Collision prediction method and device and readable storage medium

Abstract

The invention discloses a collision prediction method and device and a readable storage medium, and relates to the technical field of intelligent driving, the method is applied to a first vehicle, and the method comprises the steps: obtaining a first network packet loss rate between the first vehicle and a remote control end at a current moment; predicting a first duration of network anomaly based on the first network packet loss probability through a network prediction model; determining a collision threshold duration according to the first duration; and predicting the possibility of collision of the first vehicle according to a driving parameter of the first vehicle at the current moment, an environment parameter and the collision threshold duration. According to the invention, the influence of network abnormity on the safety of the remotely controlled vehicle can be reduced, and the prediction accuracy of the possibility of collision can be improved.

Description

Technical Field

[0001] The present application relates to the field of intelligent driving technology, and more specifically, to a collision prediction method, device and readable storage medium. Background Art

[0002] At present, with the development of electronic information technology, vehicles can be controlled through remote control terminals, thereby realizing remote driving of vehicles. Although collision prediction can be used to predict the possibility of collision, thereby avoiding collision risks as much as possible, the accuracy of predicting the possibility of collision is currently low. Summary of the invention

[0003] The present application proposes a collision prediction method, device and readable storage medium.

[0004] In a first aspect, an embodiment of the present application provides a collision prediction method, which is applied to a first vehicle, and the method includes: obtaining a first network packet loss rate between the first vehicle and a remote control terminal at a current moment; predicting a first duration of network abnormality based on the first network packet loss rate through a network prediction model; determining a collision threshold duration based on the first duration; and predicting the possibility of a collision of the first vehicle based on the first vehicle's driving parameters, environmental parameters and the collision threshold duration at a current moment.

[0005] Optionally, in a possible implementation, before predicting the first duration of the network anomaly based on the first network packet loss rate through the network prediction model, it also includes: acquiring sample data, wherein the sample data includes multiple sample moments, the amount of first sample data received by the first vehicle at each sample moment, the amount of second sample data sent by the remote control terminal at each sample moment, and the standard duration of the network anomaly corresponding to each sample moment; obtaining a predicted initial duration of the network anomaly corresponding to each sample moment based on the sample data based on the initial model; and training the initial model based on the difference between the initial duration and the standard duration to obtain the network prediction model.

[0006] Optionally, in a possible implementation, the obtaining of sample data includes: obtaining frequency parameters of communication between the first vehicle and the remote control terminal; determining multiple sample moments based on the frequency parameters; determining a first data volume received by the first vehicle corresponding to each sample moment, and a second data volume sent by the remote control terminal corresponding to each sample moment; based on multiple first data volumes and multiple second data volumes, respectively determining a first sample data volume and a second sample data volume corresponding to each sample moment; based on the first sample data volume and the second sample data volume corresponding to the sample moment, determining a standard duration of the network anomaly corresponding to the sample moment.

[0007] Optionally, in a possible implementation, determining the standard duration of the network anomaly corresponding to the sample moment based on the first sample data volume and the second sample data volume corresponding to the sample moment includes: determining the second network packet loss rate corresponding to the sample moment based on the first sample data volume and the second sample data volume corresponding to the sample moment; judging whether the sample moment corresponding to the second network packet loss rate is a network anomaly based on each of the second network packet loss rates; and determining the standard duration of the network anomaly corresponding to the sample moment based on each sample moment of the network anomaly before each sample moment.

[0008] Optionally, in a possible implementation, the second network packet loss rate includes a first downlink packet loss rate and a first uplink packet loss rate, and judging whether the network is abnormal at the sample moment corresponding to the second network packet loss rate based on each of the second network packet loss rates includes: taking the absolute value of the difference between the first downlink packet loss rate and the first uplink packet loss rate as the differential packet loss rate corresponding to the sample moment; when it is detected that the differential packet loss rate is greater than a first threshold packet loss rate, judging whether a target event has been detected a specified number of times at the sample moment, wherein the target event is an absolute value of the difference between the first differential packet loss rate and the second differential packet loss rate that is less than a second threshold packet loss rate, the first differential packet loss rate is a differential packet loss rate corresponding to any sample moment before the sample moment, and the second differential packet loss rate is a differential packet loss rate corresponding to a previous sample moment adjacent to the sample moment corresponding to the first differential packet loss rate; when the target event is detected a specified number of times, judging that the network is abnormal at the sample moment corresponding to the second network packet loss rate.

[0009] Optionally, in a possible implementation, predicting the possibility of a collision of the first vehicle based on the first vehicle's driving parameters, environmental parameters and the collision threshold duration at the current moment includes: predicting the duration of a collision of the first vehicle based on the first vehicle's driving parameters and environmental parameters at the current moment; obtaining a difference between the duration and the collision threshold duration as a first intermediate value; when the first intermediate value is less than or equal to the threshold value, predicting the possibility of a collision as a first possibility; when the first intermediate value is greater than the threshold value, predicting the possibility of a collision as a second possibility, wherein the first possibility is greater than the second possibility.

[0010] Optionally, in a possible implementation, at the current moment, there is a second vehicle ahead of the first vehicle in the driving direction, the driving parameters of the first vehicle include a first driving speed, a first acceleration, and a distance between the first vehicle and the second vehicle ahead, and the environmental parameters include a second driving speed and a second acceleration of the second vehicle; the predicting of the duration of the collision of the first vehicle based on the driving parameters and environmental parameters of the first vehicle at the current moment includes: obtaining the difference between the first driving speed and the second driving speed as a second intermediate value, obtaining the difference between the first acceleration and the second acceleration as a third intermediate value, and obtaining the difference between the distance and the body length of the first vehicle as a fourth intermediate value; when the third intermediate value is 0, obtaining the ratio of the fourth intermediate value to the second intermediate value as the duration; when the third intermediate value is not 0, performing a square root operation on the fourth intermediate value and then multiplying it with a specified value and the third intermediate value to obtain a sixth intermediate value; subtracting the second intermediate value from the sum of the square of the second intermediate value and the sixth intermediate value to obtain a seventh intermediate value; and taking the ratio of the seventh intermediate value to the third intermediate value as the duration.

[0011] Optionally, in a possible implementation, determining the collision threshold duration based on the first duration includes: obtaining a first time consumption for predicting the first duration through the network prediction model; determining a second time consumption for the first vehicle to respond to the braking instruction when the braking instruction is obtained; and taking the sum of the first time consumption, the second time consumption and the first duration as the collision threshold duration.

[0012] In the second aspect, the embodiment of the present application also provides a collision prediction device, which is applied to the first vehicle, and the device includes: an acquisition unit, a network anomaly duration prediction unit, a determination unit, and a collision result prediction unit. Among them, the acquisition unit is used to obtain the first network packet loss rate between the first vehicle and the remote control terminal at the current moment; the network anomaly duration prediction unit is used to predict the first duration of the network anomaly based on the first network packet loss rate through a network prediction model; the determination unit is used to determine the collision threshold duration according to the first duration; the collision result prediction unit is used to predict the possibility of the first vehicle colliding according to the first vehicle's driving parameters, environmental parameters and the collision threshold duration at the current moment.

[0013] In a third aspect, an embodiment of the present application further provides a computer-readable storage medium, in which a program code is stored. The program code can be called by a processor to execute the method described in the first aspect above.

