An automatic driving collision risk detection method and system based on lane information and a medium

Abstract

The application relates to a lane information-based automatic driving collision risk detection method, system and medium, the method comprising P1. In a global coordinate system, based on a vehicle perception system, real-time acquisition of obstacle boundary shape point coordinate array information and obstacle center point coordinate data information, based on a vehicle positioning system, real-time acquisition of vehicle positioning coordinate array information, based on a high-precision map, real-time acquisition of lane left boundary coordinate data information, lane right boundary coordinate data information and lane center line discrete shape point coordinate array information. The application not only directly filters out obstacle information outside the lane through the frenet coordinate system, thereby reducing the calculation power consumption, but also calculates the collision risk according to the coordinates, and the calculation amount is very low, and when a plurality of alternative paths are planned and generated, the frenet coordinates of the obstacles do not need to be recalculated, and the collision risk can be calculated only through the matching point coordinate positions in the trajectory, thereby greatly reducing the calculation power consumption.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automatic driving, in particular to an automatic driving collision risk detection method and system based on lane information and a medium. BACKGROUND

[0002] In related technologies, the traditional collision detection method usually represents a vehicle by a mass point or a vehicle circumscribed circle and represents an obstacle by a polygon box, determines whether there is a collision by judging whether the polygon boxes intersect in geometric space, and then iteratively performs collision detection by geometric space detection on discrete points of a preselected lane trajectory to achieve collision detection on a road, or performs collision detection by using sampling methods such as scan lines, quadtree or grid search. The geometric space detection itself requires a large amount of computing power, and a general planning algorithm generates a series of reference trajectories, then selects the trajectories, and iteratively performs collision detection on discrete points of the preselected lane line to exponentially increase the consumption of system resources, so these methods require high system performance to ensure the real-time data processing of the automatic driving system.

[0003] In the prior art, patent application (CN115230694A) discloses an obstacle identification method and device for an automatic driving vehicle and a vehicle. The obstacle information of at least one obstacle identified by the automatic driving vehicle based on any coordinate system is obtained. The obstacle information of the at least one obstacle is converted into collision information in a Frenet coordinate system with the road center line as the reference, and the obstacles whose collision information meets the screening condition are screened out. The collision graph of each remaining obstacle is generated according to the collision information of the remaining obstacle, and the collision result with the automatic driving vehicle is determined based on the collision graph. When the collision result is a determined collision, the obstacle with the determined collision is identified as a need-to-avoid obstacle of the automatic driving vehicle. However, the driving area needs to be planned first and then collision detection is performed. If there is a collision risk, the situation needs to be re-planned and collision detection needs to be performed again. Multiple collision detections will result in more consumption of computing power. SUMMARY

[0004] In view of the deficiencies of the above prior art, the present application provides an automatic driving collision risk detection method and system based on lane information and a medium. The method directly filters out obstacle information outside the lane by using the Frenet coordinate system, thereby reducing the consumption of computing power. Moreover, the collision risk is calculated based on the coordinates, the calculation amount is very low, and the Frenet coordinates of the obstacles do not need to be recalculated when a plurality of alternative paths are generated. The collision risk can be calculated only by the coordinate positions of the matching points in the trajectory, thereby greatly reducing the consumption of computing power.

[0005] To achieve the above object and other related objects, the technical solutions of the present application are as follows:

[0006] An automatic driving collision risk detection method based on lane information, the method comprising:

[0007] P1. In a global coordinate system, based on a vehicle perception system, real-time acquisition of obstacle boundary shape point coordinate array information and obstacle center point coordinate data information, based on a vehicle positioning system, real-time acquisition of vehicle positioning coordinate array information, based on a high-precision map, real-time acquisition of lane left boundary coordinate data information, lane right boundary coordinate data information, and lane center line discrete shape point coordinate array information;

[0008] P2. Traversing the lane center line discrete shape point coordinate array information, calculating the coordinate point with the shortest distance to the obstacle center point coordinate data information, taking the coordinate point as the center point, selecting a fixed distance along the lane direction, and outputting reference driving lane section data information;

[0009] P3. Based on the reference driving lane section data information and the obstacle boundary shape point coordinate array information, establishing a Frenet coordinate system with the lane right boundary coordinate data information as the reference line, and outputting obstacle left and right boundary coordinate information and obstacle front and rear boundary coordinate information;

[0010] P4. Based on the vehicle positioning coordinate array information, establishing a Frenet coordinate system with the lane right boundary coordinate data information as the reference line, and outputting vehicle left and right boundary coordinate information according to the vehicle width;

[0011] P5. Based on the vehicle left and right boundary coordinate information, the obstacle left and right boundary coordinate information, and the obstacle front and rear boundary coordinate information, according to a vehicle collision detection algorithm, outputting vehicle collision risk data information.

