Vehicle driving scene recognition method, vehicle and storage medium

Abstract

The present application provides a method for identifying a vehicle driving scene, a vehicle and a storage medium, and the method relates to the field of vehicles. The method comprises: obtaining target driving data of a target vehicle; determining whether the target vehicle is in a target state based on the target driving data; wherein the target state is used to indicate a state in which the vehicle cannot drive normally; if the target vehicle is in the target state, obtaining the external environment data of the target vehicle; and determining the target scene type when the target vehicle is in the target state based on the external environment data. The method can improve the accuracy of identifying the driving scene in which the vehicle is located.

Description

Technical Field

[0001] The present application relates to the field of vehicles, and more specifically, to a method for identifying a vehicle driving scene in the field of vehicles, a vehicle, and a storage medium. Background Art

[0002] With the continuous development of the automobile industry and the increasing popularity of off-road sports, more and more vehicles are designed and manufactured to cope with complex off-road terrains, such as sand, mud, swamps and wading. Although these terrain environments can bring adventure and excitement to drivers, they also pose potential dangers, especially the risk of vehicle entrapment. Once the vehicle is trapped in these complex terrains, the driver may face difficulties and find it difficult to escape in time, which may even cause vehicle damage or personal injury.

[0003] Traditional trapped scene recognition mainly relies on the driver's experience and intuition, and lacks automated recognition methods; when the driver misjudges the current scene, it will affect the judgment of the escape strategy and the timeliness of rescue, and may even endanger life safety.

[0004] Therefore, how to improve the accuracy of identifying the driving scene in which the vehicle is located is an urgent problem to be solved. Summary of the invention

[0005] The present application provides a method for identifying a vehicle driving scene, a vehicle and a storage medium, and the method can improve the accuracy of identifying the driving scene in which the vehicle is located.

[0006] In a first aspect, a method for identifying a vehicle driving scene is provided, the method comprising:

[0007] Acquire target driving data of a target vehicle;

[0008] Based on the target driving data, determining whether the target vehicle is in a target state; wherein the target state is used to indicate a state in which the target vehicle cannot drive normally;

[0009] If the target vehicle is in the target state, obtaining the external environment data of the target vehicle;

[0010] Based on the external environment data, the target scene type when the target vehicle is in the target state is determined.

[0011] In the above technical solution, based on the acquired target driving data of the target vehicle, it is determined whether the target vehicle is in a target state where it cannot drive normally; if the target vehicle is identified as being in the target state, the external environment data of the target vehicle is further acquired; and based on the external environment data, the target scene type when the target vehicle is currently in the target state is determined. The vehicle driving data is monitored, and when it is determined that the vehicle is in a state where it cannot drive normally, the specific scene type is further determined in combination with the external environment data to improve the accuracy of identifying the driving scene in which the vehicle is located.

[0012] In conjunction with the first aspect, in some possible implementations, determining whether the target vehicle is in a target state based on the target driving data includes:

[0013] Inputting the target driving data into the first recognition model to obtain a first recognition result; wherein the first recognition result is used to indicate the driving state of the target vehicle;

[0014] Based on the first recognition result, it is determined whether the target vehicle is in a target state.

[0015] In the above technical solution, the acquired target driving data is output to the trained first recognition model, and according to the first recognition result output by the first recognition model, it is determined whether the target vehicle is in the target state. Using the recognition model to determine the target state can adapt to the analysis and processing of large-scale data, improve the calculation efficiency and the accuracy of judging whether the vehicle is in the target state, and reduce the memory usage during the recognition process.

[0016] In combination with the first aspect and the above implementation, in some possible implementations, the first recognition model is obtained through the following training steps:

[0017] Obtain historical driving data of the vehicle;

[0018] Determine the data that meets the preset conditions in the historical driving data as sample driving data; wherein the sample driving data is the driving data when the vehicle is in the target state;

[0019] A first initial model is trained based on the sample driving data to obtain a trained first recognition model.

[0020] In the above technical solution, data that meets the preset conditions is screened from a large amount of historical vehicle driving data as sample driving data for model training; the first initial model is trained using the sample driving data indicating that the vehicle is in the target state to obtain the first recognition model. The accuracy of the sample data during model training is ensured so that the recognition model obtained by training can have the ability to identify whether the vehicle is in the target state, further improving the accuracy of the first recognition model in identifying the corresponding features.

[0021] In combination with the first aspect and the above implementation, in some possible implementations, obtaining historical driving data of the vehicle includes: if a vehicle rescue signal is detected, obtaining the generation time of the vehicle rescue signal and the target power-on time of the vehicle, and determining a target time period between the generation time and the target power-on time; wherein the target power-on time is a power-on time before the generation time and closest to the generation time;

[0022] If the target time period is less than or equal to the time period threshold, the driving data within the preset time period is determined as the historical driving data;

[0023] If the target period is greater than the period threshold, the driving data within the target period is determined as historical driving data.

[0024] In the above technical solution, when a vehicle rescue signal is detected, the target time period between the generation time of the rescue signal and the last power-on time is determined; if the target time period is less than or equal to the time period threshold, the driving data within the preset time period is determined as historical driving data, for example, driving data within one day, to ensure that the historical driving data contains driving data when the vehicle is in the target state, to avoid abnormal power failure of the vehicle in a short period of time, resulting in the acquisition of data that does not contain data that can indicate that the vehicle is in the target state; if the target time period is greater than the time period threshold, the driving data within the target time period can be directly determined as historical driving data. By limiting the time period for obtaining historical driving data, the amount of historical driving data obtained can be reduced, thereby reducing the amount of calculation during model training and improving the training efficiency of the recognition model.

[0025] In combination with the first aspect and the above implementations, in some possible implementations, the historical driving data includes an average speed, an average rotation speed, and a position difference of the vehicle within a preset period of time when a vehicle rescue signal is detected;

[0026] The data that meets the preset conditions in the historical driving data are determined as sample driving data, including:

[0027] The data in the historical driving data, in which the average speed is less than the speed threshold, the average rotation speed is greater than the rotation speed threshold, and the position difference is less than the position difference threshold, is determined as the sample driving data.

[0028] In the above technical solution, when the historical parameters include the average speed, average rotation speed and position difference of the vehicle in a preset period, if the vehicle is in the target state, the vehicle may not be able to move, but the user may continue to step on the accelerator pedal to get out of the target state. At this time, the average speed and position change are both low, but the average rotation speed may be high; therefore, the speed threshold, rotation speed threshold and position difference threshold are set to determine the data that meets the above conditions from a large amount of historical driving data. Improve the accuracy of screening sample driving data; and then improve the accuracy of judging whether the vehicle is in the target state.

[0029] In combination with the first aspect and the above implementations, in some possible implementations, determining the target scene type when the target vehicle is in the target state based on the external environment data includes:

[0030] Inputting the vehicle external environment data into a second recognition model to obtain a second recognition result; wherein the second recognition result is used to indicate the scene type in which the target vehicle is located;

[0031] Based on the second recognition result, a target scene type when the target vehicle is in the target state is determined.

[0032] In the above technical solution, the second recognition model is used to recognize the acquired video of the vehicle's exterior environment, and the target scene type of the target vehicle when it is in the target state is determined according to the second recognition result output by the model. The recognition model can improve the recognition efficiency of a large amount of data, and the accuracy and real-time performance of the processing process are high. The recognition model can also be continuously optimized during the application process to continuously improve the recognition efficiency.

