Collision detection method, system, medium and program product

Abstract

The application provides a collision detection method, system, medium and program product. The method comprises the following steps: obtaining collision data sent by a first vehicle when a suspected collision occurs, wherein the collision data at least comprises a suspected collision occurrence time, a suspected collision occurrence position and a suspected collision occurrence road; based on the collision data of the first vehicle and trajectory data uploaded by other vehicles, at least one type of second vehicle is obtained from the other vehicles, wherein the second vehicle is an other vehicle which drives on the suspected collision occurrence road of the first vehicle at the time of the suspected collision of the first vehicle, the driving position of the second vehicle is located before or after the suspected collision occurrence position of the first vehicle, and the driving behavior of the second vehicle meets a preset behavior condition; the number of different types of second vehicles is counted; and whether the first vehicle has a collision is determined at least according to the number of the second vehicles. The application can improve the accuracy of collision detection.

Description

Technical Field

[0001] This application relates to the field of transportation technology, and in particular to a collision detection method, system, medium, and program product. Background Technology

[0002] Traffic accidents are one of the factors affecting the life and property safety of road users. Timely detection and rapid roadside assistance after a traffic accident are of great significance in saving the lives of the injured and ensuring the travel safety of all road users.

[0003] Currently, there are some existing technologies for detecting whether a vehicle has been involved in a traffic accident. For example, during the vehicle's operation, the system can detect whether a collision has occurred based on the vehicle's driving data to determine whether a traffic accident has taken place. This is because vehicles involved in traffic accidents almost always involve collisions.

[0004] Because vehicles engage in various behaviors while driving on the road, such as sudden braking and sharp turns, but some of these behaviors do not lead to traffic accidents, providing technology that can accurately detect whether a vehicle has been involved in a traffic accident, and avoiding the waste of rescue resources caused by false detections, is a problem that needs to be solved by those skilled in the art. Summary of the Invention

[0005] This application provides a collision detection method, system, medium, and program product that can accurately detect whether a vehicle has collided, thereby accurately concluding whether a traffic accident has occurred and avoiding the waste of rescue resources caused by false detections.

[0006] In a first aspect, this application provides a collision detection method, the method comprising:

[0007] Acquire collision data sent by the first vehicle when the suspected collision occurs, wherein the collision data includes at least: the time of the suspected collision, the location of the suspected collision, and the road on which the suspected collision occurred;

[0008] Based on the collision data of the first vehicle and the trajectory data uploaded by other vehicles, at least one type of second vehicle is obtained from the other vehicles. The second vehicle is another vehicle that was driving on the road where the first vehicle was suspected to have collided at the time of the suspected collision, whose driving position was before or after the suspected collision position of the first vehicle, and whose driving behavior met the preset behavior conditions. The type of the second vehicle is related to the driving position before or after the suspected collision position of the first vehicle and the driving behavior.

[0009] Count the number of second vehicles of different types;

[0010] Whether the first vehicle was involved in a collision can be determined at least based on the number of the second vehicles.

[0011] Secondly, this application provides a collision detection method, the method comprising:

[0012] Obtain the driving data of the first vehicle within the time window;

[0013] Based on the driving data within the time window, it is detected whether the first vehicle is suspected of being involved in a collision. If so, the collision data of the first vehicle is obtained. The collision data includes at least: the suspected collision time, the suspected collision location, and the suspected collision road.

[0014] Send a collision detection request carrying the collision data to the server;

[0015] The system receives a collision detection response returned by the server based on the collision detection request. The collision detection response carries a detection result indicating whether the first vehicle has collided. The detection result is determined based on the number of second vehicles, which are other vehicles that were driving on the road where the first vehicle was suspected of colliding at the time of the suspected collision, were driving before or after the suspected collision location of the first vehicle, and whose driving behavior meets preset behavior conditions.

[0016] Thirdly, this application provides a collision detection device, the device comprising:

[0017] The first acquisition module is used to acquire collision data sent by the first vehicle when a suspected collision occurs. The collision data includes at least: the time of the suspected collision, the location of the suspected collision, and the road where the suspected collision occurred.

[0018] The second acquisition module is used to acquire at least one type of second vehicle from the other vehicles based on the collision data of the first vehicle and the trajectory data uploaded by other vehicles. The second vehicle is another vehicle that was driving on the road where the suspected collision of the first vehicle occurred at the time of the suspected collision of the first vehicle, whose driving position was before or after the suspected collision location of the first vehicle, and whose driving behavior met the preset behavior conditions. The type of the second vehicle is related to whether the driving position was before or after the suspected collision location of the first vehicle and the driving behavior.

[0019] The statistics module is used to count the number of second vehicles of different types.

[0020] A determining module is configured to determine, at least based on the number of the second vehicles, whether the first vehicle has been involved in a collision.

[0021] Fourthly, this application provides a collision detection device, the device comprising:

[0022] The first acquisition module is used to acquire the driving data of the first vehicle within the time window;

[0023] The second acquisition module is used to detect whether the first vehicle is suspected of being involved in a collision based on the driving data within the time window. If so, the collision data of the first vehicle is acquired. The collision data includes at least: the suspected collision time, the suspected collision location, and the suspected collision road.

[0024] The sending module is used to send a collision detection request carrying the collision data to the server;

[0025] The receiving module is used to receive a collision detection response returned by the server based on the collision detection request. The collision detection response carries a detection result indicating whether the first vehicle has collided. The detection result is determined based on the number of second vehicles. The second vehicles are other vehicles that were driving on the road where the first vehicle was suspected of colliding at the time of the suspected collision, were driving before or after the suspected collision location of the first vehicle, and whose driving behavior meets preset behavior conditions.

[0026] Fifthly, this application provides an electronic device, including: a processor and a memory; the processor is communicatively connected to the memory;

[0027] The memory stores computer instructions;

[0028] The processor executes computer instructions stored in the memory to implement the method as described in any one of the first aspects, and / or the method as described in any one of the second aspects.

[0029] Sixthly, this application provides a collision detection system, the system comprising: edge software and server, wherein,

[0030] The endpoint software is used to perform the method as described in any one of the second aspects;

[0031] The server is used to execute the method as described in any one of the first aspects.

[0032] In a seventh aspect, this application provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the collision detection method as described in any one of the first aspects, and / or the method as described in any one of the second aspects.

[0033] Eighthly, this application provides a computer program product, including a computer program or instructions, which, when executed by a processor, implement the method described in any one of the first and / or second aspects.

[0034] The collision detection method, system, medium, and program products provided in this application allow the edge software of the collision detection system to initially determine when a first vehicle is suspected of being involved in a collision based on its driving data within a time window. Upon determining a suspected collision, the system sends a collision detection request carrying the collision data to the server. This allows the server to obtain the number of at least one type of second vehicles from the other vehicles based on the collision data of the first vehicle and the trajectory data uploaded by other vehicles. The number of these at least one type of second vehicles is related to their driving position before or after the suspected collision location of the first vehicle and their driving behavior. Therefore, the number of these at least one type of second vehicles can reflect the surrounding road conditions at the suspected collision location of the first vehicle. This allows for a more accurate determination of whether a collision has occurred based on the surrounding road conditions, avoiding misjudgments based solely on the vehicle's own driving data, improving the accuracy of collision detection, and ultimately leading to an accurate conclusion as to whether a traffic accident has occurred, thus preventing the waste of rescue resources caused by false detections. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 A schematic flowchart illustrating the first collision detection method provided in this application embodiment;

[0037] Figure 2 A schematic flowchart illustrating the second collision detection method provided in this application embodiment;

[0038] Figure 3 A schematic flowchart illustrating the third collision detection method provided in this application embodiment;

[0039] Figure 4 A schematic diagram of a velocity-time curve provided in an embodiment of this application;

[0040] Figure 5 A schematic flowchart illustrating the fourth collision detection method provided in this application embodiment;

[0041] Figure 6A flowchart illustrating the fifth collision detection method provided in this application embodiment;

[0042] Figure 7 A flowchart illustrating the sixth collision detection method provided in this application embodiment;

[0043] Figure 8 A flowchart illustrating the seventh collision detection method provided in this application embodiment;

[0044] Figure 9 This is a schematic diagram of the structure of the first collision detection device provided in the embodiments of this application;

[0045] Figure 10 This is a schematic diagram of the structure of the second collision detection device provided in the embodiments of this application;

[0046] Figure 11 This is a schematic diagram of the structure of an electronic device 110 provided in an embodiment of this application;

[0047] Figure 12 This is a schematic diagram of the structure of a collision detection system provided in an embodiment of this application.

[0048] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0050] As mentioned earlier, vehicles involved in traffic accidents are usually involved in collisions. Therefore, whether a vehicle has been involved in a traffic accident is usually determined by detecting whether a collision has occurred.

[0051] Existing vehicle collision detection methods primarily acquire vehicle driving data using the Global Positioning System (GPS), accelerometers, and gyroscopes to determine whether a collision has occurred. In other words, they detect whether a collision has taken place during vehicle operation based on the vehicle's own driving data, thus determining whether a traffic accident has occurred.

[0052] For example, based on vehicle driving data, the negative acceleration when the vehicle speed suddenly drops, and the subsequent speed after the drop, are calculated to determine whether the negative acceleration meets a predetermined threshold and whether the subsequent speed is sufficiently small to determine whether a collision has occurred; alternatively, a pre-trained detection model is used to detect whether a collision has occurred based on the vehicle's driving data. Some existing technologies further incorporate changes in vehicle air pressure and / or the sound of the vehicle's environment. However, all of these vehicle collision detection methods suffer from low detection accuracy.

[0053] The inventors discovered through research that existing vehicle collision detection methods rely solely on the vehicle's own driving data for collision detection, without considering the surrounding road conditions. However, vehicles exhibit a variety of behaviors while driving on the road, such as sudden braking and sharp turns. Some of these behaviors resemble the characteristics of a collision but do not necessarily lead to a traffic accident. Collision detection based solely on the vehicle's own driving data is prone to misjudgment. Therefore, the surrounding road conditions are crucial for accurately determining whether a collision has occurred.

[0054] For example, the road conditions around a vehicle at the time of a collision are often significantly different from those before a collision. For instance, if a collision does not occur, other vehicles behind the vehicle are usually traveling at a higher speed; however, if a collision occurs, other vehicles behind the vehicle are often traveling at a lower speed or even coming to a standstill.

