A highway confusing event discrimination method based on radar trajectory data

Abstract

The application discloses a kind of expressway confusing event discrimination method based on radar trajectory data, collect the driving trajectory of each car and the position information of real-time vehicle on current road surface, judge whether vehicle is stopped driving, stop driving vehicle is stored in set, there is vehicle stop driving, the number of vehicle around stop driving vehicle is counted, if it meets car many constraint condition, then continue to judge the average speed of surrounding vehicle, whether it is congestion state is judged according to speed, enter step 4;If it does not meet car many constraint condition, enter step 5;Using the way of deceleration grading, improve the distinction of each event;Multi-dimensional analysis and discrimination mode is used to each event, such as traffic flow, deceleration, distance, parking time, deceleration time, heading angle, pedestrian and other factors, using hierarchical analysis method, more comprehensive and meticulous discrimination congestion, car accident, car road accident, illegal parking, realize the purpose of low cost, high reliability, high precision, all-weather, full coverage discrimination confusing event.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of highway event detection, more specifically, and particularly relates to a highway easily confused event discrimination method based on radar trajectory data. BACKGROUND

[0002] In recent years, with the development of video and AI neural network technology, cameras are generally used for perception on the road. However, there are still many problems. First, high-definition cameras are expensive and are easily affected by lighting, weather, etc., which can cause data loss, detection errors and other abnormalities. Second, there are a large number of ordinary cameras on the road, and if a large number of high-definition cameras are used, the existing perception system is not fully utilized, which can cause waste. Radar can overcome the influence of lighting and weather, and can make full use of the existing perception system. At present, the detection distance of a single radar is about 250m, and the trajectory splicing of multiple radars can realize all-weather and full-coverage detection of vehicle position.

[0003] On the highway, vehicles involved in traffic accidents generally exhibit collisions, sudden deceleration, vehicle head deviation from normal driving direction, stopping, and driver getting off the vehicle. However, these phenomena do not occur simultaneously and are often a combination of several phenomena, and often similar to other traffic events. For example, a vehicle collision may exhibit a slight collision, a slight deviation of the vehicle head from the normal driving direction, and a stop and driver getting off the vehicle, which is similar to the phenomenon of occupying the emergency lane and parking. Congestion can also affect the identification of vehicle collisions, such as when congestion occurs, a large number of vehicles are queuing in the lane and even stopping, at which time individual vehicles may exhibit deviation of the vehicle head from the normal driving direction, stopping, and even driver getting off the vehicle. Vehicle-road accidents and parking violations have similar phenomena, such as deviation of the vehicle head from the normal driving direction, stopping, and driver getting off the vehicle, and in some cases, sudden deceleration and collision are not obvious, so they cannot be simply distinguished. In view of the above, it is necessary to analyze congestion, parking violations, vehicle collisions, and vehicle-road accidents together.

[0004] However, current event recognition algorithms have some problems: some methods are only suitable for some simple events such as reverse driving, illegal lane changing, and occupying the emergency lane, and when accidents or parking violations occur, the recognition accuracy is not sufficient. Some methods use neural networks that require a large amount of training data, which is difficult to obtain.

[0005] Some situations of existing patents:

[0006] The patent "Traffic event detection method, device, equipment and storage medium" (patent application number: 202211024672.7) only inputs video frames into a neural network to obtain the probability of each event, and does not explain how to ensure the accuracy of model training and the influence of weather on video.

[0007] The patent "Event detection method, device and system based on tunnel radar and video data fusion" (patent application number: 202111218621.3) only determines that it is illegal parking by zero speed and parking time greater than a time threshold, without distinguishing between congestion and accident caused illegal parking.

[0008] The patent "Highway traffic accident identification method, module and system" (patent application number: 202111071589.0) inputs historical traffic flow parameter information and current traffic parameter information into a pre-trained traffic accident identification model to obtain a traffic accident occurrence probability, but the generality of the traffic accident identification model is not explained, and the prediction performance needs to be iteratively upgraded, even if used at the same location, it will take a long time to really play a role.

[0009] In view of the shortcomings of the prior art, a highway easily confused event discrimination method based on radar trajectory data is proposed. SUMMARY

[0010] The purpose of the present application is to solve the problems existing in the prior art, and a highway easily confused event discrimination method based on radar trajectory data is proposed.

