Expressway reconstruction and expansion construction period poor visibility section identification and risk warning method

Abstract

The application discloses a method for identifying and risk warning of a poor sight distance road section during expressway reconstruction and expansion construction, and comprises the following steps: collecting road and environment information of a target road section, and obtaining point cloud data of the target road section; performing voxelization processing on the point cloud data to obtain processed point cloud; extracting observation points along the expressway from the point cloud, and calculating driving sight distance of the observation points; judging whether the driving sight distance is less than the stopping sight distance, if yes, the target road section is a risk road section; and if no, the target road section is not a risk road section. The application can automatically evaluate the driving sight distance along the expressway, and accurately determine the risk road section with insufficient sight distance.

Description

Technical Field

[0001] This invention relates to the field of road traffic, specifically to a method for identifying and providing risk warnings for road sections with poor visibility during the reconstruction and expansion of highways. Background Technology

[0002] During the widening and reconstruction of mountain highways, if there are high slopes on the roadside, construction retaining walls (slabs) are usually erected between the original and expanded sections to ensure safety. However, these retaining walls (slabs) can shorten drivers' visibility on curves, turning traffic conditions that were previously within the driver's line of sight (such as traffic accidents or queues caused by traffic congestion) into "beyond sight" information, thus creating traffic safety risks. Figure 1 As shown in (A) and (B). Especially when the original hard shoulder section is also encroached upon by the construction area, the reduction in sight distance will be more severe, exacerbating traffic safety risks. Therefore, identifying road sections with poor sight distance to reduce traffic safety risks is of great significance.

[0003] Currently, poor visibility sections are primarily identified by assessing driving sight distance. Existing sight distance assessments typically rely on manual quantification using construction drawings and on-site measurements. Manual inspection involves drawing lines on the construction drawings and identifying obstacles between the road alignment and the line of sight—a very tedious and time-consuming process, particularly challenging during large-scale highway implementation. Furthermore, manual inspection can only handle sight distances at horizontal curves on a planar projection; changes in vertical alignment cannot be quickly assessed using this method. In addition, manual sight distance assessment using construction drawings is limited to features captured on the drawings. If dense vegetation or temporary construction retaining walls / slabs are not fully shown on the drawings, the sight distance assessment will be inaccurate, resulting in poor identification of poor visibility sections.

[0004] Therefore, to solve the above problems, a method for identifying and risk warning of poor visibility sections during the construction period of highway reconstruction and expansion is needed, which can automatically assess the driving visibility along the highway and accurately identify the risk sections with insufficient visibility. Summary of the Invention

[0005] In view of this, the purpose of this invention is to overcome the defects in the prior art and provide a method for identifying and risk warning of poor visibility road sections during the construction period of highway reconstruction and expansion, which can automatically assess the driving visibility along the highway and accurately determine the risk sections with insufficient visibility.

[0006] The method for identifying and providing risk warning of poor visibility sections during highway reconstruction and expansion construction of the present invention includes the following steps:

[0007] Collect road and environmental information of the target road segment and obtain point cloud data of the target road segment;

[0008] The point cloud data is voxelized to obtain the processed point cloud;

[0009] Observation points along the highway are extracted from the point cloud, and the driving sight distance of the observation points is calculated.

[0010] Determine if the driving visibility distance is less than the stopping visibility distance. If so, the target road segment is a risky road segment; otherwise, the target road segment is not a risky road segment.

[0011] Furthermore, the point cloud data undergoes voxelization processing, specifically including:

[0012] Point cloud data is discretized using grid-based subsampling to form a three-dimensional voxel grid;

[0013] Based on the position of points in the point cloud data in three-dimensional space, each point in the point cloud data is assigned to a voxel; wherein the voxel size is slightly larger than the spacing between the scan lines of the scanning system, the scanning system being the scanning device used to acquire information.

[0014] Furthermore, observation points along the highway are extracted from the point cloud, specifically including:

[0015] Point clouds containing road information are used as candidate point clouds, and trajectory seed points are extracted from the candidate point clouds. The trajectory seed points are a set of the starting points of the driver's line of sight.

[0016] The trajectory seed points that meet the observation point spacing are used as observation points along the highway.

