Vehicle collision detection method and device and electronic equipment

Abstract

The invention provides a vehicle collision detection method and device and electronic device.The method comprises the steps that current driving data of a vehicle and current state data of a target assembly in the vehicle are obtained, the target assembly comprises at least one of a gear, a vehicle lamp, a vehicle door and a safety belt, data detection is conducted on the current driving data, and a data detection result is obtained; wherein the data detection result includes normal or abnormal, and under the condition that the data detection result is abnormal, collision detection is performed according to the current driving data and the current state data of the target component to obtain a collision detection result; according to the invention, multi-dimensional collision detection is realized, the accuracy and reliability of collision detection are improved, and the probability of misjudgment of collision is reduced, so that the adaptability to complex operation conditions of the vehicle is enhanced, and the probability of misjudgment of collision is further reduced through a dual judgment mechanism of data detection and collision detection.

Description

Technical Field

[0001] This application relates to the field of vehicle collision technology, and in particular to a vehicle collision detection method, device and electronic equipment. Background Technology

[0002] Vehicle collisions are a major risk threatening the lives of occupants throughout a vehicle's lifecycle. Rapid and accurate collision detection and triggering of a rescue response can significantly shorten the critical rescue time and reduce the mortality rate of serious injuries. Therefore, highly reliable and accurate collision detection systems have become a core requirement in the field of automotive safety.

[0003] The relevant technology collects vehicle acceleration through acceleration sensors and uses it as the basis for judging whether a vehicle has collided. This method is difficult to adapt to complex vehicle operating conditions, such as when a vehicle goes over speed bumps, potholes, or brakes suddenly. If the vehicle is judged to have collided based on acceleration information alone in the face of such road interference, it is easy to cause false collision judgments. Summary of the Invention

[0004] This application provides a vehicle collision detection method, device, and electronic device to solve the technical problem that related collision detection technologies are prone to collision misjudgment when dealing with complex road surface interference.

[0005] This application provides a vehicle collision detection method, the method comprising: acquiring current driving data of a vehicle and current state data of a target component in the vehicle, wherein the target component includes at least one of a gear shift, headlights, doors, and seat belts; performing data detection on the current driving data to obtain a data detection result, wherein the data detection result includes normal or abnormal; and, under the condition that the data detection result is abnormal, performing collision detection based on the current driving data and the current state data of the target component to obtain a collision detection result.

[0006] In one embodiment of this application, the current driving data includes acceleration time-series data. Collision detection is performed based on the current driving data and the current state data of the target component to obtain a collision detection result. This includes: determining a hypothetical collision time based on the acceleration time-series data; extracting the state data of the target component after the hypothetical collision time from the current state data of the target component using the hypothetical collision time as a reference; and determining the collision detection result based on the state data of the target component after the hypothetical collision time. By determining the hypothetical collision time based on the acceleration time-series data, the collision critical point can be accurately locked, thereby determining a reasonable collision detection basis. Furthermore, by determining whether a collision has occurred based on the state data of at least one vehicle component, such as gear shift, headlights, doors, and seat belts, after the hypothetical collision time, the reliability of the collision detection can be improved.

[0007] In one embodiment of this application, the target component includes the gear shift, the door, and the seat belt. Determining the collision detection result based on the state data of the target component after the assumed collision time includes: determining the moment when the gear shift is in the parking position based on the state data of the gear shift after the assumed collision time, as the parking position state moment; counting the number of times the door is opened based on the state data of the door after the assumed collision time; determining the seat belt state based on the state data of the seat belt after the assumed collision time; and determining the collision detection result by combining the difference between the parking position moment and the assumed collision time, the number of door openings, and the seat belt state. By comprehensively verifying multiple dimensions including the parking position moment, the number of door openings, and the seat belt state, the accuracy of collision detection is further improved.

[0008] In one embodiment of this application, the target component includes the vehicle headlights, the vehicle door, and the seat belt. Determining the collision detection result based on the state data of the target component after the assumed collision time includes: calculating the duration of the vehicle headlights in hazard light mode based on the state data of the vehicle headlights after the assumed collision time, as the hazard light duration; calculating the number of times the vehicle door was opened based on the state data of the vehicle door after the assumed collision time; determining the seat belt status based on the state data of the seat belt after the assumed collision time; and determining the collision detection result by combining the hazard light duration, the number of times the vehicle door was opened, and the seat belt status. By comprehensively verifying multiple dimensions including the hazard light duration, the number of times the vehicle door was opened, and the seat belt status, the accuracy of collision detection is further improved.

[0009] In one embodiment of this application, data detection of the current driving data includes: normalizing the current driving data to obtain normalized current driving data; reconstructing the data based on the normalized current driving data to obtain reconstructed current driving data; calculating the difference between the normalized current driving data and the reconstructed current driving data; and determining the data detection result based on the calculated difference. By quantifying anomalies through reconstruction error, anomalies can be quickly identified, improving data detection efficiency and real-time performance. Furthermore, by normalizing different types of data in the current driving data, the scale of different types of data is unified, thereby simplifying the computational complexity of data reconstruction, further improving data detection efficiency and real-time performance. Normalization also balances the contribution of all data to the reconstruction error, reducing the possibility of some data being ignored due to differences in units of measurement.