[0011] To achieve the above purpose, the present application provides the following technical scheme: a highway easily confused event discrimination method based on radar trajectory data, comprising the following steps:

[0012] Step 1: collect the driving trajectory of each vehicle and the real-time position information of the vehicle on the current road surface;

[0013] Step 2: determine whether the vehicle is stopped, and store the stopped vehicle in set A;

[0014] Step 3: if there is a stopped vehicle, count the number of vehicles around the stopped vehicle, if the vehicle constraint condition is met, continue to determine the average speed of the surrounding vehicles, and determine whether it is a congestion state according to the speed, and go to step 4; if the vehicle constraint condition is not met, go to step 5;

[0015] Step 4: If not congested, go to Step 5; if congested, determine whether the parking time of the parking vehicle in the congested vehicle group meets the threshold value β (β can be determined in combination with the range of the sensing area and the average speed, and β is set as 60s by default; when the average speed is close to 0km / h, it is not compared with β, and the flow is ended) constraint, if the parking time is less than β, the corresponding parking vehicle is removed, the set A is updated, and Step 6 is entered;

[0016] Step 5: Take out the vehicle with a large deceleration greater than the threshold value γ (γ is set as 8 m / s 2 by default) from the set A, if there is a vehicle meeting the condition, go to Step 7; otherwise, go to Step 8;

[0017] Step 6: If the set A is empty, end the flow; otherwise, go to Step 5;

[0018] Step 7: Find the parking vehicle with a distance less than the threshold value d (d is set as 7m by default) from the current vehicle at a certain time before parking, determine it as a related vehicle, and determine the event as a vehicle-vehicle accident. If the related vehicle exists, remove the current vehicle and the related vehicle, and update the set A. If no parking vehicle close to the current vehicle is found, remove the current parking vehicle, determine the current vehicle as a vehicle-road accident, update the set A, and go to Step 6;

[0019] Step 8: Take out the vehicle with a large deceleration greater than the threshold value δ (δ is set as 5 m / s 2 by default) from the set A, if there is no vehicle meeting the condition, go to Step 9; otherwise, determine the time of the vehicle maintaining the large deceleration, if the time meets the threshold value μ (μ is set as 500ms by default), go to Step 7, otherwise, go back to Step 8;

[0020] Step 9: Take out the vehicle with a large deceleration greater than the threshold value θ (θ is set as 2 m / s 2 by default) from the set A, if there is no vehicle meeting the condition, go to Step 10; otherwise, go to Step 11;

[0021] Step 10: Determine whether the heading angle of the current vehicle deviates from the normal direction, if yes, go to Step 7; otherwise, go to Step 14;

[0022] Step 11: Determine the time of the vehicle maintaining the large deceleration, if the time does not meet the threshold value μ, go to Step 10; if the time meets the threshold value μ, go to Step 12;

[0023] Step 12: Determine whether there is a heading angle deviating from the normal direction before parking, if there is a deviation angle greater than τ (τ is set as 60 。 by default), go to Step 7; if there is no heading angle deviating from the normal direction, go to Step 13;

[0024] Step 13: Check if there are pedestrians around the vehicle, if yes, go to step 7; otherwise, go to step 14;

[0025] Step 14: Remove the current stopped vehicle, the current vehicle is determined as illegal parking, update set A, and go to step 6.

[0026] Preferably, in step 1, the traffic congestion is characterized by: many cars slow down. In the national standard, the speed is used to classify congestion, but there is no clear requirement for the number of cars. If only a small number of cars are slow or even stopped on the road, it cannot be truly said that the road is congested, so the number of slow cars needs to be limited.

[0027] Preferably, in step 2, it is determined whether the vehicle is stopped (at least meets the speed of 0 km / h for 500 ms), and the stopped vehicle is stored in set A. The deceleration value of the vehicle in set A before entering the stopped state is calculated for deceleration classification discussion. If there is no stopped vehicle, the flow is ended.

[0028] Preferably, in step 3, the maximum number of vehicles that the road can accommodate is related to the size of the evaluation area and the safety distance. The safety distance has many influencing factors and it is not easy to obtain an accurate value. When a certain number is reached, it can be said that the effectiveness of the slow-down result is valid. The car constraint is as follows:

[0029] Veh num > num

[0030] Veh num is the minimum number of vehicles required for car, and a can be set according to the actual situation, num is the number of lanes.