[0017] Furthermore, the driving sight distance at the observation point is calculated, specifically including:

[0018] Copy the observation point set I to obtain the target point set J; where the height of each observation point in I is the driver's eye level, and the height of all target points in J is the height of the vehicle's taillights.

[0019] Arrange the observation points and target points on the target road segment to establish a line of sight connecting each pair of observation points and target points in three-dimensional space;

[0020] Select multiple points along the line of sight and assign each point to a voxel unit based on its position in three-dimensional space;

[0021] Multiple points are clustered into different cluster groups. The cluster group closest to the starting point of the line of sight is taken as the first obstacle, and the voxel unit containing the first obstacle is taken as the target voxel unit.

[0022] The distance between the observation point and the centroid of the target voxel unit is taken as the line-of-sight distance of the observation point.

[0023] Furthermore, the line-of-sight equation connecting the observation point and the target point is determined according to the following formula:

[0024]

[0025] Where t represents any point on the line of sight; the coordinates of the observation point i in I are (x... i ,y i ,z i The coordinates of target point j in J are (x...). j ,y j ,z j The coordinates of point t on the line of sight are (x(t), y(t), z(t)).

[0026] Furthermore, it also includes:

[0027] If the target road segment is a high-risk segment, calculate the minimum deceleration 'a' required to avoid a collision between the two vehicles on the target road segment. min ;if Then a safety warning needs to be issued to remind drivers upstream to slow down in time; if Then no safety warning is needed; among them, This represents the maximum deceleration of the upstream vehicle.

[0028] Furthermore, the minimum deceleration a is determined according to the following formula. min :

[0029]

[0030] The speed of the upstream vehicle is V. t The maximum deceleration is The driving visibility distance at this location is ASD. t The speed of the last car downstream is V. e The maximum deceleration is The distance between the upstream vehicle and the last downstream vehicle is D. et .

[0031] Furthermore, both the upstream vehicle and the last downstream vehicle are considered as target vehicles, and the speed V of the target vehicle is determined according to the following formula:

[0032]

[0033] Where N is the region r j The total number of cameras is divided into several regions, and the camera view is used to capture images of the target vehicle; ∑v ij This is the sum of the velocities of all regions in the camera view;

[0034] ∑v i For target vehicle i in area rj Several velocities v i The sum of Δf j For target vehicle i in area r j The number of frames in the tracking track;

[0035] D j Δt represents the true distance between two pixels. i Represents the reciprocal of the frame rate;

[0036] Let s be the positional interval of target vehicle i in a pair of video frames; i (t) is the position of the target vehicle i in the current video frame, s i (t-Δt) is the position of target vehicle i in the previous frame, and Δt is the time interval between two consecutive frames.

[0037] The beneficial effects of this invention are as follows: This invention discloses a method for identifying and warning of poor visibility sections during highway reconstruction and expansion construction. It automatically assesses the visibility along the highway using vehicle-mounted lidar data to identify sections with insufficient visibility and adds temporary traffic flow sensing equipment. Based on real-time detection of traffic flow conditions in the road sections using computer vision, it determines the warning threshold by quantifying the minimum deceleration required to avoid a collision. It then reminds upstream drivers to slow down in time through methods such as displaying information on temporary roadside information screens and flashing safety warning facilities, thereby ensuring traffic safety on poor visibility sections during highway reconstruction and expansion construction. Attached Figure Description

[0038] The present invention will be further described below with reference to the accompanying drawings and embodiments:

[0039] Figure 1 (A) is a schematic diagram of sufficient sight distance before the reconstruction and expansion of the present invention;

[0040] Figure 1 (B) is a schematic diagram of insufficient sight distance under the influence of the baffle during the reconstruction and expansion construction period of the present invention;

[0041] Figure 1 (C) is a schematic diagram of traffic safety risk warning during the reconstruction and expansion construction period of the present invention;

[0042] Figure 2 This is a schematic diagram of the method for identifying and warning of poor sight distance road sections according to the present invention;

[0043] Figure 3 This is a schematic diagram illustrating the vehicle speed estimation method of the present invention.