[0033] In combination with the first aspect and the above implementation, in some possible implementations, the second recognition model is obtained through the following training steps:

[0034] Obtain historical environmental data of the vehicle;

[0035] Determine sample environmental data from historical environmental data; wherein the sample environmental data is environmental data when the vehicle is in a target state;

[0036] Classify the sample environment data and determine the classification labels of the sample environment data;

[0037] A second initial model is trained based on the sample environment data and the classification labels to obtain a trained second recognition model.

[0038] In the above technical solution, a large amount of historical environmental data obtained is classified to determine the classification labels corresponding to the historical environmental data; and the second initial model is trained using the historical environmental data and the corresponding classification labels to obtain a trained second recognition model. The second recognition model is able to recognize environmental data to determine the type of scene the vehicle is currently in; training in combination with historical environmental data can ensure the authenticity and accuracy of the data, thereby improving the recognition accuracy of the second recognition model.

[0039] In combination with the first aspect and the above implementation manner, in some possible implementation manners, the method further includes:

[0040] Based on the target scene type, determining a rescue strategy corresponding to the target scene; wherein the rescue strategy is used to rescue the target vehicle so that the target vehicle exits the target state;

[0041] Control the target vehicle to execute the rescue strategy.

[0042] In the above technical solution, after determining the target scene type when the vehicle is in the target state, a corresponding rescue strategy is determined according to the target scene type to control the vehicle so that the target vehicle can exit the target state. Since the road conditions and other environments of the vehicle are different in different scene types, the rescue strategies to be adopted are also different. Determining a rescue strategy that matches the target scene type can improve the rescue efficiency of the vehicle and avoid danger caused by adopting the wrong rescue strategy.

[0043] In combination with the first aspect and the above implementations, in some possible implementations, the vehicle is configured with a target control, the target control is used to generate a vehicle rescue signal, and the method further includes:

[0044] If a touch operation on the target control is detected, the number of touch operations and / or the operation duration are obtained;

[0045] If the number of operations reaches a number threshold, and / or the number of operations reaches a number threshold, the vehicle is controlled to generate a vehicle rescue signal.

[0046] In the above technical solution, a target control for generating a vehicle rescue signal is configured for the vehicle. When the vehicle needs to call for rescue services, the call can be made by touching the target control. When a touch operation on the target control is detected, it is determined whether the number of touches is greater than a number threshold, and / or whether the touch duration is greater than a duration threshold, to determine whether to control the vehicle to generate a vehicle rescue signal. Avoiding accidental touch operations on the target control improves the accuracy of generating a vehicle rescue signal; and then obtaining vehicle driving data through the vehicle rescue signal can improve the accuracy of data acquisition.

[0047] In a second aspect, a vehicle driving scene recognition device is provided, the recognition device comprising:

[0048] A first acquisition module, used to acquire target driving data of a target vehicle;

[0049] A first determination module is used to determine whether the target vehicle is in a target state based on the target driving data; wherein the target state is used to indicate a state in which the target vehicle cannot drive normally;

[0050] A second acquisition module is used to acquire the external environment data of the target vehicle if the target vehicle is in the target state;

[0051] The second determination module is used to determine the target scene type when the target vehicle is in a target state based on the external vehicle environment data.

[0052] In combination with the second aspect, in some possible implementations, the first determination module is further used to input the target driving data into a first recognition model to obtain a first recognition result; and based on the first recognition result, determine whether the target vehicle is in a target state.

[0053] In combination with the second aspect and the above implementation manner, in some possible implementation manners, the identification device further includes:

[0054] The first training module is used to obtain historical driving data of the vehicle; determine the data that meets preset conditions in the historical driving data as sample driving data; wherein the sample driving data indicates that the vehicle is in a target state; and train a first initial model based on the sample driving data to obtain a trained first recognition model.

[0055] In combination with the second aspect and the above-mentioned implementation methods, in some possible implementation methods, the first training module is also used to detect a vehicle rescue signal, obtain the generation time of the vehicle rescue signal and the target power-on time of the vehicle, and determine the target time period between the generation time and the target power-on time; wherein the target power-on time is the power-on time before the generation time and closest to the generation time; if the target time period is less than or equal to the time period threshold, the driving data within the preset time period is determined as historical driving data; if the target time period is greater than the time period threshold, the driving data within the target time period is determined as historical driving data.

[0056] In combination with the second aspect and the above-mentioned implementation methods, in some possible implementation methods, the historical driving data includes the average speed, average rotation speed and position difference of the vehicle within a preset time period when a vehicle rescue signal is detected; the preset conditions are that the average speed is less than the speed threshold, the average rotation speed is greater than the rotation speed threshold, and the position difference is less than the position difference threshold.

[0057] In combination with the second aspect and the above-mentioned implementation methods, in some possible implementation methods, the second determination module is also used to input the external vehicle environment data into a second recognition model to obtain a second recognition result; based on the second recognition result, determine the target scene type when the target vehicle is in the target state.

[0058] In combination with the second aspect and the above implementation manner, in some possible implementation manners, the identification device further includes:

[0059] The second training module is used to obtain historical environmental data of the vehicle; determine sample environmental data from the historical environmental data; wherein the sample environmental data indicates that the vehicle is in a target state; classify the sample environmental data and determine the classification label of the sample environmental data; train a second initial model based on the sample environmental data and the classification label to obtain a trained second recognition model.

[0060] In combination with the second aspect and the above implementation manner, in some possible implementation manners, the identification device further includes:

[0061] The third determination module is used to determine a rescue strategy corresponding to the target scene based on the target scene type; wherein the rescue strategy is used to rescue the target vehicle so that the target vehicle exits the target state;

[0062] The first control module is used to control the target vehicle to execute the rescue strategy.

[0063] In combination with the second aspect and the above implementations, in some possible implementations, the vehicle is configured with a target control, the target control is used to generate a vehicle rescue signal, and the identification device further includes:

[0064] A third acquisition module is used to acquire the number of touch operation operations and / or the operation duration if a touch operation on a target control is detected;

[0065] The second control module is used to control the vehicle to generate a vehicle rescue signal if the number of operations reaches a number threshold and / or the number of operations reaches a number threshold.

[0066] In a third aspect, a vehicle is provided, comprising a memory and a processor. The memory is used to store executable program code, and the processor is used to call and run the executable program code from the memory, so that the vehicle executes the method in the first aspect or any possible implementation of the first aspect.

[0067] In a fourth aspect, a computer program product is provided, comprising: a computer program code, which, when executed on a computer, enables the computer to execute the method in the first aspect or any possible implementation of the first aspect.

[0068] In a fifth aspect, a computer-readable storage medium is provided, which stores a computer program code. When the computer program code runs on a computer, the computer executes the method in the above-mentioned first aspect or any possible implementation manner of the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 is a schematic diagram of a vehicle driving scenario provided in an embodiment of the present application;

[0070] Figure 2 is a schematic flow chart of a method for identifying a vehicle driving scene provided in an embodiment of the present application;

[0071] Figure 3 This is a schematic diagram of a training process of a first recognition model provided in an embodiment of the present application;

[0072] Figure 4 is a schematic diagram of a training process of a second recognition model provided in an embodiment of the present application;

[0073] Figure 5 is a schematic flow chart of another method for identifying a vehicle driving scene provided in an embodiment of the present application;

[0074] Figure 6 It is a structural schematic diagram of a vehicle driving scene recognition device provided in an embodiment of the present application;

[0075] Figure 7 It is a structural schematic diagram of a vehicle provided in an embodiment of the present application. DETAILED DESCRIPTION

[0076] The technical solution in the present application will be described clearly and in detail below in conjunction with the accompanying drawings. In the description of the embodiments of the present application, unless otherwise specified, " / " means or, for example, A / B can mean A or B: "and / or" in the text is only a description of the association relationship of associated objects, indicating that there can be three relationships, for example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the description of the embodiments of the present application, "multiple" means two or more than two.

