Road traffic accident multi-source data fusion and standardized scene reconstruction method

Abstract

The invention relates to a road traffic accident multi-source data fusion and standardized scene reconstruction method, which comprises the following steps: collecting traffic accident multi-source data, and constructing an accident reconstruction data table; wherein the accident reconstruction data table comprises static data and dynamic data; constructing a static scene graph based on the static data; performing multi-source dynamic data fusion based on the static scene graph and the dynamic data; constructing a standardized accident scene based on the static scene graph and the fused multi-source dynamic data; and constructing a standardized dangerous scene based on the standardized accident scene and the fused multi-source dynamic data. According to the brand new road traffic accident multi-source data fusion and standardized scene reconstruction method provided by the invention, a perfect accident reconstruction data table is established, a data standard format is unified, multi-platform accident scene simulation is better adapted, and an accident scene is accurately restored.

Description

Technical Field

[0001] The present invention relates to the technical field of multi-source data fusion and scene reconstruction, and in particular to a method for multi-source data fusion and standardized scene reconstruction of road traffic accidents. Background Art

[0002] In recent years, with the rapid development of my country's economy and society, factors such as people, vehicles, and roads have increased rapidly, and road traffic accidents are prone to occur frequently. The determination of accident responsibility relies on the early stage of on-site investigation, evidence fixation, questioning and other links. Therefore, it is particularly important to comprehensively collect data related to road traffic accidents and accurately restore and reconstruct accident scenes for accident investigation and disposal.

[0003] In recent years, both at home and abroad, corresponding accident scenario libraries have been established to collect road traffic accident information and conduct in-depth investigation and analysis of traffic accidents. Common ones include road traffic accident databases, in-depth traffic accident research projects, fatal accident reporting systems, motor vehicle accident sampling systems, etc. In order to improve the universality of accident scenario definitions and achieve interoperability and compatible use of scenarios between different simulation tools, it is crucial to establish a digital accident scenario library based on a standard format. As the current international common format, the OpenX series of standards provide a reference for the standardized construction of road traffic accident scenario libraries.

[0004] After research and analysis, there are still some problems in the existing technology. First, data collection is not comprehensive. The occurrence of road traffic accidents is caused by a combination of multiple factors, and it is necessary to collect accident data comprehensively and objectively. The existing technology is not comprehensive and detailed in data collection, and it is not professional enough, resulting in many detailed data items that may change over time being omitted, resulting in a large difference between the subsequent analysis results and the actual results, and it is impossible to accurately reconstruct the accident scene. Second, the data format is not unified. On the basis of comprehensive data collection, the quality of the data is also crucial. Many existing technologies in China cannot unify the data format of traffic accidents, resulting in unclear data levels and non-standard names for model input, many duplicate items, and cannot be applied to general simulation platforms. The transplantability is poor and data co-construction and sharing cannot be achieved. Third, the accident scene is not perfect. There are some common reasons for the occurrence of accidents, and there are similarities in the accident forms, such as rear-end collisions, head-on collisions, collisions with pedestrians, etc., and the existing technology has established fewer scene model libraries for these typical and simpler accidents, resulting in a large workload for accident analysis and accident reconstruction, and is not conducive to overall analysis of typical accidents, and cannot prevent and reduce accidents well. In summary, although road traffic accident data fusion and process reconstruction technology have made certain progress, there is still much room for improvement, and a set of scientific and effective solutions needs to be proposed. Summary of the invention

[0005] The purpose of the present invention is to propose a method for multi-source data fusion and standardized scene reconstruction of road traffic accidents to solve the problems existing in the above-mentioned prior art. First, an accident reconstruction data table is established to provide directional guidance for accident investigation. Second, a digital accident scene library based on a standard format is established to accurately restore the accident process; increase the portability of accident scenes, and realize multi-platform and multi-angle accident analysis, which is helpful to prevent and reduce road traffic accidents. Third, dangerous scenes are expanded under accident scenes, retaining all static information, initial parameters of the parties involved in the accident, and dynamic information of other parties involved, to provide standardized dangerous scenes for research on driving behavior, vehicle design, and autonomous driving algorithms.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] A method for multi-source data fusion and standardized scene reconstruction of road traffic accidents, comprising:

