: Determining the first started target trajectory data from the mapped trajectory data; Using the interval time information of the target trajectory data as an offset; Debiasing is performed on the mapped trajectory data based on the offset to obtain the debiased trajectory data.

17. The device according to any one of claims 12 to 16, wherein: The cycle true value data includes a red light cycle true value and a green light cycle true value; and The model training module is further configured to: The target trajectory data is used as input, and the red light cycle true value and the green light cycle true value are used as output, and an initial model is trained to obtain an information determination model for determining the cycle information of a traffic light.

18. The device according to any one of claims 12 to 16, wherein: The total period determination module comprises: An extraction submodule, configured to extract the driving characteristic information of the vehicle from the vehicle trajectory data associated with the target signal light; The determination submodule is configured to determine the total signal cycle of the target signal light in different time periods based on the driving characteristic information.

19. The device according to claim 18, wherein: The extraction submodule is further configured to: Determining a target lane corresponding to the target signal light; Determine lane trajectory data corresponding to the target lane from the full trajectory data of the intersection where the target traffic light is located; Vehicle trajectory data associated with the target signal light is determined from the lane trajectory data, and driving characteristic information of the vehicle is extracted from the vehicle trajectory data.

20. The device according to claim 19, wherein The vehicle trajectory data associated with the target signal light includes: The vehicle trajectory data includes a distance from the target signal light within a preset range and a driving direction of the vehicle matching the target signal light.

21. The device according to claim 18, wherein The driving characteristic information includes time information of the start track point; and The determining submodule is further configured to: The time information of the start trajectory points collected on the same day is input into a pre-trained cycle mining model, and the total signal cycle of the target signal light in different time periods is output.

22. An information determination device, comprising: A data acquisition module, configured to acquire vehicle trajectory data associated with the signal light to be determined; The cycle determination module is configured to input the vehicle trajectory data into an information determination model, and output the red light cycle and green light cycle of the signal light to be determined, wherein the information determination model is trained using the method described in any one of claims 1-10.

23. An electronic device, comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1 to 10 or claim 11.

24. A non-transitory computer-readable storage medium storing computer instructions for causing the computer to execute the method of any one of claims 1-10 or claim 11.

25. A computer program product comprising a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 10 or claim 11.

Images