[0014] The collision prediction method, device and readable storage medium provided by the embodiment of the present application first obtain the first network packet loss rate between the first vehicle and the remote control terminal at the current moment; then predict the first duration of network anomaly based on the first network packet loss rate through the network prediction model; determine the collision threshold duration according to the first duration; and then predict the possibility of collision of the first vehicle according to the driving parameters, environmental parameters and the collision threshold duration of the first vehicle at the current moment. In the scheme of the present application, the first duration of network anomaly after the current moment is predicted based on the first network packet loss rate at the current moment, and then the collision threshold duration is predicted based on the first duration, that is, the prediction of the collision threshold duration takes into account the influence of the network packet loss rate. Then, the possibility of collision is comprehensively predicted based on the driving parameters, environmental parameters and the collision threshold duration of the first vehicle. The impact of network anomaly on the safety of remotely controlled vehicles can be reduced, and the accuracy of prediction of the possibility of collision can be improved.

[0015] Other features and advantages of the embodiments of the present application will be described in the subsequent description, and partly become apparent from the description, or can be understood by practicing the embodiments of the present application. The purposes and other advantages of the embodiments of the present application can be realized and obtained by the structures specifically pointed out in the written description, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0017] Figure 1 An application scenario diagram of the collision prediction method provided by an embodiment of the present application is shown;

[0018] Figure 2 A method flow chart of a collision prediction method provided by an embodiment of the present application is shown;

[0019] Figure 3 A method flow chart of a collision prediction method provided by another embodiment of the present application is shown;

[0020] Figure 4 A method flow chart of a collision prediction method provided by another embodiment of the present application is shown;

[0021] Figure 5 A schematic diagram showing the collision threshold duration provided by an embodiment of the present application is shown;

[0022] Figure 6A method flow chart of a collision prediction method provided by another embodiment of the present application is shown;

[0023] Figure 7 A structural block diagram of a collision prediction device provided in an embodiment of the present application is shown;

[0024] Figure 8 A structural block diagram of a computer-readable storage medium provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0025] In order to make those skilled in the art better understand the present application scheme, the technical scheme in the present application embodiment will be clearly and completely described below in conjunction with the drawings in the present application embodiment. Obviously, the described embodiment is only a part of the present application embodiment, rather than all the embodiments. The components of the present application embodiment usually described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the present application for protection, but merely represents the selected embodiment of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0026] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of this application, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0027] At present, with the development of electronic information technology, vehicles can be controlled through remote control terminals, thereby realizing remote driving of vehicles. Although collision prediction can be used to predict the possibility of collision, thereby avoiding collision risks as much as possible, the accuracy of predicting the possibility of collision is currently low. How to improve the accuracy of predicting the possibility of collision is an urgent problem to be solved.

[0028] At present, when a vehicle is remotely controlled through a remote control terminal, the operator of the remote control terminal can predict the possibility of a collision of the remotely controlled vehicle through the video display interface of the remote control terminal, and then determine whether braking is required. In addition, the operator of the remote control terminal can also take braking measures for the vehicle when the remote control of the vehicle is disconnected.

[0029] However, the inventors found in their research that the above method does not take into account the stability of the network for controlling the vehicle through the remote control terminal, and the accuracy of predicting the possibility of a collision is low.

[0030] Therefore, in order to solve the above problems, the present application provides a collision prediction method, device and readable storage medium.

[0031] See also Figure 1 , Figure 1 The application scenario diagram of the collision prediction method provided by the embodiment of the present application is shown, namely, a collision prediction scenario 100. In the collision prediction scenario 100, a first vehicle 110 and a remote control terminal 120 are included.

[0032] Among them, the remote control terminal 120 can be a server, such as a cloud server. The first vehicle 110 can access the network, for example, the first vehicle 110 can access the network through the vehicle-mounted mobile cellular network. Thus, the first vehicle 110 can establish a communication connection with the remote control terminal 120, and the remote control terminal 120 can remotely control the first vehicle 110. Exemplarily, the remote control terminal 120 can send a control instruction to the first vehicle 110, so that after the first vehicle 110 receives the control instruction, it can respond to the control instruction and perform an operation corresponding to the control instruction. For example, the control instruction can be an acceleration instruction, and the first vehicle 110 can accelerate when the acceleration instruction is obtained.

[0033] In addition, the first vehicle 110 may also send the collected data to the remote control terminal 120, thereby providing data support for the remote control terminal 120 to make control instructions. For example, the collected data may include state parameters of the first vehicle 110 or environmental parameters around the first vehicle 110, wherein the environmental parameters may include road traffic information, traffic information, etc.

[0034] Furthermore, the first vehicle 110 can also predict the possibility of a collision based on the stability of the network communicating with the remote control terminal 120, and then brake immediately when a collision is predicted to be more likely. Detailed description can be found in the subsequent embodiments.

[0035] See also Figure 2 , Figure 2 A method flow chart of a collision prediction method provided by an embodiment of the present application is shown. The collision prediction method can be applied to a first vehicle, for example, a processor in a vehicle-mounted system of the first vehicle is used as an execution subject of the collision prediction method. The collision prediction method specifically includes steps S110 to S140.

[0036] Step S110: Obtaining a first network packet loss rate between the first vehicle and the remote control terminal at the current moment.

[0037] The first vehicle and the remote control terminal can communicate via a network, so that the first vehicle and the remote control terminal can send data to each other, and the remote control terminal can send a control instruction to control the first vehicle.

[0038] It is understandable that the stability of the network may fluctuate, such as unstable communication or network jitter, which may cause the transmission time to change when the same amount of data is transmitted between the first vehicle and the remote control terminal. For example, if the network fluctuates greatly, it will take longer to transmit the same amount of data. Therefore, in order to improve the accuracy of the subsequent prediction of the possibility of the first vehicle colliding, the network stability can be used as one of the parameters to predict the possibility of collision.

[0039] In some embodiments, the stability of the network between the first vehicle and the remote control terminal can be characterized by a network packet loss rate. The network packet loss rate can represent the proportion of data lost during network transmission. Exemplarily, the stability of the network between the first vehicle and the remote control terminal at the current moment can be characterized by the first network packet loss rate.

[0040] Therefore, the first network packet loss rate between the first vehicle and the remote control terminal at the current moment can be first obtained. Exemplarily, the third data volume received by the first vehicle from the remote control terminal in the first time period before the current moment can be obtained, and the fourth data volume sent by the remote control terminal to the first vehicle in the first time period before the current moment can be obtained. Then, based on the third data volume and the fourth data volume, the first network packet loss rate corresponding to the current moment is determined.

[0041] The first network packet loss rate may include a second uplink packet loss rate and a second downlink packet loss rate. Exemplarily, the second uplink packet loss rate may be characterized as a ratio of a difference between the third data volume and the fourth data volume to the third data volume; and the second downlink packet loss rate may be characterized as a ratio of a difference between the fourth data volume and the third data volume to the fourth data volume.

[0042] Therefore, the third data volume and the fourth data volume in the first time period before the current moment may be collected, and the second uplink packet loss rate and the second downlink packet loss rate may be calculated to obtain the first network packet loss rate.

[0043] It should be noted that the third data volume and the fourth data volume can be obtained by actually collecting data from the first vehicle and the remote control terminal at the current moment and in a first time period before the current moment, and the actually collected data can also be regarded as road test data.

[0044] Step S120: predicting a first duration of a network anomaly based on the first network packet loss rate using a network prediction model.

[0045] After obtaining the first network packet loss rate, the network prediction model can be used to predict the first duration of the network anomaly according to the first network packet loss rate. It should be noted that the first duration is the predicted duration of the future network anomaly under the first network packet loss rate corresponding to the current moment, which is a predicted value rather than the actual value actually collected.

