Trajectory-based method and device for detecting traffic accident of vehicle and storage medium

By acquiring and preprocessing vehicle trajectory data in real time, extracting speed and position characteristics, and using accident detection models, timely prediction of traffic accidents is achieved, solving the problem of lag in accident discovery in the existing technology, and improving the efficiency of accident handling .

CN118887809APending Publication Date: 2024-11-01BEIJING HOLIDAY SUNSHINE GLOBAL TRAVEL AGENCY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411307868.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-19
Publication Date
2024-11-01

AI Technical Summary

Technical Problem

In the prior art, the discovery of traffic accidents has a certain lag, resulting in the escalation of the accident or the missed optimal rescue time.

Method used

By obtaining the real-time driving trajectory data of the vehicle to be detected, preprocessing is performed to extract speed characteristics and position characteristics, and input these characteristics into the pre-trained accident detection model to predict whether a vehicle has a traffic accident in real time.

Benefits of technology

It realizes the first accurate reflection of whether a vehicle has an accident, reduces the influence of human subjective factors, and solves the lag problem of traffic accident discovery.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118887809A_ABST
    Figure CN118887809A_ABST
Patent Text Reader

Abstract

The invention relates to a track-based vehicle traffic accident detection method and device and a storage medium, and is applied to the technical field of traffic safety, and the method comprises the steps: obtaining the real-time driving track data of a to-be-detected vehicle, and extracting the speed characteristics and position characteristics of the to-be-detected vehicle based on the driving track data; inputting the speed feature and the position feature of the to-be-detected vehicle into a pre-trained accident detection model, and outputting whether the to-be-detected vehicle has a traffic accident through the accident detection model; whether the to-be-detected vehicle has the traffic accident or not is predicted through the real-time driving track data of the to-be-detected vehicle, whether the vehicle has the accident or not can be accurately reflected at the first time, and compared with the prior art that a party or a passerby dials an alarm call after the accident occurs, the influence of human subjective factors is reduced, and the safety of the vehicle is improved. The problem that the discovery of traffic accidents has hysteresis is solved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of traffic safety technology, and in particular to a method, device and storage medium for detecting a traffic accident caused by a vehicle based on trajectory. Background Art

[0002] With the continuous development and popularization of Internet technology, online ride-hailing has become one of the important ways for people to travel. However, due to the large number of online ride-hailing drivers and the uncertainty of driving behavior, the incidence of online ride-hailing traffic accidents is also relatively high. However, in the existing technology, the relevant departments can only arrive at the scene to handle the accident after the parties or passers-by report the accident. This leads to the efficiency of accident handling being determined by the time of reporting the accident. In remote areas or when the parties have become comatose, it often takes a long time for the relevant departments to arrive at the scene, resulting in the escalation of the accident or missing the best rescue time. Summary of the invention

[0003] In view of this, the purpose of the present invention is to provide a method, device and storage medium for detecting vehicle traffic accidents based on trajectory, so as to solve the problem in the prior art that the discovery of traffic safety accidents has a certain lag, resulting in accident escalation or missing the best rescue time.

[0004] According to a first aspect of an embodiment of the present invention, a method for detecting a traffic accident of a vehicle based on a trajectory is provided, the method comprising: Acquire real-time driving trajectory data of the vehicle to be detected, and pre-process the real-time driving trajectory data to obtain standard driving trajectory data; Extracting speed characteristics and position characteristics of the vehicle to be detected according to the standard driving trajectory data; The speed characteristics and position characteristics of the vehicle to be detected are input into a pre-trained accident detection model, and the pre-trained accident detection model outputs whether the vehicle to be detected has a traffic accident.

[0005] Preferably, The step of obtaining the real-time driving trajectory data of the vehicle to be detected includes: The latitude and longitude, speed and direction of the vehicle to be detected are recorded in real time by a GPS or Beidou positioning device installed on the vehicle to be detected, so as to obtain the real-time driving trajectory data of the vehicle to be detected; or / and, The real-time position of the vehicle to be detected is obtained based on the communication time difference between the mobile communication base station and the mobile phone of the driver of the vehicle to be detected, thereby obtaining the real-time driving trajectory data of the vehicle to be detected.