[0031] Preferably, in step 4, when severe congestion occurs, the vehicle will be in a stopped state. At this time, the stopped vehicle caused by congestion should not be judged as illegal parking or accident. Unlike accident or illegal parking vehicles, which generally stay in the original position for a long time, the surrounding vehicles will move slowly. The stopped vehicle in congestion will move slowly with the traffic flow, and even if it is stopped for a long time, it will also be stopped for a long time with the surrounding vehicles. If the congestion is caused by an accident, the accident will be discovered before the congestion.

[0032] Preferably, the stoppage time length threshold β is different according to the congestion level, and when the time threshold β is exceeded, it can be determined that the stoppage is not caused by congestion. If severe congestion occurs (the average speed is close to 0 km / h), the vehicle will be stopped for a long time, and the stoppage time can not be compared with the time threshold β. To simplify the judgment process, and because it is difficult to calculate the accurate value of the threshold β for each congestion level, the default value of β is set to 60 s (except for the case where the average speed is close to 0 km / h).

[0033] Preferably, in step 5, from the braking ability of the vehicle, the maximum deceleration of the vehicle during emergency braking is 7.5-8 m / s 2 . When a vehicle-road collision accident occurs, the speed of the vehicle may be decelerated at the maximum deceleration and eventually become 0 km / h, and the driver gets off to check and avoid danger; or the driver may not take corresponding action, and there will be no maximum deceleration phenomenon; or in extreme cases, the vehicle collides with the road, and the maximum deceleration may exceed 8 m / s 2 ; during the accident, the heading angle may deviate from the lane for a period of time, and in some cases, it may also be similar to the phenomenon of illegal parking.

[0034] Preferably, in step 7, the deceleration is classified: the deceleration threshold values γ, δ, and θ are set to 8 m / s 2 , 5 m / s 2 , and 2 m / s 2 , respectively. When the deceleration a is greater than 8 m / s 2 , that is, exceeds the braking limit of the vehicle, it indicates that the vehicle may not only be subjected to the braking force from the vehicle itself, but also be subjected to the impact force from the outside, that is, it indicates that an accident has occurred; when the deceleration a is between 5 m / s 2 and 8 m / s 2 , it indicates that the vehicle is subjected to a force close to or equivalent to the braking limit of the vehicle, and in combination with the final stoppage state of the vehicle, it is determined that an accident has occurred; because in actual use of braking, the braking deceleration should not be greater than 1.5 m / s 2 except in emergency situations, when the deceleration a is between 2 m / s 2 and 5 m / s 2 , it indicates that the vehicle has made the passengers feel uncomfortable or caused danger or caused the goods to be unsafe, in combination with the final stoppage of the vehicle, the change of the heading angle, and the appearance of pedestrians around, the accident situation is judged; when the deceleration a is less than 2 m / s 2 , it cannot be determined that an accident has not occurred, and the heading angle needs to be combined for judgment.

[0035] Preferably, in step 10, when a vehicle-road accident occurs, the heading angle deviates from the normal driving direction, the heading angle of the driving direction of the road is set as , the deviation angle threshold is τ, and the heading angle of the vehicle during driving is ω. When , it is considered that the normal driving direction is deviated, that is, it is determined that an accident has occurred;

[0036] Unlike vehicle-road collisions, such accidents involve more than one vehicle. When a collision occurs, there may be sudden deceleration, two vehicles approaching each other at an infinite distance, speed dropping to 0 km / h, and pedestrians on the side of the road. The deceleration threshold is determined by the same rule as the vehicle-road collision;

[0037] Pedestrians on the side of the road: when the vehicle deceleration is between 2 m / s 2 and 5 m / s 2 for a period of time, it indicates that the vehicle has undergone a relatively urgent unexpected deceleration process and finally stopped, and the passengers get off the vehicle, determining that the vehicle has had an accident;

[0038] The vehicle determined to have an accident is close to or even less than the set threshold d from the vehicles in the set A at a certain time, and is determined to be a related vehicle, that is, a vehicle-vehicle accident has occurred.

[0039] Preferably, in step 14, the vehicle speed is 0 km / h. It is necessary to exclude congestion and parking caused by accidents.