[0077] In the following, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as suggesting or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features.

[0078] Figure 1It is a schematic diagram of a vehicle driving scene provided by an embodiment of the present application. During the driving process, the vehicle will inevitably encounter various terrains that may cause the vehicle to be trapped and unable to drive normally.

[0079] For example, Figure 1 As shown in (a), area 102 is a muddy area on the road surface. When the first vehicle 101 is driving, the wheels of the first vehicle 101 are stuck in the area 102. At this time, the first vehicle 101 may not be able to move forward, resulting in the vehicle being trapped in the muddy scene.

[0080] For example, Figure 1 As shown in (b), the area 104 is a gully area on the road surface. When the second vehicle 103 is driving, the wheel is stuck in the area 104, and the second vehicle 103 cannot drive normally.

[0081] It should be understood that the above uses muddy areas and gully areas as examples to illustrate the trapped scenes of vehicles during driving. The trapped scenes of vehicles may also include other scenes, and the embodiments of the present application do not limit the trapped scenes.

[0082] It should be noted that the first vehicle 101 and the second vehicle 103 may be the same vehicle or different vehicles, and are only used as examples to distinguish between the two vehicles in different trapped scenarios; the embodiments of the present application are not limited to this.

[0083] When a vehicle is trapped in complex terrain such as sand, mud, swamps, wading water, and ditches, the user is often required to get out of the vehicle to judge the current trapped scene and determine how to escape based on their own experience, or contact rescue services for rescue. In this process, the scene recognition may be inaccurate, and the rescue service may not be informed of the accurate trapped information. For non-professional users, they may not be able to determine the current trapped scene, which may affect the accuracy of scene judgment and the timeliness of rescue, and may even lead to further increased accident risks.

[0084] Therefore, the embodiments of the present application provide a method for identifying a vehicle driving scene, a vehicle, and a storage medium. By identifying the target driving data of the vehicle and the external environment data of the vehicle, it is determined whether the vehicle is in a target state where it cannot drive and the scene type when it is in the target state, so as to improve the accuracy of identifying the driving scene in which the vehicle is located.

[0085] Figure 2 It is a schematic flowchart of a method for identifying a vehicle driving scene provided in an embodiment of the present application.

[0086] For example, Figure 2 As shown, the method 200 includes:

[0087] S201, obtaining target driving data of a target vehicle.

[0088] The target driving data may include data such as the average speed, cumulative mileage, position change, average engine speed, etc. of the target vehicle during driving.

[0089] During the driving process of the target vehicle, the target driving data of the target vehicle is acquired in real time, and the target driving data is monitored and analyzed.

[0090] Optionally, in order to improve the accuracy of the judgment, a preset time may be set, for example, 10 minutes, and the target driving data within every 10 minutes is obtained.

[0091] For example, it is obtained that the average speed of the target vehicle in 10 minutes is 0, the average speed of the engine is 2000 RMP (speed unit: revolutions per minute), and the position does not change.

[0092] In one implementation, the target driving data may be obtained from sources including, but not limited to, vehicle sensors, ECU, GPS, and other information recording devices.

[0093] S202, based on the target driving data, determining whether the target vehicle is in a target state; wherein the target state is used to indicate a state in which the target vehicle cannot drive normally.

[0094] After the target driving data of the target vehicle is acquired, it is determined whether the target vehicle is in a target state where it cannot drive normally by analyzing the target driving data.

[0095] In one implementation, data thresholds corresponding to various types of data are preset, and whether the target vehicle is in the target state is determined based on the relationship between the target driving data and the data thresholds.

[0096] Exemplarily, a vehicle speed threshold of 5 km / h and an engine speed threshold of 2000 RPM are set. When the acquired target driving data indicates that the current driving speed of the target vehicle is less than the vehicle speed threshold of 5 km / h, but the engine speed is greater than the speed threshold of 2000 RPM, it can be determined that the target vehicle is in the target state.

[0097] In order to improve the accuracy and efficiency of identifying the target state, in one implementation, based on the target driving data, the process of determining whether the target vehicle is in the target state may include: inputting the target driving data into a first recognition model to obtain a first recognition result; wherein the first recognition result is used to indicate the driving state of the target vehicle; based on the first recognition result, determining whether the target vehicle is in the target state.

[0098] Exemplarily, a first recognition model is pre-trained, and the first recognition model can determine whether the vehicle is in a target state according to the target driving data of the vehicle. After acquiring the target driving data of the target vehicle, the target driving data is input into the first recognition model, and then it can be determined whether the vehicle is in the target state according to the first recognition result output by the first recognition model.

[0099] In the embodiment of the present application, the acquired target driving data is output to the trained first recognition model, and the target vehicle is determined to be in the target state according to the first recognition result output by the first recognition model. The recognition model is used to determine the target state, which can adapt to the analysis and processing of large-scale data, improve the computing efficiency and the accuracy of determining whether the vehicle is in the target state, and reduce the memory usage during the recognition process.

[0100] In order to improve the accuracy of recognition by the recognition model, in one implementation, the first recognition model can be obtained through the following training steps:

[0101] Obtain historical driving data of the vehicle;

[0102] Determine the data that meets the preset conditions in the historical driving data as sample driving data; wherein the sample driving data is the driving data when the vehicle is in the target state;

[0103] A first initial model is trained based on the sample driving data to obtain a trained first recognition model.

[0104] The first initial model may be a model trained by a gradient boosting tree algorithm; the gradient boosting tree algorithm may include an extreme gradient boosting algorithm (XGBoost), a light gradient boosting algorithm (LightGBM), or a GradientBoosted Decision Trees (GBDT), etc. The gradient boosting tree algorithm has the advantages of high computational efficiency and small memory usage, and is suitable for processing large-scale data and application scenarios with high real-time requirements; and the gradient boosting tree algorithm performs well in processing numerical features and categorical features, and can effectively learn abnormal driving patterns from a variety of sensor data and accurately identify abnormal vehicle states.

[0105] For example, most vehicles may encounter a situation where the vehicle cannot drive normally during driving, and the historical driving data of each vehicle can be uploaded to the cloud platform through the Internet or other means; the cloud platform stores a large amount of historical driving data; by processing the historical driving data, the driving data that can indicate that the vehicle is in the target state is screened out as the sample driving data for training the first initial model; because the sample driving data can indicate that the vehicle is in the target state, the first recognition model trained using the sample driving data can have the ability to identify whether the vehicle is in the target state.

[0106] Optionally, the historical driving data may be all driving data recorded after the vehicle leaves the factory; or may be driving data within a preset period of time.

[0107] Optionally, in the process of using the first recognition model to identify whether the vehicle is in the target state, as the amount of data continues to increase, the model parameters in the first recognition model can be continuously updated according to the latest collected driving data to improve the recognition efficiency and recognition accuracy of the first recognition model.