[0008] Collect multi-source data of traffic accidents and construct an accident reconstruction data table; wherein the accident reconstruction data table includes: static data and dynamic data;

[0009] Based on the static data, construct a static scene graph;

[0010] Based on the static scene graph and the dynamic data, multi-source dynamic data fusion is performed;

[0011] Based on the static scene graph and the fused multi-source dynamic data, a standardized accident scene is constructed;

[0012] Based on the standardized accident scenario and the fused multi-source dynamic data, a standardized danger scenario is constructed.

[0013] Optionally, the multi-source data of traffic accidents include: on-site investigation records, interrogation records, accident scene maps, on-site images, appraisal opinions, investigative experimental data, electronic data, road design information, automobile announcements, and motor vehicle registration information.

[0014] Optionally, constructing the accident reconstruction data table includes: a road layer, a facility layer, an environment layer, and a traffic operation layer;

[0015] The road layer includes: intersection section type, road alignment, horizontal and vertical alignment, number of lanes, road width, and road surface morphology;

[0016] The facility layer includes: traffic signs, traffic markings, guardrails and railings, and obstacles;

[0017] The environmental layer includes: lighting conditions, weather conditions;

[0018] The traffic operation layer includes: basic data of participants, dynamic parameters, and collision factors;

[0019] The static data includes the road layer, facility layer, and environment layer, and the dynamic data includes the traffic operation layer.

[0020] Optionally, constructing the static scene graph includes:

[0021] The static scene graph is drawn based on the road layer, the facility layer and the environment layer. The static scene graph also includes: a vehicle starting position, a collision point position and a stop position after the collision.

[0022] Optionally, performing multi-source dynamic data fusion based on the static scene graph and the dynamic data includes:

[0023] Import the static scene graph into PC-Crash and preset the origin position;

[0024] Add accident participants and set parameters based on the basic data of the participants;

[0025] According to the collision point position and the post-collision stop position, set and optimize the motion sequence of each participant until the post-collision stop position is within the error range, complete the post-collision simulation, and lock the post-collision trajectory;

[0026] According to the post-collision simulation results and the vehicle's starting position, the participant's motion sequence is adjusted until the participant's trajectory is within the error range, completing the pre-collision simulation;

[0027] According to the above steps, the accident collision process is restored to complete the fusion of multi-source dynamic data; and the fused multi-source dynamic data is verified.

[0028] Optionally, constructing a standardized accident scenario based on the static scene graph and the fused multi-source dynamic data includes:

[0029] Performing format conversion on the static scene graph and the fused multi-source dynamic data respectively;

[0030] After visual verification of the format-converted static scene graph and multi-source dynamic data, a standardized accident scenario is generated.

[0031] Optionally, constructing a standardized danger scenario based on the standardized accident scenario and the fused multi-source dynamic data includes:

[0032] Performing avoidance analysis on the fused multi-source dynamic data to obtain accident avoidability parameters;

[0033] Based on the accident avoidability parameters, using preset judgment conditions, judging whether the accident is avoidable;

[0034] For accidents that are avoidable, the initial parameters of the accident scene of the parties involved in the accident, all information about other parties involved, and the road traffic environment information are retained, and a standardized danger scene is constructed based on the method for constructing a standardized accident scene.

[0035] Optionally, the judgment condition is:

[0036] The prevention distance is less than the visibility of the road section where the accident occurred, and there is no line of sight obstruction;

[0037] The prevention reaction time is greater than the theoretical braking reaction time when humans are concentrating;

[0038] The preventive deceleration is less than the maximum deceleration; otherwise, the accident is determined to be unavoidable under the initial conditions.

