A method for investigating and handling traffic accident scenes based on drones

Abstract

The present application discloses a method for investigating and processing traffic accident scenes based on drones, which relates to the technical field of traffic guidance. After a traffic accident occurs, it quickly locates the accident location based on the mobile phones of the accident personnel and the accident vehicles, guiding the high-altitude drone to quickly go to the accident scene; calls the high-altitude drone for high-altitude detection to obtain the accident scene range and surrounding environment information, and temporarily intercepts the subsequent traffic; the high-altitude drone releases the evidence-taking drone to restore and generate the accident cause; constructs an accident liability judgment model to generate an accident liability recognition letter; the high-altitude drone releases the safety drone to obtain accident safety hazards and road safety hazards; based on the accident safety hazards, calls for support forces to handle the accident safety hazards; based on the road safety hazards, conducts traffic control on the road traffic. The present application has the effects of improving the speed, accuracy of traffic accident handling and rescue, and the efficiency of road traffic.

Description

Technical Field

[0001] This application relates to the technical field of traffic guidance, and in particular, to a method for investigating and handling traffic accident scenes based on drones. Background Art

[0002] A traffic accident refers to an event in which a vehicle causes personal injury, death or property damage on the road due to fault or accident. After a traffic accident occurs, it needs to be processed in a timely manner. Otherwise, the accident vehicle on the road will cause traffic jams and even trigger secondary accidents.

[0003] In the prior art, there are two ways to handle traffic accidents on the road. One is a minor accident with a clear responsible party, such as a minor fender bender, etc., which can be directly completed through the quick handling channel. The other is a more serious accident or an unclear responsible party, in which case traffic police need to be called to the accident site for handling. In the second case above, it generally takes a long time to handle, and during the handling process, it will affect road traffic, resulting in a decline in road traffic efficiency. On the other hand, during the process of reporting the accident, due to unfamiliarity with the geographical location, the description of the accident location may be unclear, resulting in a waste of time. In addition, when investigating the accident scene, for the same accident scene, different traffic police have different abilities to investigate the accident scene, and it is impossible to accurately investigate the accident scene. Different traffic police may choose different legal provisions and give different results, resulting in inaccurate handling results and affecting accident rescue at the same time. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for investigating and handling traffic accident scenes based on drones to solve the problems raised in the above background art.

[0005] This application provides a method for investigating and handling traffic accident scenes based on drones, and the method includes:

[0006] After a traffic accident occurs, quickly locate the accident location based on the mobile phones of the accident personnel and the accident vehicle, and guide a high-altitude drone to quickly fly to the accident scene according to the accident location;

[0007] Based on the accident scene, call the high-altitude drone for high-altitude detection to obtain the accident scene range and surrounding environment information, and temporarily intercept the subsequent traffic;

[0008] The high-altitude drone releases a forensics drone, and the forensics drone obtains accident information based on the accident scene range, and restores the accident according to the accident information to generate the accident cause;

[0009] Construct an accident liability judgment model, input the accident cause into the accident liability judgment model, and generate an accident liability recognition letter;

[0010] The high-altitude drone releases a safety drone, and the safety drone obtains accident safety hazards and road safety hazards according to the accident scene range and the surrounding environment information respectively;

[0011] Based on the accident safety hazards, accident hazard types are obtained, support signals are generated according to the accident hazard types, and support forces are called according to the support signals to handle the accident safety hazards;

[0012] Based on the road safety hazards, road traffic information is collected, and road traffic is controlled according to the road traffic information.

[0013] Preferably, the steps of quickly positioning the accident location according to the mobile phones of accident personnel and the accident vehicle and guiding the high-altitude drone to quickly go to the accident scene are as follows:

[0014] The accident vehicle collects the current location information when the accident occurs, and at the same time scans the surrounding of the vehicle to obtain the target mobile phones of the accident personnel;

[0015] The accident vehicle generates an alarm prompt according to the current location information and sends it to the target mobile phone;

[0016] After receiving the alarm prompt, the target mobile phone obtains the mobile phone location information, verifies the mobile phone location information with the current location information, and obtains the accident location;

[0017] The target mobile phone alarms according to the accident location, sends the accident location to the nearest high-altitude drone, and the high-altitude drone quickly goes to the accident scene according to the accident location.