[0012] Further, in step P5, the vehicle collision detection algorithm comprises:

[0013] P51. Obtaining the vehicle left boundary coordinate information (EL si , EL di ), the vehicle right boundary coordinate information (ER si , ER di ), the obstacle left boundary coordinate information (OL si , OL di ), and the obstacle right boundary coordinate information (OR si , OR di );

[0014] P52. If the function F is satisfied and the obstacle is static,

[0015] , where n is the time point and is a positive integer, then the vehicle has a collision risk, and the first risk index data information is outputted;

[0016] P53. If the function F is satisfied and the obstacle is dynamic, calculate and compare the time when the obstacle leaves the collision and the time when the vehicle arrives at the obstacle's position, and output the second risk index data information.

[0017] P54. If the function F is not satisfied and the obstacle is static, output the data information that there is no collision risk.

[0018] P55. If the function F is not satisfied and the obstacle is dynamic, predict whether the obstacle's trajectory intersects with the vehicle's trajectory, and output the third risk index data.

[0019] Furthermore, in step P53, calculating and comparing the time it takes for the obstacle to detach from the collision and the time it takes for the vehicle to reach the obstacle's location includes:

[0020] P531. Based on the left boundary coordinate information of the obstacle (OL) si ,OL di ), calculate the moving speed v of the obstacle. 障 and the angle α between the obstacle and the horizontal.

[0021] ,

[0022] ;

[0023] P532. Based on the obstacle's moving speed v 障 Calculate the time T for the obstacle to escape the collision, based on the included angle α with the horizontal plane. 脱 and the time T when the vehicle reaches the obstacle position 车 ,

[0024] ,

[0025] ,

[0026] Where v is the average speed of the vehicle;

[0027] P533. Based on the time T when the obstacle leaves the collision zone. 脱 and the time T when the vehicle reaches the obstacle position 车 If T 脱 Greater than T 车 There is no risk of collision if T 脱 Less than T 车 There is a risk of collision.

[0028] Furthermore, the average speed v of the vehicle,

[0029] .

[0030] Furthermore, in step P55, the prediction of whether the obstacle's trajectory intersects with the vehicle's trajectory includes whether there is a collision risk if there is an intersection and whether there is no collision risk if there is no intersection.

[0031] Furthermore, the method for determining the left and right boundary coordinate information of the obstacle is as follows: based on the left and right boundary values ​​in the boundary point coordinate information of the obstacle, the left lane boundary and right lane boundary of the current point are generated; all points of the left and right lane boundaries are traversed; the distance of the obstacle from the left and right road boundaries is calculated; and based on the left and right boundary distances and the lane width, the coordinate information of the left and right boundaries of the obstacle in the Frenet reference line coordinate system is calculated.

[0032] Furthermore, the method for determining the coordinates of the front and rear boundaries of the obstacle is as follows: first, traverse the discrete points from the starting point in the reference driving road segment data information, and calculate the first point where a collision occurs, which is the rear boundary of the obstacle; then, traverse the discrete points from the ending point in the reference driving road segment data information back to the starting point, and calculate the first point where a collision occurs, which is the coordinates of the front boundary of the obstacle.

[0033] Furthermore, the vehicle collision risk data information includes first risk index data information, second risk index data information, third risk index data information, and no collision risk data information.

[0034] To achieve the above and other related objectives, the present invention also provides an autonomous driving collision risk detection system based on lane information, including a computer device programmed or configured to perform the steps of any of the lane information-based autonomous driving collision risk detection methods described above.

[0035] To achieve the above and other related objectives, the present invention also provides a computer-readable storage medium storing a computer program programmed or configured to perform any of the lane information-based autonomous driving collision risk detection methods described above.

[0036] The present invention has the following positive effects:

[0037] 1. This invention directly filters out obstacle information outside the lane that does not need to be processed by using the Frenet coordinate system, thereby reducing computing power consumption.

[0038] 2. This invention obtains the coordinates of obstacles in the Frenet coordinate system of the driving lane. The Frenet coordinate information of the obstacle only needs to be calculated once in each calculation cycle, and the collision risk assessment can be made based on the relative position for direct use in planning decisions.