[0108] Optionally, the first recognition model can be deployed on the vehicle side, on the cloud platform side, or on both the vehicle side and the cloud platform side. During the recognition process, recognition can be performed on the side with less computational effort.

[0109] In the embodiment of the present application, data that meets preset conditions is screened from a large amount of historical vehicle driving data as sample driving data for model training; the first initial model is trained using sample driving data indicating that the vehicle is in a target state to obtain a first recognition model. The accuracy of the sample data during model training is ensured so that the recognition model obtained by training can have the ability to identify whether the vehicle is in the target state, further improving the accuracy of the first recognition model in identifying the corresponding features.

[0110] When there is a lot of historical data, the workload of screening sample driving data from it will increase, which may lead to a decrease in the efficiency of model training. In order to improve the training efficiency of model training, in one implementation, the historical driving data includes the average speed, average rotation speed and position difference of the vehicle within a preset period of time when the vehicle rescue signal is detected; the data that meets the preset conditions in the historical driving data is determined as sample driving data, including: the data in the historical driving data with an average speed less than a speed threshold, an average rotation speed greater than a rotation speed threshold, and a position difference less than a position difference threshold is determined as sample driving data.

[0111] For example, the speed threshold may be set to 5 km / h, the rotation speed threshold may be set to 2000 RPM, and the position difference threshold may be set to 20 meters.

[0112] Optionally, under different road conditions, the parameters of the vehicle during normal driving may be different; in order to improve the accuracy of determining the sample driving data, different preset conditions may be set for different road conditions.

[0113] Exemplarily, when a vehicle rescue signal is received, it indicates that the vehicle is currently in the target state and there is a need for vehicle rescue; the generation time of the vehicle rescue signal is determined, and the driving speed, engine speed and position information within 5 seconds before and after the generation time are collected; and then the average speed, average speed and position difference of the vehicle during this period are determined; if the average speed is less than the speed threshold, and the average speed is greater than the speed threshold, and the position difference is less than the position difference threshold, then the data can be determined as sample driving data to indicate that the vehicle is in the target state.

[0114] Optionally, the preset time period can be the time period between the moment when the vehicle power-on signal is received and the current moment during this trip; the total time period is divided into several sub-time periods, and the average speed, average rotation speed and position difference in each sub-time period are judged. If the preset conditions are met, the historical driving data in the sub-time period is used as sample driving data.

[0115] When a vehicle is trapped and cannot drive normally, the user may not initiate vehicle rescue immediately, but may initiate it at a later time; if only the driving data before and after the rescue signal is detected, it cannot be ensured that the current data can indicate that the vehicle is in the target state. Therefore, all data in the current power-on cycle can be obtained, and sample driving data that meets the preset conditions can be filtered out.

[0116] For example, the vehicle was powered on at 9:30 am on this trip, and the current time is 10:00 am. Vehicle rescue data is detected at the current time. All speed information, rotation speed information, and position information from the power-on time to the current time can be obtained. Based on the speed information, rotation speed information, and position information, the average speed, average rotation speed, and position difference within every 10 seconds can be calculated. If there are average speeds, average rotation speeds, and position differences that meet the preset conditions, these data can be determined as sample driving data.

[0117] In the embodiment of the present application, when the historical parameters include the average speed, average rotation speed and position difference of the vehicle in a preset period, if the vehicle is in the target state, the vehicle may not be able to move, but the user may continue to step on the accelerator pedal to get out of the target state. At this time, the average speed and position change are both low, but the average rotation speed may be high; therefore, the speed threshold, rotation speed threshold and position difference threshold are set to determine the data that meets the above conditions from a large amount of historical driving data. Improve the accuracy of screening sample driving data; and then improve the accuracy of judging whether the vehicle is in the target state.

[0118] In one implementation, the process of obtaining the historical driving data of the vehicle may include: if a vehicle rescue signal is detected, obtaining the generation time of the vehicle rescue signal and the target power-on time of the vehicle, and determining a target time period between the generation time and the target power-on time; wherein the target power-on time is a power-on time before the generation time and closest to the generation time;

[0119] If the target time period is less than or equal to the time period threshold, the driving data within the preset time period is determined as the historical driving data;

[0120] If the target period is greater than the period threshold, the driving data within the target period is determined as historical driving data.

[0121] Exemplarily, the time period threshold is set to 30 minutes. If the target time period is determined to be 20 minutes based on the generation time of the vehicle rescue signal and the target power-on time, which is less than the time period threshold, it may indicate that the vehicle has abnormal power failure after being in the target state, and this time period may not include driving data when the vehicle is in the target state. In this case, the driving data within 2 hours before the generation time can be determined as historical driving data; if the target time period is calculated to be 1 hour, the driving data within this 1 hour can be determined as historical driving data.

[0122] In the embodiment of the present application, when a vehicle rescue signal is detected, a target time period between the generation time of the rescue signal and the last power-on time is determined; if the target time period is less than or equal to the time period threshold, the driving data within the preset time period is determined as historical driving data, for example, driving data within one day, to ensure that the historical driving data contains driving data when the vehicle is in the target state, to avoid abnormal power failure of the vehicle in a short period of time, resulting in the acquisition of data that does not contain data that can indicate that the vehicle is in the target state; if the target time period is greater than the time period threshold, the driving data within the target time period can be directly determined as historical driving data. By limiting the time period for acquiring historical driving data, the amount of historical driving data acquired can be reduced, thereby reducing the amount of calculation during model training and improving the training efficiency of the recognition model.

[0123] In one implementation, the vehicle is configured with a target control, which is used to generate a vehicle rescue signal; if a touch operation on the target control is detected, the number of touch operations and / or the operation duration are obtained; if the number of operations reaches a number threshold and / or the number of operations reaches a number threshold, the vehicle is controlled to generate a vehicle rescue signal.

[0124] The number of operations indicates the number of consecutive operations on the target control; and the operation duration indicates the duration of consecutive operations on the target control.

[0125] Optionally, in the process of obtaining the number of operations, a time interval threshold may be set, for example, 1 second. When the time interval between two operations is less than the threshold, the number of operations is accumulated.

[0126] Exemplarily, a number threshold is set to 3 times, and a duration threshold is set to 3 seconds; when the number of operations on the target control is obtained to be greater than 3 times, it is determined that there is a need for vehicle rescue, and the vehicle is controlled to generate a vehicle rescue signal; or, when the operation duration of the target control is obtained to be greater than 3 seconds, it is determined that there is a need for vehicle rescue, and the vehicle is controlled to generate a vehicle rescue signal; or, when the number of operations on the target control is obtained to be greater than 3 times, and the operation duration is greater than 3 seconds, it is determined that there is a need for vehicle rescue, and the vehicle is controlled to generate a vehicle rescue signal.

[0127] Optionally, the target control may include a physical control and a virtual control, etc. The physical control may include a soft switch, a lever switch, etc.

[0128] Optionally, the target control can be set on a central control display screen, a vehicle steering wheel, or the inside of a main driving door, etc., so that the user can perform touch operations on it when rescue is needed.

