Vehicle behavior identification method and device fusing cloud data

Abstract

The invention relates to a cloud data fused vehicle behavior recognition method and device, and belongs to the technical field of data processing.The cloud data fused vehicle behavior recognition method comprises the steps that real-time driving data of a target vehicle is acquired, the real-time driving data comprises driving data of vehicles around the geographic position of the target vehicle and traffic cloud data; taking the preprocessed real-time driving data as the input of a vehicle behavior recognition model, and determining the real-time vehicle behavior of the target vehicle based on the output of the vehicle behavior recognition model; wherein the vehicle behavior recognition model is obtained based on historical driving data of the target vehicle and historical vehicle behavior training. According to the invention, while the influence of the vehicles around the target vehicle on the target vehicle is considered, the influence of other factors on the target vehicle is also considered in combination with the cloud data, so that the accuracy of vehicle driving behavior recognition is effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a vehicle behavior recognition method and device integrating cloud data. Background Art

[0002] In recent years, with the rapid development of intelligent, connected, shared and autonomous driving technologies in the automotive industry, building efficient and accurate vehicle driving behavior recognition models by integrating traffic cloud data has become a research hotspot in the field of autonomous driving. The driving behavior recognition model is an intelligent prediction system that can accurately reflect driving behavior and intentions. It can perceive and analyze the driving behavior of vehicles and predict the driving intentions of surrounding vehicles, thereby providing more intelligent driving decisions and behavior evaluations.

[0003] Current autonomous driving technology mainly focuses on vehicle perception and environmental decision-making, and accurate recognition of vehicle driving behavior and intention remains a challenge. Traditional vehicle driving behavior recognition models often only focus on driving behavior or driving trajectory, resulting in a lack of accuracy in the recognition of vehicle driving behavior. Therefore, how to improve the accuracy of vehicle driving behavior recognition has become a technical problem that needs to be solved urgently. Summary of the invention

[0004] In view of this, it is necessary to provide a vehicle behavior recognition method and device that integrates cloud data to solve the problem that traditional vehicle driving behavior recognition models often only focus on driving behavior or driving trajectory, resulting in a lack of accuracy in recognizing vehicle driving behavior.

[0005] In order to solve the above problems, the present invention provides a vehicle behavior recognition method integrating cloud data, comprising: Acquire real-time driving data of the target vehicle, wherein the real-time driving data includes driving data of surrounding vehicles at the geographical location of the target vehicle and traffic cloud data; Using the preprocessed real-time driving data as an input of a vehicle behavior recognition model, and determining the real-time vehicle behavior of the target vehicle based on the output of the vehicle behavior recognition model; The vehicle behavior recognition model is obtained based on the historical driving data and historical vehicle behavior training of the target vehicle.

[0006] In a possible implementation, the driving data of surrounding vehicles at the geographical location of the target vehicle includes the position, speed, acceleration and steering angle of surrounding vehicles, and the traffic cloud data at the geographical location of the target vehicle includes traffic accident conditions, weather conditions and traffic flow.

[0007] In a possible implementation, the vehicle behavior recognition model is an LSTM network with an attention mechanism layer added, and the attention mechanism layer is used to assign weights to the input of the vehicle behavior recognition model.

[0008] In a possible implementation, assigning weights to the inputs of the vehicle behavior recognition model includes: Mapping the hidden state of the LSTM network based on a fully connected layer; Weights are assigned to the input of the vehicle behavior recognition model based on the hidden state of the mapped LSTM network and the softmax function.

[0009] In a possible implementation, assigning weights to the inputs of the vehicle behavior recognition model includes: The inputs of the vehicle behavior recognition model are weighted based on the following formula:

[0010]

[0011] in, represents the hidden state of the LSTM network after mapping, represents the fully connected layer function, represents the hidden state of the LSTM network, represents the input allocation weight of the vehicle behavior recognition model, Represents the softmax function.

[0012] In a possible implementation, the real-time vehicle behavior of the target vehicle includes any one of the following: Accelerate and go straight, drive normally, slow down and go straight, turn left, turn right, or about to stop.