[0006] Preferably, The preprocessing of the real-time driving trajectory data to obtain standard driving trajectory data includes: Removing noise and abnormal values ​​from the real-time driving trajectory data by a filtering algorithm to obtain filtered real-time driving trajectory data; A smoothing algorithm is used to smooth the jitter and fluctuation in the filtered real-time driving trajectory data to obtain the standard driving trajectory data.

[0007] Preferably, The extracting of the speed characteristics of the vehicle to be detected according to the standard driving trajectory data comprises: Acquire the average driving speed of the vehicle to be detected within a preset time period according to the standard driving trajectory data; Acquire the speed change rate of the vehicle to be detected within a preset time interval according to the standard driving trajectory data, and acquire the maximum acceleration and maximum deceleration of the vehicle to be detected within a preset time period according to the speed change rate of the vehicle to be detected within the preset time interval; Acquire all deceleration events of the vehicle to be detected within a preset time period and the duration of each deceleration event according to the speed change rate of the vehicle to be detected within a preset time interval; According to the standard driving trajectory data of the vehicle to be detected and the road speed limit on the corresponding path, it is determined whether the vehicle to be detected is speeding. If it is speeding, the speeding speed of the vehicle to be detected and the corresponding road speed limit are recorded.

[0008] Preferably, The extracting of the position features of the vehicle to be detected according to the standard driving trajectory data comprises: Acquire the abnormal position and residence time of the vehicle to be detected according to the standard driving trajectory data; The road type traveled by the vehicle to be detected and the speed limit information corresponding to each road type are obtained according to the standard driving trajectory data.

[0009] Preferably, The training process of the pre-trained accident detection model includes: Acquire historical traffic accident data and historical normal driving data, and perform data cleaning on the historical traffic accident data and historical normal driving data to remove missing values, abnormal values, and duplicate data; Label the cleaned historical traffic accident data, record the accident type, location, real-time speed before the accident, and acceleration or deceleration; Extracting speed features and position features of the historical traffic accident data and historical normal driving data; Expanding the speed feature and the position feature; An optimal model framework is selected, and the speed features and position features are input into the optimal model framework for iterative training to obtain an optimal hyperparameter combination, thereby obtaining a pre-trained accident detection model.

[0010] According to a second aspect of an embodiment of the present invention, a device for detecting a traffic accident of a vehicle based on a trajectory is provided, the device comprising: Driving trajectory acquisition module: used to acquire real-time driving trajectory data of the vehicle to be detected, and pre-process the real-time driving trajectory data to obtain standard driving trajectory data; Feature extraction module: used to extract speed features and position features of the vehicle to be detected according to the standard driving trajectory data; Accident detection module: used to input the speed characteristics and position characteristics of the vehicle to be detected into a pre-trained accident detection model, and the pre-trained accident detection model outputs whether the vehicle to be detected has a traffic accident.

[0011] According to a third aspect of an embodiment of the present invention, there is provided a storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a main controller, each step in the above method is implemented.

[0012] The technical solution provided by the embodiments of the present invention may have the following beneficial effects: The present application obtains real-time driving trajectory data of the vehicle to be detected, and extracts speed characteristics and position characteristics of the vehicle to be detected based on the driving trajectory data; the speed characteristics and position characteristics of the vehicle to be detected are input into a pre-trained accident detection model, and the accident detection model outputs whether the vehicle to be detected is involved in a traffic accident; the present application predicts whether the vehicle to be detected is involved in a traffic accident through the real-time driving trajectory data of the vehicle to be detected, and can accurately reflect whether the vehicle is involved in an accident at the first time. Compared with the prior art in which the parties or passers-by call the police after the accident occurs, the influence of human subjective factors is reduced, and the problem of delayed discovery of traffic accidents is solved.

[0013] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0015] Figure 1 is a flow chart of a method for detecting a vehicle traffic accident based on a trajectory according to an exemplary embodiment; Figure 2is a system schematic diagram of an apparatus for detecting a vehicle traffic accident based on a trajectory according to another exemplary embodiment; In the attached figure: 1-driving trajectory acquisition module, 2-feature extraction module, 3-accident detection module. DETAILED DESCRIPTION

[0016] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of devices and methods consistent with some aspects of the present invention as detailed in the appended claims.