[0046] The network prediction model may be a pre-trained model. Exemplarily, the model may be pre-trained with sample data to train the initial model, thereby obtaining a network prediction model, so that the network prediction model has the ability to analyze the input network packet loss rate and output the predicted duration of network anomalies. For a detailed description, please refer to the subsequent embodiments.

[0047] Step S130: determining a collision threshold duration according to the first duration.

[0048] In some embodiments, after the first duration is obtained, the first duration may be used as one of the parameters to determine the collision threshold duration, wherein the collision threshold duration may be used to characterize the theoretical duration required to brake the first vehicle at the current moment.

[0049] Therefore, the time duration for predicting the collision of the first vehicle can be obtained later, and then the possibility of collision can be predicted by combining the time duration with the collision threshold time duration. For a detailed description, please refer to the subsequent embodiments.

[0050] Step S140: predicting the possibility of a collision of the first vehicle according to the first vehicle's driving parameters, environmental parameters and the collision threshold duration at the current moment.

[0051] It is understandable that there may be obstacles in the environment, wherein the obstacle is an object in the driving direction of the first vehicle, such as the second vehicle, a pedestrian, or an emergency sign.

[0052] In some embodiments, the first vehicle driving parameters of the first vehicle at the current moment may also be collected, wherein the first vehicle driving parameters may include a first driving speed, a first acceleration, and a distance between the first vehicle and an obstacle object in front.

[0053] In addition, the obstacle object may also correspond to an environmental parameter at the current moment. For example, the speed and acceleration of the obstacle object at the current moment may be collected as the environmental parameter.

[0054] For example, the driving parameters of the first vehicle may be acquired by the vehicle system of the first vehicle through a sensor component disposed on the first vehicle. The first driving speed may be acquired through a wheel speed sensor, the first acceleration may be acquired through a gyroscope, and the distance between the first vehicle and an obstacle in front may be acquired through a laser radar.

[0055] In addition, the environmental parameters may be calculated by acquiring images and point cloud data including obstacles in real time through cameras, lidar and other equipment configured on the first vehicle, and then fusing and analyzing the images and point cloud data.

[0056] Thus, the possibility of the first vehicle colliding can be predicted based on the first vehicle's driving parameters, environmental parameters and the collision threshold duration at the current moment. Exemplarily, the duration of the first vehicle's collision can be predicted based on the first vehicle's driving parameters and environmental parameters, and then the duration can be compared with the collision threshold duration to predict the possibility of the collision. It should be noted that the occurrence of a collision indicates that the first vehicle collides with an obstacle object; the possibility of a collision refers to the possibility of predicting a collision between the first vehicle and the obstacle object.

[0057] Optionally, after obtaining the predicted possibility of collision, the first vehicle may also perform a preset operation based on the possibility. For example, when the predicted possibility is relatively high, the first vehicle may take a braking operation, thereby reducing the probability of a collision accident.

[0058] The collision prediction method provided in the embodiment of the present application first obtains the first network packet loss rate between the first vehicle and the remote control terminal at the current moment; then predicts the first duration of network anomaly based on the first network packet loss rate through a network prediction model; determines the collision threshold duration based on the first duration; and then predicts the possibility of collision of the first vehicle based on the driving parameters, environmental parameters and the collision threshold duration of the first vehicle at the current moment. In the scheme of the present application, the first duration of network anomaly after the current moment is predicted based on the first network packet loss rate at the current moment, and then the collision threshold duration is predicted based on the first duration, that is, the prediction of the collision threshold duration takes into account the influence of the network packet loss rate. Then, the possibility of collision is comprehensively predicted based on the driving parameters, environmental parameters and the collision threshold duration of the first vehicle. The impact of network anomaly on the safety of remotely controlled vehicles can be reduced, and the accuracy of prediction of the possibility of collision can be improved.

[0059] See also Figure 3 , Figure 3A method flow chart of a collision prediction method provided by an embodiment of the present application is shown. The collision prediction method can be applied to a first vehicle, for example, a processor in a vehicle-mounted system of the first vehicle is used as an execution subject of the collision prediction method. The collision prediction method specifically includes steps S210 to S270.

[0060] Step S210: Obtaining a first network packet loss rate between the first vehicle and the remote control terminal at the current moment.

[0061] Among them, step S210 has been introduced in detail in the above embodiment and will not be repeated here.

[0062] Step S220: Obtain sample data, wherein the sample data includes multiple sample moments, the amount of first sample data received by the first vehicle at each sample moment, the amount of second sample data sent by the remote control terminal at each sample moment, and the standard duration of the network anomaly corresponding to each sample moment.

[0063] In some implementations, sample data may be obtained, so that the initial model may be trained by the sample data to obtain a network prediction model, wherein the initial model may be a prediction model established based on a neural network.

[0064] Among them, the sample data may include multiple sample moments, the first sample data volume received by the first vehicle at each sample moment, the second sample data volume sent by the remote control terminal at each sample moment, and the standard duration of the network anomaly corresponding to each sample moment.

[0065] The sample data can be obtained based on historical road test data collected when the first vehicle is driven by remote control of the remote control terminal. Exemplarily, the historical road test data may include multiple sample moments. It may also include the amount of first sample data received by the first vehicle at each sample moment. In addition, the historical road test data may also include the amount of second sample data sent by the remote control terminal at each sample moment.

[0066] Specifically, step S220 may include steps S221 to S225.

[0067] Step S221: Acquire frequency parameters of communication between the first vehicle and the remote control terminal.

[0068] Step S222: Determine a plurality of sample moments based on the frequency parameter.

[0069] Step S223: Determine the amount of first data received by the first vehicle corresponding to each of the sample moments, and the amount of second data sent by the remote control terminal corresponding to each of the sample moments.

[0070] Step S224: Based on the plurality of the first data amounts and the plurality of the second data amounts, respectively determine the first sample data amount and the second sample data amount corresponding to each sample moment.

[0071] Step S225: Based on the first sample data volume and the second sample data volume corresponding to the sample time, determine the standard duration of the network anomaly corresponding to the sample time.

[0072] It is understandable that the communication between the first vehicle and the remote control terminal may have a certain frequency, so the frequency parameter of the communication between the first vehicle and the remote control terminal may be first obtained. For example, the frequency parameter may be 10 Hz, 5 Hz, etc. The first vehicle may directly obtain the frequency parameter from the communication information between the first vehicle and the remote control terminal.

[0073] Furthermore, multiple sample moments can be determined based on the frequency parameter. For example, the reciprocal of the frequency parameter can be used as the interval duration, and then a corresponding sample moment is determined each time an interval duration is reached. In other words, the time interval between each sample moment is the interval duration. Taking the frequency parameter as 10 Hz as an example, the time interval between each sample moment is 100 ms.

[0074] Then, the amount of first data received by the first vehicle corresponding to each of the sample moments and the amount of second data sent by the remote control terminal corresponding to each of the sample moments may be determined.

[0075] Based on the first data amounts and the second data amounts, the first sample data amount and the second sample data amount corresponding to each sample moment are determined respectively. In some implementations, the sum of the first data amounts corresponding to each sample moment and all sample moments before the sample moment can be used as the first sample data amount corresponding to the sample moment, and the sum of the second data amounts corresponding to each sample moment and all sample moments before the sample moment can be used as the second sample data amount corresponding to the sample moment.

[0076] Among them, the first sample data volume can be regarded as the uplink data volume. Specifically, the first sample data volume may include the data volume used to characterize the state of the first vehicle, wherein the state of the first vehicle may include vehicle speed, gear position, and camera capture image data, etc. In addition, the second sample data volume can be regarded as the downlink data volume. Specifically, the second sample data volume may include the data volume used to characterize the control instruction, wherein the control instruction may include the steering wheel angle, gear setting, torque setting, etc.