[0013] In a possible implementation, the real-time vehicle behavior of the target vehicle is output from the vehicle behavior recognition model in the form of a binary vector.

[0014] The present invention also provides a vehicle behavior recognition device integrating cloud data, comprising: An acquisition module is used to acquire real-time driving data of a target vehicle, wherein the real-time driving data includes driving data of surrounding vehicles at the geographical location of the target vehicle and traffic cloud data; A determination module, used to use the preprocessed real-time driving data as an input of a vehicle behavior recognition model, and determine the real-time vehicle behavior of the target vehicle based on the output of the vehicle behavior recognition model; The vehicle behavior recognition model is obtained based on the historical driving data and historical vehicle behavior training of the target vehicle.

[0015] The present invention also provides an electronic device, including a memory and a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the vehicle behavior recognition method integrating cloud data as described above is implemented.

[0016] The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the vehicle behavior recognition method integrating cloud data as described above is implemented.

[0017] The beneficial effect of the present invention is that the vehicle behavior recognition method and device integrating cloud data provided by the present invention, when performing vehicle behavior recognition, not only focus on a single driving behavior or driving trajectory, but also perform vehicle behavior recognition in combination with the driving data of surrounding vehicles at the geographical location of the target vehicle and traffic cloud data. While considering the impact of the vehicles around the target vehicle on the target vehicle, it also considers the impact of other factors on the target vehicle in combination with the cloud data, thereby effectively improving the accuracy of vehicle driving behavior recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A schematic flow chart of an embodiment of a vehicle behavior recognition method integrating cloud data provided by the present invention; Figure 2 A schematic diagram of an embodiment of a data collection source provided by the present invention; Figure 3 A schematic diagram of an embodiment of a driving information input source provided by the present invention; Figure 4 A schematic diagram of a process flow of an embodiment of a driving behavior recognition model training process provided by the present invention; Figure 5 A schematic diagram of a process flow of an embodiment of a vehicle driving behavior recognition process provided by the present invention; Figure 6 A schematic diagram of an embodiment of a vehicle driving behavior visualization interface provided by the present invention; Figure 7 A schematic diagram of the structure of an embodiment of a vehicle behavior recognition device integrating cloud data provided by the present invention; Figure 8 A schematic structural diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0019] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.

[0020] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0021] In the description of the present invention, reference to an "embodiment" means that a particular feature, structure, or characteristic described in conjunction with the embodiment may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive of other embodiments. It is explicitly and implicitly understood by those skilled in the art that the described embodiments may be combined with other embodiments.

[0022] In recent years, with the rapid development of intelligent, connected, shared and autonomous driving technologies in the automotive industry, building efficient and accurate vehicle driving behavior recognition models by integrating traffic cloud data has become a research hotspot in the field of autonomous driving. The driving behavior recognition model is an intelligent prediction system that can accurately reflect driving behavior and intentions. It can perceive and analyze the driving behavior of vehicles and predict the driving intentions of surrounding vehicles, thereby providing more intelligent driving decisions and behavior evaluations.

[0023] Current autonomous driving technology mainly focuses on vehicle perception and environmental decision-making, and accurate recognition of vehicle driving behavior and intention remains a challenge. Traditional vehicle driving behavior recognition models often only focus on driving behavior or driving trajectory, resulting in a lack of accuracy in identifying vehicle driving behavior.

[0024] In order to solve the above problems, the present invention provides a vehicle behavior recognition method integrating cloud data.

[0025] The specific embodiments are described in detail below: A specific embodiment of the present invention discloses a vehicle behavior recognition method integrating cloud data, combining Figure 1 Come and see, Figure 1 A flow chart of an embodiment of a vehicle behavior recognition method integrating cloud data provided by the present invention includes steps S101 to S102, wherein: In step S101, real-time driving data of a target vehicle is obtained, wherein the real-time driving data includes driving data of surrounding vehicles at the target vehicle's geographical location and traffic cloud data; In step S102, the preprocessed real-time driving data is used as an input of a vehicle behavior recognition model, and the real-time vehicle behavior of the target vehicle is determined based on the output of the vehicle behavior recognition model; The vehicle behavior recognition model is obtained based on the historical driving data and historical vehicle behavior training of the target vehicle.