[0055] Furthermore, in some scenarios, a vehicle's driving characteristics often resemble those of a collision. For example, in traffic jams, vehicles often experience a sudden drop in speed, or even come to a complete stop. In such cases, based on the vehicle's own driving data, a collision is often presumed, even when no collision has actually occurred, leading to a misjudgment. In such situations, if the surrounding road conditions can be used to determine that the current road segment is experiencing a traffic jam rather than a collision, this misjudgment can be avoided.

[0056] In view of this, this application provides a collision detection method that can detect whether a vehicle has been involved in a collision based on the vehicle's own driving data and the road conditions around the vehicle. This method can achieve more refined detection of whether a collision has occurred, improve the accuracy of collision detection, and thus improve the accuracy of detecting whether a vehicle has been involved in a traffic accident.

[0057] The implementing entities of this application are the edge software and server in a collision detection system. The edge software can be application software with map navigation capabilities, dedicated collision detection software, or other application software integrating collision detection functions. This software can be deployed on electronic devices located in the first vehicle, such as mobile phones or in-vehicle infotainment systems. When deployed on a mobile phone, it can be, for example, an application program (APP); when deployed on an in-vehicle infotainment system, it can be, for example, pre-installed software. The server can be deployed on cloud-based, roadside equipment, or other devices capable of processing different edge data.

[0058] The technical solutions of this application will be described in detail below with reference to specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0059] The following description will first focus on the server-side perspective to illustrate the solution provided in the embodiments of this application.

[0060] Figure 1 This is a schematic flowchart illustrating a collision detection method provided in an embodiment of this application. Figure 1 As shown, the method may include the following steps:

[0061] S101. Obtain the collision data sent by the first vehicle when the suspected collision occurred.

[0062] The first vehicle mentioned above can be any vehicle suspected of being involved in a collision. The collision data mentioned above includes at least: the time of the suspected collision, the location of the suspected collision, and the road on which the suspected collision occurred.

[0063] This application does not limit the time of the suspected collision, the location of the suspected collision, or the method of representing the road where the suspected collision occurred. For example, the time of the suspected collision can be represented by dimensions such as year, month, day, hour, minute, and second; the location of the suspected collision can be represented by latitude and longitude; and the road where the first vehicle is located can be represented by an identifier, such as the ID of a road preset in the map data on which the collision detection system relies.

[0064] This application does not limit how to determine whether the first vehicle is suspected of being involved in a collision, nor does it limit the entity that performs the determination. For example, it could be that the edge software, based on the first vehicle's driving data, initially determines that the first vehicle is suspected of being involved in a collision and then sends the collision data of the first vehicle to the server. Alternatively, the server itself, based on the first vehicle's driving data, initially determines that the first vehicle is suspected of being involved in a collision and then obtains the collision data of the first vehicle based on the driving data. The aforementioned driving data may include, for example, positioning data, or positioning data and navigation path information, or positioning data and inertial navigation data, or positioning data, inertial navigation data, and navigation path information, etc.

[0065] S102. Based on the collision data of the first vehicle and the trajectory data uploaded by other vehicles, obtain at least one type of second vehicle from the other vehicles.

[0066] The trajectory data of the other vehicles mentioned above may include, for example, the location of the trajectory points of the other vehicles during their journey, the road markings, and the time when they passed through the trajectory points.

[0067] The second vehicle is any other vehicle that was traveling on the road where the suspected collision occurred at the time the first vehicle was suspected of being in contact with the first vehicle, was located before or after the location of the suspected collision, and whose driving behavior met preset behavioral conditions. The type of the second vehicle is related to its location before or after the location of the suspected collision and its driving behavior. For example, a second vehicle exhibiting one type of driving behavior can be classified as one type, or multiple types of driving behavior can be classified as one type.

[0068] It should be understood that the specific pre-defined behavioral conditions included in this application may vary depending on the type of the second vehicle. Furthermore, the specific types of the second vehicle are related to the type used in subsequent determination of whether a collision with the first vehicle occurred.

[0069] As mentioned above, the road conditions surrounding a vehicle play a crucial role in determining whether a collision has occurred. In this application, a type of second vehicle is used to characterize a relative spatiotemporal relationship between the first and second vehicles, which reflects the road conditions surrounding the vehicle. Therefore, in this step, the server can, based on the collision data of the first vehicle and the trajectory data uploaded by other vehicles, obtain at least one type of second vehicle from the other vehicles to determine the number of second vehicles with various relative spatiotemporal relationships to the first vehicle. This allows the server to ascertain the road conditions surrounding the first vehicle and accurately determine whether a collision has occurred.

[0070] For example, the server can filter out other vehicles and their trajectory data based on the collision data of the first vehicle and the trajectory data uploaded by other vehicles at the time of the suspected collision of the first vehicle, or whose driving position is before or after the location of the suspected collision of the first vehicle. Then, based on the obtained trajectory data of other vehicles, other vehicles whose driving behavior meets preset behavioral conditions are identified as second vehicles and classified as second vehicles.

[0071] Alternatively, the server can evaluate other vehicles one by one based on the collision data of the first vehicle and the trajectory data uploaded by other vehicles, and add other vehicles that meet the preset behavior conditions and belong to the corresponding second vehicle type as the second vehicle type.

[0072] S103, Count the number of second vehicles of different types.

[0073] As mentioned above, the road conditions surrounding a vehicle play a crucial role in determining whether a collision has occurred, and the number of different types of second vehicles reflects the road conditions surrounding the first vehicle. Therefore, in this step, the server counts the number of different types of second vehicles so that it can subsequently determine whether a collision has occurred based on the number of different types of second vehicles.

[0074] S104. Determine whether the first vehicle has collided, at least based on the number of the second vehicles.

[0075] One possible implementation is that the server determines whether the first vehicle has been involved in a collision based solely on the number of second vehicles. When a collision is determined to have occurred, it indicates that the first vehicle has been involved in a traffic accident.

[0076] Another possible implementation is that the server determines whether the first vehicle has collided based on the number of the second vehicles, as well as one or more of the first vehicle's inertial navigation data, initial velocity, and negative acceleration.

[0077] In either of the above implementations, the server can determine whether a collision has occurred using a threshold method or a collision detection model. This collision detection model can be any classification model, such as a collision detection model built on a random forest method or a collision detection model built on a Transformer model.

[0078] After determining whether a collision has occurred with the first vehicle, the subsequent processing actions based on the determination result are related to the actual application scenario of the collision detection method provided in this application. When the collision detection method provided in this application is applied to a roadside assistance scenario, after determining whether a collision has occurred with the first vehicle, the server can send a collision detection response to the client software. This collision detection response is used to indicate whether a collision has occurred with the first vehicle, so that the client software outputs a request to trigger roadside assistance services when a collision occurs with the first vehicle. This application does not limit the way the client software outputs the request to trigger roadside assistance services; for example, it can output to the display interface when the client software has a display interface, and / or, the above request can be output via voice.

[0079] The above method allows for timely and proactive provision of roadside assistance services to occupants of the first vehicle after a collision is confirmed. This is convenient, fast, and improves the speed at which roadside assistance services are triggered, thereby increasing the response speed and facilitating timely roadside assistance to the collided vehicle and its occupants.

[0080] In the collision detection method provided in this embodiment, the edge software of the collision detection system can initially determine that the first vehicle is suspected of being involved in a collision based on the driving data of the first vehicle within a time window. Upon determining that a collision is suspected, it sends a collision detection request carrying the collision data to the server. This allows the server to obtain the number of at least one type of second vehicles from the other vehicles based on the collision data of the first vehicle and the trajectory data uploaded by other vehicles. The number of these at least one type of second vehicles is related to their driving position before or after the suspected collision location of the first vehicle and their driving behavior. Therefore, the number of these at least one type of second vehicles can reflect the surrounding road conditions at the suspected collision location of the first vehicle. This allows for a more accurate determination of whether the first vehicle has been involved in a collision based on the surrounding road conditions, avoiding misjudgments based solely on the vehicle's own driving data, improving the accuracy of collision detection, and ultimately accurately concluding whether a traffic accident has occurred, thus avoiding the waste of rescue resources caused by false detections.

[0081] The following describes how the server obtains the collision data sent by the first vehicle when a suspected collision occurs, i.e., step S101 in the above embodiment. Specific implementation methods include, but are not limited to, the following two:

[0082] Method 1: The server receives a collision detection request from the client-side software. This request is triggered when the client-side software determines that the first vehicle is suspected of being involved in a collision based on its driving data within a time window. The collision detection request includes collision data. Specifically, the client-side software acquires the driving data of the first vehicle within the time window and, based on this data, detects whether the first vehicle is suspected of being involved in a collision. If a collision is suspected, the software acquires the collision data of the first vehicle and sends a collision detection request carrying this data to the server.

[0083] This application does not limit the specific time range covered by the time window. For example, it can be a time window of any length, including the time when the suspected collision of the first vehicle occurred, depending on the actual settings. For instance, it can be related to the timeliness of the server-side collision detection. For example, if the server can detect the collision 30 seconds after the collision occurs, then the aforementioned time window can be a time window of 30 seconds in length.

[0084] In this implementation, the edge software can make a preliminary judgment on whether a collision has occurred based on the driving data of the first vehicle. If it is determined that the first vehicle is suspected of being involved in a collision, the edge software sends the collision data of the first vehicle to the server, where the server makes further judgments. Through this method, the edge software can share some of the computational burden, reducing the amount of data uploaded to the server and alleviating the computational burden on the server, thus improving the overall efficiency of collision detection.

[0085] Optionally, the client-side software can package the collision data, add the packaged collision data to the collision detection request, and send it to the server to reduce the traffic pressure caused by collision detection.

[0086] Method 2

[0087] The server receives a collision detection request from the client-side software. This request includes the driving data of the first vehicle within a time window. When the server determines that the first vehicle is suspected of being involved in a collision based on the driving data of the first vehicle within the time window, it obtains the collision data of the first vehicle.

[0088] In this implementation, regardless of whether the first vehicle is suspected of being involved in a collision, the edge software sends the first vehicle's driving data within the time window to the server. The server then determines whether a collision has occurred based on this driving data. With this implementation, the edge software does not bear the computational burden of collision detection, meaning it has lower computational requirements, thus expanding the application scenarios of this application at the edge software level. Furthermore, since the edge software does not need to reserve computational resources for collision detection, its own performance remains unaffected.