[0010] In one embodiment of this application, data reconstruction based on the normalized current driving data includes: inputting the normalized current driving data into a trained anomaly detection model, wherein the trained anomaly detection model includes an encoder and a decoder; extracting features from the normalized current driving data using the encoder to obtain a feature vector; and mapping the feature vector back to the data space using the decoder to obtain the reconstructed current driving data. By encoding and extracting feature vectors from the current driving data, noise in the current driving data is reduced; and by decoding and reconstructing the current driving data using the feature vectors, abnormal data is effectively eliminated.

[0011] In one embodiment of this application, the target component includes the gear position. Before performing data detection on the current driving data, the method includes: performing gear shift detection on the current state data of the gear position; and performing data detection on the current driving data when it is detected that the gear position has shifted from a drive gear to a park gear. Gear shift detection enables initial screening of the current driving data and the current state data of the target component, further reducing the amount of data processing.

[0012] In one embodiment of this application, after obtaining the collision detection result, the method includes: acquiring image data of the vehicle when the collision detection result indicates a collision exists, wherein the collision detection result includes whether a collision exists or not; and verifying the collision detection result based on the image data to obtain a collision verification result. Verification using image data can further improve the accuracy of collision detection and reduce the probability of false collision detection.

[0013] This application also provides a vehicle collision detection device, the device comprising: a data acquisition module for acquiring current driving data of the vehicle and current state data of target components in the vehicle, wherein the target components include at least one of gear shift, headlights, doors, and seat belts; an information processing module for performing data detection on the current driving data, and, under the condition that the data detection result is abnormal, performing collision detection based on the current driving data and the current state data of the target components to obtain a collision detection result, wherein the data detection result includes normal or abnormal; and a push module for pushing collision alarm information, the collision alarm information being generated based on the collision detection result indicating a collision, wherein the collision detection result includes whether a collision exists or no collision is detected.

[0014] This application also provides an electronic device, the electronic device comprising: one or more processors; and a storage device for storing one or more programs, which, when executed by the one or more processors, cause the electronic device to implement the vehicle collision detection method as described above.

[0015] This application has at least the following beneficial effects: By using current driving data and the current state data of the target component as the basis for collision detection, multi-dimensional collision detection based on driving data and user behavior is achieved, improving the accuracy and reliability of collision detection, reducing the probability of false collisions, and thus enhancing adaptability to complex vehicle operating conditions. Simultaneously, by performing data detection on the current driving data to identify anomalies before collision detection, the amount of data processing required for collision detection is significantly reduced, improving the efficiency of collision detection. Furthermore, the dual judgment mechanism of data detection and collision detection further reduces the probability of false collisions. Attached Figure Description

[0016] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. It is obvious that the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0017] In the attached diagram:

[0018] Figure 1 A schematic diagram illustrating the implementation environment of a vehicle collision detection method according to an embodiment of this application; Figure 2 A schematic flowchart of a vehicle collision detection method provided in an embodiment of this application; Figure 3 for Figure 2 A flowchart illustrating step S230 in the process; Figure 4 A schematic diagram of a vehicle collision detection process provided for an exemplary embodiment of this application; Figure 5 A block diagram of a vehicle collision detection device provided in one embodiment of this application; Figure 6 This is a schematic diagram of the structure of a computer system for an electronic device provided in an embodiment of this application. Detailed Implementation

[0019] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments. Various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0020] The illustrations provided in the following embodiments are only schematic representations of the basic concept of this application. The drawings only show the components related to this application and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0021] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the present application. However, it will be apparent to those skilled in the art that embodiments of the present application may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the present application.

[0022] The embodiments of this application respectively propose a vehicle collision detection method, a vehicle collision detection device, an electronic device, a computer-readable storage medium, and a computer program product, which will be described in detail below.

[0023] Please see Figure 1 , Figure 1 This is a schematic diagram illustrating the implementation environment of a vehicle collision detection method according to an embodiment of this application, as shown below. Figure 1 As shown, the implementation environment can include vehicle 110 and server 120. Vehicle 110 can be a smart gasoline vehicle, smart electric vehicle, smart hybrid vehicle, or other types of smart vehicles. Server 120 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms. No restrictions are placed here. Vehicle 110 is equipped with various types of sensors, which can collect current driving data and the current status data of target components and upload them to server 120 for collision detection.

[0024] Schematic illustration: Server 120 acquires the current driving data of vehicle 110 and the current state data of target components in vehicle 110, wherein the target components include at least one of gear shift, headlights, doors, and seat belts; performs data detection on the current driving data to obtain data detection results, wherein the data detection results include normal or abnormal; under the condition that the data detection result is abnormal, collision detection is performed based on the current driving data and the current state data of the target components to obtain a collision detection result. It can be seen that the technical solution of this application embodiment, by using the current driving data and the current state data of the target components as the basis for collision detection, realizes multi-dimensional collision detection of driving data and user behavior, improves the accuracy and reliability of collision detection, reduces the probability of false collision judgment, and thus enhances the adaptability to complex vehicle operating conditions. Simultaneously, by performing data detection on the current driving data to identify anomalies before collision detection, the data processing volume of collision detection is significantly reduced, the efficiency of collision detection is improved, and the dual judgment mechanism of data detection and collision detection further reduces the probability of false collision judgment.