[0018] Preferably, the steps of calling the high-altitude drone for high-altitude detection based on the accident scene to obtain the accident scene range and the surrounding environment information are as follows:

[0019] The high-altitude drone conducts high-altitude detection on the accident scene, obtains the on-site photos of the accident scene, extracts the physical objects existing in the on-site photos, and extracts the physical characteristics of each physical object;

[0020] The physical objects are screened according to the physical characteristics to obtain the accident main body, and the specific location data of the accident main body and the main body size data of the accident main body are extracted;

[0021] According to the specific location data and the main body size data, the accident scene is circled to obtain the initial accident range;

[0022] Extract the entity type of the accident entity, obtain the accident extension range according to the entity type, and obtain the accident scene range according to the initial accident range and the accident extension range;

[0023] Expand the detection around the scene photo as the base point to obtain the accident expansion photo, and detect the environmental information of the accident expansion photo to obtain the surrounding environmental information.

[0024] Preferably, the high-altitude drone releases the evidence-taking drone. The steps of the evidence-taking drone obtaining accident information according to the accident scene range and generating the accident cause according to the accident information are specifically as follows:

[0025] The high-altitude drone releases the evidence-taking drone, and the evidence-taking drone scans the accident scene range to obtain the initial accident distribution;

[0026] According to the surrounding environmental information, obtain the terrain data of the accident scene range, and generate an accident evidence-taking route according to the terrain data and the initial accident distribution;

[0027] Take evidence of the accident scene according to the accident evidence-taking route to obtain evidence-taking photos and evidence-taking videos, and obtain accident characteristics and accident traces according to the evidence-taking photos and the evidence-taking videos;

[0028] Generate accident information according to the accident traces and the accident characteristics, and restore and generate the accident cause according to the accident information.

[0029] Preferably, the steps of generating an accident evidence-taking route according to the terrain data and the initial accident distribution are specifically as follows:

[0030] According to the initial accident distribution, determine the collision direction and vehicle out-of-control state at the time of the accident, and obtain the scattering direction of the accident debris according to the collision direction and the vehicle out-of-control state;

[0031] Scan the terrain data to obtain the shooting blind area of the drone within the accident scene range, and extract the blind area position of the shooting blind area;

[0032] Overlap and select the scattering direction and the blind area position to determine the target blind area that needs to be key-processed;

[0033] Extract the visual angle of the target blind area, and generate a blind area shooting point position of the drone according to the visual angle;

[0034] According to the initial accident distribution, extract the road surface state data and brake mark data within the accident scene range;

[0035] Determine the initial evidence-taking point of the evidence-taking UAV based on the road surface state data and the braking mark data, and generate an accident evidence-taking route by combining the initial evidence-taking point, the scattering direction, and the blind area shooting point.

[0036] Preferably, after the step of generating accident information according to the accident trace and the accident characteristics, it further includes:

[0037] The evidence-taking UAV is signal-connected to the accident vehicle, and the driving record data in the accident vehicle is extracted;

[0038] Divide the driving record data into time periods to obtain data before the accident, data during the accident, and data after the accident respectively;

[0039] Extract the video images and sound data in the data before the accident and the data during the accident to generate accident corroboration for the traffic accident;

[0040] Evaluate the data after the accident to obtain a value score, and determine whether the value score is greater than a preset score threshold;

[0041] If it is determined that the value score is greater than the score threshold, obtain the subsequent impact of the accident according to the data after the accident, and add the accident corroboration and the subsequent impact of the accident to the accident information.

[0042] Preferably, the step of constructing an accident liability determination model and inputting the accident cause into the accident liability determination model to generate an accident liability certificate is specifically:

[0043] Obtain historical traffic accident handling data and traffic laws and regulations, classify the historical traffic accident handling data to obtain standard accident handling data and flexible accident handling data;

[0044] Generate a traffic accident handling template according to the standard accident handling data;

[0045] Generate special factors that need to be flexibly changed during the traffic accident handling process according to the flexible accident handling data;

[0046] Construct an accident liability determination model by combining the traffic accident handling template, the traffic laws and regulations, and the special factors;

[0047] Input the accident cause into the accident liability determination model, and extract the target special factors existing in this traffic accident;

[0048] The accident liability determination model generates an initial liability certificate according to the accident cause, and then modifies the initial liability certificate according to the target special factors to generate an accident liability certificate.