[0129] In an embodiment of the present application, a target control for generating a vehicle rescue signal is configured for the vehicle. When the vehicle needs to call for rescue services, the call can be made by touching the target control. When a touch operation on the target control is detected, it is determined whether the number of touches is greater than a number threshold, and / or whether the touch duration is greater than a duration threshold, to determine whether to control the vehicle to generate a vehicle rescue signal. Avoiding accidental touch operations on the target control improves the accuracy of generating a vehicle rescue signal; and then obtaining vehicle driving data through the vehicle rescue signal can improve the accuracy of data acquisition.

[0130] S203: If the target vehicle is in the target state, obtain the external environment data of the target vehicle.

[0131] Among them, the external environmental data may include environmental images, environmental videos and environmental parameters around the vehicle, such as temperature, humidity, etc.

[0132] When the target driving data of the target vehicle indicates that the target vehicle is in a target state where it cannot drive normally, in order to determine the specific scene in which the target vehicle is currently located, the external environment data of the target vehicle is further acquired.

[0133] Exemplarily, an external camera device is used to capture an image or video of the environment outside the target vehicle; or, radar and other external sensors are combined to determine the temperature, humidity and whether there are obstacles outside the vehicle.

[0134] In one implementation, in order to protect the privacy and security of the user, when acquiring the external vehicle environment data, the sensitive information in the data is desensitized.

[0135] Exemplarily, sensitive data may include data such as portraits and license plate numbers; they may be desensitized through methods such as replacement, encryption, scrambling, generalization, and pseudo-randomization; for example, portraits may be mosaiced or replaced with other images; license plate numbers may be replaced with random numbers, etc.

[0136] Optionally, if the vehicle is in the target state, the tires of the vehicle are more likely to be in a trapped state. In order to improve the accuracy of scene recognition, an environmental image or video including the vehicle tires may be obtained; or an environmental image or video around the vehicle tires may be obtained.

[0137] S204, determining a target scene type when the target vehicle is in a target state based on the external vehicle environment data.

[0138] After acquiring the external environment data around the target vehicle, the specific target scene type in which the target vehicle is currently located is determined according to the external environment data.

[0139] For example, when sand features are detected in the video of the vehicle's exterior environment, it can be determined that the target vehicle is currently trapped in a desert scene; when wet soil features are detected, it can be determined that the target vehicle is in a muddy scene.

[0140] Optionally, the current location information of the target vehicle can also be obtained; the location information can indicate the approximate environment in which the target vehicle is currently located, such as desert areas, mountainous areas, swamp areas, urban areas, etc., which can further narrow the scope of judging the target scene type.

[0141] In order to improve the recognition efficiency of the target scene type, in one implementation, based on the external environment data, the process of determining the target scene type when the target vehicle is in the target state may include: inputting the external environment data into a second recognition model to obtain a second recognition result; wherein the second recognition result is used to indicate the scene type in which the target vehicle is located; based on the second recognition result, determining the target scene type when the target vehicle is in the target state.

[0142] Exemplarily, the second recognition model is pre-trained, and the second recognition model has the function of determining the target scene type that the vehicle is currently in according to the vehicle exterior environment data. The acquired vehicle exterior environment video is input into the second recognition model, and the second recognition model can output a second recognition result, and the second recognition result can represent the target scene type indicated by the current vehicle exterior environment video.

[0143] In the embodiment of the present application, the second recognition model is used to recognize the acquired video of the vehicle's exterior environment, and the target scene type of the target vehicle when it is in the target state is determined based on the second recognition result output by the model. The recognition model can improve the recognition efficiency of a large amount of data, and the accuracy and real-time performance of the processing process are high. The recognition model can also be continuously optimized during the application process to continuously improve the recognition efficiency.

[0144] In one implementation, the second recognition model can be obtained through the following training steps:

[0145] Obtain historical environmental data of the vehicle;

[0146] Determine sample environmental data from historical environmental data; wherein the sample environmental data is environmental data when the vehicle is in a target state;

[0147] Classify the sample environment data and determine the classification labels of the sample environment data;

[0148] A second initial model is trained based on the sample environment data and the classification labels to obtain a trained second recognition model.

[0149] Among them, the second initial model can be a deep learning model that combines visual and language information - ViLBERT (Visual-Linguistic BERT) model; this model is specially designed to process vision and speech tasks such as visual question answering (VQA), image caption generation, and visual reasoning.

[0150] The ViLBERT model can process visual and language information simultaneously, and is suitable for joint analysis of video and text descriptions. If the ViLBERT model is used, the vehicle's trapped scene can be identified and classified by analyzing environmental data and classification labels; ViLBERT also has a multi-head attention mechanism that can capture different important information in the video, improving the understanding of complex scenes and classification accuracy.

[0151] Exemplarily, a large amount of historical environmental data during vehicle driving is collected, and environmental data when the vehicle is in a target state is extracted as sample environmental data; different types of sample environmental data are classified through manual labeling and other methods to determine their classification labels, which may include sand, mud, swamps, wading, gullies, etc.; and then the data pairs consisting of the sample environmental data and the classification labels are input into the second initial model and trained to obtain a trained second recognition model, so that the second recognition model can determine the corresponding classification labels based on the acquired target environmental data to determine the target environment type in which the vehicle is currently located.

[0152] Optionally, when a vehicle rescue signal is detected, environmental data around the vehicle is synchronously acquired and bound to the target state to ensure the accuracy of the sample data.

[0153] Optionally, in the process of using the second recognition model to identify whether the vehicle is in the target state, as the amount of data continues to increase, the model parameters in the second recognition model can be continuously updated according to the latest collected external environment data to improve the recognition efficiency and recognition accuracy of the second recognition model.

[0154] Optionally, the second recognition model can be deployed on the vehicle side, on the cloud platform side, or on both the vehicle side and the cloud platform side. During the recognition process, recognition can be performed on the side with less computational effort.

[0155] In the embodiment of the present application, a large amount of historical environmental data obtained is classified to determine the classification labels corresponding to the historical environmental data; and the second initial model is trained using the historical environmental data and the corresponding classification labels to obtain a trained second recognition model. The second recognition model is able to recognize environmental data to determine the type of scene the vehicle is currently in; training in combination with historical environmental data can ensure the authenticity and accuracy of the data, thereby improving the recognition accuracy of the second recognition model.

[0156] Different scenario types require different response measures. In order to improve the efficiency of vehicle escape, in one implementation, based on the target scenario type, a rescue strategy corresponding to the target scenario is determined; wherein the rescue strategy is used to rescue the target vehicle so that the target vehicle exits the target state; and the target vehicle is controlled to execute the rescue strategy.

[0157] For example, when the target scene type is sand, the rescue strategy is determined to be to reduce the tire pressure, and then the vehicle is controlled to adjust the tire pressure; when the target scene type is a gully, the rescue strategy is determined to be to sway back and forth, and the vehicle can be controlled to slowly accelerate and move back and forth continuously.

[0158] Optionally, for scenario types that require rescue equipment to escape from distress, prompt information can be output after the rescue strategy is determined to prompt the user to use the corresponding rescue equipment to make the vehicle exit the target state.

[0159] Exemplarily, when the rescue strategy cannot automatically control the vehicle for rescue, the rescue equipment currently required is determined based on the target scene type; for example, when the target scene type is mud, the user is required to prepare towing ropes and items such as wooden boards and sand that can increase ground friction, and a prompt voice can be output: "The current vehicle is trapped in a mud scene, and you need to prepare towing ropes, wooden boards or sand to escape"; when the target scene type is a ravine, the output can be: "The current vehicle is trapped in a ravine scene, and you need to prepare a jack and pads to escape."