[0025] The vehicle collision detection method provided in this application embodiment can be specifically executed by the server 120, and correspondingly, the vehicle collision detection device can be set in the server 120.

[0026] Please see Figure 2 , Figure 2 This is a flowchart illustrating a vehicle collision detection method according to an embodiment of this application. This vehicle collision detection method can be applied to... Figure 1 The implementation environment shown is specifically executed by server 120 within that implementation environment. It should be understood that this vehicle collision detection method can also be applied to the vehicle itself and executed by electronic devices such as processors and processing chips on the vehicle side. This vehicle collision detection method can also be applied to other exemplary implementation environments and executed by devices in other implementation environments. This embodiment does not limit the implementation environment to which this vehicle collision detection method is applicable. Figure 2 As shown, in an exemplary embodiment, the vehicle collision detection method includes at least steps S210 to S230, which are described in detail below: Step S210: Obtain the current driving data of the vehicle and the current status data of the target components in the vehicle.

[0027] Step S220: Perform data detection on the current driving data to obtain the data detection results.

[0028] Step S230: If the data detection result is abnormal, perform collision detection based on the current driving data and the current state data of the target component to obtain the collision detection result.

[0029] In step S210, the current driving data refers to the real-time, dynamic driving data of the vehicle during its current driving process, including at least one of the following: vehicle speed, steering wheel angle, braking torque, braking status, accelerator pedal opening, roll rate, longitudinal acceleration, lateral acceleration, vertical acceleration, pitch angle, roll angle, yaw rate, left front wheel speed, right front wheel speed, left rear wheel speed, and right rear wheel speed. The target component refers to a pre-specified vehicle component, including at least one of the following: gear position, headlights, doors, and seatbelts. The current state data of the target component refers to the real-time, dynamic state data of the target component during the vehicle's current driving process, such as at least one of the following: gear position, headlight on / off status, door on / off status, and seatbelt locked / unlocked status at different times, used to represent user behavior. The current driving data and the current state data of the target component are at the same time or time period.

[0030] In some embodiments, sensors can collect vehicle speed, steering wheel angle, braking torque, braking status, accelerator pedal opening, roll rate, longitudinal acceleration, lateral acceleration, vertical acceleration, pitch angle, roll angle, yaw rate, left front wheel speed, right front wheel speed, left rear wheel speed, and right rear wheel speed within a preset duration during the current driving process, as current driving data.

[0031] In some embodiments, sensors can be used to collect current state data of the vehicle's gear position, doors, and seat belts within a preset time period during the current driving process, at a preset collection frequency, as current state data of the target components; In some embodiments, sensors can be used to collect current state data of the vehicle's lights, doors, and seat belts within a preset time period during the current driving process, at a preset collection frequency, as current state data of the target components. In some embodiments, sensors can be used to collect current state data of the vehicle's gear position, headlights, doors, and seat belts within a preset time period during the current driving process, as current state data of the target components. In some embodiments, the preset sampling frequency can be 1Hz, that is, once per second.

[0032] In some embodiments, the preset duration can be 60 seconds.

[0033] In step S220, the data detection result includes normal or abnormal. During normal vehicle operation, the changes in vehicle driving data over a period of time are smooth. Therefore, by extracting the change features of the current driving data, it is possible to identify whether there are any anomalies in the current driving data based on these features.

[0034] In some embodiments, various types of data in the current driving data can be fitted, and abnormal data can be identified based on the fitted curve. If the peak or trough of a type of data exceeds the corresponding threshold, the data detection result can be determined to be abnormal; otherwise, the data detection result can be determined to be normal.

[0035] In some embodiments, if the change in the same type of data in the current driving data exceeds the corresponding preset change amount between two adjacent moments, the data detection result can be determined to be abnormal; otherwise, the data detection result can be determined to be normal. For example, if the change in steering wheel angle between two adjacent moments exceeds the corresponding preset change amount, the steering wheel angle can be considered abnormal; if the difference in braking torque between two adjacent moments exceeds the corresponding preset change amount, the braking torque can be considered abnormal; and so on. If the difference in acceleration in a certain direction between two adjacent moments exceeds the corresponding preset change amount, the acceleration in that direction can be considered abnormal. When any type of data is abnormal, the data detection result can be determined to be abnormal; when no type of data is abnormal, the data detection result can be determined to be normal.

[0036] In some embodiments, if the difference between some or all of the data in the current driving data and the reference data is greater than a preset threshold, the data detection result can be determined to be abnormal; otherwise, the data detection result can be determined to be normal. The reference data can be preset or determined based on the current driving data.

[0037] The ideal signal acquisition frequency for vehicle collision detection is 20Hz~50Hz, while the commonly used acquisition frequency for driving data is 1Hz. Therefore, compared to the ideal acquisition frequency, the commonly used acquisition frequency for driving data is relatively low. Thus, for driving data acquired at a lower acquisition frequency, the anomalies detected by the data may not be actual collisions.