[0026] During implementation, the real-time driving data of the target vehicle, including the driving data of the surrounding vehicles at the target vehicle's geographical location and the traffic cloud data, can be obtained first. The driving data of the surrounding vehicles at the target vehicle's geographical location can be obtained through the global positioning system, vehicle communication system, vehicle-mounted laser radar and millimeter wave radar on the target vehicle, and the traffic cloud data at the target vehicle's geographical location is obtained from the traffic cloud data platform.

[0027] The real-time driving data of the target vehicle can then be preprocessed, including denoising, filtering, calibration, and normalization operations to ensure the quality and accuracy of the data. The preprocessed real-time driving data is then used as the input of the vehicle behavior recognition model, and the real-time vehicle behavior of the target vehicle is determined through the output of the vehicle behavior recognition model, thereby completing the vehicle behavior recognition of the target vehicle.

[0028] The vehicle behavior recognition model can be obtained through the historical driving data and historical vehicle behavior training of the target vehicle. During training, the historical driving data of the target vehicle can be used as input and the historical vehicle behavior can be used as a label to train the vehicle behavior recognition model.

[0029] The vehicle behavior recognition method integrating cloud data provided by the present invention can be applied to vehicle automatic driving scenarios or vehicle assisted driving scenarios, and the present invention does not make any specific limitation on this.

[0030] Compared with the prior art, the vehicle behavior recognition method integrating cloud data provided in this embodiment does not only focus on a single driving behavior or driving trajectory when performing vehicle behavior recognition, but also combines the driving data of surrounding vehicles at the geographical location of the target vehicle and the traffic cloud data to perform vehicle behavior recognition. While considering the impact of the vehicles around the target vehicle on the target vehicle, it also considers the impact of other factors on the target vehicle in combination with the cloud data, thereby effectively improving the accuracy of vehicle driving behavior recognition.

[0031] Exemplarily, the driving data of surrounding vehicles at the geographical location of the target vehicle includes the position, speed, acceleration and steering angle of surrounding vehicles, and the traffic cloud data at the geographical location of the target vehicle includes traffic accident conditions, weather conditions and traffic flow.

[0032] Specifically, the driving data of surrounding vehicles at the target vehicle's geographical location may include the location, speed, acceleration and steering angle of surrounding vehicles, so as to accurately determine the future driving trajectory of surrounding vehicles and improve the accuracy of target vehicle driving behavior recognition. The traffic cloud data at the target vehicle's geographical location may include traffic accident conditions, weather conditions and traffic flow. These cloud data are also highly correlated with driving behavior, which helps to further improve the accuracy of target vehicle driving behavior recognition.

[0033] Exemplarily, the vehicle behavior recognition model is an LSTM network with an attention mechanism layer added, and the attention mechanism layer is used to assign weights to the input of the vehicle behavior recognition model.

[0034] Specifically, the vehicle behavior recognition model used for vehicle behavior recognition can be a long short-term memory (LSTM) network with an attention mechanism layer added. At the same time, in order to improve the accuracy of vehicle behavior recognition, the present invention also adds an attention mechanism layer to the model for assigning weights to the input of the vehicle behavior recognition model.

[0035] Exemplarily, the assigning weights to the inputs of the vehicle behavior recognition model includes: Mapping the hidden state of the LSTM network based on a fully connected layer; Weights are assigned to the input of the vehicle behavior recognition model based on the hidden state of the mapped LSTM network and the softmax function.

[0036] Specifically, when assigning weights to the input of the vehicle behavior recognition model, the hidden state of the LSTM network can be mapped through the fully connected layer, and then the weights are assigned to the input of the vehicle behavior recognition model according to the hidden state of the mapped LSTM network and the softmax function.