[0039] Optionally, the accident avoidable parameters are: maximum deceleration, maximum allowed speed, prevention speed, prevention distance, prevention reaction time, and prevention deceleration.

[0040] The beneficial effects of the present invention are:

[0041] The present invention proposes a new method for multi-source data fusion and standardized scene reconstruction of road traffic accidents, establishes a complete accident reconstruction data table, unifies the data standard format, better adapts to multi-platform accident scene simulation, and accurately restores the accident scene. In terms of economy, it can reduce the time and manpower cost of accident handling, provide a scientific basis for determining the cause of the accident and the responsibility of the parties, and can also help prevent road traffic accidents and reduce economic losses caused by accidents and disputes involving compensation for damages caused by traffic accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0043] Figure 1 The present invention is a flowchart of a method for multi-source data fusion and standardized scene reconstruction of a road traffic accident according to an embodiment of the present invention. DETAILED DESCRIPTION

[0044] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0046] Related terms introduction:

[0047] 0047 Road traffic accident multi-source data fusion and scene reconstruction technology refers to first collecting multi-source data at the accident scene, including roads, traffic infrastructure, temporary control facilities, traffic participants, on-site environment, digital maps and other information, and then processing, processing and fusing the above data to generate a standardized data format, and then modeling and analyzing the standardized data through a simulation platform to build an accident scene reconstruction model.

[0048] like Figure 1 As shown, this embodiment proposes a method for multi-source data fusion and standardized scene reconstruction of road traffic accidents, including:

[0049] Collect multi-source data of traffic accidents and construct accident reconstruction data tables; the accident reconstruction data tables include: static data and dynamic data;

[0050] Based on static data, a static scene graph is constructed;

[0051] Based on static scene graphs and dynamic data, multi-source data fusion is performed;

[0052] Construct standardized accident scenarios based on static scene graphs and fused multi-source data;

[0053] Construct standardized danger scenarios based on standardized accident scenarios and integrated multi-source data.

[0054] Specifically, in this embodiment, the collected multi-source information on traffic accidents is used to accurately restore the accident process and generate quantitative countermeasures for accident prevention. A five-stage standardized reconstruction and application framework for traffic accident scenarios is proposed, including data integration, static scenario construction, multi-source dynamic data fusion, standardized accident scenario construction, and standardized danger scenario construction.

[0055] Furthermore, the multi-source data of traffic accidents include: on-site investigation records, interrogation records, accident scene maps, on-site images, appraisal opinions, investigative experimental data, electronic data, road design information, automobile announcements, and motor vehicle registration information.

[0056] Furthermore, the accident reconstruction data table is constructed including: road layer, facility layer, environment layer, and traffic operation layer;

[0057] The road layer includes: intersection section type, road alignment, horizontal and vertical alignment, number of lanes, road width, and road surface morphology;

[0058] The facility layer includes: traffic signs, traffic markings, guardrails and railings, and obstacles;

[0059] The environmental layer includes: lighting conditions, weather conditions;

[0060] The traffic operation layer includes: basic data of participants, dynamic parameters, and collision factors;

[0061] Static data includes road layer, facility layer, and environment layer, while dynamic data includes traffic operation layer.

[0062] Specifically, in this embodiment, the main steps of the data integration stage include: (1) Data collection: Collect raw data from the data sources in Table 1. (2) Data processing: The recommended data sources for each data item are given in Table 2, and the collection and processing of data should have evidentiary effect. In addition, 1) For video image data, such as DVR, monitoring, etc., the speed of the participants should be calculated in accordance with the standard of "Technical Appraisal of Vehicle Speed ​​Based on Video Images" (GA / T 1133-2014); 2) For on-board electronic data, such as EDR, DVR, VDR, etc., its extraction should comply with the "Technical Specifications for Extraction of On-board Electronic Data of Automobiles" (GA / T 1998-2022) to ensure data integrity. (3) Establish an accident reconstruction data table: For each accident, establish a data table as shown in Table 2, and fill in the processed data into the table accordingly.