[0038] In step S230, when a collision occurs, the occupants usually get out of the vehicle to check. Therefore, by using the current driving data as a basis and combining the current state data of at least one vehicle component, such as the gear shift, headlights, doors, and seat belts, to perform collision detection, the problem of low data collection frequency can be compensated for, the accuracy of collision detection can be improved, and the false collision judgment can be reduced.

[0039] In some embodiments, no processing is required when the data detection result is normal. When the data detection result is abnormal, it can be determined whether a collision has occurred based on the acceleration data in the current driving data, and secondary verification can be performed based on the current state data of the target component.

[0040] In one embodiment of this application, before step S220, the method includes: performing gear shift detection on the current state data of the gear, wherein the target component includes the gear; and performing data detection on the current driving data when a shift from drive to park is detected. Gear shift detection enables initial screening of the current driving data and the current state data of the target component, further reducing the amount of data processing.

[0041] In some embodiments, Flink (a stream processing framework) can be used as a real-time consumption computing engine. The real-time consumption computing engine performs real-time calculations on the data generated by the vehicle (including current driving data and current state data of target components). When the real-time consumption computing engine collects a preset amount of data generated by the vehicle, it performs gear shift detection on the data. If there is a gear shift event in the data from D (Drive) to P (Park), it further checks whether there are any anomalies in the current driving data. This can significantly reduce the amount of computation, improve computational efficiency, and thus ensure the real-time nature of the data.

[0042] In one embodiment of this application, step S220 includes: normalizing the current driving data to obtain normalized current driving data; reconstructing the data based on the normalized current driving data to obtain reconstructed current driving data; calculating the difference between the normalized current driving data and the reconstructed current driving data; and determining the data detection result based on the calculated difference.

[0043] In this embodiment, data reconstruction serves to reduce noise. When anomalies exist in the current driving data, data reconstruction can remove these anomalies. This results in a significant difference between the current driving data before and after reconstruction, allowing the determination of whether anomalies existed in the original data based on the difference. Therefore, quantifying anomalies through reconstruction error enables rapid anomaly identification, improving data detection efficiency and real-time performance. Furthermore, normalizing different types of data in the current driving data unifies the scale of different data types, simplifying the computational complexity of data reconstruction and further improving data detection efficiency and real-time performance. Normalization also balances the contribution of all data to the reconstruction error, reducing the possibility of some data being ignored due to differences in units of measurement.

[0044] In some embodiments, the current driving data can be reconstructed using a neural network model, thereby improving the efficiency of data reconstruction.

[0045] In some embodiments, when the calculated difference is greater than a preset threshold, it indicates that the current driving data is abnormal, and the data detection result can be determined to be abnormal; conversely, the data detection result can be determined to be normal. For example, the preset threshold can be 0.001 or other values, which are not limited here.

[0046] In one embodiment of this application, data reconstruction based on normalized current driving data includes: inputting the normalized current driving data into a trained anomaly detection model, wherein the trained anomaly detection model includes an encoder and a decoder; extracting features from the normalized current driving data through the encoder to obtain a feature vector; and mapping the feature vector back to the data space through the decoder to obtain the reconstructed current driving data.

[0047] In this embodiment, the normalized current driving data is encoded by an encoder, mapping the normalized data from a high-dimensional space to a low-dimensional space. Key features are retained as feature vectors, while redundant and noisy data are discarded, thereby eliminating abnormal data. The feature vectors are then decoded by a decoder, mapping them back from the low-dimensional space to the high-dimensional space, i.e., the data space, to obtain the reconstructed current driving data. By reducing noise through encoding and completing data reconstruction through decoding, abnormal data is effectively eliminated.

[0048] The data space can be the original representation space of the normalized current driving data, i.e., the original data space, or a data space that approximates the original data space. The trained anomaly detection model can be a supervised anomaly detection model, an unsupervised anomaly detection model, or a semi-supervised anomaly detection model; there are no restrictions here.

[0049] In real-world scenarios, vehicle collision data is relatively scarce. Supervised training of anomaly detection models suffers from a severely uneven distribution of positive and negative samples. Therefore, unsupervised training of the anomaly detection model can be performed using the massive historical driving data generated by the Internet of Vehicles (IoV), resulting in a trained anomaly detection model, also known as an unsupervised anomaly detection model. Historical driving data refers to the dynamic driving data of a vehicle during its historical driving process.

[0050] In some embodiments, the training and application process of an unsupervised anomaly detection model is as follows: Collect massive amounts of historical driving data from different vehicles that are in normal operation. The data can be in tabular form. Please refer to Table 1, which is a schematic table of historical driving data provided by an exemplary embodiment of this application. As shown in Table 1, all historical driving data are real values, including vehicle speed, steering wheel angle, braking torque, braking status, accelerator pedal opening, roll rate, longitudinal acceleration, lateral acceleration, vertical acceleration, pitch angle, roll angle, yaw rate, left front wheel speed, right front wheel speed, left rear wheel speed, and right rear wheel speed. Each piece of historical driving data also has a corresponding timestamp and vehicle number.