[0044] Figure 4 This is a schematic diagram illustrating the traffic safety early warning principle of the present invention. Detailed Implementation

[0045] The present invention will be further described below with reference to the accompanying drawings, as shown in the figures:

[0046] This embodiment discloses a method for identifying and providing risk warnings for road sections with poor visibility during the reconstruction and expansion of highways, including the following steps:

[0047] Collect road and environmental information of the target road segment and obtain point cloud data of the target road segment;

[0048] The point cloud data is voxelized to obtain the processed point cloud;

[0049] Observation points along the highway are extracted from the point cloud, and the driving sight distance of the observation points is calculated.

[0050] Determine if the driving visibility distance is less than the stopping visibility distance. If so, the target road segment is a risky road segment; otherwise, the target road segment is not a risky road segment.

[0051] This invention, after acquiring and recording lidar scans and generating point cloud data, includes a line-of-sight assessment process that identifies line-of-sight-limiting objects, such as construction retaining walls, along an expanded or reconstructed highway based on the scan data. This is achieved by analyzing the driving line-of-sight distances between multiple observation points and target points along a virtual highway formed by the scan. Once the location of the line-of-sight-limiting point is determined, points representing obstacles are extracted, and then the distance is calculated.

[0052] In this embodiment, the first step is to drive an inspection vehicle equipped with a lidar on the highway during the reconstruction and expansion construction period. During the vehicle's journey, the vehicle-mounted lidar is used to scan the road and environment, collect road and environmental information of the target road section, and then generate point cloud data.

[0053] In this embodiment, the point cloud data undergoes voxelization processing, specifically including:

[0054] First, the point cloud data is discretized using grid-based subsampling to form a 3D voxel mesh. This process involves defining the 3D voxel mesh based on the data range and specific voxel dimensions. A 3D voxel mesh is a method for representing and processing 3D spatial data by dividing 3D space into a series of small cubic units called voxels.

[0055] Then, the points in the point cloud are assigned to voxel units based on their positions in three-dimensional space. The voxel size used is slightly larger than the spacing between the scan lines of the laser scanning system. This setting ensures that the voxel size can include potential obstacles while avoiding including obstacles that are not actually in the line of sight.

[0056] In this embodiment, extracting observation points along the highway from the point cloud includes:

[0057] Point clouds containing road information are used as candidate point clouds. Trajectory seed points, which are sets of the driver's line-of-sight starting points, are extracted from these candidate point clouds. Trajectory seed points that meet the required observation point spacing are used as observation points along the highway. Specifically, after voxelizing the point cloud, trajectory seed points are extracted and superimposed on the results of the highway laser scan. These seed points represent observation points and target points along the highway, and are sets of the driver's line-of-sight starting points. These points are parallel to the road centerline and cover the entire assessed road segment, obtained by filtering points located in the lowest plane of the laser scanning system. These points are resampled to meet a predefined observation point spacing; for example, if the driving sight distance is assessed every 30 meters on the road segment, the observation point spacing is 30 meters.

[0058] In this embodiment, calculating the driving sight distance at the observation point specifically includes:

[0059] The set of observation points I is copied to obtain the set of target points J; where the height of each observation point in I is the driver's eye level, and the height of all target points in J is the height of the vehicle's taillights; both I and J represent points from the point cloud ρ∈R. N×3 A set of trajectory seed points, where I = {i1, i2, ..., i n}, J={j1,j2,…,j n Let} represent the number of equidistant observation points and target points, and the number of observation points and target points must be equal. Set J is a copy of set I at different heights. Specifically, the height of each observation point in I is adjusted to the eye level of the car driver, for example, 1.05 meters. Similarly, the height of all target points in J is adjusted to the height of the vehicle's taillights, for example, 0.38 meters.

[0060] Arrange the observation points and target points on the target road segment to establish a line of sight connecting each pair of observation points and target points in three-dimensional space; the line of sight between a single observation point i and a single target point j can be represented by the following set of parametric equations:

[0061]

[0062] Where t represents any point on the line of sight; the coordinates of the observation point i in I are (x... i ,y i ,z i The coordinates of target point j in J are (x...). j ,y j ,z j The coordinates of point t on the line of sight are (x(t), y(t), z(t)).

[0063] Select a plurality of points along the line of sight, and assign the points to voxel units based on the position of each point in the three-dimensional space; that is, locate the voxel units falling on the line of sight. Once the voxel units are found, it is possible to analyze whether the voxel units along the line of sight contain potential obstacles.