[0039] 3. The collision risk calculated based on coordinates in this invention has a very low computational load, and the planning and generation of multiple alternative paths does not require recalculating the Frenet coordinates of obstacles. The collision risk can be calculated only by the coordinate positions of the matching points in the trajectory, which greatly reduces the consumption of computing power. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of the method flow of the present invention;

[0041] Figure 2 This is a schematic diagram of the obstacle distribution according to the present invention. Detailed Implementation

[0042] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0043] Example 1: As Figure 1 As shown, an autonomous driving collision risk detection method based on lane information is provided, the method comprising:

[0044] P1. In the global coordinate system, based on the vehicle perception system, the coordinate array information of obstacle boundary points and the coordinate data information of obstacle center point are obtained in real time. Based on the vehicle positioning system, the vehicle positioning coordinate array information is obtained in real time. Based on the high-precision map, the coordinate data information of the left boundary of the lane, the coordinate data information of the right boundary of the lane, and the coordinate array information of the discrete points of the lane centerline are obtained in real time.

[0045] P2. Traverse the discrete point coordinate array information of the lane centerline, calculate the coordinate point with the shortest distance to the coordinate data information of the obstacle center point, take the coordinate point as the center point, select a fixed distance along the lane direction, and output the reference driving road segment data information;

[0046] P3. Based on the reference driving road segment data information and the obstacle boundary point coordinate array information, establish the Frenet coordinate system with the right boundary coordinate data information of the lane as the reference line, and output the left and right boundary coordinate information and the front and rear boundary coordinate information of the obstacle;

[0047] P4. Based on the vehicle positioning coordinate array information, establish a Frenet coordinate system with the right boundary coordinate data of the lane as the reference line, and output the left and right boundary coordinate information of the vehicle according to the vehicle width;

[0048] P5. Based on the vehicle's left and right boundary coordinates, the obstacle's left and right boundary coordinates, and the obstacle's front and rear boundary coordinates, output vehicle collision risk data information according to the vehicle collision detection algorithm.

[0049] In this embodiment, in step P5, the vehicle collision detection algorithm includes:

[0050] P51. Obtain the left boundary coordinate information of the vehicle (EL) si ,EL di The vehicle's right boundary coordinate information (ER) si ,ER di The left boundary coordinate information of the obstacle (OL) si ,OL di ) and the right boundary coordinate information of the obstacle (OR) si OR di );

[0051] P52. If the function F is satisfied and the obstacle is static.

[0052] If n is a time point and is a positive integer, then the vehicle is at risk of collision, and the first risk index data information is output.

[0053] P53. If the function F is satisfied and the obstacle is dynamic, calculate and compare the time when the obstacle leaves the collision and the time when the vehicle arrives at the obstacle's position, and output the second risk index data information.

[0054] P54. If the function F is not satisfied and the obstacle is static, output the data information that there is no collision risk.

[0055] P55. If the function F is not satisfied and the obstacle is dynamic, predict whether the obstacle's trajectory intersects with the vehicle's trajectory, and output the third risk index data.

[0056] Furthermore, in step P53, calculating and comparing the time it takes for the obstacle to detach from the collision and the time it takes for the vehicle to reach the obstacle's location includes:

[0057] P531. Based on the left boundary coordinate information of the obstacle (OL) si ,OL di ), calculate the moving speed v of the obstacle. 障 and the angle α between the obstacle and the horizontal.

[0058] ,

[0059] ;

[0060] P532. Based on the obstacle's moving speed v障 Calculate the time T for the obstacle to escape the collision, based on the included angle α with the horizontal plane. 脱 and the time T when the vehicle reaches the obstacle position 车 ,

[0061] ,

[0062] ,

[0063] Where v is the average speed of the vehicle;

[0064] P533. Based on the time T when the obstacle leaves the collision zone. 脱 and the time T when the vehicle reaches the obstacle position 车 If T 脱 Greater than T 车 There is no risk of collision if T 脱 Less than T 车 There is a risk of collision.

[0065] Furthermore, the average speed v of the vehicle,

[0066] .

[0067] In this embodiment, in step P55, whether the predicted trajectory of the obstacle intersects with the trajectory of the vehicle includes whether there is a collision risk if there is an intersection and whether there is no collision risk if there is no intersection.

[0068] In this embodiment, the method for determining the left and right boundary coordinate information of the obstacle is to generate the left lane boundary and right lane boundary of the current shape point based on the left and right boundary values ​​in the obstacle boundary shape point coordinate information, traverse all points of the left and right lane boundaries, calculate the distance of the obstacle from the left and right road boundaries, and calculate the coordinate information of the left and right boundaries of the obstacle in the Frenet reference line coordinate system based on the left and right boundary distances and the lane width.