[0049] Preferably, based on the accident safety hazards, the accident hazard type is obtained, a support signal is generated according to the accident hazard type, and the steps of invoking support forces to handle the accident safety hazards according to the support signal are specifically as follows:

[0050] Based on the accident safety hazards, the hazard manifestations of the accident safety hazards are extracted, and the accident hazard type and the accident hazard degree value are obtained according to the hazard manifestations;

[0051] A current accident risk value and a secondary accident risk value are generated according to the accident hazard type and the accident hazard degree value;

[0052] An emergency degree value for accident handling is generated according to the current accident risk value and the secondary accident risk value, and a support signal is generated according to the emergency degree value;

[0053] The safety drone selects the corresponding support force according to the accident hazard type, sends the support signal to the support force, and invokes the support force to handle the accident safety hazards.

[0054] Preferably, based on the road safety hazards, the road traffic information is collected, and the steps of controlling the road traffic according to the road traffic information are specifically as follows:

[0055] Based on the road safety hazards, the road traffic information is collected, and the congestion state of the road is obtained according to the road traffic information;

[0056] Combining the road safety hazards and the congestion state, an emergency degree value for traffic recovery is generated;

[0057] According to the accident scene range and the surrounding environment information, a temporary traffic lane is constructed, and it is judged whether the temporary traffic lane is within the accident scene range;

[0058] If it is judged that the temporary traffic lane is within the accident scene range, wait for the evidence-taking drone to complete the evidence-taking work and then release it;

[0059] If it is judged that the temporary traffic lane is not within the accident scene range, directly release it;

[0060] The position and shape data of the temporary traffic lane are obtained, and the safety drone projects the temporary traffic lane on the ground according to the position and shape data;

[0061] According to the congestion state, a passing / stopping instruction is generated and projected in front of the temporary traffic lane, and subsequent vehicles are guided to pass orderly according to the passing / stopping instruction.

[0062] In summary, the present application includes at least one of the following beneficial technical effects:

[0063] By cooperating with the vehicle and the mobile phone, the accident information and location information are sent to the high-altitude unmanned aerial vehicle (UAV). After receiving the accident information and location information, the high-altitude UAV goes to the accident scene, conducts high-altitude detection on the accident scene, and temporarily intercepts the traffic of the subsequent vehicles at the accident scene. The high-altitude UAV releases the evidence-taking UAV and the safety UAV respectively. The evidence-taking UAV takes evidence within the accident scene range, obtains the accident information and restores it to generate the accident cause, and inputs the accident cause into the pre-constructed accident liability judgment model to generate the accident liability recognition letter. The safety UAV identifies and analyzes the accident scene and the surrounding environment information to obtain the accident safety hazards and road safety hazards. For the accident safety hazards, the safety UAV sends support information to call for support forces to handle the safety problems existing in this accident. For the road safety hazards, it collects the road traffic information, generates a temporary traffic passage and a traffic flow / stop indication, and guides the subsequent vehicles to detour or pass through the accident scene in an orderly manner. This improves the speed, accuracy of traffic accident handling and rescue, and the efficiency of road traffic. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 is a flowchart of the steps of a method for on-site investigation and handling of traffic accidents based on UAVs provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0065] The following is a further detailed description of the present application in conjunction with Figure 1 This is not limited to the embodiments of the present invention.

[0066] An embodiment of the present application discloses a method for on-site investigation and handling of traffic accidents based on UAVs.

[0067] In this embodiment, a method for on-site investigation and handling of traffic accidents based on UAVs includes:

[0068] S100: After a traffic accident occurs, quickly locate the accident location based on the mobile phones of the accident personnel and the accident vehicle, and guide the high-altitude UAV to quickly go to the accident scene according to the accident location;

[0069] S200: Based on the accident scene, call the high-altitude UAV for high-altitude detection to obtain the accident scene range and the surrounding environment information, and temporarily intercept the subsequent traffic;

[0070] S300: The high-altitude UAV releases the evidence-taking UAV, and the evidence-taking UAV takes evidence according to the accident scene range to obtain the accident information, and restores the accident according to the accident information to generate the accident cause;

[0071] S400: Build an accident liability judgment model, input the accident cause into the accident liability judgment model, and generate an accident liability recognition form.

[0072] S500: Release a safety drone by a high-altitude drone. The safety drone obtains accident safety hazards and road safety hazards respectively according to the accident scene range and the surrounding environment information.

[0073] S600: Based on the accident safety hazards, obtain the accident hazard types, generate support signals according to the accident hazard types, and call support forces to handle the accident safety hazards according to the support signals.

[0074] S700: Based on the road safety hazards, collect road traffic information and control the road traffic according to the road traffic information.

[0075] It should be noted that the above modules are only the basic steps of this embodiment. In the specific implementation process, on the premise of not affecting the overall implementation effect, some steps can be appropriately added, reduced or modified.