[0077] Furthermore, based on the first sample data volume and the second sample data volume corresponding to the sample time, the standard duration of the network anomaly corresponding to the sample time may be determined.

[0078] In some implementations, the second network packet loss rate may be first determined based on the first sample data volume and the second sample data volume corresponding to each sample moment, and then the standard duration corresponding to the sample moment may be determined based on the second network packet loss rate.

[0079] Specifically, step S225 may include step S2251 to step S2253.

[0080] Step S2251: Based on the first sample data volume and the second sample data volume corresponding to the sample time, determine the second network packet loss rate corresponding to the sample time.

[0081] Step S2252: Based on each of the second network packet loss rates, determine whether the sample time corresponding to the second network packet loss rate is network abnormal.

[0082] Step S2253: Determine the standard duration of the network anomaly corresponding to each sample moment based on the sample moment of each network anomaly before each sample moment.

[0083] First, based on the first sample data volume and the second sample data volume corresponding to the sample time, the second network packet loss rate corresponding to the sample time can be determined. In some embodiments, the second network packet loss rate may include a first downlink packet loss rate and a first uplink packet loss rate. The first uplink packet loss rate can be used to characterize the packet loss rate corresponding to the data received by the first vehicle, and the first downlink packet loss rate can be used to characterize the packet loss rate corresponding to the data sent by the remote control terminal.

[0084] In some implementations, the first uplink packet loss rate may be characterized as a ratio of a difference between the first data volume and the second data volume to the first data volume; and the second downlink packet loss rate may be characterized as a ratio of a difference between the second data volume and the first data volume to the second data volume.

[0085] Optionally, the first uplink packet loss rate and the first downlink packet loss rate corresponding to multiple moments within a certain time period before the sample moment may also be obtained. Then, based on the multiple first uplink packet loss rates corresponding to the sample moment, an average uplink packet loss rate is calculated as the first uplink packet loss rate corresponding to the sample moment; based on the multiple first downlink packet loss rates corresponding to the sample moment, an average downlink packet loss rate is calculated as the first downlink packet loss rate corresponding to the sample moment.

[0086] For example, the Beta distribution concept can be used to find the average value. Specifically, in the Beta distribution concept, the average value can be represented by taking four times the maximum value, the minimum value, and the middle value of multiple data, and then finding the weighted average value. Specifically, it can be represented by the following formula (1).

[0087] θ 平均值 =(θ max +θ min +θ middle x4) / 6 (1)

[0088] Therefore, if through θ 下 Characterizes the average downlink packet loss rate, θ 上 The average uplink packet loss rate is represented by the first downlink packet loss rate corresponding to the sample time, and the first downlink packet loss rate corresponding to the sample time can be represented by the following formula (2).

[0089] θ 下平均值 =(θ 下max +θ 下min +θ 下middle x4) / 6 (2)

[0090] Among them, θ 下平均值 That is, it represents the average downlink packet loss rate; θ 下max Used to represent the maximum value of multiple first downlink packet loss rates corresponding to the sample time; θ 下min Used to represent the minimum value among multiple first downlink packet loss rates corresponding to the sample time; θ 下middle Used to represent the middle value of multiple first downlink packet loss rates corresponding to the sample time.

[0091] Similarly, the first uplink packet loss rate corresponding to the sample time can be expressed as the following formula (3).

[0092] θ 上平均值 =(θ 上max +θ 上min +θ 上middle x4) / 6 (3)

[0093] Among them, θ 上平均值 That is, it represents the average uplink packet loss rate; θ 上max Used to represent the maximum value of multiple first uplink packet loss rates corresponding to the sample time; θ 上min Used to represent the minimum value of multiple first uplink packet loss rates corresponding to the sample time; θ 上middle Used to represent the middle value of multiple first uplink packet loss rates corresponding to the sample time.

[0094] The average downlink packet loss rate corresponding to the sample time obtained above can be used to characterize the first downlink packet loss rate corresponding to the sample time, and the average uplink packet loss rate corresponding to the sample time can be used to characterize the first uplink packet loss rate corresponding to the sample time.

[0095] Further, based on each of the second network packet loss rates, it is determined whether the sample time corresponding to the second network packet loss rate is network abnormal. In some embodiments, it is possible to determine whether the sample time corresponding to the second network packet loss rate is network abnormal based on the difference between the first downlink packet loss rate and the first uplink packet loss rate. Specifically, step S2252 may include steps S2254 to S2256.

[0096] Step S2254: taking the absolute value of the difference between the first downlink packet loss rate and the first uplink packet loss rate as the difference packet loss rate corresponding to the sample time.

[0097] Step S2255: When it is detected that the difference packet loss rate is greater than the first threshold packet loss rate, determine whether a target event has been detected a specified number of times at the sample moment, wherein the target event is that the absolute value of the difference between the first difference packet loss rate and the second difference packet loss rate is less than the second threshold packet loss rate, the first difference packet loss rate is the difference packet loss rate corresponding to any sample moment before the sample moment, and the second difference packet loss rate is the difference packet loss rate corresponding to the previous sample moment adjacent to the sample moment corresponding to the first difference packet loss rate.

[0098] Step S2256: When a target event is detected a specified number of times, it is determined that the network is abnormal at the sample time corresponding to the second network packet loss rate.

[0099] First, the absolute value of the difference between the first downlink packet loss rate and the first uplink packet loss rate can be used as the differential packet loss rate corresponding to the sample time. Specifically, the differential packet loss rate can be represented by the following formula (4).

[0100] Δθ 上下行差 =|θ 下平均值 -θ 上平均值 | (4)

[0101] Among them, Δθ 上下行差 That is, it is used to characterize the differential packet loss rate.

[0102] Furthermore, the size of the differential packet loss rate may be detected. If the differential packet loss rate is large, a network anomaly may exist at the sample time corresponding to the differential packet loss rate.

[0103] In some embodiments, the difference packet loss rate can be compared with a first threshold packet loss rate. Thus, when it is detected that the difference packet loss rate is greater than the first threshold packet loss rate, a network anomaly may exist at the sample moment corresponding to the difference packet loss rate. At this time, it can be further determined whether a target event has been detected a specified number of times at the sample moment.

[0104] In some implementations, the first threshold packet loss rate may be 0.01, so that it can be determined whether formula (4) is greater than 0.01.

[0105] Among them, the target event is that the absolute value of the difference between the first difference packet loss rate and the second difference packet loss rate is less than the second threshold packet loss rate, the first difference packet loss rate is the difference packet loss rate corresponding to any sample moment before the sample moment, and the second difference packet loss rate is the difference packet loss rate corresponding to the previous sample moment adjacent to the sample moment corresponding to the first difference packet loss rate.

[0106] For example, if θ n Characterizes the first difference packet loss rate, θ n-1 Characterizes the second difference packet loss rate. The first difference packet loss rate and the second difference packet loss rate are obtained in a similar manner to the difference packet loss rate, and will not be described in detail here.

[0107] In some implementations, the second threshold packet loss rate may be 0.005. Thus, whether the target event is satisfied may be determined by the following formula (5).

[0108] Δθ=|(θ n -θ n-1 )|<0.005 (5)

[0109] Here, Δθ is the absolute value of the difference between the first difference packet loss rate and the second difference packet loss rate.

[0110] Therefore, when a target event is detected a specified number of times, it can be determined that the network is abnormal at the sample time corresponding to the second network packet loss rate.