[0064] Cluster the plurality of points into different clustering groups, take the clustering group closest to the starting point of the line of sight as the first obstacle, and take the voxel unit containing the first obstacle as the target voxel unit; wherein, since a plurality of obstacles may exist along the same line of sight, it is first necessary to separate the obstacles at different points on the line of sight, so as to identify the first obstacle.

[0065] The present invention adopts the density-based clustering algorithm DBSCAN, and clusters the points into different groups based on two parameters, which are the proximity between points in a cluster and the minimum allowable number of points in a cluster. Specifically, an unvisited point is first selected, and then it is checked whether the number of points within its proximity range is greater than or equal to the minimum allowable number. If yes, it is regarded as a core point, and all points within its proximity range are added to the same clustering group. If not, it is marked as a noise point or a boundary point.

[0066] After the first obstacle is identified and separated from the line of sight, the stopping sight distance for driving of each observation point can be calculated, and the distance between the observation point i and the centroid of the target voxel unit is taken as the stopping sight distance for driving of the observation point i. After the above operation is performed on all observation points in the set I, the available sight distance ASD of all observation points along the expressway can be obtained.

[0067] In this embodiment, if ASD < SSD, the target road section is a risk road section. Wherein, SSD is the stopping sight distance, and the value of the stopping sight distance SSD is shown in Table 1 below:

[0068] Table 1

[0069] Highway speed limit (km / h) 120 100 80 60 Parking sight distance (m) 210 160 110 75

[0070] In this embodiment, the present invention further comprises:

[0071] If the target road section is a risk road section, calculate the minimum deceleration a for two vehicles in the target road section to avoid collision min ; if it indicates that the deceleration effect brought by the deceleration of the upstream vehicle is weak, the upstream vehicle needs greater deceleration to ensure safety, and a safety warning is required, which reminds upstream drivers to decelerate in time by means of publishing information through a roadside temporary information screen and flashing of safety warning facilities (flash lamp array); if it indicates that the upstream vehicle can safely avoid collision through the current deceleration, and no safety warning is required; wherein, This represents the maximum deceleration of the upstream vehicle.

[0072] like Figure 4 As shown, suppose at a certain moment, the speed V of the upstream vehicle in the road segment with insufficient visibility (risk section) is... t Maximum deceleration The driving visibility distance at this location is ASD. t The speed V of the last car downstream e Maximum deceleration The maximum deceleration depends on the vehicle type and is generally a constant; the distance between the two vehicles is D. et .

[0073] For the upstream vehicle, the minimum deceleration 'a' required to avoid a collision between the two vehicles. min Calculated by the following formula:

[0074]

[0075] The speed of the upstream vehicle is V. t The maximum deceleration is The driving visibility distance at this location is ASD. t The speed of the last car downstream is V. e The maximum deceleration is The distance between the upstream vehicle and the last downstream vehicle is D. et .

[0076] For velocity V t or V e The calculation can be performed by converting the image plane displacement into real-world vehicle speed measurements.

[0077] Both the upstream vehicle and the last downstream vehicle are considered as target vehicles. The speed V(V) of the target vehicle is determined according to the following formula. t or V e ):

[0078]

[0079] Where N is the region r j The total number of cameras is divided into several regions, and the camera view is used to capture images of the target vehicle; ∑v ij The sum of the velocities of each region in the camera view; such as Figure 3 The virtual view of the camera is described, which is divided into three regions (r1, r2, and r3), each with distances d1, d2, and d3. The number of regions is not strictly fixed and depends on the specific configuration of the camera, including factors such as camera panning and tilt.

[0080] In region r jIn the middle, the vehicle's speed v ij The estimate is:

[0081] ∑v i For target vehicle i in area r j Several velocities v i The sum of Δf j For target vehicle i in area r j The number of frames in the tracking track;

[0082] Vehicle speed v in two consecutive frames i for:

[0083] D j Δt represents the true distance between two pixels. i Represents the reciprocal of the frame rate;

[0084] Each pair of frames uses a tracking track to generate motion adjacency, which can be expressed as follows:

[0085] Let s be the positional interval of target vehicle i in a pair of video frames; i (t) is the position of the target vehicle i in the current video frame, s i (t-Δt) represents the position of target vehicle i in the previous frame, and Δt is the time interval between two consecutive frames. Existing vehicle tracking algorithms can be used to perform trajectory analysis on the vehicle, which will not be elaborated upon here.