[0017] Embodiment 1 Figure 1 is a flow chart of a method for detecting a vehicle traffic accident based on a trajectory according to an exemplary embodiment. Figure 1 As shown, the method includes: S1, obtaining real-time driving trajectory data of a vehicle to be detected, and preprocessing the real-time driving trajectory data to obtain standard driving trajectory data; S2, extracting speed characteristics and position characteristics of the vehicle to be detected according to the standard driving trajectory data; S3, inputting the speed characteristics and position characteristics of the vehicle to be detected into a pre-trained accident detection model, and the pre-trained accident detection model outputs whether the vehicle to be detected has a traffic accident; It can be understood that the present application collects the driving trajectory data of the vehicle through positioning devices such as GPS, and pre-processes the driving trajectory data, including denoising, filtering, smoothing and other operations to eliminate the influence of noise and abnormal points. According to the pre-processed trajectory data, the vehicle's speed and position and other features closely related to traffic accidents are extracted to predict whether the vehicle has a traffic accident. By inputting the extracted feature information into a pre-trained accident detection model, the model is used to determine whether there is a possibility of a traffic accident, and the detection results are output to the customer service or relevant management department of the online car-hailing platform to guide subsequent processing and prevention work. The prediction results can also be directly sent to the relevant processing department with the permission of the relevant department; the specific plan includes: Collect vehicle trajectory data: Vehicle-mounted GPS equipment: By installing a GPS or Beidou positioning device on the vehicle, the latitude and longitude, speed, direction and other information of the vehicle can be recorded in real time. In this embodiment, the Beidou positioning device is preferred. Compared with GPS in China, the Beidou positioning device has better stability and more accurate coordinate information. The movement trajectory of the vehicle is obtained based on the latitude and longitude, speed, direction and other information of the vehicle, which has the characteristics of high precision and strong real-time performance; Mobile communication base station positioning: In vehicles that do not have GPS or Beidou positioning equipment installed, the positioning service provided by the mobile communication base station can be used to obtain the approximate location information of the vehicle, thereby obtaining the vehicle's driving trajectory data; the principle is: mobile location service refers to a communication service that calculates your current approximate location based on the communication time difference between the base station and the mobile phone (the mobile phone here is the driver's mobile phone), and does not require the activation of GPRS; China Mobile has base stations in various regions, and the base stations are responsible for the call functions of the mobile phone terminal. When the mobile phone is turned on, it will contact the base station at fixed intervals to confirm the current status of the mobile phone. When using the query location service, the base station will communicate with the queried mobile phone, and then calculate the current location based on the feedback time of the signal; compared with GPS or Beidou positioning devices, the accuracy of this method will be slightly worse, so it is used as an alternative solution; Trajectory data preprocessing: After collecting vehicle trajectory data, preprocessing is usually required to improve the quality and usability of the data, including: Filtering algorithms: used to remove noise and outliers in data, such as Kalman filtering and median filtering. These algorithms can help reduce errors and fluctuations in data. Smoothing algorithm: used to smooth jitter and fluctuation in trajectory data to make the trajectory smoother and more continuous. Commonly used smoothing algorithms include sliding average method, polynomial fitting, etc. Feature extraction: In the preprocessed trajectory data, features that are highly relevant to accident detection need to be extracted. This embodiment uses features in two dimensions, speed and position, including: Get the average speed of the vehicle within a preset time period. The preset time period here generally refers to the time taken for the entire driving trajectory. It can also be a specified time period, and each time period is predicted separately. For example, with one hour as the boundary, predict whether a traffic accident will occur within each hour; Obtain the speed change rate of the vehicle within a preset time interval, and obtain the maximum acceleration and maximum deceleration according to the speed change rate of the vehicle within the preset time interval. The preset time interval here is generally very short, usually set to 3-10 seconds, that is, to obtain the speed change rate of the vehicle in a short period of time, and the speed change rate reflects the acceleration or deceleration; Because drivers usually slow down before a traffic accident occurs, it is necessary to focus on the deceleration events in the vehicle speed characteristics (such as emergency braking) and the duration of each deceleration event; In existing navigation software, there is usually road speed limit information for each route. Combining the navigation software and the real-time speed of the vehicle, it can be determined whether the vehicle is speeding on the road. If speeding, the speeding speed of the vehicle and the corresponding road speed limit are recorded; Location Features: Abnormal location stay: Determine whether the vehicle stays in an unreasonable location (such as outside the road, no-parking zone, etc.) for a long time.