[0051] Table 1

[0052] Using a sliding window approach, historical driving data is grouped into a single data record every 60 seconds. All data records are preprocessed according to a preset data format, which can be (m, 60, 16), where m represents the total number of data records, 60 represents the time step, and 16 represents the number of various signals such as vehicle speed, steering wheel angle, braking torque, and longitudinal acceleration in the driving data.

[0053] The preprocessed historical driving data is normalized so that all values ​​are between 0 and 1, ensuring the stability of subsequent model training.

[0054] A network with an encoder-decoder structure, such as TimesNet (a deep learning-based time series prediction model), can be used as the base model. Normalized historical driving data is used as the input of the model. The MSELoss (Mean Squared Error Loss) of the input and output is calculated, and the base model is trained through the backpropagation algorithm to obtain an unsupervised anomaly detection model. The data input to the model is in the form of (batch, 60, 16), and the data output of the model after a series of calculations is also in the form of (batch, 60, 16).

[0055] After the model training is completed, the unsupervised anomaly detection model is deployed. After deployment, a normalized current driving data (1, 60, 16) is input into the unsupervised anomaly detection model so that the unsupervised anomaly detection model outputs reconstructed current driving data. The error value (i.e., difference) loss_value between the input and output is calculated. When the error value loss_value is greater than the preset threshold, it indicates that there is an anomaly in this data and the abnormal data detection result is returned; otherwise, the normal data detection result is returned.

[0056] Please see Figure 3 , Figure 3 for Figure 2The flowchart of step S230 is shown below. Figure 3 As shown, step S230 includes: Step S231: Determine the assumed collision time based on acceleration time series data; Step S232: Based on the assumed collision time, extract the target component's state data after the assumed collision time from the target component's current state data; Step S233: Determine the collision detection result based on the state data of the target component after the assumed collision time.

[0057] By determining the assumed collision time based on acceleration time series data, the collision critical point can be accurately locked, thereby determining a reasonable basis for collision detection.

[0058] In step S231, the assumed collision time refers to the assumed moment when the vehicles collide. The current driving data includes acceleration time-series data, which includes at least one of longitudinal acceleration time-series data, lateral acceleration time-series data, and vertical acceleration time-series data. The longitudinal acceleration time-series data refers to the longitudinal acceleration at multiple consecutive moments within a preset time period; the lateral acceleration time-series data refers to the lateral acceleration at multiple consecutive moments within a preset time period; and the vertical acceleration time-series data refers to the vertical acceleration at multiple consecutive moments within a preset time period.

[0059] During normal vehicle operation, acceleration tends to stabilize or change slowly. For example, when a vehicle travels at a constant speed, the longitudinal acceleration is close to zero. During normal acceleration or deceleration, the acceleration is positive or negative, but the rate of change is usually small. When a collision occurs, the impact force causes the acceleration to change drastically in a very short time (usually on the order of milliseconds). In a frontal collision, the longitudinal acceleration changes drastically, while in a side collision, the lateral acceleration increases significantly. A collision may also cause a brief change in vertical acceleration. Therefore, the moment when the amplitude of longitudinal, lateral, or vertical acceleration abruptly changes (i.e., the amplitude exceeds a preset value) can be identified based on the characteristics of the changes in longitudinal, lateral, and vertical acceleration, and used as the hypothetical collision moment.

[0060] In some embodiments, step S231 includes: selecting the minimum longitudinal acceleration from the longitudinal acceleration time series data, and taking the moment of the minimum longitudinal acceleration as the assumed collision moment, wherein the acceleration time series data includes longitudinal acceleration time series data.

[0061] In step S232, the current state data of the target component includes the state information of the target component at multiple consecutive moments within a preset time period. The state information at different moments has a corresponding timestamp, and the state data of the target component after the assumed collision moment can be filtered out according to the timestamp.

[0062] In step S233, the state data of the target component after the assumed collision time includes at least one of the state data of the gear position, the headlights, the doors, and the seat belts after the assumed collision time. This data can directly or indirectly reflect the behavior of the occupants after a vehicle collision. Therefore, determining whether a collision has occurred based on the state data of at least one of the vehicle components (gear position, headlights, doors, and seat belts) after the assumed collision time can improve the reliability of collision detection.

[0063] In one embodiment of this application, step S233 includes: determining the moment when the gear is in the parking gear based on the state data of the gear after the assumed collision time, as the parking gear state moment; counting the number of times the doors are opened based on the state data of the doors after the assumed collision time; determining the seat belt state based on the state data of the seat belt after the assumed collision time; and determining the collision detection result by combining the difference between the parking gear time and the assumed collision time, the number of door openings, and the seat belt state.

[0064] In this embodiment, the target components include a gear shifter, a door, and a seatbelt. The moment the gear shifter is in the parking position refers to the moment when the gear shifter first enters the parking position after the assumed collision. The seatbelt status includes whether it is unfastened or locked. By comprehensively verifying multiple dimensions such as the parking position moment, the number of door openings, and the seatbelt status, the accuracy of collision detection is further improved.