[0089] Optionally, the client-side software can package the driving data of the first vehicle within the time window, and then add the packaged driving data of the first vehicle within the time window to the collision detection request and send it to the server to reduce the traffic pressure caused by collision detection.

[0090] Both of the above implementation methods require determining whether the first vehicle has been suspected of being involved in a collision based on its driving data within the time window to which the collision occurred. The difference lies in that method one is executed by the client-side software, while method two is executed by the server. The following explanation, using client-side software as the execution subject, illustrates how to detect whether the first vehicle has been suspected of being involved in a collision based on its driving data within the time window to which the suspected collision occurred. It should be understood that, for method two, when the server determines that the first vehicle has been suspected of being involved in a collision based on its driving data within the time window, the method for obtaining the collision data of the first vehicle can refer to the implementation method described below using client-side software as the execution subject, and will not be repeated here.

[0091] One possible implementation is that the driving data includes the location data of the first vehicle within a time window. The terminal software determines whether the speed change of the first vehicle within the time window meets the speed change behavior of a vehicle collision based on the location data. If it does, the software determines that the first vehicle is suspected of being involved in a collision.

[0092] Figure 2 This is a flowchart illustrating the second collision detection method provided in the embodiments of this application, as shown below. Figure 2 As shown, in this implementation, step S101 of the above embodiment may include the following steps:

[0093] S201. Based on the positioning data within the time window, determine whether the speed change of the first vehicle within the time window meets the speed change performance required for a vehicle collision.

[0094] If a vehicle collides, its speed changes before and after the collision often exhibit typical characteristics, depending on the driving scenario, such as high-speed or low-speed driving. Taking a high-speed driving scenario as an example, the speed changes before and after a collision may include the following characteristics: a relatively high initial speed, a sudden drop in speed, no recovery for a certain period after the drop, and a subsequent sustained low speed. It should be understood that the above is merely an example of one type of speed change characteristic classification after a collision. In actual implementation, more feature classifications may exist, or only one or more features may be used for judgment.

[0095] Optionally, the edge software can determine whether a collision has occurred for a single driving scenario, or it can first determine the current scenario and then use the speed change judgment method corresponding to that scenario to determine whether a collision has occurred. For example, for scenarios distinguished by speed, the scenario can be determined based on the vehicle's initial speed.

[0096] It should be noted that this application does not limit how the vehicle speed is obtained from the positioning data. For example, the positioning data may include the vehicle speed, or the positioning data may include the vehicle's position at each moment, thereby determining the vehicle's speed at those multiple moments based on the vehicle's position at multiple moments and the time span. For example, the edge software obtains the distance traveled by the vehicle based on the positioning data of two adjacent moments, and divides this distance by the time interval between the two adjacent moments to obtain the vehicle's speed at those two moments.

[0097] The following example uses these four features to illustrate how to determine whether the speed change of the first vehicle within the time window meets the speed change behavior of a vehicle collision based on the positioning data within the time window.

[0098] Figure 3 This is a flowchart illustrating the third collision detection method provided in an embodiment of this application. Figure 4 This application provides a schematic diagram of a speed-time curve, which is a speed-time curve obtained based on all positioning data within a time window, such as... Figure 3 and Figure 4 As shown, this implementation method may include the following steps:

[0099] S2011. Based on the first to xth positioning data within the time window, determine whether the initial speed of the first vehicle within the time window is greater than the first preset speed threshold.

[0100] The above x is an integer greater than or equal to 1. Based on this, this application does not limit x, and the specific value of the first preset speed threshold can be determined by those skilled in the art according to the actual situation.

[0101] For example, such as Figure 4 As shown, the edge software can determine the initial speed of the first vehicle within the time window based on the first to xth positioning data within the time window, and then determine whether the initial speed is greater than a first preset speed threshold. If so, the requirement for the initial speed in the speed change leading to a collision is met, and step S2022 is executed.

[0102] If not, then it is determined that the speed change of the first vehicle within the time window does not meet the speed change required for a collision, and therefore it is determined that the first vehicle did not collide.

[0103] S2012. Based on the y-th positioning data to the last positioning data of the time window, determine whether the speed of the first vehicle at the end of the time window is less than the second preset speed threshold.

[0104] The aforementioned y is an integer greater than or equal to 1. Based on this, this application does not limit y, and the specific value of the second preset speed threshold can be determined by those skilled in the art according to the actual situation. Wherein, y is greater than x.

[0105] For example, continue to refer to Figure 4 The edge software can obtain the speed of the first vehicle at the end of the time window based on the y-th positioning data to the last positioning data, and then determine whether the end speed is less than the second preset speed threshold.

[0106] Typically, if the first vehicle collides, its speed, as determined by the location data, will drop sharply, even to zero, in the time following the collision. Therefore, in this step, it is determined whether the speed of the first vehicle at the end of the time window is less than the second preset speed threshold. If so, it indicates that the driving characteristics of the first vehicle meet the requirements for the final speed in the speed change after a collision, that is, it will remain at a very low value for a long period of time. Then, step S2023 is executed.

[0107] If not, then it is determined that the speed change of the first vehicle within the time window does not meet the speed change required for a collision, and therefore it is determined that the first vehicle did not collide.

[0108] S2013. Determine whether target positioning data exists based on the (x+1)th to (x+n)th positioning data within the time window.

[0109] Where n is an integer greater than or equal to 2. The target positioning data is the positioning data of the first vehicle when the negative acceleration is greater than or equal to the third preset speed threshold.

[0110] This application does not limit the specific value of the third preset speed threshold, which can be determined by those skilled in the art based on the actual situation.

[0111] One possible implementation is that the edge software can acquire the negative acceleration corresponding to any two adjacent positioning data points from the (x+1)th to the (x+n)th positioning data point within the time window. When a negative acceleration is greater than or equal to a third preset velocity threshold, the positioning data corresponding to that negative acceleration is taken as the target positioning data. For example, it could be the later positioning data point among the two adjacent positioning data points.

[0112] Alternatively, the edge software can perform interpolation on the (x+1)th to (x+n)th positioning data; based on the interpolated positioning data, determine whether the target positioning data exists among the (x+1)th to (x+n)th positioning data.

[0113] Taking x+1 as 3 and x+n as 5 as an example, the edge software interpolates at the midpoint between the 3rd and 4th positioning data points, i.e., adding the 3.5th positioning data point; it also adds the 4.5th positioning data point at the midpoint between the 4th and 5th positioning data points, thus implementing positioning data interpolation. Then, it can obtain the negative accelerations corresponding to the 3rd and 4th positioning data points, the 3.5th and 4.5th positioning data points, and the 4th and 5th positioning data points, and determine whether the maximum value among these three negative accelerations is greater than or equal to a third preset velocity threshold. If so, the positioning data corresponding to the maximum value is used as the target positioning data; if the positioning data corresponding to the maximum value is interpolated positioning data, then the original positioning data following that positioning data is used as the target positioning data. For example, if the negative acceleration corresponding to the 3rd and 4th positioning data points is the maximum value and greater than or equal to the third preset velocity threshold, then the 4th positioning data point can be used as the target positioning data. Taking the negative acceleration corresponding to the 3.5th and 4.5th positioning data as the maximum value and greater than or equal to the third preset velocity threshold as an example, the 5th positioning data can be used as the target positioning data.

[0114] It should be noted that this application does not limit the interpolation frequency; for example, multiple positioning data can be interpolated between two adjacent positioning data points. The method by which the edge software obtains the negative acceleration of the first vehicle during that time period based on the interpolated positioning data can refer to existing technologies and will not be elaborated here.

[0115] As mentioned above, if the first vehicle collides, its speed will typically drop sharply. Therefore, in this step, the edge software checks whether target location data exists within the time window. If it exists, it indicates that the first vehicle experienced a sharp speed drop within the time window, and then step S2014 is executed.

[0116] If it does not exist, then it is determined that the speed change of the first vehicle within the time window does not meet the speed change required for a collision, and therefore it is determined that the first vehicle did not collide.

[0117] It should be noted that the execution order of steps S2011-S2013 in this application can be adjusted according to actual needs; that is, the execution of these three steps is not in any particular order.

[0118] S2014. Based on the target positioning data up to the (y-1)th positioning data within the time window, determine whether the speed of the first vehicle during this time period is less than the fourth preset speed threshold.

[0119] For example, continue to refer to Figure 4 That is, the terminal software determines whether the speed during the time period is less than the fourth preset speed threshold based on the fourth judgment window. The fourth preset speed threshold is greater than the second preset speed threshold. Based on this, this application does not limit the specific value of the fourth preset speed threshold, and those skilled in the art can determine it according to the actual situation.

[0120] Optionally, the edge software can also receive threshold configuration information, which is used to configure the threshold used when determining that the first vehicle is suspected of being involved in a collision; the threshold includes at least one of the following: a first preset speed threshold, a second preset speed threshold, a third preset speed threshold, and a fourth preset speed threshold. Then, the edge software can configure the corresponding threshold according to the threshold configuration information.

[0121] Normally, if the first vehicle is involved in a collision, its speed will not recover for a short period of time after the sudden drop in speed. Instead, it will continue to decrease in speed or remain at a low speed level. Therefore, if the first vehicle's speed does not recover after the sudden drop in speed, that is, it does not recover within a certain period of time after the sudden drop, then step S2015 is executed.

[0122] If not, then it is determined that the speed change of the first vehicle within the time window does not meet the speed change required for a collision.

[0123] S2015. Determine that the speed change of the first vehicle within the time window meets the speed change behavior of the vehicle collision, and take the time and position corresponding to the (x+1)th positioning data as the suspected collision time and position of the first vehicle.

[0124] Since the speed change of the first vehicle within the time window satisfies the judgment conditions of steps S2011-S2015 above, the terminal software determines in this step that the speed change of the first vehicle within the time window meets the speed change required for a vehicle collision. Furthermore, since the first vehicle begins to experience a sudden speed drop at the (x+1)th positioning data point, the time and location corresponding to the (x+1)th positioning data point are used as the suspected collision time and location for the first vehicle. If it is determined that the speed change of the first vehicle within the time window meets the speed change required for a vehicle collision, then step S202 is executed.