[0076] The steps of quickly positioning the accident location based on the mobile phones of the accident personnel and the accident vehicle and guiding the high-altitude drone to quickly go to the accident scene are specifically as follows:

[0077] The accident vehicle collects the current location information when the accident occurs, and at the same time scans the surrounding of the vehicle to obtain the target mobile phones of the accident personnel.

[0078] The accident vehicle generates an alarm prompt according to the current location information and sends it to the target mobile phones.

[0079] After receiving the alarm prompt, the target mobile phone obtains the mobile phone location information, verifies the mobile phone location information with the current location information, and obtains the accident location.

[0080] The target mobile phone alarms according to the accident location, sends the accident location to the nearest high-altitude drone, and the high-altitude drone quickly goes to the accident scene according to the accident location.

[0081] During operation, taking the rear-end collision accident on the ring expressway of a certain city as an example, vehicle A (SUV) obtains the current location information (latitude 31.23 degrees north, longitude 121.47 degrees east) through in-vehicle GPS at the moment of collision. At the same time, it discovers driver's mobile phone B (model HUAWEI Mate 50) through Bluetooth scanning. Vehicle A generates an alarm prompt containing the accurate coordinates and sends it to mobile phone B. After receiving it, mobile phone B starts dual-location verification: the location obtained through the built-in GPS is latitude 31.2301 degrees north, longitude 121.4698 degrees east, with an error of only 15 meters from the vehicle coordinates. The system automatically fuses the two coordinates to generate the accident center point (latitude 31.23005 degrees north, longitude 121.4699 degrees east). Mobile phone B immediately triggers an emergency alarm, and the nearest high-altitude unmanned aerial vehicle (number UAV-057) is dispatched to fly to the scene at a speed of 120 km / h. The flight path is planned to avoid high-rise building clusters, and it is expected to arrive at the accident point in 6 minutes and 30 seconds.

[0082] Based on the accident scene, the steps of using a high-altitude unmanned aerial vehicle for high-altitude detection to obtain the accident scene range and surrounding environment information are as follows:

[0083] The high-altitude unmanned aerial vehicle conducts high-altitude detection on the accident scene, obtains the on-site photos of the accident scene, extracts the physical objects existing in the on-site photos, and extracts the physical characteristics of each physical object;

[0084] Screen the physical objects according to the physical characteristics to obtain the accident main body, and extract the specific position data and the main body size data of the accident main body;

[0085] Circumscribe the accident scene according to the specific position data and the main body size data to obtain the initial accident range;

[0086] Extract the main body type of the accident main body, obtain the accident extension range according to the main body type, and obtain the accident scene range according to the initial accident range and the accident extension range;

[0087] Taking the on-site photo as the base point, expand the detection around to obtain the accident expansion photo, and conduct environmental information detection on the accident expansion photo to obtain the surrounding environment information.

[0088] During operation, taking the rear-end collision accident on the ring expressway of a certain city as an example, after UAV-057 arrives, it starts the multi-spectral scanner (resolution 0.5 m) and takes panoramic photos to identify the main bodies of the accident: the front of the SUV is severely deformed (length 2.3 m × width 1.8 m), the rear of the sedan is dented (area 0.8 ㎡), and bumper fragments are scattered within a range of 20 m (maximum size 45 cm). The system determines the position of the SUV (coordinate point P1) and the position of the sedan (coordinate point P2) through feature matching, and expands 5 m outward based on the line connecting the two points to form an initial range (area approximately 300 ㎡). Combining the vehicle type (the self-weight of the SUV is 2.1 tons), it calculates the kinetic energy extension range, and finally demarcates the accident site as an elliptical area (major axis 35 m, minor axis 22 m, area 605 ㎡). At the same time, 3 surrounding surveillance cameras (distances are 50 m, 80 m, and 120 m respectively) are detected and included in the environmental information database.

[0089] The high-altitude UAV releases the evidence-taking UAV. The evidence-taking UAV obtains accident information through evidence-taking according to the accident site range, and the steps of generating the accident cause according to the accident information are as follows:

[0090] The high-altitude UAV releases the evidence-taking UAV. The evidence-taking UAV scans the accident site range to obtain the initial accident distribution;

[0091] According to the surrounding environmental information, obtain the terrain data of the accident site range, and generate an accident evidence-taking route based on the terrain data and the initial accident distribution;

[0092] Take evidence of the accident site according to the accident evidence-taking route to obtain evidence-taking photos and evidence-taking videos, and obtain accident characteristics and accident traces based on the evidence-taking photos and the evidence-taking videos;

[0093] Generate accident information based on the accident traces and accident characteristics, and restore and generate the accident cause according to the accident information.