[0111] Optionally, the network at the sample moment corresponding to the second network packet loss rate is determined to be abnormal only when the target event is detected a specified number of times in succession. It should be noted that the target event is detected a specified number of times in succession, which means that the detection is performed for a plurality of consecutive sample moments, and the target condition needs to be met for a specified number of consecutive sample moments.

[0112] For some implementations, the specified number may be a positive integer, such as 3, 4, or 5.

[0113] By determining that the network is abnormal at the sample time corresponding to the second network packet loss rate when a specified number of target events are detected, it is possible to avoid prediction errors caused by calculation errors at a certain time and improve the accuracy of the determination. In some implementations, this method may also be referred to as an anti-shake strategy.

[0114] Thus, through the above steps, it can be determined whether each sample moment before each sample moment is network abnormal. Further, the standard duration of the network abnormality corresponding to each sample moment can be determined based on each sample moment of network abnormality before each sample moment.

[0115] Exemplarily, a reference duration may be determined based on the frequency parameter; then the number of sample moments of each anomaly before the sample moment is obtained; and the product of the number and the reference duration is used as the standard duration of the network anomaly corresponding to the sample moment.

[0116] For example, the reference duration may be the inverse of the frequency parameter. If the frequency parameter is 10 Hz, the reference duration may be determined to be 100 ms. Thus, if the number of sample moments of each abnormality before the sample moment is 5, the standard duration of the network abnormality corresponding to the sample duration may be determined to be 5 x 100 ms = 500 ms.

[0117] Step S230: obtaining a predicted initial duration of the network anomaly corresponding to each of the sample moments based on the sample data based on the initial model.

[0118] Step S240: training the initial model based on the difference between the initial duration and the standard duration to obtain the network prediction model.

[0119] After the initial model is obtained, the initial duration of the network anomaly corresponding to each sample moment can be predicted based on the sample data based on the initial model. Specifically, multiple sample moments in the sample data, the amount of first sample data received by the first vehicle at each sample moment, and the amount of second sample data sent by the remote control terminal at each sample moment can be used as input parameters of the initial model to obtain the initial duration of the initial model output.

[0120] Then, the difference between the initial duration and the standard duration is obtained, so as to train the initial model through the difference to minimize the difference. After the initial model is trained, a network prediction model can be obtained. Exemplarily, the initial model can be repeatedly trained until the difference converges to an allowable range. The allowable range can be a pre-set range.

[0121] It should be noted that the difference should be determined by the initial duration and the standard duration corresponding to the same sample moment.

[0122] Optionally, the network prediction model may be characterized by F(θ,T), wherein θ is used to characterize the first network packet loss rate, and T is used to characterize the first duration.

[0123] Step S250: predicting a first duration of a network anomaly based on the first network packet loss rate using a network prediction model.

[0124] Step S260: determining a collision threshold duration according to the first duration.

[0125] Step S270: predicting the possibility of a collision of the first vehicle based on the first vehicle's driving parameters, environmental parameters and the collision threshold duration at the current moment.

[0126] Among them, step S250 to step S270 have been introduced in detail in the above embodiments and will not be repeated here.

[0127] In the collision prediction method provided in the embodiment of the present application, the initial model is trained by acquiring sample data, and then a network prediction model is obtained. Subsequently, the first duration of the network anomaly is predicted based on the first network packet loss rate through the network prediction model; the collision threshold duration is determined according to the first duration; and the possibility of the first vehicle colliding is predicted in combination with the collision threshold duration. The impact of network anomalies on the safety of remotely controlled vehicles can be reduced, and the accuracy of the prediction of the possibility of a collision can be improved. The first vehicle can predict the possible network anomaly in advance, reserve enough buffer time for the first vehicle to deal with the network anomaly, improve the reliability of the remotely controlled vehicle, and effectively ensure the safe operation of the first vehicle without a safety officer in the vehicle. In addition, in the process of acquiring sample data, by detecting a specified number of target events, the network anomaly at the sample time corresponding to the second network packet loss rate can be determined, which can avoid prediction errors caused by calculation errors at a certain moment and improve the accuracy of the determination. In some embodiments, this method can also be referred to as an anti-shake strategy.

[0128] See also Figure 4 , Figure 4 A method flow chart of a collision prediction method provided by an embodiment of the present application is shown. The collision prediction method can be applied to a first vehicle, for example, a processor in a vehicle-mounted system of the first vehicle is used as an execution subject of the collision prediction method. The collision prediction method specifically includes steps S310 to S390.

[0129] Step S310: Obtain a first network packet loss rate between the first vehicle and the remote control terminal at the current moment.

[0130] Step S320: predicting a first duration of a network anomaly based on the first network packet loss rate using a network prediction model.

[0131] Among them, step S310 and step S320 have been introduced in detail in the above embodiments and will not be repeated here.

[0132] Step S330: Obtain a first time consumption for predicting the first duration through the network prediction model.

[0133] Step S340: determining a second time taken for the first vehicle to respond to the braking instruction when the braking instruction is obtained.

[0134] Step S350: taking the sum of the first time consumption, the second time consumption and the first duration as the collision threshold duration.

[0135] It is understandable that it takes a certain amount of time for the first vehicle to predict the first duration through the network prediction model, and thus, the first time consumed for predicting the first duration through the network prediction model can be obtained.

[0136] In addition, even if the first vehicle obtains the designated instruction, it still takes a certain amount of time to respond to the braking instruction and perform the braking operation corresponding to the braking instruction. Therefore, the second time consumption of the first vehicle in responding to the braking instruction when the braking instruction is obtained can also be determined.

[0137] In some implementations, the second time may include a response time of an actuator in the first vehicle and a deceleration increase time.

[0138] Therefore, the first time consumption, the response time consumption of the actuator in the first vehicle, the deceleration increase time consumption and the first duration may be summed to obtain the collision threshold duration.

[0139] Optionally, the driver's reaction time may be obtained, so that the reaction time is used as one of the reference factors for determining the collision threshold time, so as to improve the redundancy of the determined collision threshold time. The driver's reaction time may be used to characterize the time from when the driver receives the braking command to when the driver performs the braking operation on the first vehicle. For some embodiments, multiple experiments may be performed in advance to obtain multiple reaction times, and then the arithmetic average of the multiple reaction times may be calculated.

[0140] Thus, the first time duration, the response time duration of the actuator in the first vehicle, the deceleration growth time duration, the first duration, and the reaction duration may be summed to obtain the collision threshold duration. For example, the collision threshold duration may be represented by the following formula (6).

[0141] ΔTTC=t1+t2+t3+t4+T (6)

[0142] Among them, ΔTTC is used to characterize the collision threshold time; t1 is used to characterize the first time; t2 is used to characterize the reaction time; t3 is used to characterize the response time of the actuator in the first vehicle; t4 is used to characterize the deceleration growth time; T is used to characterize the first time. In some embodiments, the collision threshold time can also be characterized as ΔQ(T, ΔTTC), thereby highlighting that the duration of the network anomaly is used as one of the influencing factors to obtain the collision threshold time.

[0143] For further understanding of the collision threshold duration, see Figure 5 , Figure 5 A schematic diagram of the collision threshold duration provided by an embodiment of the present application is shown. Figure 5 In the figure, multiple stages are shown, including: information processing stage, driver reaction stage, brake coordination stage, deceleration growth stage, network abnormality stage and continuous braking stage. Among them, the information processing stage corresponds to the first time t1; the driver reaction stage corresponds to the reaction time t2; the brake coordination stage corresponds to the response time t3; the deceleration growth stage corresponds to the deceleration growth time t4; the network abnormality stage corresponds to the first time T. In the continuous braking stage, it is characterized that the first vehicle is continuously braked.