[0125] S202, It is determined that the first vehicle is suspected of being involved in a collision.

[0126] In this way, the edge software can determine whether the first vehicle is suspected of being involved in a collision, thus achieving a preliminary judgment of collision detection.

[0127] Optionally, before step S202, the terminal software may first determine whether the first vehicle is located within a preset type of area. This step may be performed before S201, after S201, or simultaneously with S201.

[0128] For example, the edge software can determine whether the suspected collision location of the first vehicle is within a preset type of area. For instance, the edge software can pre-store identifiers for preset type areas, and then determine whether the first vehicle is within the preset type of area based on the suspected collision location and the road identifier where the first vehicle is located. If it is determined that the vehicle is within the preset type of area, it is determined that the first vehicle has not collided; if it is determined that the vehicle is not within the preset type of area, it can be determined whether a collision has occurred based on the result obtained from speed changes. In this scenario, the aforementioned preset type of area may include one or more of the following scenarios where the first vehicle's speed would change: highway service areas, gas stations, toll booths, etc.

[0129] Because vehicle driving characteristics and collision-related driving characteristics are similar in certain areas such as highway service areas and gas stations—for example, both experience a rapid decrease in speed—the above method can filter collision data. Subsequent actions are only performed when the first vehicle is not located within a preset type of area. This saves computational resources and further avoids the influence of similar data on the judgment results, improving the accuracy of determining whether a collision has occurred.

[0130] And / or, the edge software can determine whether the first vehicle is located within a preset type of area based on the first vehicle's location data within the time window and the navigation path. If it is determined that the first vehicle is located within the preset type of area, it is determined that the first vehicle has not collided. If it is determined that the first vehicle is not located within the preset type of area, it can be determined whether the first vehicle has collided based on the results obtained from speed changes. For example, it can determine whether the first vehicle is located within the preset type of area based on the last location data within the time window and the navigation path. In this scenario, the aforementioned preset type of area may include one or more of the following: roads of a preset road level, the start and end points of the first vehicle's current navigation path, etc. This application does not limit the specific level of the aforementioned preset road level; for example, it may be a road other than highways and urban expressways.

[0131] It should be noted that the types of information included in the navigation route information are related to the preset type of area. For example, if the preset type of area includes roads of a preset road level, then the navigation route information includes the road level of the road where the first vehicle is located within the time window.

[0132] Typically, when the first vehicle is traveling on highways or urban expressways, it rarely exhibits driving characteristics indicative of a collision, such as a sudden drop in speed. However, on some rural roads, where there are no speed limits, driving speeds can vary significantly, leading to situations where a collision does not occur, but the driving behavior is similar to that of a vehicle at the time of a collision. Furthermore, when the first vehicle is on the starting and ending points of the current navigation route, it may also exhibit driving characteristics similar to those of a collision, such as a sudden drop in speed. This application does not consider the above-mentioned vehicle driving scenarios similar to those of a collision; therefore, the areas encompassed by such scenarios can be added as preset types of areas.

[0133] In this way, the driving data of the first vehicle in scenarios similar to the collision scenario can be eliminated first. This reduces computational burden and improves collision detection accuracy. If not, the edge software / server can determine whether a collision is suspected based on whether the speed change of the first vehicle within the time window meets the speed change behavior of a collision.

[0134] It should be understood that in scenarios where the client-side software performs an initial screening to determine whether the first vehicle is suspected of being involved in a collision, the client-side software can either determine whether the first vehicle is located in a preset type of area corresponding to that scenario based solely on the suspected collision location, or it can make no such determination, or the server can make the determination. Alternatively, the client-side software can choose not to determine whether the vehicle is located in a preset type of area, leaving all processing to the server, or neither the client-side software nor the server can make any determinations. The specific settings can be configured according to actual needs.

[0135] In addition, the above-mentioned preset type of area is only an example. In actual implementation, one or more preset type areas can be divided according to actual needs.

[0136] Taking the server-side execution of the above judgment as an example, for instance, the server can determine whether the suspected collision location of the first vehicle is within a preset type of area before obtaining the number of at least one type of second vehicle. If it is determined that the location is within the preset type of area, then it is determined that the first vehicle has not collided, and the actions S102 and S103 are no longer executed.

[0137] After determining that the first vehicle is suspected of being involved in a collision, how to generate the collision data for the first vehicle is related to the data required for subsequent collision detection by the server.

[0138] For example, the collision data only includes: the suspected collision time and location of the first vehicle, and the road where the first vehicle is located. In this implementation, based on the aforementioned positioning data within the time window, it is possible to determine whether the speed change performance of the first vehicle within the time window meets the speed change required for a collision, the suspected collision time and location of the first vehicle, and the suspected collision road determined based on the suspected collision location and navigation path information, thus obtaining the collision data.

[0139] For example, in addition to the suspected collision time, suspected collision location, and suspected collision road, the collision data may also include one or more of the following: the first vehicle's inertial navigation data within the time window, the first vehicle's initial velocity within the time window, and the negative acceleration at the time of the suspected collision. Taking the inclusion of the aforementioned inertial navigation data, initial velocity, and negative acceleration as an example, in this implementation, the collision data can be further obtained by determining whether the speed change of the first vehicle within the time window meets the speed change behavior determined when the vehicle collides with the collision, based on the aforementioned positioning data within the time window, and by using the inertial navigation data collected by the edge software itself. The aforementioned inertial navigation data may include, for example, one or more of the following: the acceleration data of the first vehicle at various time points within the time window, and angular velocity data.

[0140] Optionally, when the edge software determines that the first vehicle is suspected of being involved in a collision based on the driving data within the time window, it can also acquire the driving data of the first vehicle within the time window before acquiring the collision data of the first vehicle. For example, the edge software can acquire acceleration data through a built-in accelerometer, acquire angular velocity data through a built-in gyroscope, and thus acquire the aforementioned inertial navigation data; acquire positioning data through GPS; and acquire navigation path information from the server.

[0141] The following example illustrates how, based on the collision data of the first vehicle and the trajectory data uploaded by other vehicles, at least one type of second vehicle can be obtained from other vehicles, i.e., the above-mentioned real-time step S102. Figure 5 A flowchart illustrating the fourth collision detection method provided in this application embodiment is shown below. Figure 5 As shown, step S102 may include the following steps:

[0142] S301. Based on the collision data of the first vehicle and the trajectory data uploaded by other vehicles, filter out other vehicles and their trajectory data that were traveling on the road where the first vehicle was suspected to have collided at the time of the suspected collision, or whose driving position was before or after the suspected collision location of the first vehicle.

[0143] Since the collision data of the first vehicle includes at least the suspected collision time, the suspected collision location, and the suspected collision road, and the trajectory data uploaded by other vehicles includes the location of trajectory points during their travels, the road markings, and the time they passed those trajectory points, the server can use the collision data of the first vehicle and the trajectory data uploaded by other vehicles to filter out other vehicles and their trajectory data that were traveling on the road where the first vehicle's suspected collision occurred at the time of the suspected collision, or whose travel positions were before or after the suspected collision location of the first vehicle. This method allows for an initial screening of other vehicles, avoiding the computational burden of directly determining the type of the second vehicle based on all other vehicle data.

[0144] For example, the server can determine the time range for trajectory filtering based on the suspected collision time of the first vehicle, and the location range for trajectory filtering based on the suspected collision location of the first vehicle. In this way, the spatiotemporal range of trajectory filtering can be obtained.

[0145] This application does not limit the specific time range covered by the trajectory filtering time range. For example, it can be any time range covering the suspected collision time of the first vehicle, such as 60 seconds before the suspected collision time to 30 seconds after the suspected collision time. This application does not limit the specific spatial range covered by the trajectory filtering location range. For example, it can be any spatial range covering the suspected collision location of the first vehicle, such as a rectangular area with a preset side length centered on the suspected collision location of the first vehicle. This application does not limit the specific value of the preset side length; for example, it can be 2 kilometers.

[0146] Then, the server retrieves the trajectory data of other vehicles whose trajectory time and location fall within the trajectory filtering time and location range from the trajectory data uploaded by other vehicles. Next, it performs road network matching on the retrieved other vehicles and their trajectory data to determine the roads that match the retrieved other vehicles. This road network matching refers to determining the travel routes of other vehicles within the trajectory filtering time range based on their trajectories. Specific implementation methods can be found in existing technologies and will not be elaborated upon here.

[0147] If the matched road is the same road where the first vehicle's suspected collision occurred, then it is determined that the other vehicles were traveling on the same road where the first vehicle's suspected collision occurred at the time of the first vehicle's suspected collision, and their driving positions were either before or after the first vehicle's suspected collision location.

[0148] Optionally, the server can first perform road network matching based on other vehicles and their trajectory data to determine the roads that match other vehicles, and obtain the trajectory data of other vehicles whose matched roads are the same as the road where the first vehicle's suspected collision occurred. Then, the server filters out other vehicles and their trajectory data whose trajectory time and trajectory position fall within the trajectory filtering time range and trajectory filtering position range, based on the obtained other vehicles and their trajectory data, to determine whether the other vehicle was traveling on the road where the first vehicle's suspected collision occurred at the time of the first vehicle's suspected collision, or whether its travel position was before or after the location of the first vehicle's suspected collision.

[0149] S302. Based on the acquired data of other vehicles and their trajectories, identify other vehicles whose driving behavior meets the preset behavior conditions as second vehicles and classify the second vehicles.

[0150] For example, Table 1 provides six types of second vehicles, as shown in Table 1:

[0151] Table 1

[0152]

[0153]

[0154] Table 1 illustrates the vehicle status, judgment method, and vehicle characteristics of six types of second vehicles. It should be noted that the types of second vehicles shown in Table 1 are merely examples. This application does not limit the classification of second vehicle types or the number of types; these can be set according to actual needs during implementation.

[0155] Taking the first type as an example, the determination method for this type is to identify other vehicles whose driving behavior meets the criteria of being in front of the suspected collision location of the first vehicle and traveling at high speed at the time of the suspected collision as second vehicles of the first type. That is, the vehicle status of second vehicles of this type is a vehicle traveling at high speed in front of the vehicle involved in the accident at the time of the accident. Other vehicles whose driving behavior does not meet the above behavioral conditions in the sixth type can include, for example, vehicles that are behind the suspected collision location of the first vehicle at the time of the suspected collision and whose speed is zero.