[0160] Optionally, when the vehicle cannot be exited from the target state by multiple methods, a rescue service may be automatically contacted to seek professional help.

[0161] In the embodiment of the present application, after determining the target scene type when the vehicle is in the target state, a corresponding rescue strategy is determined according to the target scene type to control the vehicle so that the target vehicle can exit the target state. Since the road conditions and other environments of the vehicle are different under different scene types, the rescue strategies to be adopted are also different. Determining a rescue strategy that matches the target scene type can improve the rescue efficiency of the vehicle and avoid danger caused by adopting the wrong rescue strategy.

[0162] Optionally, Figure 2 The method shown may be applied to a vehicle; or, a processor in a vehicle; or, a chip in a vehicle.

[0163] Optionally, Figure 2 The method shown can be applied to cloud platforms.

[0164] Optionally, Figure 2 The above process can be performed in the vehicle and the cloud platform. For example, the operations of obtaining target driving data and obtaining vehicle external environment data in S201 and S203 can be performed by the vehicle; the vehicle sends the above data to the cloud platform; and then S202 and S204 can be performed by the cloud platform.

[0165] In the embodiment of the present application, based on the acquired target driving data of the target vehicle, it is determined whether the target vehicle is in a target state where it cannot drive normally; if the target vehicle is identified as being in the target state, the external environment data of the target vehicle is further acquired; and based on the external environment data, the target scene type when the target vehicle is currently in the target state is determined. The driving data of the vehicle is monitored, and when it is determined that the vehicle is in a state where it cannot drive normally, the specific scene type is further determined in combination with the external environment data to improve the accuracy of identifying the driving scene in which the vehicle is located; and accurate scene information can be provided for rescue to avoid affecting the accuracy and timeliness of the rescue strategy due to scene judgment errors.

[0166] Figure 3 This is a schematic diagram of a training process of a first recognition model provided in an embodiment of the present application. It should be understood that the first recognition model can be deployed in a vehicle and / or a cloud platform.

[0167] For example, Figure 3 As shown, the process includes:

[0168] S301, obtaining historical driving data of the vehicle.

[0169] The historical driving data may include the average speed, average rotation speed and position difference of the vehicle within a preset time period when the vehicle rescue signal is detected.

[0170] For example, if a vehicle rescue signal is detected, the generation time of the vehicle rescue signal is determined, and the driving speed, engine speed and position information in the target period between the vehicle power-on time and the generation time in the same power-on cycle are obtained; and then the average speed, average speed and position difference of multiple time periods in the target period are determined. For example, various data in the target period are calculated with a period of 10s.

[0171] Optionally, the implementation of S301 can refer to Figure 2 The relevant description of S202 in the present application embodiment will not be repeated here.

[0172] S302, determine whether there is data that meets the preset conditions in the historical driving data; if so, execute S303; if not, execute S304.

[0173] The preset condition may be that the average speed is less than a speed threshold, the average rotation speed is greater than a rotation speed threshold, and the position difference is less than a position difference threshold.

[0174] The average speed, average rotation speed and position difference obtained during the target period are judged to determine whether there is driving data that meets the preset conditions.

[0175] For example, the speed threshold may be set to 5 km / h, the rotation speed threshold may be set to 2000 RPM, and the position difference threshold may be set to 20 meters.

[0176] Exemplarily, when it is detected that the average speed within a preset time period is less than a speed threshold, and the average rotational speed is greater than a rotational speed threshold, and the position difference is less than a position difference threshold, it indicates that the vehicle may be in a state where it cannot drive normally, and it can be determined that there is data in the historical driving data that meets the preset conditions; if each driving data does not meet the above preset conditions, it is determined that there is no data in the historical driving data that meets the preset conditions.

[0177] Optionally, under different road conditions, the parameters of the vehicle during normal driving may be different; in order to improve the accuracy of determining the sample driving data, different preset conditions may be set for different road conditions.

[0178] Optionally, the implementation of S302 can refer to Figure 2 The relevant description of S202 in the present application embodiment will not be repeated here.

[0179] S303: Determine the data that meets the preset conditions as sample driving data.

[0180] The sample driving data is driving data when the vehicle is in a target state, and the target state is used to indicate a state where the target vehicle cannot drive normally.

[0181] Exemplarily, when there is driving data meeting preset conditions in a large amount of historical driving data, the data is determined as sample driving data for training the recognition model.

[0182] Optionally, the implementation of S303 can refer to Figure 2 The relevant description of S202 in the present application embodiment will not be repeated here.

[0183] S304, remove historical driving data.

[0184] Exemplarily, if there is no driving data meeting the preset conditions in the historical driving data, the collected historical driving data is removed to reduce the occupation of storage space.

[0185] S305: Train the first initial model using the sample driving data to obtain a trained first recognition model.

[0186] The first initial model may be a model trained by a gradient boosting tree algorithm; the gradient boosting tree algorithm may include XGBoost, LightGBM, or CatBoost, etc.

[0187] Exemplarily, the sample driving data can indicate that the vehicle is in a target state where it cannot drive normally. Based on the screened sample driving data, the first initial model can be continuously trained to obtain a trained first recognition model. At this time, the first recognition model already has the function of determining whether the vehicle is in the target state based on the target driving data during the vehicle's driving process.

[0188] Optionally, the implementation of S305 can refer to Figure 2 The relevant description of S202 in the present application embodiment will not be repeated here.

[0189] In an embodiment of the present application, preset conditions are set to filter out sample driving data that can reflect that the vehicle is in a target state from historical driving data as training samples for the first recognition model, thereby ensuring the accuracy of the sample driving data and its relevance to the target state, and further improving the accuracy and efficiency of the trained first recognition model in identifying the target state.

[0190] Figure 4 is a schematic diagram of a training process of a second recognition model provided in an embodiment of the present application. It should be understood that the second recognition model can be deployed in a vehicle and / or a cloud platform.

[0191] For example, Figure 4 As shown, the process includes:

[0192] S401, obtaining historical environmental data of the vehicle.

[0193] The historical environmental data may include images, videos, environmental parameters, etc. around the vehicle during its historical driving process; or images, videos, environmental parameters, etc. around the vehicle tires.

[0194] Exemplarily, video data of a vehicle traveling within a month is obtained.

[0195] Optionally, the implementation of S401 can refer to Figure 2 The relevant description of S204 in the present application embodiment is not repeated here.

[0196] S402, determining sample environmental data from historical environmental data.

[0197] The sample environment data is the environment data when the vehicle is in the target state.

[0198] Exemplarily, the historical environmental data acquired of the vehicle includes a majority of driving scenes, and it is necessary to filter out sample environmental data that can indicate that the vehicle is trapped from the historical environmental data.

[0199] Optionally, the implementation of S402 can refer to Figure 2 The relevant description of S204 in the present application embodiment is not repeated here.

[0200] S403, classify the sample environment data and determine the classification label of the sample environment data.

[0201] For example, the sample environmental data includes environmental data of multiple different scenes, and the sample environmental data can be classified by manual labeling or other methods to determine the classification labels corresponding to the sample environmental data. For example, the classification label corresponding to image A is a sandy scene; the classification label corresponding to video B is a muddy scene, etc.

[0202] Optionally, the implementation of S403 can refer to Figure 2 The relevant description of S204 in the present application embodiment is not repeated here.