[0040] Technical effects and advantages of the present application: the highway easily confused event discrimination method based on radar trajectory data provided by the present application has the following advantages compared with the prior art:

[0041] 1. The deceleration is classified, and the distinction of each event is improved;

[0042] 2. The vehicle flow, deceleration, distance, parking time, deceleration time, heading angle, pedestrian and other factors are analyzed and distinguished in multiple dimensions, and the hierarchical analysis method is used to more comprehensively and carefully distinguish congestion, vehicle-vehicle accident, vehicle-road accident, and illegal parking to achieve the purpose of low cost, high reliability, high precision, all-weather, full coverage and easy-to-confuse event discrimination.

[0043] Other features and advantages of the present application will be described in the following description, and some will become apparent from the description, or will be understood by those skilled in the art. The purpose and other advantages of the present application can be achieved and obtained by the structure indicated in the specification and the drawings. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 The step flowchart of the present application. DETAILED DESCRIPTION

[0045] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to specific examples. It should be understood that the specific examples described herein are only used to explain the present application and should not be used to limit the present application. Based on the examples in the present application, all other examples obtained by those of ordinary skill in the art without creative effort belong to the scope of protection of the present application.

[0046] The present application provides a highway confusing event discrimination method based on radar trajectory data as shown in Figure 1 The present application provides a highway confusing event discrimination method based on radar trajectory data as shown in

[0047] The present application provides a highway confusing event discrimination method based on radar trajectory data as shown in

[0048] Step 1: Collect the driving trajectory of each vehicle and the real-time position information of the vehicle on the current road surface;

[0049] Step 2: Determine whether the vehicle is stopped, and store the stopped vehicle in set A;

[0050] Step 3: If there is a stopped vehicle, count the number of vehicles around the stopped vehicle, if the vehicle constraint condition is met, continue to determine the average speed of the surrounding vehicles, and determine whether it is a congestion state according to the speed, and enter step 4; if the vehicle constraint condition is not met, enter step 5;

[0051] Step 4: If it is not a congestion state, enter step 5; if it is a congestion state, determine whether the stopped time of the stopped vehicle in the congestion vehicle group meets the threshold β (β can be determined in combination with the size of the perception area range and the average speed, and β is set to 60s by default; when the average speed is close to 0km / h, it is not compared with β, and it is determined as congestion, and the process is ended) constraint, if the stay time is less than β, the corresponding stopped vehicle is removed, set A is updated, and step 6 is entered;

[0052] Step 5: Take out the vehicle with a large deceleration greater than the threshold γ (γ is set to 8 m / s 2 by default) from set A, if there is a vehicle meeting the condition, enter step 7; otherwise, enter step 8;

[0053] Step 6: If set A is empty, end the process; otherwise, enter step 5;

[0054] Step 7: Find the stopped vehicle which is less than a threshold d (d is set to 7m by default) from the current vehicle at a certain time before the current vehicle stops, determine it as the relevant vehicle, and determine the event as a vehicle-vehicle accident. If the relevant vehicle exists, remove the current vehicle and the relevant vehicle, and update the set A. If no stopped vehicle is found near the current vehicle, remove the current stopped vehicle, determine the current vehicle as a vehicle-road accident, update the set A, and go to step 6;

[0055] Step 8: Take the vehicle with a sharp deceleration greater than a threshold δ (δ is set to 5 m / s by default) from the set A. If no vehicle meets the condition, go to step 9; otherwise, determine the time for which the vehicle maintains a sharp deceleration. If the time meets a threshold μ (μ is set to 500ms by default), go to step 7, otherwise, go back to step 8; 2

[0056] Step 9: Take the vehicle with a sharp deceleration greater than a threshold θ (θ is set to 2 m / s by default) from the set A. If no vehicle meets the condition, go to step 10; otherwise, go to step 11; 2

[0057] Step 10: Determine whether the heading angle of the current vehicle deviates from the normal direction. If yes, go to step 7; otherwise, go to step 14;

[0058] Step 11: Determine the time for which the vehicle maintains a sharp deceleration. If the time does not meet the threshold μ, go to step 10; if the time meets the threshold μ, go to step 12;

[0059] Step 12: Determine whether the vehicle has a heading angle deviating from the normal direction before stopping. If there is a deviation angle greater than τ (τ is set to 60 。 by default), go to step 7; if there is no deviation from the normal direction, go to step 13;

[0060] Step 13: Check whether there are pedestrians around the vehicle. If there are, go to step 7; otherwise, go to step 14;

[0061] Step 14: Remove the current stopped vehicle, determine the current vehicle as a parking violation, update the set A, and go to step 6.