[0069] In this embodiment, the method for determining the coordinate information of the front and rear boundaries of the obstacle is to first traverse the discrete points from the starting point in the reference driving road segment data information, calculate the first point where a collision occurs, which is the rear boundary of the obstacle, and then traverse the discrete points from the ending point to the starting point in the reference driving road segment data information, calculate the first point where a collision occurs, which is the coordinate of the front boundary of the obstacle.

[0070] In this embodiment, the vehicle collision risk data information includes first risk index data information, second risk index data information, third risk index data information, and no collision risk data information.

[0071] Example 2: Based on the lane information-based autonomous driving collision risk detection method in Example 1, the present invention will be further explained and described below.

[0072] like Figure 2 As shown, based on the polygonal obstacle information perceived by the autonomous driving system, the current positioning coordinate information, and the high-precision map information, the obstacle information is processed to calculate the obstacle collision risk judgment result in the current driving reference path.

[0073] The positioning information includes: coordinate information gaussX / gaussY, heading angle azimuth, etc.

[0074] The high-precision map information includes lane lines, lane center lines, lane line attributes, and intersection attributes. The lane center line information is an array of lane center point data from back to front. Each point data includes: point ID, x and y coordinates in Gaussian coordinates, longitudinal coordinate s in Frenet coordinates for the current lane, orientation of the point within the current lane, distance from the left edge of the road, distance from the right edge of the road, distance from the left edge of the current lane, and distance from the right edge of the current lane.

[0075] The perceived polygon obstacle includes: obstacle ID, obstacle label (pedestrian, vehicle, etc.), obstacle center point coordinates, obstacle speed, obstacle direction, and obstacle boundary point coordinate array, with each boundary point containing coordinate information (local or global coordinates).

[0076] First, the nearest shape point in the driving path lane is matched with the target obstacle. Based on the lane boundary information in the shape point information, the left and right boundary coordinates and front and rear boundary coordinates of the perceived polygon obstacle in the Frenet coordinate system with the right boundary of the lane as the reference are calculated. Based on the lane boundary information in the shape point information, the left and right boundary coordinates of the vehicle when it travels to this shape point in the Frenet coordinate system with the right boundary of the lane as the reference are calculated. It is determined whether it is within the range of the left and right boundaries of the corresponding obstacle. If it is within the range, there is a risk of collision. If it is not, the minimum difference between the vehicle and the left and right boundaries of the obstacle is calculated as the closest distance to the obstacle. At the same time, the speed and direction of the obstacle are considered for the planning decision to make a risk assessment of the obstacle. If the speed and direction are the same and the speed is greater than the current vehicle speed, there is no risk of collision. If the speed is 0, there is a risk of collision when the vehicle travels to the current position.

[0077] Based on the lateral collision risk of obstacles and their lateral and longitudinal relative distances, planning decisions can be made by detecting collision risks and then taking appropriate actions. For example, although there is a collision risk at the lateral position, the risk of collision with a distant obstacle is very low based on the distance. If there is no collision at the lateral position, but the lateral distance is 0.1m and the longitudinal relative distance is 10m, then although the driving trajectory will not collide, there is still a significant risk because the driving trajectory is only 0.1m from the obstacle. Planning decisions need to take appropriate actions against this obstacle to ensure driving safety.

[0078] To achieve the above and other related objectives, the present invention also provides an autonomous driving collision risk detection system based on lane information, including a computer device programmed or configured to perform the steps of any of the lane information-based autonomous driving collision risk detection methods described above.

[0079] To achieve the above and other related objectives, the present invention also provides a computer-readable storage medium storing a computer program programmed or configured to perform any of the lane information-based autonomous driving collision risk detection methods described above.

[0080] Any references to memory, storage, database, or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0081] In summary, this invention not only directly filters out obstacle information outside the lane that does not need to be processed through the Frenet coordinate system, reducing computational power consumption, but also calculates collision risk based on coordinates with very low computational load. Furthermore, the planning and generation of multiple alternative paths does not require recalculating the Frenet coordinates of obstacles; collision risk can be calculated only through the coordinates of matching points in the trajectory, greatly reducing computational power consumption.