[0203] S404: Train a second initial model based on the sample environment data and the classification labels to obtain a trained second recognition model.

[0204] The second initial model may be a deep learning model combining visual and language information - ViLBERT (Visual-Linguistic BERT) model.

[0205] Exemplarily, the data pairs between the classified sample environment data and the classification labels are input into the second initial model, which is trained to obtain a trained second recognition model, so that the second recognition model has the ability to recognize the type of scene in which the vehicle is located.

[0206] Optionally, the implementation of S404 can refer to Figure 2 The relevant description of S204 in the present application embodiment is not repeated here.

[0207] In an embodiment of the present application, by pre-processing the collected historical environmental data, sample environmental data that can indicate that the vehicle is in a target state is screened out and classified to obtain sample data for training a recognition model. This ensures the accuracy of the sample data and its relevance to the target state, and further improves the accuracy of the second recognition model in identifying the target scene type.

[0208] Take the example where the first recognition model is deployed on a vehicle and the second recognition model is deployed on a cloud platform. Figure 5 It is a schematic flowchart of another method for identifying a vehicle driving scene provided in an embodiment of the present application.

[0209] For example, Figure 5 As shown, the method includes:

[0210] S501, obtaining target driving data of a target vehicle.

[0211] The target driving data may include data such as the average speed, cumulative mileage, position change, average engine speed, etc. of the target vehicle during driving.

[0212] For example, various sensors obtain that the average speed of the target vehicle in 10 minutes is 5 km / h, the average engine speed is 2000 RMP, and the position has not changed.

[0213] Optionally, the implementation of S501 can refer to Figure 2S201 and related descriptions; the embodiments of the present application are not described in detail here.

[0214] S502, inputting the target driving data into a first recognition model to determine whether the vehicle is in a target state.

[0215] The target state is used to indicate a state where the target vehicle cannot drive normally.

[0216] Exemplarily, the first recognition model can be deployed locally in the vehicle, and the acquired driving data such as average speed, average rotation speed, position change, mileage, etc. are input into the pre-trained first recognition model to determine whether the vehicle is in the target state.

[0217] Optionally, the implementation of S502 can refer to Figure 2 S202 and related descriptions; the embodiments of the present application are not described in detail here.

[0218] S503: If the target vehicle is in the target state, obtain the external environment data.

[0219] Exemplarily, when the output result of the first recognition model indicates that the target vehicle is in the target state, current external vehicle environment data is further acquired, such as video information or image information within a preset range around the target vehicle.

[0220] Optionally, the implementation of S503 can refer to Figure 2 The relevant description of S203 in the present application embodiment will not be repeated here.

[0221] S504, uploading the vehicle external environment data.

[0222] Exemplarily, the vehicle acquires external environment data through a camera device, and uploads the external environment data to a cloud platform to analyze the external environment data.

[0223] S505, inputting the external environment data into a second recognition model to determine the target scene type where the target vehicle is located.

[0224] Exemplarily, the second recognition model can be deployed on a cloud platform, and the external environment data uploaded by the vehicle is input into the second recognition model to determine the target scene type in which the target vehicle is currently located.

[0225] Optionally, the implementation of S505 can refer to Figure 2 The relevant description of S204 in the present application embodiment is not repeated here.

[0226] S506: Determine a rescue strategy based on the target scene type.

[0227] Exemplarily, after the current target scene type of the target vehicle is determined, different rescue strategies need to be determined according to different scene types.

[0228] It should be noted that after the cloud platform determines the target scene type, the target scene type can be sent to the vehicle first, and the vehicle side further determines the rescue strategy; for the execution end of this step, the embodiment of the present application does not limit this.

[0229] Optionally, the implementation of S506 can refer to Figure 2 The relevant description of S204 in the present application embodiment is not repeated here.

[0230] S507, controlling the target vehicle to execute the rescue strategy.

[0231] Exemplarily, the cloud platform sends the determined rescue strategy that matches the target scenario type to the target vehicle, and the target vehicle executes the corresponding rescue strategy to make the vehicle exit the target state.

[0232] Optionally, the implementation of S507 can refer to Figure 2 The relevant description of S204 in the present application embodiment is not repeated here.

[0233] In an embodiment of the present application, a first recognition model and a second recognition model are used in turn to identify whether the vehicle is in a target state, and when in the target state, the specific target scene type, thereby improving the accuracy of scene recognition; the two recognition models are deployed on the vehicle side and the cloud platform side, respectively, to avoid performing calculations entirely on the vehicle side or the cloud platform side, which results in excessive calculations, thereby improving the efficiency of vehicle driving scene recognition.

[0234] Combination of the above Figures 1 to 5 The vehicle driving scene recognition method provided by the embodiment of the present application is described in detail; Figure 6 and Figure 7 The device embodiments of the present application are described in detail. It should be understood that the device in the embodiments of the present application can execute the various methods of the aforementioned embodiments of the present application, that is, the specific working processes of the following various products can refer to the corresponding processes in the aforementioned method embodiments.

[0235] Figure 6 It is a structural schematic diagram of a vehicle driving scene recognition device provided in an embodiment of the present application.

[0236] For example, Figure 6 As shown, the identification device 600 includes:

[0237] The first acquisition module 601 is used to acquire target driving data of a target vehicle;

[0238] A first determination module 602 is used to determine whether the target vehicle is in a target state based on the target driving data; wherein the target state is used to indicate a state in which the target vehicle cannot drive normally;

[0239] The second acquisition module 603 is used to acquire the external environment data of the target vehicle if the target vehicle is in the target state;

[0240] The second determination module 604 is used to determine the target scene type when the target vehicle is in the target state based on the external vehicle environment data.

[0241] In one possible implementation, the first determination module 602 is also used to input the target driving data into the first recognition model to obtain a first recognition result; wherein the first recognition result is used to indicate the driving state of the target vehicle; based on the first recognition result, determine whether the target vehicle is in the target state.

[0242] In one possible implementation, the recognition device 600 also includes: a first training module, used to obtain historical driving data of the vehicle; determine data in the historical driving data that meets preset conditions as sample driving data; wherein the sample driving data is driving data when the vehicle is in a target state; and train a first initial model based on the sample driving data to obtain a trained first recognition model.

[0243] In one possible implementation, the first training module is also used to detect a vehicle rescue signal, obtain the generation time of the vehicle rescue signal and the target power-on time of the vehicle, and determine a target time period between the generation time and the target power-on time; wherein the target power-on time is the power-on time before the generation time and closest to the generation time; if the target time period is less than or equal to the time period threshold, the driving data within the preset time period is determined as historical driving data; if the target time period is greater than the time period threshold, the driving data within the target time period is determined as historical driving data.

[0244] In one possible implementation, the historical driving data includes the average speed, average rotation speed and position difference of the vehicle within a preset time period when a vehicle rescue signal is detected; the preset conditions are that the average speed is less than a speed threshold, the average rotation speed is greater than a rotation speed threshold, and the position difference is less than a position difference threshold.

[0245] In one possible implementation, the second determination module 604 is also used to input the external environment data into a second recognition model to obtain a second recognition result; wherein the second recognition result is used to indicate the scene type in which the target vehicle is located; based on the second recognition result, the target scene type when the target vehicle is in the target state is determined.