[0063] If the same data item has different values ​​from different sources, (1) for static data such as the road layer, facility layer, and environment layer, the final parameters are determined in accordance with the order of effectiveness of evidence; (2) for dynamic data at the traffic operation layer, multiple values ​​are retained until the next step of fusion.

[0064] Table 1 Available data sources

[0065]

[0066]

[0067] Table 2 Accident reconstruction data table

[0068]

[0069]

[0070]

[0071]

[0072]

[0073]

[0074]

[0075] Furthermore, constructing a static scene graph includes:

[0076] A static scene graph is drawn based on the road layer, facility layer, and environment layer. The static scene graph also includes: the vehicle starting position, the collision point position, and the stop position after the collision.

[0077] Specifically, in this embodiment, a static scene map in DXF or JPEG format is produced based on the accident reconstruction data table to restore the static information of the accident section at a 1:1 ratio. The static scene map is drawn in accordance with 4.7.3 of "Drawing of Road Traffic Accident Scene Map" (GA / T 49-2019), and should include (1) road traffic environment elements. The drawing data refers to the road layer, facility layer, environment layer and other information in the accident reconstruction data table; in addition, it should also include (2) elements of each participant, including the collision point location and the stop position after the collision.

[0078] Furthermore, based on the static scene graph and dynamic data, multi-source data fusion is performed including:

[0079] Import the static scene graph into PC-Crash and preset the origin position;

[0080] Add accident participants and set parameters based on the basic data of the participants;

[0081] According to the collision point position and the post-collision stop position, set and optimize the motion sequence of each participant until the post-collision stop position is within the error range, complete the post-collision simulation, and lock the post-collision trajectory;

[0082] According to the post-collision simulation results and the vehicle's starting position, adjust the participant's motion sequence until the participant's trajectory is within the error range to complete the pre-collision simulation; the pre-collision time must be greater than 5s, and the participant can be ahead of the preset starting position in the figure;

[0083] According to the above steps, the accident collision process is restored to complete the fusion of multi-source dynamic data; and the fused multi-source dynamic data is verified.

[0084] Specifically, in this embodiment, the multi-source data fusion stage uses a trajectory-based reconstruction method to restore the accident process, and uses PC-Crash traffic accident reconstruction software to achieve redundant data fusion of traffic operation layer parameters based on simulation. The main steps include: (1) Importing a static scene graph: Import the static scene graph into PC-Crash, and set the upper left corner of the graph as the origin (0,0). (2) Restore the accident collision process and complete data fusion: Input the redundant multi-source parameters as initialization parameters, take the collision point position and the post-collision stop position as optimization targets, use PC-Crash to optimize the accident collision process, select the parameter set with the smallest error as the final data item value, and complete data fusion. (3) Verification of reconstruction results: Refer to the verification method in 7.2 of the "Technical Guidelines for Reconstruction of Motor Vehicle Road Traffic Accident Processes" (DB51-T1772Y2014). ① If there are vehicle traces as verification indicators, the maximum relative error between the simulated vehicle trajectory and the vehicle trajectory obtained by investigation can be calculated; ② If the recognized collision point and collision impulse direction obtained by other methods (such as speed identification, etc.) are used as verification indicators, the maximum relative error between the simulated accident vehicle collision point position and velocity vector and the recognized vehicle collision point position and velocity vector can be calculated. If the maximum relative error does not exceed 10%, it passes the verification and the traffic operation layer data of the accident reconstruction data table is updated at the same time. The calculation method of the maximum relative error is as follows:

[0085]

[0086] Furthermore, based on the static scene graph and the fused multi-source dynamic data, the standardized accident scenario is constructed including:

[0087] Perform format conversion on the static scene graph and the fused multi-source dynamic data respectively;

[0088] After visual verification of the format-converted static scene graph and multi-source dynamic data, a standardized accident scenario is generated.

[0089] The fused elements are filled into the accident reconstruction data table. A static graph is made according to the element values ​​corresponding to each item in the table, and the initial conditions of the simulation are input. For example, if the "road adhesion coefficient" in the table is 0.7, then select 0.7 for the road adhesion coefficient in pccrash.