[0144] It can be understood that the first duration is used to characterize the duration of the network anomaly. Therefore, in the embodiment of the present application, the duration of the network anomaly is used as one of the influencing factors to obtain the collision threshold duration, thereby optimizing the collision threshold duration and improving the accuracy of predicting the collision threshold duration.

[0145] Step S360: predicting the duration of a collision of the first vehicle based on the driving parameters and environmental parameters of the first vehicle at the current moment.

[0146] It is understandable that there may be an obstacle object in the environment, and the obstacle object may correspond to an environmental parameter. Exemplarily, the environmental parameter may include parameters such as the driving speed and acceleration of the obstacle object. The obstacle object may be located in front of the driving direction of the first vehicle. Thus, the duration of the collision between the first vehicle and the second vehicle can be predicted based on the first vehicle driving parameters at the current moment and the environmental parameters of the second vehicle at the current moment.

[0147] In some embodiments, there may be a second vehicle ahead of the first vehicle in the driving direction at the current moment, that is, the obstacle object in the environment is the second vehicle. At this time, the second vehicle can be detected by the first vehicle, thereby obtaining environmental parameters.

[0148] For example, the first vehicle may detect the second vehicle currently in front of it in the driving direction. For example, the second driving speed and the second driving acceleration of the second vehicle currently in front of the first vehicle may be detected, or the distance between the first vehicle and the second vehicle in front of it may be detected.

[0149] The second vehicle may be a vehicle in front of the first vehicle in the driving direction and in the same lane as the first vehicle. In some embodiments, the first vehicle may obtain the second driving speed, the second driving acceleration, the current distance between the first vehicle and the second vehicle in front of it, etc. of the second vehicle through a sensor component such as a camera or a laser radar, and then use the second driving speed and the second driving acceleration as environmental parameters.

[0150] Therefore, for some embodiments, the first vehicle driving parameter may include a first driving speed, a first acceleration, and a distance between the first vehicle and a second vehicle in front, and the environmental parameter may include a second driving speed and a second acceleration. Specifically, step S360 may include steps S361 to S365.

[0151] Step S361: Obtain the difference between the first driving speed and the second driving speed as a second intermediate value, obtain the difference between the first acceleration and the second acceleration as a third intermediate value, and obtain the difference between the distance and the body length of the first vehicle as a fourth intermediate value.

[0152] Step S362: when the third intermediate value is 0, obtain the ratio of the fourth intermediate value to the second intermediate value as the duration.

[0153] Step S363: when the third intermediate value is not 0, perform square root operation on the fourth intermediate value and multiply the resultant with the designated value and the third intermediate value to obtain a sixth intermediate value.

[0154] Step S364: subtract the second intermediate value from the sum of the square of the second intermediate value and the sixth intermediate value to obtain a seventh intermediate value.

[0155] Step S365: taking the ratio of the seventh intermediate value to the third intermediate value as the duration.

[0156] In some embodiments, the difference between the first driving speed and the second driving speed can be first obtained as a second intermediate value, the difference between the first acceleration and the second acceleration can be obtained as a third intermediate value, and the difference between the distance and the body length of the first vehicle can be obtained as a fourth intermediate value.

[0157] The second intermediate value can be represented by the following formula (7).

[0158] ΔV= V0- V1 (7)

[0159] Among them, ΔV is used to represent the second intermediate value; V0 is used to represent the first driving speed; and V1 is used to represent the second driving speed.

[0160] In addition, the third intermediate value can be represented by the following formula (8).

[0161] Δa= a0- a1 (8)

[0162] Among them, Δa is used to represent the third intermediate value; a0 is used to represent the first acceleration; and a1 is used to represent the second acceleration.

[0163] Furthermore, the fourth intermediate value can also be represented by Sd. S is used to represent the distance between the first vehicle and the second vehicle in front; d is used to represent the body length of the first vehicle. The body length of the first vehicle can be captured by the vehicle system of the first vehicle in the configuration information of the first vehicle, or can be pre-input into the vehicle system, which is not specifically limited in the embodiment of the present application.

[0164] In some embodiments, the duration can be predicted by establishing a collision time model. For example, the collision event model can be represented by the following formula (9).

[0165] V0(TTC)+1 / 2a0(TTC) 2 +d=S+V1(TTC)+1 / 2a1(TTC) 2 (9)

[0166] Among them, TTC is used to represent the duration.

[0167] Thus, by solving TTC in equation (9), the duration can be obtained. Optionally, equation (9) can also be represented by Q(S, TTC), that is, Q(S, TTC) can be used to characterize the collision event model.

[0168] It is understandable that the value of the duration is related to the size of the third intermediate value, specifically, whether the third intermediate value is 0. Therefore, the duration can be calculated according to whether the third intermediate value is 0.

[0169] Specifically, when the third intermediate value is 0, the ratio of the fourth intermediate value to the second intermediate value can be obtained as the duration. Therefore, when the third intermediate value is 0, the duration can be calculated by the following formula (10).

[0170] TTC=(Sd) / ΔV,Δa=0 (10)

[0171] When the third intermediate value is not 0, the square root of the fourth intermediate value is calculated and then multiplied by the specified value and the third intermediate value to obtain a sixth intermediate value; then, the second intermediate value is subtracted from the sum of the square of the second intermediate value and the sixth intermediate value to obtain a seventh intermediate value; and the ratio of the seventh intermediate value to the third intermediate value is used as the duration. Therefore, when the third intermediate value is not 0, the duration can be obtained by the following formula (11).

[0172] TTC=[-ΔV+[ΔV 2 +2Δa(Sd)] 1 / 2 ] / Δa,Δa≠0 (11)

[0173] It can be seen that in the embodiment provided in the present application, the designated value is 2, so the square root operation is performed on the fourth intermediate value and the product of the designated value and the third intermediate value is obtained to obtain 2Δa(Sd)] 1 / 2 , which is the sixth intermediate value.

[0174] Then, the second intermediate value is subtracted from the sum of the square of the second intermediate value and the sixth intermediate value to obtain -ΔV+[ΔV 2 +2Δa(Sd)] 1 / 2 , which is the seventh intermediate value.

[0175] Further, the ratio of the seventh intermediate value to the third intermediate value, that is, [-ΔV+[ΔV 2 +2Δa(Sd)] 1 / 2 ] / Δa, as the duration, that is, the duration is characterized by TTC.

[0176] Step S370: Obtain a difference between the duration and the collision threshold duration as a first intermediate value.

[0177] Further, a collision prediction function may be constructed based on the duration and the collision threshold duration. Exemplarily, the collision prediction function may be characterized by f(T, TTC). In some embodiments, the collision prediction function may be characterized by the following formula (12).

[0178] f(T,TTC)=Q(S,TTC)-ΔQ(T,ΔTTC)=D (12)

[0179] It can be seen that the difference between the duration and the collision threshold duration can be obtained as the first intermediate value, that is, the first intermediate value can be characterized by D. Therefore, the possibility of a collision can be predicted by comparing the first intermediate value with the threshold value. In some embodiments, the threshold value can be 0.

[0180] Step S371: Whether the first intermediate value is greater than the threshold value.

[0181] Specifically, when the first intermediate value is less than or equal to the threshold value, the process may jump to step S380 ; and when the first intermediate value is greater than the threshold value, the process may jump to step S390 .

[0182] Step S380: When the first intermediate value is less than or equal to the threshold value, the possibility of the collision is predicted to be the first possibility.