[0065] In some embodiments, the collision detection result is determined by combining the difference between the parking position time and the assumed collision time, the number of door openings, and the seat belt status. This includes: determining that a collision exists if the difference between the parking position time and the assumed collision time, the number of door openings, and the seat belt status meet the first behavior rule; and determining that no collision was detected if the difference between the parking position time and the assumed collision time, the number of door openings, and the seat belt status do not meet the first behavior rule. The first behavior rule includes that the difference between the parking position time and the assumed collision time is less than a preset time difference threshold, the number of door openings is greater than a preset number, and the seat belt status is in an unfastened state.

[0066] For example, the preset time difference threshold can be 10 seconds, and the preset number of times can be 0; neither is restricted here.

[0067] In another embodiment of this application, step S233 includes: calculating the duration of the vehicle lights in hazard light mode based on the state data of the vehicle lights after the assumed collision time, and using this as the hazard light duration; calculating the number of times the vehicle doors are opened based on the state data of the vehicle doors after the assumed collision time; determining the seat belt status based on the seat belt status data after the assumed collision time; and determining the collision detection result by combining the hazard light duration, the number of times the vehicle doors are opened, and the seat belt status.

[0068] In this embodiment, the target components include headlights, doors, and seat belts. Headlights in hazard light mode refers to the vehicle's turn signals (left and right turn signals) flashing simultaneously at a high frequency. By comprehensively verifying multiple dimensions, including hazard light duration, door opening frequency, and seat belt status, the accuracy of collision detection is further improved.

[0069] In some embodiments, the collision detection result is determined by combining the hazard light duration, the number of door openings, and the seat belt status, including: determining that a collision exists when the hazard light duration, the number of door openings, and the seat belt status meet the second behavior rule; and determining that no collision is detected when the hazard light duration, the number of door openings, and the seat belt status do not meet the second behavior rule. The second behavior rule includes that the hazard light duration is greater than a preset hazard light duration threshold, the number of door openings is greater than a preset number, and the seat belt status is in an unfastened state.

[0070] For example, the preset dual-flash duration threshold can be 30 seconds, or other duration values, which are not limited here.

[0071] In one embodiment of this application, after step S230, the method includes: acquiring image data of the vehicle under the condition that the collision detection result is that a collision exists, wherein the collision detection result includes whether a collision exists or no collision is detected; verifying the collision detection result based on the image data to obtain a collision verification result.

[0072] In this embodiment, no processing is required when the collision detection result indicates no collision was detected. When the collision detection result indicates a collision exists, the collision detection result can be output to downstream users for further processing, such as generating collision alarm information based on the collision detection result. Furthermore, image data such as video and pictures can be obtained based on the assumed collision time, and the collision detection result can be verified, which can further improve the accuracy of collision detection and reduce the probability of false collision detection.

[0073] In some embodiments, verifying the collision detection result based on image data includes: determining the collision direction based on acceleration time-series data at the assumed collision time; determining the target detection region in the image data at the assumed collision time based on the collision direction; detecting whether a target object exists in the target detection region using a target detection model, where the target object includes vehicles, pedestrians, animals, obstacles, etc.; if a target object exists, determining whether the target object is a collision object based on the distance between the target object's position and the vehicle's position, or the area overlap between the target object and the vehicle; if the target object is a collision object, then the collision verification result is determined to be a real collision; otherwise, the collision verification result is determined to be a collision misjudgment.

[0074] In some embodiments, when the collision verification result is a false collision, the current driving data and the current state data of the target component corresponding to the false collision can be collected as target vehicle signal data. The behavior rules can then be corrected based on the target vehicle signal data, such as by adding or subtracting behavior rules, or adjusting preset hazard light duration thresholds or preset time difference thresholds. Through iterative optimization of the behavior rules, the accuracy of collision detection can be further improved.

[0075] Please see Figure 4 , Figure 4 Provided for an exemplary embodiment of this application Figure 4 A schematic diagram of a vehicle collision detection process provided in an exemplary embodiment of this application is shown below. Figure 4 As shown, vehicle collision detection comprises four parts. The first part is unsupervised model training. Driving data is acquired through vehicle signal acquisition and stored in a data warehouse to obtain historical driving data for different vehicles. An unsupervised anomaly detection model is trained based on this historical data and then deployed. The second part is a real-time computing program. Real-time driving data and target component status data are acquired through vehicle signal acquisition and preprocessed using a sliding window to obtain window signal data containing current driving data and the current status data of the target component. If the current status data of the target component shows a shift from D to P gear, the system performs a real-time calculation. The data undergoes data detection, and when an anomaly is detected, it proceeds to the third part for further processing. The third part involves expert experience judgment. For vehicles with anomalies in the current driving data, the moment of minimum longitudinal acceleration in the current driving data is obtained as the assumed collision moment. Behavioral rules are retrieved from the expert experience base, and the state data of the target component after the assumed collision moment is compared with the behavioral rules to determine whether a collision has occurred. If a collision is determined, the result is output and transferred to the fourth part for further processing. The fourth part involves expert experience optimization. For vehicles that have experienced collisions, video verification is used to determine whether the collision was real. If it is determined that the collision was not real, i.e., a false collision judgment, the behavioral rules in the expert experience base are optimized.