[0094] During operation, taking the rear-end collision accident on the ring expressway of a certain city as an example, the evidence-taking UAV (serial number FJ-12) starts three-dimensional laser scanning, establishes a point cloud model of the accident site (accuracy ±2 cm), and identifies the distribution of the main scattered objects: the windshield fragments are scattered in a fan shape (angle 120°, maximum radius 8 m), and the oil stain penetration area (area 2.4 ㎡). Combining the terrain data to identify the shooting blind area: the drainage ditch area on the right side of the accident vehicle (length 3 m × depth 0.6 m). The system generates a spiral progressive evidence-taking route: starting from the starting point of the braking mark (coordinate Q1), setting evidence-taking points every 2 m along the scattering direction (a total of 18), and especially setting 3 hovering points above the drainage ditch (height 1.5 m, shooting at an inclination of 45°). The total shooting duration is 23 minutes, obtaining 487 high-definition photos and 35 minutes of video, and finding that there are abnormal wear marks on the left rear wheel of the sedan (groove depth 2.3 mm, exceeding the standard value of 1.6 mm).

[0095] The steps for generating an accident evidence collection route based on terrain data and the initial accident distribution are specifically as follows:

[0096] Based on the initial accident distribution, determine the collision direction and vehicle out-of-control state at the time of the accident. According to the collision direction and vehicle out-of-control state, obtain the scattering direction of accident debris;

[0097] Scan the terrain data to obtain the shooting blind spots of the unmanned aerial vehicle (UAV) within the accident scene range, and extract the blind spot positions of the shooting blind spots;

[0098] Overlap and select the scattering direction and the blind spot positions to determine the target blind spots that need to be key processed;

[0099] Extract the visual angles of the target blind spots, and generate blind spot shooting points for the UAV according to the visual angles;

[0100] Based on the initial accident distribution, extract the road surface state data and brake mark data within the accident scene range;

[0101] Determine the initial evidence collection points of the evidence collection UAV according to the road surface state data and brake mark data. Combine the initial evidence collection points, the scattering direction, and the blind spot shooting points to generate an accident evidence collection route.

[0102] In operation, taking a rear-end collision accident on the ring expressway of a certain city as an example, systematic analysis of the scattered object distribution found that 30% of the debris fell into the drainage ditch blind spot. The evidence collection UAV started adaptive path adjustment: deployed a micro ground penetrating radar (frequency 800 MHz) at a height of 1.2 m above the ditch, scanned and found 13 metal debris pieces (maximum size 18 cm) in the ditch. At the same time, adjusted the pan-tilt angle to a 60° depression angle, and carried out macro shooting (magnification 20x) on the concealed impact point (the VIN code area of the car chassis), and successfully obtained the vehicle identification code. The road surface state analysis showed that there was an 8-m long rubber drag mark (width 15 cm, depth 0.3 mm) in the second lane. The system corrected the brake mark model accordingly, and finally generated an optimized route including 32 key evidence collection points.

[0103] After the steps of generating accident information based on accident traces and accident characteristics, it further includes:

[0104] The evidence collection UAV is signal-connected to the accident vehicle, and extracts the driving record data in the accident vehicle;

[0105] Divide the driving record data by time periods to obtain the data before the accident, the data during the accident, and the data after the accident respectively;

[0106] Extract the video and audio data in the data before the accident and the data during the accident to generate accident corroboration for the traffic accident;

[0107] Evaluate the data after the accident to obtain a value score, and determine whether the value score is greater than a preset score threshold;

[0108] If it is determined that the value score is greater than the score threshold, obtain the subsequent impact of the accident based on the data after the accident, and add the accident corroboration and the subsequent impact of the accident to the accident information.

[0109] During operation, taking the rear-end collision accident on the ring expressway of a certain city as an example, FJ-12 connects to the accident vehicle A through the V2X protocol and extracts the data of the driving recorder (timestamp 14:23:17 - 14:25:49). The system divides the time axis: 42 seconds before the accident (the vehicle speed drops suddenly from 82 km / h to 32 km / h), 3 seconds at the moment of collision (the lateral acceleration reaches 1.2g), and 34 seconds after the collision (the record of the airbag deployment). Extract the key frame images (select the 683rd and 687th frames out of 25 frames per second) to show the sudden lane-changing trajectory of the car. The system evaluates the value of the later data: a secondary scraping sound is recorded 18 seconds after the collision (the voiceprint feature matches the guardrail material), and after being judged by the scoring system (87 points > the threshold of 75 points), this section of data is marked as key evidence and incorporated into the accident information.