[0037] Exemplarily, the assigning weights to the inputs of the vehicle behavior recognition model includes: The inputs of the vehicle behavior recognition model are weighted based on the following formula:

[0038]

[0039] in, represents the hidden state of the LSTM network after mapping, represents the fully connected layer function, represents the hidden state of the LSTM network, represents the input allocation weight of the vehicle behavior recognition model, Represents the softmax function.

[0040] Specifically, weights may be assigned to the inputs of the vehicle behavior recognition model according to the above formula.

[0041] Exemplarily, the real-time vehicle behavior of the target vehicle includes any one of the following: Accelerate and go straight, drive normally, slow down and go straight, turn left, turn right, or about to stop.

[0042] Specifically, the real-time vehicle behavior of the target vehicle may be any one of accelerating and going straight, driving normally, decelerating and going straight, turning left, turning right, or about to stop.

[0043] Exemplarily, the real-time vehicle behavior of the target vehicle is output from the vehicle behavior recognition model in the form of a binary vector.

[0044] Specifically, the real-time vehicle behavior of the target vehicle can be output from the vehicle behavior recognition model in the form of a binary vector, and the vehicle behavior recognition result can be subsequently visualized by decoding the binary vector to facilitate real-time monitoring or subsequent analysis by a user or system operator.

[0045] The technical solution of the present invention is better described below with reference to a specific embodiment: Combination Figure 2 Come and see, Figure 2 The schematic diagram of the data collection source of the present invention provides an embodiment. First, the image data of the road environment on which the vehicle is traveling can be collected through the vehicle-mounted camera and uploaded to the vehicle-mounted computing center. The surrounding vehicles, pedestrians, and traffic signs can be identified through image analysis and computer vision. At the same time, the timestamp, the location, speed, acceleration, and steering angle of the surrounding vehicles are obtained through the GPS receiver, vehicle-mounted laser radar, and millimeter-wave radar. The data is uploaded to the vehicle-mounted computing center and a visual interface is generated. Then, the vehicle geographic location data and the traffic cloud data platform can be used to obtain the traffic events, weather conditions, and traffic flow of the road on which the vehicle is traveling at the current moment, and the data obtained in the cloud are analyzed and processed to extract key information, such as event type, location, and timestamp. The data is structured and stored in the vehicle database for subsequent query and use.

[0046] The vehicle location is the vehicle's geographic coordinates, with a latitude range of -90-90 and a longitude range of 180-180, and the data acquisition format is (latitude, longitude); traffic flow represents the number of vehicles passing through the current road within a unit time (per hour); the event type is a text string (yes / no), which is encoded as 1 / 0. The traffic accident situation on the current road is sent through the traffic cloud information platform according to GPS positioning and uploaded to the on-board computing unit for storage; the weather data is a text string with three type options: sunny, rainy, and snowy, and the encoding characters are 001, 010, and 100.

[0047] Combination Figure 3 Come and see, Figure 3 This is a schematic diagram of an embodiment of a driving information input source provided by the present invention, that is, the driving information input is mainly provided by surrounding vehicles, vehicle-mounted sensors and cloud traffic data.

[0048] Combination Figure 4 Come and see, Figure 4 The flowchart of an embodiment of the driving behavior recognition model training process provided by the present invention first pre-processes the collected original vehicle driving data, including denoising, filtering, calibration and normalization operations, to ensure the quality and accuracy of the data.

[0049] The vehicle data processing center then divides the data into time units and batch sizes, and assigns corresponding vehicle driving behavior labels (accelerating and going straight, normal driving, slowing and going straight, turning left, turning right, and stopping) to the vehicle position, speed, acceleration, steering angle, weather data, traffic flow information, and road traffic accident data of each time unit, and forms training sets, validation sets, and test sets. Furthermore, cloud information (weather data, traffic flow information, and road traffic conditions), acceleration, speed, and steering angle can be used to label the data set.