[0076] For specific details regarding the vehicle collision detection process, please refer to the descriptions in the aforementioned embodiments, which will not be repeated here. The technical solution of this embodiment can detect whether a vehicle is in collision in real time, improving vehicle rescue efficiency; it uses an unsupervised training method to train the basic model, effectively utilizing the massive amount of signal data generated by the vehicle, thereby solving the problem of severely uneven distribution of positive and negative samples in vehicle collision scenarios; during the collision detection process, an anomaly detection model is used to determine whether there are anomalies in the current driving data. Based on the detection of anomalies, the determination of whether a collision has occurred is further refined based on behavioral rules. This two-stage detection effectively solves the problem of low signal acquisition frequency and improves collision detection accuracy; before data detection, data filtering is performed through gear shift detection, which can significantly reduce the amount of data processing and improve computational efficiency; after collision detection, video data is used to verify the authenticity of the collision, and behavioral rules are optimized after a false collision judgment, further improving the collision detection accuracy.

[0077] Please see Figure 5 , Figure 5 This is a block diagram of a vehicle collision detection device according to an embodiment of this application. This device can be applied to… Figure 1 The implementation environment shown is specifically configured in server 120. This device can also be applied to other exemplary implementation environments and specifically configured in other devices. This embodiment does not limit the implementation environment to which this device is applicable.

[0078] like Figure 5 As shown, the exemplary vehicle collision detection device includes: a data acquisition module 510, used to acquire the current driving data of the vehicle and the current state data of target components in the vehicle, wherein the target components include at least one of gear shift, headlights, doors, and seat belts; an information processing module 520, used to perform data detection on the current driving data, and under the condition that the data detection result is abnormal, to perform collision detection based on the current driving data and the current state data of the target components to obtain a collision detection result, wherein the data detection result includes normal or abnormal; and a push module 530, used to push collision alarm information, the collision alarm information being generated based on the collision detection result indicating that a collision exists, wherein the collision detection result includes whether a collision exists or no collision is detected.

[0079] In some embodiments, the data acquisition module 510 can be various types of sensors inside the vehicle.

[0080] In other embodiments, the data acquisition module 510 may be a hardware device for collecting data from various sensors within the vehicle.

[0081] In other embodiments, the data acquisition module 510 includes various types of sensors in the vehicle, as well as hardware devices for collecting data from the sensors in the vehicle.

[0082] In some embodiments, the information processing module 520 may be a computer, computing cluster, microcomputer, embedded computer, neural network computer, processor, etc., or a microprocessor or chip such as MCU (Microcontroller Unit) or ECU (Electronic Control Unit), or a physical server, server cluster, cloud server, etc.

[0083] In some embodiments, the push module 530 may be an in-vehicle networking module, an in-vehicle networking control unit, an in-vehicle infotainment system, a physical server, a server cluster, a cloud server, etc.

[0084] In some embodiments, the information processing module 520 may be configured in the vehicle's collision sensor or airbag sensor.

[0085] In some embodiments, the data acquisition module 510 can be configured on the vehicle side, and the information processing module 520 and the push module 530 can be configured in the cloud.

[0086] It should be noted that the vehicle collision detection device and the vehicle collision detection method provided in the above embodiments belong to the same concept. The specific operation methods of each module and unit have been described in detail in the method embodiments and will not be repeated here. In practical applications, the vehicle collision detection device provided in the above embodiments 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. This is not a limitation here.

[0087] In one embodiment of this application, an electronic device is also provided, including: one or more processors; and a storage device for storing one or more programs, which, when executed by one or more processors, cause the electronic device to implement the vehicle collision detection method provided in the above embodiments.

[0088] Please see Figure 6 , Figure 6 This is a schematic diagram of the structure of a computer system for an electronic device provided in an embodiment of this application. Figure 6 The computer system 600 of the electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0089] like Figure 6As shown, the computer system 600 includes a central processing unit 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory 602 or a program loaded from a storage section 608 into a random access memory 603, such as performing the methods described in the above embodiments. The random access memory 603 also stores various programs and data required for system operation. The central processing unit 601, the read-only memory 602, and the random access memory 603 are interconnected via a bus 604. An input / output interface 605 is also connected to the bus 604.

[0090] The following components are connected to the input / output interface 605: an input section 606 including a keyboard, mouse, etc.; an output section 607 including CRT (Cathode Ray Tube), LCD (Liquid Crystal Display), and speakers, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN (Local Area Network) card and a modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the input / output interface 605 as needed. A removable medium 611, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 610 as needed so that computer programs read from it can be installed into the storage section 608 as needed.

[0091] Specifically, according to embodiments of this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program including a computer program for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 609, and / or installed from removable medium 611. When the computer program is executed by central processing unit 601, it performs the various functions defined in the apparatus of this application.

