CN114445576A - Three-dimensional dynamic sight distance evaluation system and method based on laser radar point cloud data - Google Patents

Abstract

The invention discloses a three-dimensional dynamic sight distance evaluation system and method based on laser radar point cloud data. The invention provides a theoretical basis for dynamic visual range inspection and road driving safety improvement based on a road three-dimensional scene.

Description

3D dynamic line-of-sight assessment system and method based on lidar point cloud data

technical field

The invention relates to the field of highway safety evaluation, in particular to a three-dimensional dynamic line-of-sight evaluation system and method based on laser radar point cloud data.

Background technique

Due to the complex driving environment on the road, the actual running speed of the vehicle changes continuously with the changes of road conditions, driving techniques and vehicle performance. Therefore, the static two-dimensional design line of sight of the separated horizontal and vertical roads cannot accurately reflect the real driving environment. There is a certain deviation in the line-of-sight estimation for the road, which makes it difficult to determine the validity of the line-of-sight evaluation result.

SUMMARY OF THE INVENTION

In order to overcome the shortcomings and deficiencies of the prior art, the present invention provides a three-dimensional dynamic line-of-sight evaluation system and method based on lidar point cloud data, specifically based on the running speed prediction model to accurately estimate the available line-of-sight of the road three-dimensional space.

The present invention adopts following technical scheme:

A three-dimensional dynamic line-of-sight evaluation method based on lidar point cloud data, comprising the following steps:

Obtain vehicle trajectory and point cloud data;

Preprocess and separate point cloud data to form ground point cloud data and point cloud data higher than the ground;

Delaunay triangulation algorithm is used to form a TIN triangulation network for the ground point cloud data, and the linear data of the road is extracted by the TIN triangulation network; it is segmented according to the road alignment combination to obtain the three-dimensional geometric feature set of the road; based on the three-dimensional linear geometric features The running speed prediction model of the vehicle is used to obtain the running speed of the vehicle, and the driving angle range of the driver at each moment is determined according to the running speed;

Classify the point cloud data above the ground to form homogeneous objects and non-homogeneous objects, and further obtain a 3D environment that only contains road information;

The most unfavorable position of the line of sight of the vehicle is the most unfavorable position of the line of sight on one side of the one-way road, and the most favorable position of the line of sight is the other side, and the smaller of the two is the three-dimensional view of the driver in the real three-dimensional environment. Dynamic available viewing distance.

Further, it also includes a line of sight evaluation step, calculating the driving line of sight of the vehicle at this position, and comparing it with the three-dimensional dynamic available line of sight at the position, if the three-dimensional dynamic usable line of sight is greater than the driving line of sight, the requirement is satisfied, otherwise it is not satisfied. .

Further, the most unfavorable position of the line of sight on one side of the one-way road is the most unfavorable position of the line of sight, and the other side is the most favorable position of the line of sight. The smaller of the two is that the driver is in a real three-dimensional environment. The 3D dynamic available viewing distance of , specifically:

When the driver of the vehicle is in the most unfavorable line-of-sight vehicle position, using the view cone ellipse, the farthest position of the lane edge line that the driver can see, and the route arc length of the driver's current position corresponding to the lane edge line as the three-dimensional dynamic line of sight;

When the vehicle driver is at the vehicle position with the most favorable sight distance, obtain the three-dimensional dynamic sight distance in the same way as the above method;

The three-dimensional dynamic visual distance obtained from the two positions is compared, and the smaller value is the three-dimensional dynamic available visual distance of the driver in the real three-dimensional environment.

Further, the RANSAC segmentation algorithm is performed on the preprocessed point cloud data to form ground data and other point cloud data higher than the ground.

Further, the moving angle of view is the maximum horizontal and vertical angle of view that the driver can see in the dynamic field of view state, and the elliptical viewing cone of each driver's position is determined by this viewing angle, and its projection to the vertical two-dimensional space is the viewing cone. Ellipse.

A system for a three-dimensional dynamic line-of-sight evaluation method includes a lidar and an inertial navigation system, wherein the inertial navigation system includes an inertial navigation unit and a GPS unit.

Further, the lidar is arranged at the top of the vehicle and maintains a horizontal position.

Further, the inertial navigation unit is installed on the center line of the rear wheel of the vehicle and the Y-axis direction faces the forward direction of the vehicle.