[0110] Construct an accident liability determination model, and input the accident cause into the accident liability determination model to generate the steps of the accident liability determination letter, specifically:

[0111] Obtain the historical traffic accident handling data and traffic laws and regulations, classify the historical traffic accident handling data to obtain standard accident handling data and flexible accident handling data;

[0112] Generate a traffic accident handling template based on the standard accident handling data;

[0113] Generate special factors that need to be flexibly changed during the traffic accident handling process based on the flexible accident handling data;

[0114] Construct an accident liability determination model by combining the traffic accident handling template, traffic laws and regulations, and special factors;

[0115] Input the accident cause into the accident liability determination model to extract the target special factors existing in this traffic accident;

[0116] The accident liability determination model generates an initial liability determination letter based on the accident cause, and then modifies the initial liability determination letter according to the target special factors to generate the accident liability determination letter.

[0117] In operation, taking a rear-end collision accident on the ring expressway of a certain city as an example, the system calls the data of 1,367 similar accidents in the past three years, including 892 standard cases (full responsibility for rear-end collisions) and 475 special cases (such as emergency avoidance). At the same time, the special factors in the 475 special cases are extracted as references. The model identifies the special factors of this accident: when the car changed lanes, the distance from the vehicle in front was only 1.2 seconds (lower than the safety standard of 2.5 seconds), but the SUV had an overloading record (the approved load was 5 people, but actually carried 7 people). According to Article 43 of the Road Traffic Safety Law of the People's Republic of China, it is initially judged that the car is mainly responsible (70%). The model detects the mitigating factor of overloading and finally revises the liability division: the car is 60% and the SUV is 40%, generating a liability determination letter containing 23 pieces of evidence citation.

[0118] Based on the accident safety hazards, obtain the types of accident hazards, generate support signals according to the types of accident hazards, and call support forces to handle the accident safety hazards according to the support signals. The specific steps are as follows:

[0119] Based on the accident safety hazards, extract the manifestations of the accident safety hazards, and obtain the types of accident hazards and the accident hazard degree values according to the manifestations;

[0120] Generate the current accident risk value and the secondary accident risk value according to the types of accident hazards and the accident hazard degree values;

[0121] Generate the emergency degree value for accident handling according to the current accident risk value and the secondary accident risk value, and generate support signals according to the emergency degree value;

[0122] The safety drone selects the corresponding support forces according to the types of accident hazards, sends the support signals to the support forces, and calls the support forces to handle the accident safety hazards.

[0123] In operation, taking a rear-end collision accident on the ring expressway of a certain city as an example, the safety drone (number AQ-09) detected that the SUV's fuel was leaking (flow rate 30 ml / min). According to the GB / T 28921-2012 standard, the current risk value was evaluated to reach level R3 (may cause a fire), and the secondary risk value was level R2 (there is a gas station within 30 meters). The system generated a red support signal (code 1107) and automatically dispatched an explosion-proof fire truck (arriving within 5 minutes) and a first aid unit. At the same time, it was found that the car door was deformed (opening gap < 30 cm), and a jacking rescue was implemented through the robotic arm (applying a force value of 5,000 N) to create a safety passage (width 55 cm). The entire process synchronously sent the disposal progress to the command center and updated the risk heat map every 30 seconds.

[0124] Based on the road safety hazards, collect road traffic information, and the steps for controlling road traffic according to the road traffic information are as follows:

[0125] Collect road traffic information based on road safety hazards, and obtain the congestion status of the road according to the road traffic information;

[0126] Combine road safety hazards and congestion status to generate an emergency level value for traffic recovery;

[0127] Construct a temporary passing lane based on the accident scene range and surrounding environment information, and determine whether the temporary passing lane is within the accident scene range;

[0128] If it is determined that the temporary passing lane is within the accident scene range, wait for the evidence-taking drone to complete the evidence-taking work and then release it;

[0129] If it is determined that the temporary passing lane is not within the accident scene range, release it directly;

[0130] Obtain the position and shape data of the temporary passing lane, and the safety drone projects the temporary passing lane on the ground according to the position and shape data;

[0131] Generate a passing / stopping instruction according to the congestion status, project it in front of the temporary passing lane, and guide the subsequent vehicles to pass orderly according to the passing / stopping instruction.