[0062] In step 1, the characteristics of traffic congestion are: many vehicles and slow driving. The national standard classifies congestion by speed, but does not explicitly require the number of vehicles. If only a small number of vehicles are slowly driving or even stopped on the road, it cannot truly indicate that the road is congested, so the number of slow driving vehicles needs to be limited.

[0063] ​​In step 2, it is judged whether the vehicle is stopped (at least the speed is 0 km / h for 500 ms), the stopped vehicle is stored in set A, and the deceleration value of the vehicle in set A before entering the stopped state is calculated for deceleration classification discussion; if no vehicle is stopped, the judgment process ends.

[0064] In step 3, the maximum number of vehicles that the road can accommodate is related to the size of the evaluation area and the safety distance. The safety distance has many influencing factors and is not easy to obtain an accurate value. When a certain number is reached, it can be explained that the effectiveness of the slow-moving result. The vehicle constraint is as follows:

[0065] Veh num > num

[0066] Veh num The minimum number of vehicles required for vehicle constraint is α, which can be set according to the actual situation, num The number of lanes. The congestion speed level can be set according to the national standard.

[0067] In step 4, when severe congestion occurs, the stopped vehicle caused by congestion should not be judged as a violation or an accident. Unlike accidents or violations, vehicles generally stay in the same place for a long time, and surrounding vehicles move slowly. The stopped vehicle in congestion will move slowly with the traffic flow, and even if it is stopped for a long time, it will also be stopped for a long time with the surrounding vehicles. If the congestion is caused by an accident, the accident will be discovered before the congestion.

[0068] According to the different levels of congestion, the stopped time length threshold β will be different. If the time threshold β is exceeded, it can be judged as a non-congestion caused by parking. If severe congestion (average speed close to 0 km / h) occurs, the vehicle will be stopped for a long time, and the stopped time can not be compared with the time threshold β. To simplify the judgment process, and it is difficult to calculate the accurate value of each level for each congestion level, the default value of β is set to 60 s (except for the case where the average speed is close to 0 km / h).

[0069] In step 5, from the braking ability of the vehicle, the maximum deceleration of the vehicle during emergency braking is 7.5-8 m / s 2 . When a vehicle-road collision accident occurs, the vehicle speed may decrease to 0 km / h with the maximum deceleration, and the driver may get off to check and avoid danger; it may not have the maximum deceleration phenomenon if the driver does not take corresponding action; in extreme cases, the vehicle may collide with the road, and the maximum deceleration may exceed 8 m / s 2The heading angle may deviate from the lane for a period of time during the accident process, and in some cases, it may be similar to the phenomenon of illegal parking.

[0070] In step 7, the deceleration is graded: set the deceleration threshold γ, δ, θ, the default values are 8 m / s 2 , 5 m / s 2 , and 2 m / s 2 respectively. When the deceleration a > 8 m / s 2 , i.e. exceeds the braking limit of the vehicle, it indicates that the vehicle may not only be subjected to the braking force from the vehicle itself, but should also be subjected to the impact force from the outside, i.e. it indicates that an accident has occurred; when the deceleration a is between 5 m / s 2 and 8 m / s 2 , it indicates that the vehicle is subjected to a force close to or equivalent to the braking limit of the vehicle, and in combination with the final stop state of the vehicle, it is determined that an accident has occurred; in actual use of braking, the braking deceleration should not be greater than 1.5 m / s 2 under normal circumstances except in emergency situations; when the deceleration a is between 2 m / s 2 and 5 m / s 2 , it indicates that the vehicle has made the passengers feel uncomfortable or dangerous or caused unsafe problems of goods, in combination with the final stop of the vehicle, the change of the heading angle and the appearance of pedestrians around, the accident situation is judged; when the deceleration a < 2 m / s 2 , it cannot be determined that no accident has occurred, and the heading angle needs to be combined for judgment.

[0071] In step 10, when a vehicle-road accident occurs, the heading angle will deviate from the normal driving direction, the heading angle of the driving direction of the road is set as , the deviation angle threshold is τ, and the heading angle of the vehicle during driving is ω. When , it is considered that the normal driving direction is deviated, i.e. it is determined that an accident has occurred.