[0050] Combination Figure 5 Come and see, Figure 5The present invention provides a flow chart of an embodiment of a vehicle driving behavior recognition process. Based on vehicle driving status data and cloud traffic data, an LSTM neural network is introduced, and a three-dimensional tensor is used as input. The shape is (batch_size, time_steps, input_dim), wherein batch_size represents the number of samples in each training batch, time_steps represents the number of time steps of each sequence data, input_dim represents the number of features of each time step, and each sample contains data of n timestamps, and each time period contains 7 features, namely, vehicle position, speed, acceleration, steering angle, weather data, traffic flow information, and road traffic accidents; the steering angle range is [-90, 90], wherein the value of turning left is negative, the value of turning right is positive, Assume that the vehicle position is (longitude 37, latitude -121), speed is 60km / h, acceleration is , steering angle +30, sunny weather, no accidents, traffic flow of 1000 vehicles / hour and timestamp of 3 seconds, its feature vector is [37, -121, 60, -2, 30, 1, 0, 0, 1, 1000].

[0051] The batch_size is set to 32, and the number of time steps is 30 timestamps, that is, each sequence data contains complete data with a time series, and each timestamp contains 7 feature data, namely vehicle position, speed, acceleration, steering angle, weather data, traffic flow information, and road traffic accidents.

[0052] speed( ) ,in represents the time interval between two laser radar measurements of the vehicle and surrounding vehicles The distance changes within represents the time interval; acceleration is the rate of change of velocity, which can be calculated by the change of velocity and time as acceleration ( ) ,in represents the speed change between two speed measurements, Represents the time interval, and the steering angle is obtained by calculating the heading angle of the target vehicle. The calculation formula is: ,in , are the lateral and longitudinal positions of the vehicle, respectively. , are the lateral and longitudinal velocities of the target vehicle, respectively.

[0053] Weather data, traffic flow information, and traffic accidents ahead are transmitted from the cloud to intelligent connected vehicles by the Car-Road-Cloud Urban Traffic Platform. Weather data includes (sunny, rainy, and snowy days); traffic flow information is transmitted by the on-board vehicle GPS to the Car-Road-Cloud Urban Traffic Platform through the cloud. After the vehicle's real-time geographic location information is sent to the Car-Road-Cloud Urban Traffic Platform, the platform will promptly feedback to the vehicle the real-time traffic flow conditions and traffic accident situations on the current road.

[0054] Add an attention mechanism layer to the vehicle driving behavior model to sort the feature vectors according to the importance of the information. Weight allocation is performed to improve prediction accuracy.

[0055] For each time step t, the hidden state of the LSTM is (for each time step in the output sequence) and the feature vector as the input of the attention mechanism layer; then a fully connected layer is used Mapped to a score; introduce the softmax function to calculate the weight of each feature vector, the formula is as follows:

[0056]

[0057] Collect data at a certain time step t, where the LSTM hidden state in the vehicle driving behavior model =[0.1, 0.2, 0.3], eigenvector Use a linear layer to initialize the weights and bias of the linear layer to [37, -121, 60, -2, 30, 1, 0, 0, 1, 1000], and set the hidden state Mapped to a score, the calculated score is [0.05, 0.1, 0.15, 0.2, 0.25, 0.3, 0.35]. Then the softmax function is used to calculate the weight of each feature vector and convert the score into probability. The formula is:

[0058] The calculated Attention Weight(t)=[0.0246, 0.0302, 0.0371, 0.0456, 0.0561, 0.0689, 0.0840] indicates that at time step t, the driving behavior recognition model calculates the feature vector The importance of each feature in predicting vehicle driving behavior is different. From the case results, the seventh feature has the highest importance at this time step.

[0059] After the speed, acceleration, steering angle, weather data, traffic flow information, and traffic accident information ahead of the target vehicle are input into the neural network of the constructed vehicle driving behavior recognition model for training, the output vehicle driving behavior includes (accelerating and going straight, driving normally, decelerating and going straight, turning left, turning right, and about to stop).

[0060] One-Hot Encoding is used to convert each driving behavior into a binary vector output from the neural network, and a decoding method is used to visualize the recognition results so that users or system operators can conduct real-time monitoring or subsequent analysis. Figure 6 Come and see, Figure 6 A schematic diagram of an embodiment of a vehicle driving behavior visualization interface provided by the present invention.