[0086] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for identifying and providing risk warning of poor visibility sections during the construction period of highway reconstruction and expansion, characterized in that: Includes the following steps: Collect road and environmental information of the target road segment and obtain point cloud data of the target road segment; The point cloud data is voxelized to obtain the processed point cloud; Extract observation points along the highway from the processed point cloud and calculate the driving sight distance of the observation points; Calculating the sight distance at the observation point specifically includes: For the set of observation points Copy the data to obtain the target point set. ;in, The height of each observation point is at the driver's eye level. The height of all target points in the diagram is the height of the vehicle's taillights; Arrange the observation points and target points on the target road segment to establish a line of sight connecting each pair of observation points and target points in three-dimensional space; Select multiple points along the line of sight and assign each point to a voxel unit based on its position in three-dimensional space; Multiple points are clustered into different cluster groups. The cluster group closest to the starting point of the line of sight is taken as the first obstacle, and the voxel unit containing the first obstacle is taken as the target voxel unit. The distance between the observation point and the centroid of the target voxel unit is taken as the driving line of sight of the observation point. Determine if the driving visibility distance is less than the stopping visibility distance. If so, the target road segment is a risky road segment; otherwise, the target road segment is not a risky road segment.

2. The method for identifying and providing risk warning of poor visibility sections during highway reconstruction and expansion construction as described in claim 1, characterized in that: The point cloud data is voxelized, specifically including: Point cloud data is discretized using grid-based subsampling to form a three-dimensional voxel grid; Based on the position of points in the point cloud data in three-dimensional space, each point in the point cloud data is assigned to a voxel; wherein the voxel size is slightly larger than the spacing between the scan lines of the scanning system, the scanning system being the scanning device used to acquire information.

3. The method for identifying and providing risk warning of poor visibility sections during highway reconstruction and expansion construction as described in claim 1, characterized in that: Extracting observation points along the highway from the processed point cloud specifically includes: The point cloud containing road information in the processed point cloud is used as a candidate point cloud, and trajectory seed points are extracted from the candidate point cloud. The trajectory seed points are a set of the starting points of the driver's line of sight. The trajectory seed points that meet the observation point spacing are used as observation points along the highway.

4. The method for identifying and providing risk warning of poor visibility sections during highway reconstruction and expansion construction as described in claim 1, characterized in that: The equation of the line of sight connecting the observation point and the target point is determined according to the following formula: ； in, Indicates any point in the line of sight; Central observation point The coordinates are ; Middle target point The coordinates are ; Point of view The coordinates are .

5. The method for identifying and providing risk warning of poor visibility sections during highway reconstruction and expansion construction as described in claim 1, characterized in that: Also includes: If the target road segment is a high-risk segment, calculate the minimum deceleration required for the two vehicles to avoid a collision on the target road segment. ;if In this case, a safety warning needs to be issued to remind drivers upstream to slow down in time; if In this case, no safety warning is required; among them, This represents the maximum deceleration of the upstream vehicle.

6. The method for identifying and providing risk warning of poor visibility sections during highway reconstruction and expansion construction as described in claim 1, characterized in that: The minimum deceleration is determined using the following formula. : ； The speed of the upstream vehicle is The maximum deceleration is The visibility distance at the current location is The speed of the last vehicle downstream was The maximum deceleration is The distance between the upstream vehicle and the last downstream vehicle is .

7. The method for identifying and providing risk warning of poor visibility sections during highway reconstruction and expansion construction as described in claim 1, characterized in that: Both the upstream vehicle and the last downstream vehicle are considered as the target vehicle, and the speed of the target vehicle is determined according to the following formula. : ； in, For the region The total number of cameras is divided into several areas, and the camera view is used to capture the target vehicle. This is the sum of the velocities of all regions in the camera view; ; For the target vehicle In the region Several speeds The sum of the two, For the target vehicle In the region The number of frames in the tracking track; ; This represents the actual distance between two pixels. Represents the reciprocal of the frame rate; ; For the target vehicle Positional interval within a pair of video frames; The target vehicle in the current video frame Location, The target vehicle in the previous frame Location, It is the time interval between two consecutive frames.

Images