[0246] In a possible implementation, the identification device 600 further includes:

[0247] The second training module is used to obtain historical environmental data of the vehicle; determine sample environmental data from the historical environmental data; wherein the sample environmental data is environmental data when the vehicle is in a target state; classify the sample environmental data and determine classification labels of the sample environmental data; train a second initial model based on the sample environmental data and the classification labels to obtain a trained second recognition model.

[0248] In a possible implementation, the identification device 600 further includes:

[0249] The third determination module is used to determine a rescue strategy corresponding to the target scene based on the target scene type; wherein the rescue strategy is used to rescue the target vehicle so that the target vehicle exits the target state;

[0250] The first control module is used to control the target vehicle to execute the rescue strategy.

[0251] In a possible implementation, the vehicle is configured with a target control, the target control is used to generate a vehicle rescue signal, and the identification device 600 further includes:

[0252] A third acquisition module is used to acquire the number of touch operation operations and / or the operation duration if a touch operation on a target control is detected;

[0253] The second control module is used to control the vehicle to generate a vehicle rescue signal if the number of operations reaches a number threshold and / or the number of operations reaches a number threshold.

[0254] It should be noted that the above identification device 600 is embodied in the form of a functional unit. The term "module" here can be implemented in the form of software and / or hardware, and is not specifically limited to this.

[0255] For example, a "module" may be a software program, a hardware circuit, or a combination of the two that implements the above functions. The hardware circuit may include an application specific integrated circuit (ASIC), an electronic circuit, a processor (such as a shared processor, a dedicated processor, or a group processor, etc.) and a memory for executing one or more software or firmware programs, a combined logic circuit, and / or other suitable components that support the described functions.

[0256] Therefore, the units of each example described in the embodiments of the present application can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present application.

[0257] Figure 7 It is a structural schematic diagram of a vehicle provided in an embodiment of the present application.

[0258] For example, Figure 7 As shown, the vehicle 700 includes: a memory 701 and a processor 702, wherein the memory 701 stores an executable program code 703, and the processor 702 is used to call and execute the executable program code 703 to perform a method for identifying a vehicle driving scene.

[0259] In addition, an embodiment of the present application also protects a device, which may include a memory and a processor, wherein the memory stores an executable program code, and the processor is used to call and execute the executable program code to execute a vehicle driving scene recognition method provided by an embodiment of the present application.

[0260] In this embodiment, the functional modules of the device can be divided according to the above method example. For example, each functional module can be corresponded, or two or more functions can be integrated into one processing module. The above integrated module can be implemented in the form of hardware. It should be noted that the division of modules in this embodiment is schematic and is only a logical function division. There may be other division methods in actual implementation.

[0261] In the case of dividing each functional module according to each function, the device may further include a first acquisition module, a first determination module, a second acquisition module, a second determination module, etc. It should be noted that all relevant contents of each step involved in the above method embodiment can be referred to the functional description of the corresponding functional module, which will not be repeated here.

[0262] It should be understood that the device provided in this embodiment is used to execute the above-mentioned method for identifying a vehicle driving scene, and thus can achieve the same effect as the above-mentioned implementation method.

[0263] In the case of an integrated unit, the device may include a processing module and a storage module. When the device is applied to a vehicle, the processing module may be used to control and manage the actions of the vehicle. The storage module may be used to support the vehicle in executing mutual program codes, etc.

[0264] The processing module may be a processor or a controller, which may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of this application. The processor may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of digital signal processing (DSP) and a microprocessor, etc. The storage module may be a memory.

[0265] In addition, the device provided in the embodiments of the present application may specifically be a chip, a component or a module, and the chip may include a connected processor and a memory; wherein the memory is used to store instructions, and when the processor calls and executes the instructions, the chip can execute a vehicle driving scene recognition method provided in the above-mentioned embodiment.

[0266] This embodiment also provides a computer-readable storage medium, in which a computer program code is stored. When the computer program code is executed on a computer, the computer executes the above-mentioned related method steps to implement a vehicle driving scene recognition method provided in the above embodiment.

[0267] This embodiment also provides a computer program product. When the computer program product is run on a computer, the computer is enabled to execute the above-mentioned related steps to implement a vehicle driving scene recognition method provided by the above-mentioned embodiment.

[0268] Among them, the device, computer-readable storage medium, computer program product or chip provided in this embodiment is used to execute the corresponding method provided above. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding method provided above, and will not be repeated here.

[0269] Through the description of the above implementation methods, technical personnel in the relevant field can understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0270] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of modules or units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0271] The above contents are only specific implementation methods of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for identifying a vehicle driving scene, characterized in that: The method comprises: Acquire target driving data of a target vehicle; Based on the target driving data, determining whether the target vehicle is in a target state; wherein the target state is used to indicate a state in which the target vehicle cannot drive normally; If the target vehicle is in the target state, obtaining the external environment data of the target vehicle; Determining a target scene type when the target vehicle is in the target state based on the external vehicle environment data; Based on the target scene type, determining a rescue strategy corresponding to the target scene; wherein the rescue strategy is used to rescue the target vehicle so that the target vehicle exits the target state; Controlling the target vehicle to execute the rescue strategy; Wherein, determining the target scene type when the target vehicle is in the target state based on the external vehicle environment data includes: Inputting the vehicle external environment data into a second recognition model to obtain a second recognition result; wherein the second recognition result is used to indicate the scene type in which the target vehicle is located; Based on the second recognition result, determining a target scene type when the target vehicle is in the target state; The second recognition model is obtained through the following training steps: Obtain historical environmental data of the vehicle; Determining sample environmental data from the historical environmental data; wherein the sample environmental data is environmental data when the vehicle is in the target state; Classifying the sample environmental data and determining a classification label for the sample environmental data; A second initial model is trained based on the sample environment data and the classification labels to obtain the trained second recognition model.

2. The method according to claim 1, characterized in that The step of determining whether the target vehicle is in a target state based on the target driving data includes: Inputting the target driving data into a first recognition model to obtain a first recognition result; wherein the first recognition result is used to indicate the driving state of the target vehicle; Based on the first recognition result, it is determined whether the target vehicle is in the target state.

3. The method according to claim 2, characterized in that The first recognition model is obtained through the following training steps: Obtain historical driving data of the vehicle; Determine the data satisfying the preset conditions in the historical driving data as sample driving data; wherein the sample driving data is the driving data when the vehicle is in the target state; A first initial model is trained based on the sample driving data to obtain the trained first recognition model.

4. The method according to claim 3, characterized in that The historical driving data includes the average speed, average rotation speed and position difference of the vehicle within a preset period of time when the vehicle rescue signal is detected; The step of determining the data in the historical driving data that meets a preset condition as sample driving data includes: The data in the historical driving data, in which the average speed is less than a speed threshold, the average rotation speed is greater than a rotation speed threshold, and the position difference is less than a position difference threshold, is determined as the sample driving data.

5. The method according to claim 4, characterized in that The vehicle is configured with a target control, the target control is used to generate the vehicle rescue signal, and the method further includes: If a touch operation on the target control is detected, the number of operations and / or the operation duration of the touch operation are obtained; If the number of operations reaches a number threshold, and / or the number of operations reaches a number threshold, the vehicle is controlled to generate the vehicle rescue signal.

6. A vehicle, characterized in that: The vehicle comprises: A memory for storing executable program codes; A processor, configured to call and run the executable program code from the memory so that the vehicle executes the method according to any one of claims 1 to 5.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed, the method according to any one of claims 1 to 5 is implemented.

Images