[0132] In operation, taking a rear-end collision accident on the ring expressway of a certain city as an example, the system analyzes that there is congestion (vehicle speed < 15 km / h) 4 kilometers away and generates an orange diversion instruction. A temporary lane (width 3.5 m, length 200 m) is opened on the north side of the accident scene, and the passing area is delimited by laser projection (brightness 3000 lumens per square meter). The dynamic indicator lights are adjusted according to the traffic flow density: alternating traffic is implemented during peak hours (17:00 - 18:30) (changing directions every 15 seconds), and stroboscopic warning is enabled at night (frequency 2 Hz). 127 vehicles are guided to detour safely through V2I communication, and the average delay time is reduced to 3 minutes and 15 seconds. The overall traffic efficiency during the accident handling period remains 68% (42% higher than that of conventional accidents).

[0133] The above are all preferred embodiments of this application, and the protection scope of this application is not limited accordingly. Therefore, all equivalent changes made according to the structure, shape, and principle of this application should be covered within the protection scope of this application.

Claims

1. A method for investigating and handling traffic accident scenes based on drones, characterized in that, It includes the following steps: After a traffic accident occurs, quickly locate the accident location based on the mobile phones of the accident personnel and the accident vehicle, and guide the high-altitude unmanned aerial vehicle to quickly go to the accident scene according to the accident location; The step of quickly locating the accident location based on the mobile phones of the accident personnel and the accident vehicle, and guiding the high-altitude unmanned aerial vehicle to quickly go to the accident scene is specifically as follows: When the accident vehicle has an accident, it collects the current location information and scans the surrounding area of the vehicle to obtain the target mobile phones of the accident personnel; The accident vehicle generates an alarm prompt according to the current location information and sends it to the target mobile phone; After receiving the alarm prompt, the target mobile phone obtains the mobile phone location information, verifies the mobile phone location information with the current location information, and obtains the accident location; The target mobile phone alarms according to the accident location, sends the accident location to the nearest high-altitude unmanned aerial vehicle, and the high-altitude unmanned aerial vehicle quickly goes to the accident scene according to the accident location; Based on the accident scene, call the high-altitude unmanned aerial vehicle for high-altitude detection to obtain the accident scene range and surrounding environment information, and temporarily intercept the subsequent traffic; The high-altitude unmanned aerial vehicle releases a forensics unmanned aerial vehicle. The forensics unmanned aerial vehicle obtains accident information by forensics according to the accident scene range, and restores the accident according to the accident information to generate the accident cause; The step of the high-altitude unmanned aerial vehicle releasing a forensics unmanned aerial vehicle, the forensics unmanned aerial vehicle obtaining accident information by forensics according to the accident scene range, and restoring the accident according to the accident information to generate the accident cause is specifically as follows: The high-altitude unmanned aerial vehicle releases a forensics unmanned aerial vehicle, and the forensics unmanned aerial vehicle scans the accident scene range to obtain the initial accident distribution; According to the surrounding environment information, obtain the terrain data of the accident scene range, and generate an accident forensics route according to the terrain data and the initial accident distribution; Conduct forensics on the accident scene according to the accident forensics route to obtain forensics photos and forensics videos, and obtain accident characteristics and accident traces according to the forensics photos and the forensics videos; Generate accident information according to the accident traces and the accident characteristics, and restore the accident cause according to the accident information; The step of generating an accident forensics route according to the terrain data and the initial accident distribution is specifically as follows: According to the initial accident distribution, determine the collision direction and vehicle out-of-control state when the accident occurs, and obtain the scattering direction of the accident debris according to the collision direction and the vehicle out-of-control state; Scan the terrain data to obtain the shooting blind area of the unmanned aerial vehicle within the accident scene range, and extract the blind area position of the shooting blind area; Overlap and select the scattering direction and the blind area position to determine the target blind area that needs to be key processed; Extract the visual angle of the target blind area, and generate a blind area shooting point position of the unmanned aerial vehicle according to the visual angle; According to the initial accident distribution, extract the road surface state data and braking trace data within the accident scene range; Determine the initial evidence-taking position of the evidence-taking drone based on the road surface state data and the braking mark data, and generate an accident evidence-taking route by combining the initial evidence-taking position, the scattering direction, and the blind area shooting position; Construct an accident liability determination model, input the accident cause into the accident liability determination model, and generate an accident liability recognition form; The steps of constructing an accident liability determination model, inputting the accident cause into the accident liability determination model, and generating an accident liability recognition form are specifically as follows: Obtain historical traffic accident handling data and traffic laws and regulations, classify the historical traffic accident handling data, and obtain standard accident handling data and flexible accident handling data; Generate a traffic accident handling template according to the standard accident handling data; Generate special factors that need to be flexibly changed during the traffic accident handling process according to the flexible accident handling data; Construct an accident liability determination model by combining the traffic accident handling template, the traffic laws and regulations, and the special factors; Input the accident cause into the accident liability determination model, and extract the target special factors existing in the current traffic accident; The accident liability determination model generates an initial liability recognition form according to the accident cause, and then corrects the initial liability recognition form according to the target special factors to generate an accident liability recognition form; The high-altitude drone releases a safety drone, and the safety drone obtains accident safety hazards and road safety hazards according to the accident scene range and the surrounding environment information; Based on the accident safety hazards, obtain the accident hazard type, generate a support signal according to the accident hazard type, and call support forces to handle the accident safety hazards according to the support signal; Based on the road safety hazards, collect road traffic information and control the road traffic according to the road traffic information.