[0072] Unlike vehicle-road collision, this type of accident involves more than one vehicle. When the collision occurs, there may be sudden deceleration, the distance between the two vehicles approaching infinity, the speed dropping to 0 km / h, and pedestrians on the side of the road. The deceleration threshold is judged by the same rule as the vehicle-road collision;

[0073] Pedestrians on the side of the road: when the vehicle deceleration is between 2 m / s 2 and 5 m / s 2 and lasts for a period of time, it indicates that the vehicle has gone through a relatively emergency unexpected deceleration process and finally stopped, and the passengers get off the vehicle, it is determined that the vehicle has had an accident;

[0074] If the distance between the vehicle in which the accident occurs and the vehicle in the set A of stopped vehicles is close to or less than the set threshold d at a certain time, the vehicle is determined as a related vehicle, i.e., a vehicle in which a vehicle-vehicle accident occurs.

[0075] In step 14, the vehicle speed is 0 km / h. It is necessary to exclude the parking caused by congestion and accidents.

[0076] Working principle: Firstly, the real-time trajectories of vehicles are collected, and a set of stopped vehicles is stored. When there is a stopped vehicle in the set, it is determined whether there is congestion around the stopped vehicle. If there is congestion, the stopped vehicle in the congestion scene is analyzed. If there is no congestion, the deceleration of the stopped vehicle before stopping is analyzed. For different degrees of deceleration, targeted analysis is performed. Finally, the heading angle and surrounding pedestrians are comprehensively analyzed to achieve the purpose of distinguishing congestion, vehicle-vehicle accidents, vehicle-road accidents, and illegal parking.

[0077] Finally, it should be noted that the above only describes the preferred embodiments of the present application and is not intended to limit the present application. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent replacements to some technical features. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for identifying confusing events on a highway based on radar trajectory data, characterized in that: The method comprises the following steps: Step 1: Collecting the driving track of each vehicle and the real-time position information of the vehicle on the current road surface; Step 2: Judging whether the vehicle is stopped, and storing the stopped vehicle into a set A; Step 3: When the vehicle is stopped, counting the number of vehicles around the stopped vehicle, if the number of vehicles meets the constraint condition, then judging the average speed of the vehicles around the stopped vehicle, judging whether the traffic is congested according to the speed, and entering step 4; if the number of vehicles does not meet the constraint condition, entering step 5; Step 4: If the traffic is not congested, entering step 5; if the traffic is congested, judging whether the stopped time of the stopped vehicle in the congested vehicle group meets the threshold β constraint, if the stopped time is less than β, then removing the corresponding stopped vehicle, updating the set A, and entering step 6; Step 5: Removing the vehicle with the deceleration greater than the threshold γ from the set A, if there is a vehicle meeting the condition, entering step 7; otherwise, entering step 8; Step 6: If the set A is empty, ending the process; otherwise, entering step 5; Step 7: Finding the stopped vehicle which is close to the current vehicle at a certain time before the current vehicle is stopped, determining the stopped vehicle as a related vehicle, and determining the event as a vehicle-vehicle accident; if the related vehicle exists, removing the current vehicle and the related vehicle, and updating the set A; If no stopped vehicle close to the current vehicle is found, removing the current stopped vehicle, determining the current vehicle as a vehicle-road accident, updating the set A, and entering step 6; Step 8: Removing the vehicle with the deceleration greater than the threshold δ from the set A, if there is no vehicle meeting the condition, entering step 9; otherwise, judging the time of the vehicle maintaining the deceleration, if the time meets the threshold μ, entering step 7, otherwise, returning to step 8; Step 9: Removing the vehicle with the deceleration greater than the threshold θ from the set A, if there is no vehicle meeting the condition, entering step 10; otherwise, entering step 11; Step 10: Judging whether the heading angle of the current vehicle deviates from the normal direction, if yes, entering step 7; otherwise, entering step 14; Step 11: Judging the time of the vehicle maintaining the deceleration, if the time does not meet the threshold μ, entering step 10; if the time meets the threshold μ, entering step 12; Step 12: Judging whether the vehicle has a heading angle deviating from the normal direction before the vehicle is stopped, if there is a deviation angle greater than τ, entering step 7; if there is no heading angle deviating from the normal direction, entering step 13; Step 13: Judging whether there is a pedestrian around the vehicle, if yes, entering step 7; otherwise, entering step 14; Step 14: Removing the current stopped vehicle, determining the current vehicle as a parking violation, updating the set A, and entering step 6. 2.The method of claim 1, wherein: In the step 1, the feature of traffic congestion is that there are many vehicles and the vehicles are driving slowly. 3.The method of claim 1, wherein: In the step 2, whether the vehicle is stopped is judged, the stopped vehicle is stored into the set A, the deceleration value of the vehicle in the set A before entering the stopped state is calculated, which is used for deceleration classification discussion; if there is no vehicle stopped, the process is ended.