[0018] Road type and speed limit: record the road type and corresponding speed limit information of the vehicle; Accident detection model training: Expand the data set: collect more historical traffic accident data and normal driving data, including diverse data such as different road types, weather conditions, and time periods, to enhance the generalization ability of the model; Data cleaning: Thoroughly clean the data to remove missing values, outliers, and duplicate data to ensure the quality of data input into the model; Data labeling: Accurately label the collected data, especially traffic accident data, which requires detailed recording of the accident type, driving speed before the accident, acceleration and deceleration, location of the accident, etc. Feature selection: Based on domain knowledge and data analysis, the most critical features for accident detection are selected for extraction. In this embodiment, the above-mentioned speed features and position features are selected; Feature engineering: further processing of the extracted features, such as feature scaling, encoding, dimensionality reduction, etc. (that is, the feature expansion mentioned above) to improve the learning efficiency and performance of the model; Selection of the optimal model framework: Try to use different machine learning algorithms (such as logistic regression, random forest, gradient boosting machine, neural network, etc.) to train the model separately. According to the results of model training, compare their performance and select the model with the best performance as the optimal model framework. You can also consider using model fusion techniques, such as voting and stacking, to combine the prediction results of multiple models to improve the accuracy and stability of the overall prediction. Hyperparameter tuning: Use methods such as cross-validation and grid search to tune the hyperparameters of the optimal model framework and find the best parameter combination; Regularization: During model training, regularization techniques (such as L1 regularization and L2 regularization) are used to prevent overfitting and improve the generalization ability of the model. Early stopping method: Use early stopping method during training to stop training when the performance on the validation set starts to decline to avoid overfitting; Real-time data processing: ensure that the vehicle trajectory data collected in real time (that is, the vehicle's real-time speed characteristics and location characteristics) can be quickly and accurately input into the model for accident prediction or detection; Performance monitoring: monitor the real-time detection performance of the model, regularly check the model's accuracy and recall rate and other indicators to ensure stable model performance; Feedback mechanism: Establish a user feedback mechanism to collect user feedback on accident detection to further optimize the model and improve user experience; Data update: Continuously collect new traffic accident data to update and enrich the training data set, and continuously update and train the model; Model iteration: Based on new data and user feedback, the model is regularly iterated and optimized to adapt to new environments and needs.

[0019] The present application obtains real-time driving trajectory data of the vehicle to be detected, and extracts speed characteristics and position characteristics of the vehicle to be detected based on the driving trajectory data; the speed characteristics and position characteristics of the vehicle to be detected are input into a pre-trained accident detection model, and the accident detection model outputs whether the vehicle to be detected is involved in a traffic accident; the present application predicts whether the vehicle to be detected is involved in a traffic accident through the real-time driving trajectory data of the vehicle to be detected, and can accurately reflect whether the vehicle is involved in an accident at the first time. Compared with the prior art in which the parties or passers-by call the police after the accident occurs, the influence of human subjective factors is reduced, and the problem of delayed discovery of traffic accidents is solved.

[0020] Embodiment 2 Figure 2 is a system schematic diagram of an apparatus for detecting a vehicle traffic accident based on a trajectory according to another exemplary embodiment, comprising: Driving trajectory acquisition module 1: used to acquire real-time driving trajectory data of the vehicle to be detected, and pre-process the real-time driving trajectory data to obtain standard driving trajectory data; Feature extraction module 2: used to extract the speed feature and position feature of the vehicle to be detected according to the standard driving trajectory data; Accident detection module 3: used to input the speed characteristics and position characteristics of the vehicle to be detected into a pre-trained accident detection model, and the pre-trained accident detection model outputs whether the vehicle to be detected has a traffic accident.

[0021] Embodiment three: This embodiment provides a storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a main controller, each step in the above method is implemented; It is understandable that the storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.

[0022] It can be understood that the same or similar parts of the above embodiments can be referenced to each other, and the contents not described in detail in some embodiments can refer to the same or similar contents in other embodiments.

[0023] It should be noted that, in the description of the present invention, the terms "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. In addition, in the description of the present invention, unless otherwise specified, the meaning of "plurality" refers to at least two.