[0183] Step S390: When the first intermediate value is greater than the threshold value, the possibility of the predicted collision is a second possibility, wherein the first possibility is greater than the second possibility.

[0184] Thus, if the first intermediate value is less than or equal to the threshold value, the characterization duration is less than or equal to the collision threshold duration, and at this time, there is a greater possibility of a collision, and the possibility of the collision can be predicted to be the first possibility. If the first intermediate value is greater than the threshold value, the characterization duration is greater than the collision threshold duration, and at this time, there is a greater possibility that a collision will not occur, and the possibility of the collision can be predicted to be the second possibility. Among them, the first possibility is greater than the second possibility.

[0185] Optionally, when the predicted possibility of collision is the first possibility, the first vehicle may be directly controlled to brake through the vehicle computer system of the first vehicle, or a warning message may be displayed through the vehicle computer system of the first vehicle.

[0186] For example, see Figure 6 , Figure 6 A method flow chart of a collision prediction method provided by an embodiment of the present application is shown. The collision prediction method can be applied to a first vehicle, for example, a processor in a vehicle-mounted system of the first vehicle is used as an execution subject of the collision prediction method. The collision prediction method specifically includes steps S410 to S430.

[0187] Step S410: data acquisition.

[0188] In the data acquisition step, the third data volume and the fourth data volume can be acquired by actually collecting drive test data, so as to calculate the first network packet loss rate, and then the first duration of the network anomaly is predicted based on the first network packet loss rate by using a network prediction model.

[0189] In addition, in the data acquisition step, the first vehicle driving parameters and the perceived environment parameters of the first vehicle may also be acquired.

[0190] Step S420: prediction.

[0191] Furthermore, the possibility of the first vehicle colliding may be predicted based on the first vehicle driving parameters, environmental parameters and the collision threshold duration of the first vehicle at the current moment. In some embodiments, the possibility of a collision may be predicted using a collision prediction function.

[0192] Step S430: Decision making.

[0193] After the probability is predicted, the decision execution model can be used to determine the subsequent decisions to be executed based on the probability. For example, decisions such as braking and displaying warning information can be executed.

[0194] In the collision prediction method provided in the embodiment of the present application, the duration of the network anomaly is used as one of the influencing factors to obtain the collision threshold duration, thereby optimizing the collision threshold duration and improving the accuracy of predicting the collision threshold duration. In addition, the collision threshold duration with higher accuracy is subsequently added to the collision prediction function to predict the possibility of a collision. On the one hand, the accuracy of the first vehicle's prediction of the possibility of a collision is improved; on the other hand, the negative impact of network stability on the remotely driven first vehicle can also be effectively eliminated or weakened. Furthermore, in the collision prediction method provided in the embodiment of the present application, the predicted possibility of a collision can be subsequently provided to upstream perception and downstream decision-making, increasing the data source and redundancy for dangerous target identification and improving the accuracy and robustness of subsequent algorithms.

[0195] See also Figure 7 , Figure 7 A structural block diagram of a collision prediction device 700 provided in an embodiment of the present application is shown, which is applied to a first vehicle. The collision prediction device 700 includes: an acquisition unit 710, a network abnormality duration prediction unit 720, a determination unit 730 and a collision result prediction unit 740.

[0196] The acquisition unit 710 is used to acquire a first network packet loss rate between the first vehicle and the remote control terminal at a current moment.

[0197] The network anomaly duration prediction unit 720 is configured to predict a first duration of the network anomaly based on the first network packet loss rate using a network prediction model.

[0198] Optionally, the network anomaly duration prediction unit 720 can also be used to obtain sample data, wherein the sample data includes multiple sample moments, the amount of first sample data received by the first vehicle at each sample moment, the amount of second sample data sent by the remote control terminal at each sample moment, and the standard duration of the network anomaly corresponding to each sample moment; based on the initial model, the initial duration of the network anomaly corresponding to each sample moment is predicted according to the sample data; and the initial model is trained based on the difference between the initial duration and the standard duration to obtain the network prediction model.

[0199] Optionally, the network anomaly duration prediction unit 720 can also be used to obtain frequency parameters of communication between the first vehicle and the remote control terminal; determine multiple sample moments based on the frequency parameters; determine the first data amount received by the first vehicle corresponding to each sample moment, and the second data amount sent by the remote control terminal corresponding to each sample moment; based on multiple first data amounts and multiple second data amounts, respectively determine the first sample data amount and the second sample data amount corresponding to each sample moment; based on the first sample data amount and the second sample data amount corresponding to the sample moment, determine the standard duration of the network anomaly corresponding to the sample moment.

[0200] Optionally, the network anomaly duration prediction unit 720 can also be used to determine the second network packet loss rate corresponding to the sample moment based on the first sample data volume and the second sample data volume corresponding to the sample moment; determine whether the sample moment corresponding to the second network packet loss rate is a network anomaly based on each of the second network packet loss rates; and determine the standard duration of the network anomaly corresponding to the sample moment based on each sample moment of the network anomaly before each sample moment.

[0201] Optionally, the network anomaly duration prediction unit 720 can also be used to use the absolute value of the difference between the first downlink packet loss rate and the first uplink packet loss rate as the differential packet loss rate corresponding to the sample moment; when it is detected that the differential packet loss rate is greater than the first threshold packet loss rate, determine whether a target event has been detected a specified number of times at the sample moment, wherein the target event is that the absolute value of the difference between the first differential packet loss rate and the second differential packet loss rate is less than the second threshold packet loss rate, the first differential packet loss rate is the differential packet loss rate corresponding to any sample moment before the sample moment, and the second differential packet loss rate is the differential packet loss rate corresponding to the previous sample moment adjacent to the sample moment corresponding to the first differential packet loss rate; when the target event is detected a specified number of times, determine that the network is abnormal at the sample moment corresponding to the second network packet loss rate.

[0202] The determination unit 730 is configured to determine a collision threshold duration according to the first duration.

[0203] Optionally, the determination unit 730 can also be used to obtain a first time consumption to predict the first time duration through the network prediction model; determine a second time consumption for the first vehicle to respond to the braking instruction when the braking instruction is obtained; and use the sum of the first time consumption, the second time consumption and the first time duration as the collision threshold duration.

[0204] The collision result prediction unit 740 is used to predict the possibility of the first vehicle colliding according to the first vehicle's driving parameters, environmental parameters and the collision threshold duration at the current moment.

[0205] Optionally, the collision result prediction unit 740 can also be used to predict the duration of the collision of the first vehicle based on the driving parameters and environmental parameters of the first vehicle at the current moment; obtain the difference between the duration and the collision threshold duration as a first intermediate value; when the first intermediate value is less than or equal to the threshold value, predict the possibility of collision as a first possibility; when the first intermediate value is greater than the threshold value, predict the possibility of collision as a second possibility, wherein the first possibility is greater than the second possibility.

[0206] Optionally, the collision result prediction unit 740 can also be used to obtain the difference between the first driving speed and the second driving speed as a second intermediate value, obtain the difference between the first acceleration and the second acceleration as a third intermediate value, and obtain the difference between the distance and the body length of the first vehicle as a fourth intermediate value; when the third intermediate value is 0, obtain the ratio of the fourth intermediate value to the second intermediate value as the duration; when the third intermediate value is not 0, perform a square root operation on the fourth intermediate value and multiply it with a specified value and the third intermediate value to obtain a sixth intermediate value; subtract the second intermediate value from the sum of the square of the second intermediate value and the sixth intermediate value to obtain a seventh intermediate value; and use the ratio of the seventh intermediate value to the third intermediate value as the duration.