[0090] Specifically, in this embodiment, the main steps of the standardized scenario construction phase include: (1) Static data format conversion: Use CAD2Shape software to convert the static scene diagram in CAD format into an XLSX format file, and then use CIDASDataConverter software to convert it into an OpenDRIVE format file. (2) Dynamic data format conversion: Use PCM2XOSC software to convert the traffic operation layer data of the accident reconstruction data table as input into an OpenSCENARIO format file. (3) Visual verification: Import the OpenDRIVE and OpenSCENARIO files into simulation platforms that support the OpenX standard, such as VTD, 51Sim-One, TAD Sim, PreScan, CARLA, or the esmini visualization player for visual verification. The file can run normally and the vehicle's driving trajectory is correct, which means that the verification is passed. The OpenDRIVE and OpenSCENARIO generated in this way constitute a standardized accident scene for a case.

[0091] Furthermore, based on the standardized accident scenarios and the integrated multi-source data, the standardized danger scenarios are constructed, including:

[0092] Conduct avoidance analysis on the fused multi-source data to obtain accident avoidance parameters;

[0093] Based on the accident avoidability parameters, using the preset constraints, determine whether the accident is avoidable;

[0094] For accidents that are avoidable, the initial parameters of the accident scene of the parties at fault, all information of other parties involved, and road traffic environment information are retained, and a standardized danger scene is constructed based on the method of constructing standardized accident scenes.

[0095] Dangerous scenarios are derived from standardized scenarios. For avoidable scenarios, the dynamic information of the accident party except the start time of the accident simulation is removed (i.e., the initial position, direction, speed, etc. of the party at the 0th second of the simulation is retained), which is a dangerous scenario.

[0096] Specifically, in this embodiment, the main steps of the dangerous scenario construction stage include: (1) Accident avoidability analysis: Use the avoidance analysis tool in PC-Crash to obtain the accident avoidability parameters (see Table 3) to determine whether the accident is avoidable. The accident avoidability should meet one of the following conditions: 1) The prevention distance is less than the visibility of the accident section, and there is no line of sight obstruction; 2) The prevention reaction time is greater than the theoretical braking reaction time when human attention is concentrated (generally 0.57s); 3) The prevention deceleration is less than the maximum deceleration; otherwise, the accident is determined to be unavoidable under the initial conditions. The initial conditions here refer to the various states of the accident parties at the 0th second of the simulation start. (2) Dangerous scenario construction: 1) For accidents that are avoidable, retain the accident scenario initial parameters of the accident parties, all information of other parties, and road traffic environment information, and refer to the standardized accident scenario construction method to construct a standardized dangerous scenario. 2) For accidents that are not avoidable, no dangerous scenario construction is performed.

[0097] Table 3 Avoidance analysis parameters

[0098]

[0099]

[0100] This embodiment proposes a new method for multi-source data fusion and standardized scene reconstruction of road traffic accidents, establishes a complete accident reconstruction data table, unifies the data standard format, better adapts to multi-platform accident scene simulation, and accurately restores the accident scene. Economically, it can reduce the time and labor costs of accident handling, provide a scientific basis for determining the cause of the accident, and also assist in the prevention of road traffic accidents, reduce economic losses caused by accidents and disputes involving compensation for damages caused by traffic accidents. The above-described embodiments are only descriptions of the preferred embodiments of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solution of the present invention by ordinary technicians in this field should fall within the scope of protection determined by the claims of the present invention.

Claims

1. A method for multi-source data fusion and standardized scene reconstruction of road traffic accidents, characterized in that: include: Collect multi-source data on traffic accidents and construct accident reconstruction data tables; The accident reconstruction data table includes: static data and dynamic data; Based on the static data, construct a static scene graph; Based on the static scene graph and the dynamic data, multi-source dynamic data fusion is performed; Constructing a standardized accident scenario based on the static scene graph and the fused multi-source dynamic data; Based on the standardized accident scenario and the fused multi-source dynamic data, a standardized danger scenario is constructed.