Further, the time stamps of the lidar and inertial navigation systems are synchronized, and the origin of the point cloud collection coordinate system starts at the lidar position. The inertial navigation system provides the position of the vehicle in the geodetic coordinate system.

Further, intra-frame correction and feature extraction are performed by the LOAM algorithm.

Beneficial effects of the present invention:

A real three-dimensional environment can be established through the point cloud data of lidar technology, and a three-dimensional dynamic line-of-sight evaluation method of the road is formed by the driver's view cone space at different operating speeds, which is a method for dynamic line-of-sight inspection and improvement of road based on the three-dimensional scene of the road. Driving safety provides theoretical basis and analytical calculation methods.

Description of drawings

Fig. 1 is the schematic diagram of the three-dimensional dynamic available line-of-sight of the highway of the present invention;

FIG. 2 is the position of the vehicle in the most unfavorable line-of-sight state of the road in the road section of the present invention.

Figure 3 is a flow chart of the present invention.

The figure shows:

1. Driver's viewpoint; 2. Lane centerline; 3. Lane edge line; 4. Maximum moving angle of view in horizontal direction; 5. Maximum moving angle of view in vertical direction; 6. Centerline of viewpoint; 7. Position of target point; 8. Road Side obstacle area; 9. The most unfavorable sight distance vehicle position; 10. Lane boundary line; 11. Lane edge line; 12. Road marking line; 13. Obstacle area.

Detailed ways

The present invention will be described in further detail below with reference to the embodiments and the accompanying drawings, but the embodiments of the present invention are not limited thereto.

Example

As shown in Figures 1-3, a three-dimensional dynamic line-of-sight assessment system and method based on lidar point cloud data, the system part includes lidar and an inertial navigation system, and the inertial navigation system includes an inertial navigation unit and a GPS. Synchronize the time stamps of the lidar and the inertial navigation system, perform intra-frame correction and feature extraction through the LOAM algorithm, and at the same time, the origin of the point cloud coordinate system starts at the position of the lidar, so that the inertial navigation system and lidar can be used to provide on-vehicle lidar. The trajectory of the point cloud is finally determined accurately, so that the data can be converted into the geodetic coordinate system.

Specifically, the lidar is installed on the roof and maintained in a horizontal position. The lidar has the advantages of long measurement distance, high scanning frequency, high sampling rate of ranging and high measurement resolution. The higher scanning frequency can ensure installation in fast-moving vehicles. The lidar on the car can still obtain a clear 3D point cloud. However, when the sampling frequency is fixed, the faster scanning frequency will reduce the angle resolution, so generally only a moderate scanning frequency can be selected.

Lidar collects point cloud data frame by frame, the collection angle is 360 degrees, and the line of sight is formed by the horizontal angle α and the vertical angle β of the driver and the target point (view cone center) to form an elliptical view cone.

The inertial navigation system contains a positioning and a directional antenna. The inertial navigation system can generate 150 measurements per second, provide the geodetic coordinates of each measurement of the vehicle, and obtain the trajectory of the vehicle. Lidar collects point clouds 20 times per second. The point cloud collected each time is based on the laser radar as the coordinate origin, but not the point cloud data in the geodetic coordinate system. By synchronizing the time and space of the laser radar and the inertial navigation system, you can Convert the point cloud data to the geodetic coordinate system.

The specific evaluation method includes the following steps:

S1 obtains vehicle trajectory and point cloud data through lidar and inertial navigation system, and accurately determines the pose of the point cloud according to the vehicle trajectory.

S2 preprocesses and separates point cloud data to form ground point cloud data and point cloud data higher than the ground;

Specifically:

Due to the measurement principle, the resolution of lidar will also decrease sharply with the increase of the distance of the detected object. The point cloud data should be preprocessed first, and the point cloud data close to the scanning center area should be preserved, that is, the point cloud data is relatively low. The concentrated area should contain most of the point clouds, and the points contained have the characteristics of clear imaging, small error, and high density, so as to eliminate the distant and cluttered point clouds in the lidar point cloud.

The retained point cloud data is separated by the RANSAC segmentation algorithm to form the ground point cloud data and the point cloud data above the ground.