[0156] This application does not limit the criteria for determining the aforementioned fast driving behavior and slow driving behavior. For example, a driving speed exceeding 80 km / h can be considered fast driving behavior, while a driving speed not exceeding 20 km / h and greater than 0 km / h can be considered slow driving behavior. It should be noted that the criteria for determining slow driving behavior and fast driving behavior can be the same or different for different types, and this application does not limit them.

[0157] It should be noted that the specific value of the aforementioned set duration is related to the time coverage range after the suspected collision time of the first vehicle within the trajectory filtering time range. That is, the set duration is less than the time coverage range after the suspected collision time of the first vehicle within the trajectory filtering time range. Furthermore, the set durations for different types can be the same or different; for example, they can all be 30 seconds. This application does not limit this. This application does not limit the specific values ​​of the first preset distance threshold and the second preset distance threshold; those skilled in the art can determine them according to the actual situation.

[0158] For example, the server can, based on the acquired data of other vehicles and their trajectories, determine other vehicles located ahead of the suspected collision location of the first vehicle at the time of the suspected collision. Then, for the other vehicles ahead of the first vehicle, based on their trajectory data, it can determine whether their driving speed at the time of the suspected collision of the first vehicle meets the criteria of fast driving behavior or slow driving behavior.

[0159] Subsequently, the server can classify the second vehicle into Type 1 and Type 2 according to the classification method shown in Table 1. Alternatively, based on this, the server can determine other vehicles located behind the suspected collision location of the first vehicle when the suspected collision occurred, based on the acquired data of other vehicles and their trajectories. Then, for the other vehicles located behind, based on their trajectory data, the server can determine the relationship between their driving position after a set time period following the suspected collision of the first vehicle and the suspected collision location of the first vehicle, as well as whether their driving speed meets the criteria for fast or slow driving behavior, and further classify the second vehicle into the 6 types shown in Table 1.

[0160] Alternatively, the server can acquire at least one type of second vehicle based on the acquired data of other vehicles and their trajectories, as well as a pre-trained classification model.

[0161] In this embodiment, the server first filters out other vehicles and their trajectory data based on the collision data of the first vehicle and the trajectory data uploaded by other vehicles at the time of the suspected collision, or those traveling before or after the suspected collision location. This initial screening avoids the problems of directly classifying other vehicles, which involves numerous judgment conditions, high computational load, and low judgment accuracy. Subsequently, based on the acquired trajectory data of other vehicles, the server identifies and classifies other vehicles whose driving behavior meets preset behavioral conditions as second vehicles. Through this method, the server can obtain different types of second vehicles and their trajectory data to further determine the surrounding road conditions at the suspected collision location of the first vehicle, thereby accurately determining whether the first vehicle was involved in a collision.

[0162] The following describes how to determine whether the first vehicle has collided, based at least on the number of the second vehicles, i.e., step S103 in the above embodiment. Specific implementation methods include, but are not limited to, the following two implementation methods.

[0163] Method 1: The collision data includes one or more of the following: the inertial navigation data of the first vehicle within the time window to which the suspected collision occurred, the initial speed of the first vehicle within the time window, and the negative acceleration at the time of the suspected collision. The following example uses all three.

[0164] In this implementation, the collision detection model can be used to determine whether the first vehicle has collided based on the number of second vehicles, inertial navigation data, initial velocity, and negative acceleration, or a threshold method can be used to determine whether the first vehicle has collided.

[0165] Before using these dimensions of data to determine whether a collision has occurred, some dimensions of the data can be preprocessed, depending on the specific settings. For example, inertial navigation data can be preprocessed to obtain inertial navigation characteristic parameter values.

[0166] Figure 6 This is a flowchart illustrating the fifth collision detection method provided in this application. This embodiment uses the type of the second vehicle, including categories one through six, as an example. Figure 6 As shown, step S103 in the above embodiment may include the following steps:

[0167] S401. Based on the inertial navigation data, obtain the inertial navigation characteristic parameter values ​​of the first vehicle.

[0168] The aforementioned inertial navigation characteristic parameter values ​​are used to characterize the inertial navigation characteristics of the first vehicle. When the inertial navigation data includes acceleration data and angular velocity data, the inertial navigation characteristic parameter values ​​obtained from the acceleration data may include, for example: a) the maximum value of the three-axis amplitude of the accelerometer; b) the standard deviation of the smaller three-quarters of the three-axis amplitude of the accelerometer; c) the maximum value of the amplitude of the two horizontal axes of the accelerometer; d) the standard deviation of the smaller three-quarters of the amplitude of the two horizontal axes of the accelerometer.

[0169] The standard deviation of the smaller 3 / 4 of the accelerometer triaxial amplitude values ​​mentioned above is obtained by selecting the smaller 3 / 4 of the accelerometer triaxial amplitude values ​​of the first vehicle within the time window and then calculating the standard deviation of that portion of the data, and so on.

[0170] Inertial navigation characteristic parameter values ​​obtained from angular velocity data may include, for example: e, the maximum value of the gyroscope's three-axis amplitude; f, the standard deviation of the smaller three-quarters of the gyroscope's three-axis amplitude data; g, the maximum value of the gyroscope's two horizontal axis amplitudes; h, the standard deviation of the smaller three-quarters of the gyroscope's two horizontal axis amplitude data.

[0171] It should be noted that this application does not limit the specific content of the inertial navigation characteristic parameter values ​​of the first vehicle; that is, the inertial navigation parameter values ​​may include any one or more of the above-mentioned ah values, or other characteristic parameter values. Furthermore, this application does not limit the specific data content included in the above-mentioned items b, d, g, and h, specifically the accelerometer's three-axis amplitude data, accelerometer's two-horizontal-axis amplitude data, gyroscope's three-axis amplitude data, and the gyroscope's two-horizontal-axis amplitude data from which the standard deviation is obtained. For example, item b could also be the standard deviation of the smaller 3 / 5 of the accelerometer's three-axis amplitude data, and so on.

[0172] S402. Based on the inertial navigation characteristic parameter values, the number of second vehicles in categories one through six, the initial velocity, and the negative acceleration, determine whether the first vehicle has collided.

[0173] One possible implementation is that the server can set up a collision detection model corresponding to each dimension of data. Therefore, the server can input the data of each dimension into the corresponding collision detection model to obtain the initial collision detection results, and obtain the final collision detection results based on each initial collision detection result.

[0174] For example, the number of vehicles in categories 1 to 6 (second category) is input into the corresponding collision detection model to obtain an initial collision detection result; the inertial navigation feature parameter values ​​are input into the corresponding collision detection model to obtain another initial collision detection result; the initial velocity and negative acceleration are input into the corresponding collision detection model to obtain another initial collision detection result. Then, the three initial collision detection results are weighted and summed to obtain the final collision detection result.

[0175] Another possible implementation is that the server can set up a collision detection model. Therefore, the server can fuse the features of data from various dimensions and then input the collision detection model to obtain the collision detection results.

[0176] It should be understood that the collision detection model described above can be any classification model.

[0177] For example, the server takes the inertial navigation feature parameter values, the number of second vehicles in categories one through six, and the initial velocity and negative acceleration as input, and uses a collision detection model built based on the random forest algorithm to output a judgment result representing whether the first vehicle has collided.

[0178] This application does not limit the specific form of the collision detection model built based on the random forest algorithm; those skilled in the art can build one according to actual conditions. The specific model construction method, and the method for determining whether the first vehicle has collided using the aforementioned collision detection model, can refer to existing technologies and will not be elaborated here.

[0179] Collision detection models built using the random forest method are lightweight, and their training process is often simple and easy to operate, as well as easy to deploy. Furthermore, compared to formulaic methods, the random forest method offers higher accuracy and robustness; therefore, collision detection models built using the random forest method are highly usable for collision detection.

[0180] For example, the server determines whether the first vehicle has collided based on the inertial navigation feature parameter values, the number of second vehicles in categories one through six, and the initial velocity and negative acceleration using a collision detection model built on the Transformer model.

[0181] Since neural network models such as Transformer often have high accuracy, collision detection using collision detection models built on Transformer models can further improve detection accuracy.

[0182] Another possible implementation involves using a threshold method to determine whether the first vehicle has collided, based on inertial navigation characteristic parameter values, the number of second vehicles in categories one through six, initial velocity, and negative acceleration. That is, each dimension of data has a corresponding judgment condition. For example, if all four data points meet their corresponding judgment conditions, it is determined that the first vehicle has collided; if any condition is not met, it is determined that the first vehicle has not collided. The specific judgment conditions for each dimension of data, and under what circumstances a collision is determined, can be set according to the actual application scenario.

[0183] It should be noted that when the server uses the classification model to determine whether the first vehicle has collided, this application does not limit the specific training method of the classification model used. For example, it can construct a sample dataset with positive sample data (i.e., data that actually collided) and negative sample data (data that did not actually collide), and then use the sample dataset to train the model. The specific implementation method can refer to the existing technology, which will not be elaborated here.

[0184] This application does not limit the methods for obtaining the positive sample data and negative sample data. For example, positive sample data could be obtained by acquiring the time and location of the vehicles involved in the actual collision, then filtering vehicles matching the time and location from historically stored vehicle-reported trajectory data, and further filtering these vehicles (e.g., using the aforementioned method for determining whether a first vehicle is suspected of being involved in a collision). The filtered vehicles would then be manually confirmed, and the data of the finally manually confirmed vehicles would be used as positive sample data. This application does not limit the specific method for obtaining the time and location of the vehicles involved in the actual collision; for example, it could be obtained through online searches or relevant announcements. For negative sample data, for example, it could be vehicle-related data obtained from historically stored vehicle-reported trajectory data. Through the above methods, a highly reliable dataset can be used to train the model, further improving the accuracy of the model's output.

[0185] Method 2: Determine whether the first vehicle was involved in a collision based solely on the number of the second vehicles.

[0186] In this scenario, the collision data does not need to include one or more of the following: inertial navigation data, initial velocity, and negative acceleration.

[0187] It should be noted that the specific types of second vehicles obtained by the server in the above embodiments are consistent with the types used by the server in this embodiment to determine whether the first vehicle has collided. Alternatively, the server may select one or more of the aforementioned types of second vehicles for subsequent judgment.