2. The method for investigating and processing a traffic accident scene based on a drone according to claim 1, characterized in that, The steps of calling the high-altitude drone for high-altitude detection based on the accident scene to obtain the accident scene range and the surrounding environment information are specifically as follows: The high-altitude drone conducts high-altitude detection on the accident scene, obtains on-site photos of the accident scene, extracts the physical objects existing in the on-site photos, and extracts the physical characteristics of each physical object; Screen the physical objects according to the physical characteristics to obtain the accident subjects, and extract the specific position data of the accident subjects and the subject size data of the accident subjects; Circle the accident scene according to the specific position data and the subject size data to obtain the initial accident range; Extract the subject type of the accident subject, obtain the accident extension range according to the subject type, and obtain the accident scene range according to the initial accident range and the accident extension range; Expand the detection around the on-site photo as the base point to obtain an accident extended photo, and conduct environmental information detection on the accident extended photo to obtain the surrounding environment information.

3. The method for investigating and processing a traffic accident scene based on a drone according to claim 2, wherein, After the step of generating accident information according to the accident trace and the accident characteristics, it further includes: The evidence-taking drone is signal-connected to the accident vehicle, and the driving record data in the accident vehicle is extracted; Divide the driving record data by time period to obtain data before the accident, data during the accident, and data after the accident respectively; Extract the video images and audio data from the data before the accident and the data during the accident to generate evidence for the traffic accident; Evaluate the data after the accident to obtain a value score, and determine whether the value score is greater than a preset score threshold; If it is determined that the value score is greater than the score threshold, obtain the subsequent impact of the accident based on the data after the accident, and add the accident evidence and the subsequent impact of the accident to the accident information.

4. The method for investigating and processing a traffic accident scene based on a drone according to claim 3, wherein The steps of obtaining the accident hazard type based on the accident safety hazard, generating a support signal according to the accident hazard type, and calling support forces to handle the accident safety hazard according to the support signal are specifically as follows: Based on the accident safety hazard, extract the hazard manifestations of the accident safety hazard, and obtain the accident hazard type and the accident hazard degree value according to the hazard manifestations; Generate the current accident risk value and the secondary accident risk value according to the accident hazard type and the accident hazard degree value; Generate the emergency degree value for accident handling according to the current accident risk value and the secondary accident risk value, and generate a support signal according to the emergency degree value; The safety drone selects the corresponding support force according to the accident hazard type, sends the support signal to the support force, and calls the support force to handle the accident safety hazard.

5. A method for investigating and processing a traffic accident scene based on a drone according to claim 4, characterized in that, The steps of collecting road traffic information based on the road safety hazard and controlling road traffic according to the road traffic information are specifically as follows: Based on the road safety hazard, collect road traffic information, and obtain the congestion status of the road according to the road traffic information; Combine the road safety hazard and the congestion status to generate the emergency degree value for traffic recovery; Construct a temporary traffic lane according to the accident scene range and the surrounding environment information, and determine whether the temporary traffic lane is within the accident scene range; If it is determined that the temporary traffic lane is within the accident scene range, wait for the evidence-taking drone to complete the evidence-taking work and then release it; If it is determined that the temporary traffic lane is not within the accident scene range, release it directly; Obtain the position and shape data of the temporary traffic lane, and the safety drone projects the temporary traffic lane on the ground according to the position and shape data; Generate a passing / stopping instruction according to the congestion status, project it in front of the temporary traffic lane, and guide the subsequent vehicles to pass orderly according to the passing / stopping instruction.

Images