S3 uses the Delaunay triangulation algorithm to form a TIN triangular network for the ground point cloud data, and uses the TIN triangular network to extract the linear data of the road. First, it is segmented according to the road linear combination form, and the three-dimensional features of the road are extracted and calculated: the unit tangent vector θ, The main normal vector β, the secondary normal vector γ, the three-dimensional curvature k, and the three-dimensional torsion τ, finally form the three-dimensional geometric feature set of the road.

Further, the Delaunay triangulation consists of a series of connected but non-overlapping triangles whose circumcircles do not contain any other points in the area.

Further, the road alignment combination mainly refers to the alignment combination of the road plane and longitudinal section. The plane alignment is divided into straight line, circular curve and easing curve, and the longitudinal section is divided into longitudinal slope section and vertical curve section.

Using the existing three-dimensional running speed prediction model, the three-dimensional features of the five types of roads are substituted to obtain the running speed of the vehicle on the road.

The three-dimensional running speed prediction model can refer to the paper: Research on the running speed prediction method of passenger cars based on the linear three-dimensional spatial geometric characteristics of highways.

For different common running speeds, linear interpolation can be performed in combination with the value distribution of the maximum horizontal viewing angle 4 and the vertical maximum viewing angle 5 of the moving viewing angle to determine the range of the viewing cone ellipse formed by the corresponding viewing angle at the running speed.

S4 uses a classification algorithm to classify and reconstruct the point cloud data above the ground to form a defined three-dimensional object, that is, a homogeneous object, and other point cloud data to generate multi-patch files to form heterogeneous objects. Eliminate irrelevant point cloud parameters that do not belong to road information, such as other vehicles, pedestrians and other information. A 3D environment containing only road information can be seen. The road information includes edge lines on both sides of the road, and information on obstacles that affect the visual cone of the road, such as guardrails.

For S5, the most unfavorable position of the line of sight is the vehicle's deviation to one side of the one-way road, and the most favorable position of the line of sight is to the other side. 3D dynamic available viewing distance.

The three-dimensional dynamic line of sight is that in the three-dimensional alignment of the highway, when the vehicle is running at a certain speed, the driver is within the dynamic view, along the extending direction of the road, and starts searching from the first point of the nearer lane edge line that can be seen. Up to the first point that cannot be seen, the distance from this point to the observation point along the route is the three-dimensional dynamic sight distance of the observation point.

Through the specification, the offset position of the driver's viewpoint relative to the center of the vehicle can be determined under different standard models. Generally, the position where the target point can be seen by default is 0.1m high on the edge line of the near side lane.

The center line of the driver's line of sight and the tangent to the lane line should be parallel and consistent.

The line-of-sight relationship between the ellipse formed by the driver's point of view and the target point is derived from the principle of two-point line-of-sight in space. If the target point is blocked by obstacles such as the road itself and objects formed by surrounding facilities, it is not visible, otherwise it is in a visible state.

Starting from the lane edge line corresponding to the center point of the vehicle, the distance between the target object points is 5m, forming a view cone between the driver's point of view and the target object point, and advancing forward until the first non-visible target object point, the target object point The distance between the previous target point and the driver is the three-dimensional dynamic available sight distance of the vehicle at this position, and the viewpoint positions of different drivers are sequentially calculated along the vehicle's forward direction every 5m to determine the available sight distance of the entire road section.

Specifically: on the road containing the information of the lane dividing line 10 and the road marking line 12, the driver of the vehicle is in the most unfavorable line of sight vehicle position 9, and the vehicle is close to the lane edge line 3 of the road at this time. Position, under the combined occlusion of roadside obstacle areas 8 and 13 established by point cloud data and road alignment, the driver's viewpoint 1 is combined with the maximum moving angle of view 4 in the horizontal direction and the maximum moving angle of view 5 in the vertical direction at the running speed to form a view along the center of the viewpoint. Line 6 is the view cone at the center position, and the position 7 of the target object point at the farthest position of the lane center line 2 and the long distance of the curve projected from the driver's viewpoint 1 to the lane edge line 11 along the tangential direction of the road are taken as 3D available viewing distance.

Set the left and right lane edge lines of the vehicle to the one-way road as the most unfavorable position and the most favorable position of the line of sight, and obtain the 3D dynamic line of sight at the two positions. Dynamic available viewing distance.