[0188] Furthermore, as mentioned above, in this implementation, the server can use a threshold method or any other classification method to determine whether the first vehicle has collided. The following example, using the first and second types of the second vehicle as examples, illustrates how to use a threshold method to determine whether the first vehicle has collided based on the number of second vehicles.

[0189] Figure 7 A flowchart illustrating the sixth collision detection method provided in this application embodiment is shown below. Figure 7 As shown, step S103 may include the following steps:

[0190] S501. Based on the number of second vehicles, determine whether the number of second vehicles in the first category is greater than or equal to a first preset threshold, and whether the number of second vehicles in the second category is less than or equal to a second preset threshold.

[0191] As described in Table 1 above, the first type of second vehicle refers to other vehicles whose driving behavior conforms to the condition of being in front of the suspected collision location of the first vehicle and traveling at high speed when the suspected collision of the first vehicle occurs. The second type of second vehicle refers to other vehicles whose driving behavior conforms to the condition of being in front of the suspected collision location of the first vehicle and traveling slowly when the suspected collision of the first vehicle occurs. This application does not limit the specific value of the first preset threshold, nor does it limit the specific value of the second preset threshold, which can be determined by those skilled in the art based on the actual situation.

[0192] As mentioned above, the driving patterns of vehicles in traffic jams are often quite similar to those during a collision. For example, in a collision, a vehicle's speed typically drops sharply within a short period, and the speed usually remains within a low threshold range after the collision. Similarly, in traffic jams, vehicles also experience a rapid decrease in speed due to blocked access, and may even come to a complete stop in the congestion, similar to the driving patterns during a collision. Therefore, in traffic jams, relying solely on one's own vehicle's driving data can easily lead to a misjudgment of a collision.

[0193] However, the driving conditions of other vehicles around a vehicle differ when a collision occurs versus when there is congestion. For example, in a congested situation, most of the other vehicles in front of the vehicle are traveling at slower speeds, with fewer vehicles moving at high speeds; in a collision, most of the other vehicles in front of the vehicle are typically traveling at normal speeds.

[0194] Therefore, in this embodiment, it is determined whether the number of second vehicles of the first type is greater than or equal to a first preset threshold, and whether the number of second vehicles of the second type is less than or equal to a second preset threshold. If so, it indicates that at the time of the suspected collision of the first vehicle, in front of the suspected collision location of the first vehicle, most of the second vehicles were driving normally, that is, there was no congestion. It can be determined that the first vehicle suspected of colliding did indeed collide, and then step S502 is executed.

[0195] If not, it indicates that at the time of the suspected collision of the first vehicle, in front of the suspected collision location of the first vehicle, the second vehicle was traveling at a slower speed, indicating congestion. If the first vehicle is determined to have been suspected of colliding due to congestion, then step S503 is executed.

[0196] S502, It is determined that the first vehicle has collided.

[0197] S503, It is determined that the first vehicle did not collide.

[0198] In this implementation, the server uses a threshold method to determine whether the first vehicle has collided. There is no need for a complex model training process in the early stage, the judgment process is simple, and the collision detection efficiency is improved.

[0199] The method of this application embodiment will be described below through a specific example. In the following example, the edge software of the electronic device deployed on the first vehicle is a map application of the collision detection system. By running the map application, the electronic device can obtain driving data of the first vehicle within a time window, and detect whether the first vehicle is suspected of being involved in a collision based on the driving data within the time window, that is, to perform an initial screening for whether the first vehicle is suspected of being involved in a collision. When the initial screening determines that the first vehicle is suspected of being involved in a collision, the collision data of the first vehicle is obtained and uploaded to the server of the collision detection system along with the collision detection request. The edge software can collect data through GPS, accelerometer sensors, and gyroscope sensors installed on the electronic device, or some sensors can be deployed on the first vehicle and communicate with the edge software to send the collected data to the edge software for processing.

[0200] Accordingly, the server deploys a collision detection service. Upon receiving a collision detection request carrying collision data of the first vehicle from the client software, the server can make a final judgment on whether the first vehicle has been involved in a collision based on the collision data of the first vehicle and the number of second vehicles that were traveling on the road where the suspected collision occurred at the time of the suspected collision, were traveling before or after the suspected collision location, and whose driving behavior meets preset behavioral conditions. That is, the final judgment is made in conjunction with the road conditions surrounding the first vehicle. This application does not limit the specific form of the collision detection service provided by the server; for example, it can be an APP or a web service.

[0201] The following describes another embodiment of the collision detection method provided in this application. This embodiment is applied to a road rescue scenario. The same or similar content as the above embodiments can be referred to the above embodiments, and will not be repeated here. Figure 8 A flowchart illustrating the seventh collision detection method provided in this application embodiment is shown below. Figure 8 As shown, this embodiment may include the following steps:

[0202] S601: The terminal software determines whether the first vehicle is located in the corresponding preset type area based on the navigation path information and the positioning data within the time window.

[0203] like Figure 8 As shown, the navigation path information includes the road class of the road where the first vehicle is located in the time window, as well as the origin and destination paths of the current navigation path.

[0204] For example, the preset type of area involved in this step may include one or more of the following: roads of a preset road level, the start and end points of the first vehicle's current navigation path, etc. If the first vehicle is not located within the corresponding preset type of area, then step S602 is executed. If it is located within the corresponding preset type of area, then the process ends.

[0205] S602. The end-side software determines whether the speed change of the first vehicle within the time window meets the speed change performance required for a vehicle collision based on the positioning data within the time window.

[0206] It should be noted that the execution order of the above steps S601 and S602 is not important.

[0207] If the conditions are met, proceed to step S603.

[0208] S603, the end-side software determines that the first vehicle is suspected of being involved in a collision and obtains collision data.

[0209] The collision data includes: the suspected time of the collision, the suspected location of the collision, the suspected road where the collision occurred, the inertial navigation data of the first vehicle within the time window, the initial velocity of the first vehicle within the time window, and the negative acceleration at the time of the suspected collision. The aforementioned inertial navigation data includes acceleration data and angular velocity data.

[0210] S604: The client-side software packages the collision data and sends the packaged collision data to the server in the collision detection request.

[0211] Accordingly, the server receives the collision detection request.

[0212] This application does not limit the packaging method. For example, it can be compressed and packaged using zstd compression and base64 encoding technology to reduce data transmission costs.

[0213] This application does not limit the method of transmission. For example, when the collision detection service is in the form of a web service, the packaged collision data can be sent to the server in the form of an HTTP protocol.

[0214] S605. The server unpacks the packaged collision data in the collision detection request and obtains the collision data.

[0215] S606. The server determines whether the suspected collision location of the first vehicle is within the corresponding preset type of area.

[0216] For example, the preset type of area involved in this step may include one or more of the scenarios in which the first vehicle's speed will change, such as highway service areas, gas stations, and toll stations.

[0217] S607. The server obtains at least one type of second vehicle based on the suspected collision time, suspected collision location, and suspected collision road of the first vehicle in the collision data, as well as the trajectory data uploaded by other vehicles. Figure 8 This is a schematic diagram using the aforementioned six types of relative spatiotemporal relationships as examples.

[0218] S608. The server obtains the inertial navigation characteristic parameter values ​​of the first vehicle based on the inertial navigation data.

[0219] The inertial navigation characteristic parameter values ​​of the first vehicle mentioned above include the ah characteristic parameter values ​​described in the above embodiments.

[0220] S609. The server determines whether the first vehicle has collided based on the inertial navigation characteristic parameter values, the number of second vehicles, the initial speed, and the negative acceleration.

[0221] In this step, the server can use a collision detection model built based on the random forest method, or a threshold method, to determine whether the first vehicle has collided.

[0222] S610, the server sends a collision detection response to the client software.

[0223] Correspondingly, the edge software receives the collision detection response.

[0224] S611: The terminal software determines whether to output a request to trigger roadside assistance service based on the collision detection response.

[0225] If the collision detection response indicates that the first vehicle has collided, then step S612 is executed.

[0226] S612, Output whether to trigger a request for roadside assistance service.

[0227] In this embodiment, on the one hand, the server determines whether the first vehicle has collided based on its own collision data and the trajectory data of the second vehicles surrounding the first vehicle within the time window. As mentioned above, this method can more comprehensively detect whether a collision has occurred, improving detection accuracy. On the other hand, in this embodiment, the client-side software performs a collision detection based on its own vehicle's driving data. Only when it determines that the first vehicle is suspected of having collided will it send the collision data to the server for further determination. This method can reduce the computational burden on the client-side software, reduce the amount of data uploaded to the cloud, and improve the overall collision detection efficiency.

[0228] Figure 9 This is a schematic diagram of a collision detection device provided in an embodiment of this application. Figure 9As shown, the device includes: a first acquisition module 11, a second acquisition module 12, a statistics module 13, and a determination module 14. Optionally, the device may further include the following module: a sending module 15.

[0229] The first acquisition module 11 is used to acquire collision data sent by the first vehicle when a suspected collision occurs. The collision data includes at least: the time of the suspected collision, the location of the suspected collision, and the road where the suspected collision occurred.

[0230] The second acquisition module 12 is used to acquire at least one type of second vehicle from the other vehicles based on the collision data of the first vehicle and the trajectory data uploaded by other vehicles. The second vehicle is another vehicle that was driving on the road where the first vehicle was suspected to have collided at the time of the suspected collision, whose driving position was before or after the suspected collision position of the first vehicle, and whose driving behavior met the preset behavior conditions. The type of the second vehicle is related to the driving position before or after the suspected collision position of the first vehicle and the driving behavior.

[0231] The statistics module 13 is used to count the number of second vehicles of different types.

[0232] The determining module 14 is used to determine whether the first vehicle has collided, at least based on the number of the second vehicles.

[0233] One possible implementation is that the second acquisition module 12 is specifically used to filter out other vehicles and their trajectory data that were driving on the road where the first vehicle was suspected of colliding at the time of the suspected collision of the first vehicle and whose driving position was before or after the suspected collision location of the first vehicle, based on the collision data of the first vehicle and the trajectory data uploaded by other vehicles; based on the acquired other vehicles and their trajectory data, determine other vehicles whose driving behavior meets the preset behavior conditions as the second vehicles and classify the second vehicles.