4. The method of claim 1, wherein: In the step 3, the maximum number of vehicles that the road can accommodate is related to the size of the evaluation area and the safety distance; The constraint of the number of vehicles is as follows: Veh num > num Veh num The minimum number of vehicles required to satisfy the vehicle, and α is set according to the actual situation, num The number of lanes.

5. The method of claim 1, wherein: In the step 4, when serious congestion occurs, the stopped vehicle caused by the congestion should not be determined as a parking violation or an accident.

6. The method of claim 5, wherein: According to the congestion level, the stop time length threshold β is different. If the time threshold β is exceeded, it is determined that the stop is not caused by congestion. If there is serious congestion, the vehicle will stop for a long time, and the stop time is not compared with the time threshold β.

7. The method of claim 1, wherein: The maximum deceleration of the vehicle is 7.5-8 m / s 2 when the vehicle collides with the road, the speed of the vehicle may be decelerated at the maximum deceleration and eventually become 0 km / h, the driver gets off to check and avoid danger, or the driver may not take corresponding action, and there is no maximum deceleration phenomenon, or in extreme cases, the vehicle collides with the road, the speed sharply decreases, and the maximum deceleration may exceed 8 m / s 2 The heading angle may deviate from the lane for a period of time during the accident process, and in some cases, it may not be excluded from the similar parking violation phenomenon.

8. The method of claim 1, wherein: The step 7, deceleration grading: set deceleration threshold γ, δ, θ, the default values are 8 m / s 2 , 5 m / s 2 , 2 m / s 2 ; when the deceleration a>8 m / s 2 , that is, exceeds the automobile braking limit, it indicates that the vehicle not only may be subjected to the braking force from the vehicle itself, but also should be subjected to the impact force from the outside, that is, it indicates that an accident has occurred; when the deceleration a is between 5 m / s 2 and 8 m / s 2 , it indicates that the vehicle is subjected to a force close to or equivalent to the vehicle braking limit, and in combination with the final stop state of the vehicle, it is determined that an accident has occurred; because in actual use of braking, except in emergency situations, the braking deceleration should not be greater than 1.5 m / s 2 , when the deceleration a is between 2 m / s 2 and 5 m / s 2 , it indicates that the vehicle has made the passengers feel uncomfortable or dangerous or caused the goods to be unsafe, and in combination with the final stop, the change of the heading angle and the appearance of pedestrians around, the accident situation is judged; when the deceleration a<2 m / s 2 , it needs to be judged in combination with the heading angle.

9. The method of claim 1, wherein: In step 10, when a vehicle-road accident occurs, the heading angle deviates from the normal driving direction, the heading angle of the driving direction of the road is set as , the deviation angle threshold is τ, and the heading angle of the vehicle during driving is ω. When , it is considered that the normal driving direction is deviated, that is, it is determined that an accident has occurred. Unlike car-road collisions, such accidents involve more than one vehicle. When the collision occurs, there may be sudden deceleration, the collision distance between the two vehicles is infinitely close, the speed is reduced to 0 km / h, and there are pedestrians on the side of the road. The deceleration threshold is the same as that for car-road collisions. Car side has pedestrian: when the vehicle deceleration is between 2m / s 2 and 5m / s 2 for a period of time, it indicates that the vehicle has gone through a relatively emergency unexpected deceleration process and finally stops, and the people in the vehicle get off, and it is determined that the vehicle has an accident; If the distance between the vehicle that determines the accident and the vehicles in the stop vehicle set A is close to or less than the set threshold d at a certain time, it is determined that the vehicle is related, i.e. a car-car accident occurs.

10. The method of claim 1, wherein: In step 14, the vehicle speed is 0 km / h, and it is necessary to exclude the stop caused by congestion and accidents.

Images