[0061] The embodiment of the present invention also provides a vehicle behavior recognition device integrating cloud data, Figure 7 Come and see, Figure 7 This is a schematic diagram of the structure of an embodiment of a vehicle behavior recognition device integrating cloud data provided by the present invention. The vehicle behavior recognition device integrating cloud data 700 includes: An acquisition module 701 is used to acquire real-time driving data of a target vehicle, wherein the real-time driving data includes driving data of surrounding vehicles at the target vehicle's geographical location and traffic cloud data; A determination module 702, configured to use the preprocessed real-time driving data as an input of a vehicle behavior recognition model, and determine the real-time vehicle behavior of the target vehicle based on the output of the vehicle behavior recognition model; The vehicle behavior recognition model is obtained based on the historical driving data and historical vehicle behavior training of the target vehicle.

[0062] The specific implementation methods of each module of the vehicle behavior recognition device integrating cloud data can refer to the description of the above-mentioned vehicle behavior recognition method integrating cloud data, and have similar beneficial effects, which will not be repeated here.

[0063] It should be noted that the vehicle behavior recognition device integrating cloud data can be set in a vehicle-mounted device or in a cloud server, and the present invention does not make specific limitations on this.

[0064] The embodiment of the present invention further provides an electronic device, Figure 8 Come and see, Figure 8 This is a structural diagram of an embodiment of an electronic device provided by the present invention. The electronic device 800 includes a processor 801, a memory 802, and a computer program stored in the memory 802 and executable on the processor 801. When the processor 801 executes the program, the vehicle behavior recognition method integrating cloud data as described above is implemented.

[0065] As a preferred embodiment, the electronic device 800 further includes a display 803 for displaying that the processor 801 executes the vehicle behavior recognition method integrating cloud data as described above.

[0066] Exemplarily, the computer program may be divided into one or more modules / units, one or more modules / units are stored in the memory 802, and are executed by the processor 801 to complete the present invention. One or more modules / units may be a series of computer program instruction segments that can complete specific functions, and the instruction segments are used to describe the execution process of the computer program in the electronic device 800. For example, the computer program may be divided into the acquisition module 701 and the determination module 702 in the above embodiment, and the specific functions of each module are as described above, and are not described one by one here.

[0067] The electronic device 800 may be a desktop computer, a notebook, a PDA or a smart phone with an adjustable camera module.

[0068] Among them, the processor 801 may be an integrated circuit chip with signal processing capabilities. The above-mentioned processor 801 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The disclosed methods, steps and logic block diagrams in the embodiments of the present invention can be implemented or executed. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0069] The memory 802 may be, but is not limited to, a random access memory (RAM), a read only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electric erasable programmable read-only memory (EEPROM), etc. The memory 802 is used to store programs, and the processor 801 executes the program after receiving the execution instruction. The process definition method disclosed in any of the embodiments of the present invention may be applied to the processor 801 or implemented by the processor 801.

[0070] The display 803 may be an LCD display screen or an LED display screen, for example, a display screen on a mobile phone.

[0071] Understandably, Figure 8 The structure shown is only a schematic diagram of a structure of the electronic device 800. The electronic device 800 may also include Figure 8 More or fewer components as shown. Figure 8 Each component shown in the figure can be implemented by hardware, software or a combination thereof.

[0072] The electronic device provided according to the above-mentioned embodiment of the present invention can be implemented with reference to the specific description of the vehicle behavior identification method integrating cloud data as described above according to the present invention, and has similar beneficial effects as the vehicle behavior identification method integrating cloud data as described above, which will not be repeated here.

[0073] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the vehicle behavior recognition method integrating cloud data as described above is implemented.

[0074] Generally speaking, the computer instructions for implementing the method of the present invention can be carried by any combination of one or more computer-readable storage media. Non-transitory computer-readable storage media can include any computer-readable media except for the signal itself that is temporarily propagating.

[0075] Computer-readable storage media may be, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more conductors, 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), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium may be any tangible medium that contains or stores a program that may be used by or in conjunction with an instruction execution system, apparatus, or device.