[0234] For example, the second acquisition module 12 is specifically used to determine the trajectory filtering time range based on the suspected collision time of the first vehicle; determine the trajectory filtering location range based on the suspected collision location of the first vehicle; acquire other vehicles and their trajectory data whose trajectory time and trajectory location fall within the trajectory filtering time range and trajectory filtering location range from the trajectory data uploaded by other vehicles; perform road network matching on the acquired other vehicles and their trajectory data to determine the road matched by the acquired other vehicles; if the matched road is the same road as the suspected collision location of the first vehicle, then it is determined that the acquired other vehicles were traveling on the suspected collision location of the first vehicle at the suspected collision time of the first vehicle, and their travel position was before or after the suspected collision location of the first vehicle.

[0235] For example, the second acquisition module 12 is specifically used to determine, based on the acquired other vehicles and their trajectory data, other vehicles located in front of the suspected collision location of the first vehicle when the suspected collision occurred; for other vehicles located in front of the first vehicle, based on their trajectory data, determine whether their driving speed at the time of the suspected collision of the first vehicle meets the criteria of fast driving behavior or slow driving behavior; identify other vehicles whose driving behavior conforms to being in front of the suspected collision location of the first vehicle and driving fast as first-class second vehicles; and identify other vehicles whose driving behavior conforms to being in front of the suspected collision location of the first vehicle and driving slowly as second-class second vehicles.

[0236] One possible implementation is that the determining module 14 is specifically used to determine whether the number of the first type of second vehicles is greater than or equal to a first preset threshold, and whether the number of the second type of second vehicles is less than or equal to a second preset threshold; if not, then determine that the first vehicle has not collided; if yes, then determine that the first vehicle has collided.

[0237] In one possible implementation, the second acquisition module 12 is further configured to, based on the acquired other vehicles and their trajectory data, determine other vehicles located behind the suspected collision location of the first vehicle at the time of the suspected collision; for the other vehicles located behind the first vehicle, based on their trajectory data, determine the relationship between their driving position after a set time period following the suspected collision of the first vehicle and the suspected collision location of the first vehicle, and whether their driving speed meets the criteria of fast driving behavior or slow driving behavior; and specify the driving behavior that conforms to being located behind the suspected collision location of the first vehicle at the time of the suspected collision, and whose distance from the suspected collision location of the first vehicle after a set time period following the suspected collision of the first vehicle is greater than the first vehicle's suspected collision location. Other vehicles with a preset distance threshold are identified as Category 3 second vehicles; other vehicles whose driving behavior meets the following criteria are identified as: being located behind the suspected collision location of the first vehicle when the suspected collision occurred, being less than the second preset distance threshold after a set time period following the suspected collision, and driving at high speed; other vehicles whose driving behavior meets the following criteria are identified as: being located behind the suspected collision location of the first vehicle when the suspected collision occurred, driving slowly, and remaining behind the suspected collision location after a set time period following the suspected collision; and other vehicles whose driving behavior does not meet the above criteria are identified as Category 6 second vehicles.

[0238] In one possible implementation, the collision data further includes: inertial navigation data of the first vehicle within the time window to which the suspected collision occurred, the initial speed of the first vehicle within the time window, and the negative acceleration at the time of the suspected collision.

[0239] For example, the determining module 14 is specifically used to obtain the inertial navigation feature parameter value of the first vehicle based on the inertial navigation data; and to determine whether the first vehicle has collided based on the inertial navigation feature parameter value, the number of second vehicles of the first to sixth categories, the initial velocity, and the negative acceleration.

[0240] For example, the determining module 14 is specifically used to determine whether the first vehicle has collided based on the inertial navigation feature parameter values, the number of second vehicles in the first to sixth categories, the initial velocity, and the negative acceleration, using a collision detection model constructed based on the random forest algorithm.

[0241] One possible implementation is that the first acquisition module 11 is specifically used to receive a collision detection request sent by the terminal software. The collision detection request is triggered when the terminal software determines that the first vehicle is suspected of being involved in a collision based on the driving data of the first vehicle within a time window. The collision detection request includes the collision data.

[0242] In one possible implementation, the sending module 15 is used to send a collision detection response to the end-side software. The collision detection response is used to indicate whether a collision has occurred with the first vehicle, so that the end-side software outputs a request to trigger roadside assistance service when the first vehicle has collided.

[0243] The collision detection device provided in this application embodiment can execute the collision detection method executed by the server in the above method embodiment. Its implementation principle and technical effects are similar, and will not be repeated here. It should be noted that the above... Figure 9 The division of modules shown is merely illustrative. This application does not limit the division of modules or the naming of modules.

[0244] Figure 10 This is a schematic diagram of a collision detection device provided in an embodiment of this application. Figure 10 As shown, the device includes: a first acquisition module 21, a second acquisition module 22, a sending module 23, and a receiving module 24. Optionally, the device may also include the following module: an output module 25.

[0245] The first acquisition module 21 is used to acquire the driving data of the first vehicle within the time window.

[0246] The second acquisition module 22 is used to detect whether the first vehicle is suspected of being involved in a collision based on the driving data within the time window. If so, it acquires the collision data of the first vehicle. The collision data includes at least: the suspected collision time, the suspected collision location, and the suspected collision road.

[0247] The sending module 23 is used to send a collision detection request carrying the collision data to the server.

[0248] The receiving module 24 is used to receive a collision detection response returned by the server based on the collision detection request. The collision detection response carries a detection result indicating whether the first vehicle has collided. The detection result is determined based on the number of second vehicles. The second vehicles are other vehicles that were driving on the road where the first vehicle was suspected of colliding at the time of the suspected collision, were driving before or after the suspected collision location of the first vehicle, and whose driving behavior meets preset behavior conditions.

[0249] One possible implementation is that the driving data within the time window includes: location data and navigation path information. The second acquisition module 22 is specifically used to determine whether the first vehicle is located in a preset type of area based on the navigation path information and the location data within the time window; if the first vehicle is not located in the preset type of area, then based on the location data within the time window, determine whether the speed change of the first vehicle within the time window meets the speed change performance of a vehicle collision; if it does, then determine that the first vehicle is suspected of having been in a collision.

[0250] For example, the second acquisition module 22 is specifically used to determine whether the initial speed of the first vehicle within the time window is greater than a first preset speed threshold based on the first to xth positioning data within the time window; if so, it determines whether the final speed of the first vehicle within the time window is less than a second preset speed threshold based on the yth to the last positioning data within the time window; if so, it determines whether target positioning data exists based on the (x+1)th to (x+n)th positioning data within the time window; if so, it determines whether the speed of the first vehicle in that time period is less than a fourth preset speed threshold based on the target positioning data to the (y-1)th positioning data within the time window; if so, it determines that the speed change of the first vehicle within the time window meets the speed change behavior of a vehicle collision, and uses the time and location corresponding to the (x+1)th positioning data as the suspected collision time and location of the first vehicle. Wherein, y > x; the target positioning data is the positioning data of the first vehicle when the negative acceleration is greater than or equal to a third preset speed threshold; n is an integer greater than or equal to 2; and the fourth preset speed threshold is greater than the second preset speed threshold.

[0251] For example, the second acquisition module 22 is specifically used to perform interpolation processing on the (x+1)th to (x+n)th positioning data of the time window; and to determine whether there is target positioning data among the (x+1)th to (x+n)th positioning data based on the interpolated positioning data.

[0252] In one possible implementation, the collision data further includes: inertial navigation data of the first vehicle within the time window, the initial speed of the first vehicle within the time window, and the negative acceleration at the time of the suspected collision.

[0253] One possible implementation is that the output module 25 is configured to output a request if the collision detection response indicates that the first vehicle has collided; the request is used to inquire whether to trigger roadside assistance service.

[0254] The collision detection device provided in this application embodiment can execute the collision detection method executed by the end-side software in the above method embodiment. Its implementation principle and technical effects are similar, and will not be repeated here. It should be noted that the above... Figure 10 The division of modules shown is merely illustrative. This application does not limit the division of modules or the naming of modules.

[0255] Figure 11 This is a schematic diagram of the structure of an electronic device 110 provided in an embodiment of this application. Figure 11 As shown, the electronic device 110 may include at least one processor 111 and a memory 112.

[0256] The memory 112 is used to store programs. Specifically, the program may include program code, which includes computer operation instructions.

[0257] The memory 112 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0258] The processor 111 is used to execute computer execution instructions stored in the memory 112 to implement the collision detection method executed by the server and / or the collision detection method executed by the client software as described in the foregoing method embodiments. The processor 111 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application.

[0259] The electronic device 110 may also include a communication interface 113, through which it can communicate and interact with external devices, such as other electronic devices (e.g., mobile phones, navigators) or servers. In specific implementations, if the communication interface 113, memory 112, and processor 111 are implemented independently, they can be interconnected via a bus to complete communication. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc., but this does not imply that there is only one bus or one type of bus.

[0260] Optionally, in a specific implementation, if the communication interface 113, memory 112, and processor 111 are integrated on a single chip, then the communication interface 113, memory 112, and processor 111 can communicate through an internal interface.

[0261] Figure 12 This is a schematic diagram of a collision detection system provided in an embodiment of this application. Figure 12 As shown, the collision detection system includes: end-side software 31 and server-side software 32.

[0262] The edge software 31 is used to execute the collision detection method executed by the edge software as described in the foregoing method embodiments.

[0263] Server 32 is used to execute the collision detection method executed by the server as described in the foregoing method embodiments.

[0264] The collision detection system provided in this application embodiment can execute the collision detection method in the above method embodiment, and its implementation principle and technical effect are similar, so they will not be described again here. It should be noted that the above... Figure 12 The division of modules shown is merely illustrative. This application does not limit the division of modules or the naming of modules.

[0265] This application also provides a computer-readable storage medium, which may include various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk. Specifically, the computer-readable storage medium stores program instructions, which are used in the collision detection method described in the above embodiments.

[0266] This application also provides a computer program product including executable instructions stored in a readable storage medium. At least one processor of an electronic device can read the executable instructions from the readable storage medium, and the at least one processor executes the executable instructions to cause a server to implement the collision detection methods provided in the various embodiments described above.