[0207] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described devices and units can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here.

[0208] In several embodiments provided in the present application, the coupling between the units may be electrical, mechanical or other forms of coupling. In addition, each functional unit in each embodiment of the present application may be integrated into a processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0209] See also Figure 8 , which shows a structural block diagram of a computer-readable storage medium provided in an embodiment of the present application. The computer-readable medium 800 stores program codes, which can be called by a processor to execute the method described in the above method embodiment.

[0210] The computer readable storage medium 800 can be an electronic memory such as a flash memory, an EEPROM (electrically erasable programmable read-only memory), an EPROM, a hard disk, or a ROM. Optionally, the computer readable storage medium 800 includes a non-transitory computer-readable storage medium. The computer readable storage medium 800 has storage space for program codes 810 that perform any method steps in the above method. These program codes can be read from or written to one or more computer program products. The program code 810 can be compressed, for example, in an appropriate form.

[0211] Among them, the processor may include one or more processing cores. The processor uses various interfaces and lines to connect to the readable storage medium 800, thereby running or executing instructions, programs, code sets or instruction sets in the readable storage medium 800, and calling data stored in the readable storage medium 800 to perform various functions and process data. Optionally, the processor can be implemented in at least one hardware form of digital signal processing (DSP), field programmable gate array (FPGA), and programmable logic array (PLA). The processor can integrate one or more combinations of a central processing unit (CPU), a graphics processor (GPU), and a modem. Among them, the CPU mainly processes the operating system, user interface, and computer programs; the GPU is responsible for rendering and drawing display content; and the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor, but may be implemented separately through a communication chip. Specifically, the method described in the above embodiment can be executed by one or more processors.

[0212] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A collision prediction method, characterized in that: Applied to a first vehicle, the method comprises: Obtaining a first network packet loss rate between the first vehicle and the remote control terminal at a current moment; Predicting a first duration of network anomaly based on the first network packet loss rate using a network prediction model; Determine a collision threshold duration according to the first duration; The possibility of a collision of the first vehicle is predicted according to the first vehicle's driving parameters, environmental parameters and the collision threshold duration at the current moment.

2. The method according to claim 1, characterized in that Before predicting the first duration of the network anomaly based on the first network packet loss rate using the network prediction model, the method further includes: Acquire sample data, wherein the sample data includes a plurality of sample moments, a first sample data volume received by the first vehicle at each sample moment, a second sample data volume sent by the remote control terminal at each sample moment, and a standard duration of a network anomaly corresponding to each sample moment; Based on the initial model and according to the sample data, the initial duration of the network anomaly corresponding to each of the sample moments is predicted; The initial model is trained based on the difference between the initial duration and the standard duration to obtain the network prediction model.

3. The method according to claim 2, characterized in that The obtaining of sample data comprises: Acquiring frequency parameters of communication between the first vehicle and the remote control terminal; determining a plurality of sample moments based on the frequency parameter; Determine the amount of first data received by the first vehicle corresponding to each of the sample moments, and the amount of second data sent by the remote control terminal corresponding to each of the sample moments; Based on the plurality of the first data amounts and the plurality of the second data amounts, respectively determine a first sample data amount and a second sample data amount corresponding to each sample moment; Based on the first sample data volume and the second sample data volume corresponding to the sample moment, a standard duration of the network anomaly corresponding to the sample moment is determined.

4. The method according to claim 3, characterized in that The determining, based on the first sample data volume and the second sample data volume corresponding to the sample time, a standard duration of the network anomaly corresponding to the sample time, includes: Determine a second network packet loss rate corresponding to the sample time based on the first sample data volume and the second sample data volume corresponding to the sample time; Based on each of the second network packet loss rates, determining whether a network is abnormal at a sample time corresponding to the second network packet loss rate; The standard duration of the network anomaly corresponding to each sample moment is determined based on the sample moment of each network anomaly before each sample moment.

5. The method according to claim 4, characterized in that The second network packet loss rate includes a first downlink packet loss rate and a first uplink packet loss rate, and judging whether the network is abnormal at a sample time corresponding to the second network packet loss rate based on each of the second network packet loss rates includes: Taking the absolute value of the difference between the first downlink packet loss rate and the first uplink packet loss rate as the difference packet loss rate corresponding to the sample time; In the case where it is detected that the difference packet loss rate is greater than a first threshold packet loss rate, determining whether a target event has been detected a specified number of times at the sample moment, wherein the target event is that the absolute value of the difference between the first difference packet loss rate and the second difference packet loss rate is less than the second threshold packet loss rate, the first difference packet loss rate is the difference packet loss rate corresponding to any sample moment before the sample moment, and the second difference packet loss rate is the difference packet loss rate corresponding to a previous sample moment adjacent to the sample moment corresponding to the first difference packet loss rate; When the target event is detected a specified number of times, it is determined that the network is abnormal at the sample time corresponding to the second network packet loss rate.

6. The method according to claim 1, characterized in that The predicting the possibility of the first vehicle colliding according to the current driving parameters of the first vehicle, the environmental parameters and the collision threshold duration includes: Predicting the duration of a collision of the first vehicle based on the driving parameters and environmental parameters of the first vehicle at the current moment; Obtaining a difference between the duration and the collision threshold duration as a first intermediate value; When the first intermediate value is less than or equal to the threshold value, the possibility of the collision is predicted to be a first possibility; When the first intermediate value is greater than the threshold value, the possibility of the predicted collision is a second possibility, wherein the first possibility is greater than the second possibility.

7. The method according to claim 6, characterized in that At the current moment, there is a second vehicle ahead of the first vehicle in the driving direction, the driving parameters of the first vehicle include a first driving speed, a first acceleration, and a distance between the first vehicle and the second vehicle ahead, and the environmental parameters include a second driving speed and a second acceleration of the second vehicle; The predicting of the duration of the collision of the first vehicle based on the driving parameters and environmental parameters of the first vehicle at the current moment includes: obtaining a difference between the first driving speed and the second driving speed as a second intermediate value, obtaining a difference between the first acceleration and the second acceleration as a third intermediate value, and obtaining a difference between the distance and a body length of the first vehicle as a fourth intermediate value; When the third intermediate value is 0, obtaining a ratio of the fourth intermediate value to the second intermediate value as the duration; When the third intermediate value is not 0, performing a square root operation on the fourth intermediate value and then multiplying the result with the specified value and the third intermediate value to obtain a sixth intermediate value; subtract the second intermediate value from the sum of the square of the second intermediate value and the sixth intermediate value to obtain a seventh intermediate value; The ratio of the seventh intermediate value to the third intermediate value is used as the duration.

8. The method according to claim 1, characterized in that The determining the collision threshold duration according to the first duration includes: Obtaining a first time consumption for predicting the first duration by using the network prediction model; determining a second time taken for the first vehicle to respond to the braking instruction when the first vehicle obtains the braking instruction; The sum of the first time consumption, the second time consumption and the first duration is used as the collision threshold duration.

9. A collision prediction device, characterized in that: Applied to a first vehicle, the device comprises: An acquisition unit, configured to acquire a first network packet loss rate between the first vehicle and the remote control terminal at a current moment; A network anomaly duration prediction unit, configured to predict a first duration of the network anomaly based on the first network packet loss rate using a network prediction model; a determining unit, configured to determine a collision threshold duration according to the first duration; The collision result prediction unit is used to predict the possibility of the first vehicle colliding according to the first vehicle's driving parameters, environmental parameters and the collision threshold duration at the current moment.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program codes, which can be called by a processor to execute the method according to any one of claims 1 to 8.

Images