[0076] Computer program codes for performing the operations of the present invention may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages, in particular, Python suitable for neural network computing and platform frameworks based on TensorFlow, PyTorch, etc. may be used. The program code may be executed entirely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0077] Those skilled in the art will appreciate that all or part of the processes of the above-mentioned embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, wherein the computer-readable storage medium is a disk, an optical disk, a read-only storage memory, or a random access memory, etc.

[0078] The present invention discloses a vehicle behavior recognition method and device integrating cloud data. When performing vehicle behavior recognition, the method and device not only focus on a single driving behavior or driving trajectory, but also combine the driving data of surrounding vehicles at the geographical location of the target vehicle and traffic cloud data to perform vehicle behavior recognition. While considering the impact of the vehicles around the target vehicle on the target vehicle, the cloud data is combined to consider the impact of other factors on the target vehicle, thereby effectively improving the accuracy of vehicle driving behavior recognition.

[0079] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A vehicle behavior recognition method integrating cloud data, characterized in that: include: Acquire real-time driving data of the target vehicle, wherein the real-time driving data includes driving data of surrounding vehicles at the geographical location of the target vehicle and traffic cloud data; Using the preprocessed real-time driving data as an input of a vehicle behavior recognition model, and determining the real-time vehicle behavior of the target vehicle based on the output of the vehicle behavior recognition model; The vehicle behavior recognition model is obtained based on the historical driving data and historical vehicle behavior training of the target vehicle.

2. The vehicle behavior recognition method integrating cloud data according to claim 1 is characterized in that: The driving data of the surrounding vehicles at the geographical location of the target vehicle includes the position, speed, acceleration and steering angle of the surrounding vehicles, and the traffic cloud data at the geographical location of the target vehicle includes traffic accident conditions, weather conditions and traffic flow.

3. The vehicle behavior recognition method integrating cloud data according to claim 1 is characterized in that: The vehicle behavior recognition model is an LSTM network with an attention mechanism layer added, and the attention mechanism layer is used to assign weights to the input of the vehicle behavior recognition model.

4. The vehicle behavior recognition method integrating cloud data according to claim 3 is characterized in that: The step of assigning weights to the inputs of the vehicle behavior recognition model comprises: Mapping the hidden state of the LSTM network based on a fully connected layer; Weights are assigned to the input of the vehicle behavior recognition model based on the hidden state of the mapped LSTM network and the softmax function.

5. The vehicle behavior recognition method integrating cloud data according to claim 4 is characterized in that: The step of assigning weights to the inputs of the vehicle behavior recognition model comprises: The inputs of the vehicle behavior recognition model are weighted based on the following formula: in, represents the hidden state of the LSTM network after mapping, represents the fully connected layer function, represents the hidden state of the LSTM network, represents the input allocation weight of the vehicle behavior recognition model, Represents the softmax function.

6. The vehicle behavior recognition method integrating cloud data according to claim 1 is characterized in that: The real-time vehicle behavior of the target vehicle includes any of the following: Accelerate and go straight, drive normally, slow down and go straight, turn left, turn right, or about to stop.

7. The vehicle behavior recognition method integrating cloud data according to claim 6 is characterized in that: The real-time vehicle behavior of the target vehicle is output from the vehicle behavior recognition model in the form of a binary vector.

8. A vehicle behavior recognition device integrating cloud data, characterized in that: include: An acquisition module is used to acquire real-time driving data of a target vehicle, wherein the real-time driving data includes driving data of surrounding vehicles at the geographical location of the target vehicle and traffic cloud data; A determination module, used to use the preprocessed real-time driving data as an input of a vehicle behavior recognition model, and determine the real-time vehicle behavior of the target vehicle based on the output of the vehicle behavior recognition model; The vehicle behavior recognition model is obtained based on the historical driving data and historical vehicle behavior training of the target vehicle.

9. An electronic device, characterized in that: The invention comprises a memory and a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the vehicle behavior recognition method integrating cloud data according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, the vehicle behavior recognition method integrating cloud data as described in any one of claims 1 to 7 is implemented.

Images