[0267] The term "multiple" in this document refers to two or more. The term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. Furthermore, the character " / " in this document generally indicates an "or" relationship between the preceding and following related objects; in formulas, " / " indicates a "division" relationship. Additionally, it should be understood that in the description of this application, words such as "first" and "second" are used only for descriptive purposes and should not be construed as indicating or implying relative importance or order.

[0268] It is understood that the various numerical designations used in the embodiments of this application are merely for descriptive convenience and are not intended to limit the scope of the embodiments of this application.

[0269] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A collision detection method, characterized in that, The method includes: Acquire collision data sent by the first vehicle when the suspected collision occurs, wherein the collision data includes at least: the time of the suspected collision, the location of the suspected collision, and the road on which the suspected collision occurred; Based on the collision data of the first vehicle and the trajectory data uploaded by other vehicles, at least one type of second vehicle is obtained from the other vehicles. The second vehicle includes other vehicles that are driving on the road where the first vehicle is suspected of colliding at the time of the suspected collision, driving in front of the suspected collision location of the first vehicle, and whose driving behavior meets preset behavior conditions. The types of the second vehicles include a first type and a second type. The first type of second vehicle is other vehicles that are in front of the suspected collision location of the first vehicle and are driving fast at the time of the suspected collision. The second type of second vehicle is other vehicles that are in front of the suspected collision location of the first vehicle and are driving slowly at the time of the suspected collision. The number of second vehicles of different types is counted, and the number of second vehicles of different types is used to reflect the surrounding road conditions of the first vehicle; Whether the first vehicle was involved in a collision can be determined at least based on the number of the second vehicles.

2. The method according to claim 1, characterized in that, The second vehicle also includes other vehicles that were traveling on the road where the suspected collision occurred at the time of the suspected collision of the first vehicle, were located after the suspected collision location of the first vehicle, and whose driving behavior met the preset behavioral conditions.

3. The method according to claim 1, characterized in that, The step of obtaining at least one type of second vehicle from the other vehicles based on the collision data of the first vehicle and the trajectory data uploaded by other vehicles includes: Based on the collision data of the first vehicle and the trajectory data uploaded by other vehicles, other vehicles and their trajectory data that were traveling on the road where the first vehicle was suspected to have collided at the time of the suspected collision of the first vehicle, or whose driving position was before or after the suspected collision location of the first vehicle. Based on the acquired data of other vehicles and their trajectories, other vehicles whose driving behavior meets the preset behavioral conditions are identified as second vehicles and then classified.

4. The method according to claim 3, characterized in that, The process, based on the collision data of the first vehicle and trajectory data uploaded by other vehicles, filters out other vehicles and their trajectory data that were traveling on the road where the suspected collision occurred at the time of the suspected collision of the first vehicle, or whose driving positions were before or after the suspected collision location of the first vehicle. This includes: Based on the suspected collision time of the first vehicle, the time range for trajectory screening is determined; Based on the suspected collision location of the first vehicle, determine the range of trajectory filtering locations; From the trajectory data uploaded by other vehicles, obtain the trajectory time and trajectory location of other vehicles and their trajectory data that fall within the trajectory filtering time range and trajectory filtering location range; The acquired other vehicles and their trajectory data are matched with the road network to determine the roads that match the acquired other vehicles. If the matched road is the same road as the suspected collision location of the first vehicle, then it is determined that the other vehicles were traveling on the suspected collision location of the first vehicle at the time of the suspected collision, and their travel positions were before or after the suspected collision location of the first vehicle.

5. The method according to claim 4, characterized in that, Based on the acquired data of other vehicles and their trajectories, other vehicles whose driving behavior conforms to preset behavioral conditions are identified as second vehicles and classified, including: Based on the acquired data of other vehicles and their trajectories, other vehicles located in front of the suspected collision location of the first vehicle at the time of the suspected collision are identified. For other vehicles located in front of the first vehicle, based on their trajectory data, determine whether their driving speed at the time of the suspected collision with the first vehicle meets the criteria for fast driving behavior or slow driving behavior. Other vehicles whose driving behavior is consistent with that of the first vehicle at the time of the suspected collision and that are traveling at high speed are identified as the second vehicles in the first category. Other vehicles whose driving behavior is consistent with that of the first vehicle at the time of the suspected collision and that are moving slowly in front of the first vehicle at the suspected collision location are identified as second vehicles in the second category.

6. The method according to claim 5, characterized in that, Determining whether the first vehicle has been involved in a collision, based at least on the number of the second vehicles, includes: Determine whether the number of second vehicles in the first category is greater than or equal to a first preset threshold, and whether the number of second vehicles in the second category is less than or equal to a second preset threshold; If not, then it is determined that the first vehicle did not collide; If so, then it is determined that the first vehicle has collided.

7. The method according to claim 5, characterized in that, The method further includes: Based on the acquired data of other vehicles and their trajectories, other vehicles located behind the suspected collision location of the first vehicle at the time of the suspected collision are identified. For other vehicles located behind the first vehicle, based on their trajectory data, determine the relationship between their driving position at a set time after the suspected collision with the first vehicle and the suspected collision location of the first vehicle, as well as whether their driving speed meets the requirements of fast driving behavior or slow driving behavior. Other vehicles whose driving behavior is consistent with being located behind the suspected collision location of the first vehicle at the time of the suspected collision, and whose distance from the suspected collision location of the first vehicle is greater than a first preset distance threshold after a set time period following the suspected collision, are identified as the third type of second vehicle. Other vehicles whose driving behavior meets the criteria of being located behind the suspected collision location of the first vehicle at the time of the suspected collision, being less than a second preset distance threshold from the suspected collision location of the first vehicle after a set time period following the suspected collision, and being traveling at high speed, are identified as the fourth type of second vehicle. Other vehicles whose driving behavior is consistent with being slowly driving behind the suspected collision location of the first vehicle at the time of the suspected collision and still being behind the suspected collision location of the first vehicle after a set time period following the suspected collision are identified as Class 5 Second Vehicles. Other vehicles whose driving behavior does not meet the above-mentioned behavioral conditions are classified as Class 6, Second Vehicle.

8. The method according to claim 7, characterized in that, The collision data also includes: the inertial navigation data of the first vehicle within the time window to which the suspected collision occurred, the initial speed of the first vehicle within the time window, and the negative acceleration at the time of the suspected collision; Determining whether the first vehicle has been involved in a collision, based at least on the number of the second vehicles, includes: Based on the inertial navigation data, obtain the inertial navigation characteristic parameter values ​​of the first vehicle; Based on the inertial navigation characteristic parameter values, the number of second vehicles in categories one through six, the initial velocity, and the negative acceleration, it is determined whether the first vehicle has collided.

9. The method according to claim 8, characterized in that, The step of determining whether the first vehicle has collided based on the inertial navigation characteristic parameter values, the number of second vehicles in categories one to six, the initial velocity, and the negative acceleration includes: Based on the inertial navigation feature parameter values, the number of second vehicles in categories one through six, and the initial velocity and negative acceleration, a collision detection model based on the random forest algorithm is used to determine whether the first vehicle has collided.

10. A collision detection method, characterized in that, The method includes: Obtain the driving data of the first vehicle within the time window; Based on the driving data within the time window, it is detected whether the first vehicle is suspected of being involved in a collision. If so, the collision data of the first vehicle is obtained. The collision data includes at least: the suspected collision time, the suspected collision location, and the suspected collision road. Send a collision detection request carrying the collision data to the server; The system receives a collision detection response from the server based on the collision detection request. The collision detection response carries a detection result indicating whether the first vehicle has collided. The detection result is determined based on the number of different types of second vehicles. The second vehicles are at least one type of vehicle obtained from other vehicles based on the collision data of the first vehicle and the trajectory data uploaded by other vehicles. The second vehicles are other vehicles that were traveling on the road where the first vehicle was suspected of colliding at the time of the suspected collision, were traveling in front of the suspected collision location of the first vehicle, and whose driving behavior meets preset behavior conditions. The number of different types of second vehicles is used to reflect the surrounding road conditions of the first vehicle. The types of second vehicles include a first type and a second type. The first type of second vehicle is other vehicles that were in front of the suspected collision location of the first vehicle and were traveling at high speed when the first vehicle was suspected of colliding. The second type of second vehicle is other vehicles that were in front of the suspected collision location of the first vehicle and were traveling slowly when the first vehicle was suspected of colliding.

11. The method according to claim 10, characterized in that, The driving data within the time window includes: location data and navigation route information. The step of detecting whether the first vehicle is suspected of being involved in a collision based on the driving data within the time window includes: Based on the navigation path information and the location data within the time window, determine whether the first vehicle is located in an area of ​​a preset type; If the first vehicle is not located in a preset area, then based on the positioning data within the time window, it is determined whether the speed change of the first vehicle within the time window meets the speed change performance of a vehicle collision. If the conditions are met, it is determined that the first vehicle is suspected of being involved in a collision.

12. The method according to claim 11, characterized in that, The step of determining whether the speed change of the first vehicle within the time window meets the speed change behavior required for a vehicle collision based on the positioning data within the time window includes: Based on the first to xth location data within the time window, determine whether the initial speed of the first vehicle within the time window is greater than a first preset speed threshold. If so, then based on the y-th positioning data to the last positioning data within the time window, determine whether the final speed of the first vehicle within the time window is less than the second preset speed threshold; y > x; If so, then determine whether there is target positioning data based on the (x+1)th to (x+n)th positioning data of the time window; the target positioning data is the positioning data of the first vehicle when the negative acceleration is greater than or equal to the third preset speed threshold; n is an integer greater than or equal to 2; If it exists, then based on the target positioning data up to the (y-1)th positioning data within the time window, determine whether the speed of the first vehicle in that time period is less than the fourth preset speed threshold, wherein the fourth preset speed threshold is greater than the second preset speed threshold. If so, the speed change of the first vehicle within the time window is determined to meet the speed change behavior of a vehicle collision, and the time and location corresponding to the (x+1)th positioning data are taken as the suspected collision time and location of the first vehicle.

13. A collision detection system, characterized in that, The system includes: client-side software and server-side software, wherein... The endpoint software is used to perform the method as described in any one of claims 10-12; The server is used to perform the method as described in any one of claims 1-9.

14. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-12.

15. A computer program product, characterized in that, It includes a computer program or instructions that, when executed by a processor, implement the method of any one of claims 1-12.

Images