[0092] The computer-readable medium shown in the embodiments of this application may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, RAM (Random Access Memory), ROM (Read Only Memory), EPROM (Erasable Programmable Read Only Memory), flash memory, optical fiber, CD-ROM (Compact Disc Read-Only Memory), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying a computer-readable computer program. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The computer program contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to wireless, wired, etc., or any suitable combination thereof.

[0093] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of this application. Each block in the flowchart or block diagram may represent a module, segment, or portion of code, which includes one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagram or flowchart, and combinations of blocks in the block diagram or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0094] Another aspect of this application provides a computer-readable storage medium storing a computer program thereon, which, when executed by a computer's processor, causes the computer to perform the vehicle collision detection methods provided in the various embodiments described above. This computer-readable storage medium may be included in the electronic devices described in the above embodiments, or it may exist independently and not incorporated into the electronic devices.

[0095] Another aspect of this application provides a computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the vehicle collision detection method provided in the various embodiments described above.

[0096] The above embodiments are merely illustrative of the principles and effects of this application and are not intended to limit this application. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of this application. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in this application should still be covered by the claims of this application.

Claims

1. A vehicle collision detection method, characterized in that, The method includes: Acquire the vehicle's current driving data and the current status data of target components in the vehicle, wherein the target components include at least one of gear shift, headlights, doors, and seat belts; The current driving data is subjected to data detection to obtain data detection results, wherein the data detection results include normal or abnormal; If the data detection result is abnormal, collision detection is performed based on the current driving data and the current state data of the target component to obtain the collision detection result.

2. The vehicle collision detection method according to claim 1, characterized in that, The current driving data includes acceleration time-series data. Collision detection is performed based on the current driving data and the current state data of the target component to obtain collision detection results, including: The assumed collision time is determined based on the acceleration time series data; Based on the assumed collision time, extract the state data of the target component after the assumed collision time from the current state data of the target component; The collision detection result is determined based on the state data of the target component after the assumed collision time.

3. The vehicle collision detection method according to claim 2, characterized in that, The target components include the gear shift, the door, and the seatbelt. Based on the state data of the target components after the assumed collision moment, the collision detection result is determined, including: Based on the state data of the gear position after the assumed collision time, the moment when the gear position is in parking gear is determined as the parking gear state moment; Based on the state data of the vehicle door after the assumed collision time, count the number of times the vehicle door was opened; The seat belt status is determined based on the seat belt status data after the assumed collision time; The collision detection result is determined by combining the difference between the parking position time and the assumed collision time, the number of times the door was opened, and the seat belt status.

4. The vehicle collision detection method according to claim 2, characterized in that, The target components include the vehicle lights, the vehicle doors, and the seat belts. Based on the state data of the target components after the assumed collision time, the collision detection result is determined, including: Based on the state data of the vehicle lights after the assumed collision time, the duration for which the vehicle lights are in hazard flashing mode is calculated and used as the hazard flashing duration. Based on the state data of the vehicle door after the assumed collision time, count the number of times the vehicle door was opened; The seat belt status is determined based on the seat belt status data after the assumed collision time; The collision detection result is determined by combining the duration of the hazard lights, the number of times the vehicle door was opened, and the seat belt status.

5. The vehicle collision detection method according to claim 1, characterized in that, Data detection is performed on the current driving data, including: The current driving data is normalized to obtain normalized current driving data; Based on the normalized current driving data, data reconstruction is performed to obtain the reconstructed current driving data; The difference between the normalized current driving data and the reconstructed current driving data is calculated, and the data detection result is determined based on the calculated difference.

6. The vehicle collision detection method according to claim 5, characterized in that, Data reconstruction is performed based on the normalized current driving data, including: The normalized current driving data is input into the trained anomaly detection model, wherein the trained anomaly detection model includes an encoder and a decoder; The encoder extracts features from the normalized current driving data to obtain a feature vector; The decoder maps the feature vector back to the data space to obtain the reconstructed current driving data.

7. The vehicle collision detection method according to claim 1, characterized in that, The target component includes the gear position. Before performing data detection on the current driving data, the method includes: Perform gear shift detection on the current status data of the gear position; When the gear is detected to have switched from drive to park, the current driving data is detected.

8. The vehicle collision detection method according to claim 1, characterized in that, After obtaining the collision detection results, the method includes: If the collision detection result indicates that a collision exists, image data of the vehicle is acquired, wherein the collision detection result includes whether a collision exists or no collision is detected; The collision detection results are verified based on the image data to obtain the collision verification results.

9. A vehicle collision detection device, characterized in that, The device includes: The data acquisition module is used to acquire the current driving data of the vehicle and the current status data of target components in the vehicle, wherein the target components include at least one of gear position, headlights, doors and seat belts; The information processing module is used to perform data detection on the current driving data. If the data detection result is abnormal, collision detection is performed based on the current driving data and the current state data of the target component to obtain a collision detection result. The data detection result includes normal or abnormal. The push module is used to push collision alarm information. The collision alarm information is generated based on the collision detection result indicating that a collision exists. The collision detection result includes whether a collision exists or no collision is detected.

10. An electronic device, characterized in that, The electronic device includes: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the electronic device to implement the vehicle collision detection method as described in any one of claims 1-8.

Images