2. The method for multi-source data fusion and standardized scene reconstruction of road traffic accidents according to claim 1 is characterized in that: The multi-source data of traffic accidents include: on-site investigation records, interrogation records, accident scene maps, on-site images, appraisal opinions, investigative experimental data, electronic data, road design information, automobile announcements, and motor vehicle registration information.

3. The method for multi-source data fusion and standardized scene reconstruction of road traffic accidents according to claim 1 is characterized in that: Constructing the accident reconstruction data table includes: road layer, facility layer, environment layer, and traffic operation layer; The road layer includes: intersection section type, road alignment, horizontal and vertical alignment, number of lanes, road width, and road surface morphology; The facility layer includes: traffic signs, traffic markings, guardrails and railings, and obstacles; The environmental layer includes: lighting conditions, weather conditions; The traffic operation layer includes: basic data of participants, dynamic parameters, and collision factors; The static data includes the road layer, facility layer, and environment layer, and the dynamic data includes the traffic operation layer.

4. The method for multi-source data fusion and standardized scene reconstruction of road traffic accidents according to claim 3 is characterized in that: Constructing the static scene graph includes: The static scene graph is drawn based on the road layer, the facility layer and the environment layer. The static scene graph also includes: a vehicle starting position, a collision point position and a stop position after the collision.

5. The method for multi-source data fusion and standardized scene reconstruction of road traffic accidents according to claim 4 is characterized in that: Based on the static scene graph and the dynamic data, performing multi-source dynamic data fusion includes: Import the static scene graph into PC-Crash and preset the origin position; Add accident participants and set parameters based on the basic data of the participants; According to the collision point position and the post-collision stop position, set and optimize the motion sequence of each participant until the post-collision stop position is within the error range, complete the post-collision simulation, and lock the post-collision trajectory; According to the post-collision simulation results and the vehicle's starting position, the participant's motion sequence is adjusted until the participant's trajectory is within the error range, completing the pre-collision simulation; According to the above steps, the accident collision process is restored to complete the fusion of multi-source dynamic data; and the fused multi-source dynamic data is verified.

6. The method for multi-source data fusion and standardized scene reconstruction of road traffic accidents according to claim 1 is characterized in that: Based on the static scene graph and the fused multi-source dynamic data, constructing a standardized accident scene includes: Performing format conversion on the static scene graph and the fused multi-source dynamic data respectively; After visual verification of the format-converted static scene graph and multi-source dynamic data, a standardized accident scenario is generated.

7. The method for multi-source data fusion and standardized scene reconstruction of road traffic accidents according to claim 6 is characterized in that: Based on the standardized accident scenario and the fused multi-source dynamic data, the standardized danger scenario is constructed including: Performing avoidance analysis on the fused multi-source dynamic data to obtain accident avoidability parameters; Based on the accident avoidability parameters, using preset judgment conditions, judging whether the accident is avoidable; For accidents that are avoidable, the initial parameters of the accident scene of the parties involved in the accident, all information about other parties involved, and the road traffic environment information are retained, and a standardized danger scene is constructed based on the method for constructing a standardized accident scene.

8. The method for multi-source data fusion and standardized scene reconstruction of road traffic accidents according to claim 7 is characterized in that: The judgment conditions are: The prevention distance is less than the visibility of the road section where the accident occurred, and there is no line of sight obstruction; The prevention reaction time is greater than the theoretical braking reaction time when humans are concentrating; The preventive deceleration is less than the maximum deceleration; otherwise, the accident is determined to be unavoidable under the initial conditions.

9. The method for multi-source data fusion and standardized scene reconstruction of road traffic accidents according to claim 7 is characterized in that: The accident avoidable parameters are: maximum deceleration, maximum allowed speed, preventive speed, preventive distance, preventive reaction time, and preventive deceleration.

Images