Calculate the driving sight distance of the vehicle at this position according to the formula provided in the specification, and compare it with the 3D dynamic available sight distance calculated at the position. If the 3D dynamic available sight distance is greater than the required driving sight distance, the sight distance is satisfied. Check calculation requirements, and vice versa is not satisfied.

The above-mentioned embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited by the described embodiments, and any other changes, modifications, substitutions, and combinations made without departing from the spirit and principle of the present invention , simplification, all should be equivalent replacement modes, and are all included in the protection scope of the present invention.

Claims (10)

1. a three-dimensional dynamic line-of-sight assessment method based on lidar point cloud data, is characterized in that, comprises the steps: Obtain vehicle trajectory and point cloud data; Preprocess and separate point cloud data to form ground point cloud data and point cloud data higher than the ground; Using the Delaunay triangulation algorithm to form a TIN triangulation network for the ground point cloud data, and using the TIN triangulation network to extract the road alignment data; and segmenting according to the road alignment combination to obtain the 3D geometric feature set of the road; based on the 3D linear geometric features The running speed prediction model of the vehicle is used to obtain the running speed of the vehicle, and the driving angle range of the driver at each moment is determined according to the running speed; Classify the point cloud data above the ground to form homogeneous objects and non-homogeneous objects, and further obtain a 3D environment that only contains road information; The most unfavorable position of the line of sight of the vehicle is the most unfavorable position of the line of sight on one side of the one-way road, and the most favorable position of the line of sight is the other side, and the smaller of the two is the three-dimensional view of the driver in the real three-dimensional environment. Dynamic available viewing distance. 2. The three-dimensional dynamic line-of-sight evaluation method according to claim 1, further comprising a line-of-sight evaluation step, calculating the driving line-of-sight distance of the vehicle at this position, and comparing with the three-dimensional dynamic available line-of-sight distance of the position, if If the three-dimensional dynamic available sight distance is greater than the driving sight distance, the requirement is satisfied, otherwise it is not satisfied. 3. The three-dimensional dynamic line-of-sight evaluation method according to claim 1 is characterized in that, taking the edge line of one side of the vehicle deviating to one side of the road as the most unfavorable position of the line of sight, then deflecting to the other side is the most favorable position of the line of sight , the smaller of the two is the 3D dynamic available sight distance of the driver in the real 3D environment, specifically: When the driver of the vehicle is in the most unfavorable line-of-sight vehicle position, using the view cone ellipse, the farthest position of the lane edge line that the driver can see, and the route arc length of the driver's current position corresponding to the lane edge line as the three-dimensional dynamic line of sight; When the vehicle driver is at the vehicle position with the most favorable sight distance, obtain the three-dimensional dynamic sight distance in the same way as the above method; The three-dimensional dynamic visual distance obtained from the two positions is compared, and the smaller value is the three-dimensional dynamic available visual distance of the driver in the real three-dimensional environment. 4 . The three-dimensional dynamic line-of-sight evaluation method according to claim 1 , wherein a RANSAC segmentation algorithm is performed on the preprocessed point cloud data to form ground data and other point cloud data higher than the ground. 5 . 5. three-dimensional dynamic visual distance evaluation method according to claim 1 is characterized in that, described dynamic angle of view is the horizontal and vertical maximum angle of view that driver can see under dynamic field of view state, and each driver is determined by this angle of view The elliptical view frustum of the position, which is projected to the vertical two-dimensional space as the view frustum ellipse. 6. A system for implementing the three-dimensional dynamic line-of-sight assessment method according to any one of claims 1-5, characterized in that it comprises a lidar and an inertial navigation system, and the inertial navigation system comprises an inertial navigation unit and a GPS unit. 7. The system according to claim 6, wherein the lidar is arranged at the top of the vehicle and maintains a horizontal position. 8 . The system according to claim 6 , wherein the inertial navigation unit is installed on the center line of the rear wheel of the vehicle and the Y-axis direction is toward the forward direction of the vehicle. 9 . 9. The system according to any one of claims 6-8, wherein the time stamps of the lidar and the inertial navigation system are synchronized, and the origin of the point cloud collection coordinate system starts at the position of the lidar, and the inertial navigation system starts at the position of the lidar. The system provides the position of the vehicle in the geodetic coordinate system. 10. The system according to claim 6, wherein intra-frame correction and feature extraction are performed by LOAM algorithm.

Images