[0024] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present invention includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present invention belong.

[0025] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0026] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0027] In addition, each functional unit in each embodiment of the present invention may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0028] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.

[0029] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0030] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A method for detecting a vehicle traffic accident based on a trajectory, characterized in that: The method comprises: Acquire real-time driving trajectory data of the vehicle to be detected, and pre-process the real-time driving trajectory data to obtain standard driving trajectory data; Extracting speed characteristics and position characteristics of the vehicle to be detected according to the standard driving trajectory data; The speed characteristics and position characteristics of the vehicle to be detected are input into a pre-trained accident detection model, and the pre-trained accident detection model outputs whether the vehicle to be detected has a traffic accident.

2. The method according to claim 1, characterized in that The step of obtaining the real-time driving trajectory data of the vehicle to be detected includes: The latitude and longitude, speed and direction of the vehicle to be detected are recorded in real time by a GPS or Beidou positioning device installed on the vehicle to be detected, so as to obtain the real-time driving trajectory data of the vehicle to be detected; or / and, The real-time position of the vehicle to be detected is obtained based on the communication time difference between the mobile communication base station and the mobile phone of the driver of the vehicle to be detected, thereby obtaining the real-time driving trajectory data of the vehicle to be detected.

3. The method according to claim 2, characterized in that The preprocessing of the real-time driving trajectory data to obtain standard driving trajectory data includes: Removing noise and abnormal values ​​from the real-time driving trajectory data by a filtering algorithm to obtain filtered real-time driving trajectory data; A smoothing algorithm is used to smooth the jitter and fluctuation in the filtered real-time driving trajectory data to obtain the standard driving trajectory data.

4. The method according to claim 3, characterized in that The extracting of the speed characteristics of the vehicle to be detected according to the standard driving trajectory data comprises: Acquire the average driving speed of the vehicle to be detected within a preset time period according to the standard driving trajectory data; Acquire the speed change rate of the vehicle to be detected within a preset time interval according to the standard driving trajectory data, and acquire the maximum acceleration and maximum deceleration of the vehicle to be detected within a preset time period according to the speed change rate of the vehicle to be detected within the preset time interval; Acquire all deceleration events of the vehicle to be detected within a preset time period and the duration of each deceleration event according to the speed change rate of the vehicle to be detected within a preset time interval; According to the standard driving trajectory data of the vehicle to be detected and the road speed limit on the corresponding path, it is determined whether the vehicle to be detected is speeding. If it is speeding, the speeding speed of the vehicle to be detected and the corresponding road speed limit are recorded.

5. The method according to claim 4, characterized in that The extracting of the position features of the vehicle to be detected according to the standard driving trajectory data comprises: Acquire the abnormal position and residence time of the vehicle to be detected according to the standard driving trajectory data; The road type traveled by the vehicle to be detected and the speed limit information corresponding to each road type are obtained according to the standard driving trajectory data.

6. The method according to claim 5, characterized in that The training process of the pre-trained accident detection model includes: Acquire historical traffic accident data and historical normal driving data, and perform data cleaning on the historical traffic accident data and historical normal driving data to remove missing values, abnormal values, and duplicate data; Label the cleaned historical traffic accident data, record the accident type, location, real-time speed before the accident, and acceleration or deceleration; Extracting speed features and position features of the historical traffic accident data and historical normal driving data; Expanding the speed feature and the position feature; An optimal model framework is selected, and the speed features and position features are input into the optimal model framework for iterative training to obtain an optimal hyperparameter combination, thereby obtaining a pre-trained accident detection model.

7. A device for detecting a vehicle traffic accident based on a trajectory, characterized in that: The device comprises: Driving trajectory acquisition module: used to acquire real-time driving trajectory data of the vehicle to be detected, and pre-process the real-time driving trajectory data to obtain standard driving trajectory data; Feature extraction module: used to extract speed features and position features of the vehicle to be detected according to the standard driving trajectory data; Accident detection module: used to input the speed characteristics and position characteristics of the vehicle to be detected into a pre-trained accident detection model, and the pre-trained accident detection model outputs whether the vehicle to be detected has a traffic accident.

8. A storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by the main controller, each step of the method for detecting a traffic accident occurring to a vehicle based on trajectory as described in any one of claims 1 to 6 is implemented.