[0082] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A collision risk detection method for autonomous driving based on lane information, characterized in that, The method includes: P1. In the global coordinate system, based on the vehicle perception system, the coordinate array information of obstacle boundary points and the coordinate data information of obstacle center point are obtained in real time. Based on the vehicle positioning system, the vehicle positioning coordinate array information is obtained in real time. Based on the high-precision map, the coordinate data information of the left boundary of the lane, the coordinate data information of the right boundary of the lane, and the coordinate array information of the discrete points of the lane centerline are obtained in real time. P2. Traverse the discrete point coordinate array information of the lane centerline, calculate the coordinate point with the shortest distance to the coordinate data information of the obstacle center point, take the coordinate point as the center point, select a fixed distance along the lane direction, and output the reference driving road segment data information; P3. Based on the reference driving road segment data information and the obstacle boundary point coordinate array information, establish the Frenet coordinate system with the right boundary coordinate data information of the lane as the reference line, and output the left and right boundary coordinate information and the front and rear boundary coordinate information of the obstacle; P4. Based on the vehicle positioning coordinate array information, establish a Frenet coordinate system with the right boundary coordinate data of the lane as the reference line, and output the left and right boundary coordinate information of the vehicle according to the vehicle width; P5. Based on the vehicle's left and right boundary coordinates, the obstacle's left and right boundary coordinates, and the obstacle's front and rear boundary coordinates, output vehicle collision risk data information according to the vehicle collision detection algorithm; In step P5, the vehicle collision detection algorithm includes: P51. Obtain the left boundary coordinate information of the vehicle (EL) si ,EL di The vehicle's right boundary coordinate information (ER) si ,ER di The left boundary coordinate information of the obstacle (OL) si ,OL di ) and the right boundary coordinate information of the obstacle (OR) si OR di ); P52. If the function F is satisfied and the obstacle is static. Where n is a time point and is a positive integer. If the vehicle is at risk of collision, output the first risk index data. P53. If the function F is satisfied and the obstacle is dynamic, calculate and compare the time when the obstacle leaves the collision and the time when the vehicle arrives at the obstacle's position, and output the second risk index data information. P54. If the function F is not satisfied and the obstacle is static, output the data information that there is no collision risk. P55. If the function F is not satisfied and the obstacle is dynamic, predict whether the obstacle's trajectory intersects with the vehicle's trajectory, and output the third risk index data. In step P53, calculating and comparing the time it takes for the obstacle to detach from the collision and the time it takes for the vehicle to reach the obstacle's location includes: P531. Based on the left boundary coordinate information of the obstacle (OL) si ,OL di ), calculate the moving speed v of the obstacle. 障 and the angle α between the obstacle and the horizontal. ， ； P532. Based on the obstacle's moving speed v 障 Calculate the time T for the obstacle to escape the collision, based on the included angle α with the horizontal plane. 脱 and the time T when the vehicle reaches the obstacle position 车 , ， ， Where v is the average speed of the vehicle; P533. Based on the time Tdetach of the obstacle leaving the collision and the time Tvelocity of the vehicle reaching the obstacle's position. 车 If T 脱 Greater than T 车 There is no risk of collision if T 脱 Less than T 车 There is a risk of collision.

2. The autonomous driving collision risk detection method based on lane information according to claim 1, characterized in that: The average speed v of the vehicle, 。 3. The autonomous driving collision risk detection method based on lane information according to claim 1, characterized in that, In step P55, whether the predicted trajectory of the obstacle intersects with the trajectory of the vehicle includes whether there is a collision risk if there is an intersection and whether there is no collision risk if there is no intersection.

4. The autonomous driving collision risk detection method based on lane information according to claim 1, characterized in that, The method for determining the left and right boundary coordinate information of the obstacle is as follows: based on the left and right boundary values ​​in the boundary point coordinate information of the obstacle, the left lane boundary and right lane boundary of the current point are generated; all points of the left and right lane boundaries are traversed; the distance of the obstacle from the left and right road boundaries is calculated; and based on the left and right boundary distances and the lane width, the coordinate information of the left and right boundaries of the obstacle in the Frenet reference line coordinate system is calculated.

5. The autonomous driving collision risk detection method based on lane information according to claim 1, characterized in that: The method for determining the coordinates of the front and rear boundaries of the obstacle is as follows: first, traverse the discrete points from the starting point in the reference driving road segment data information, and calculate the first point where a collision occurs, which is the rear boundary of the obstacle; then, traverse the discrete points from the ending point in the reference driving road segment data information back to the starting point, and calculate the first point where a collision occurs, which is the coordinates of the front boundary of the obstacle.

6. The autonomous driving collision risk detection method based on lane information according to claim 1, characterized in that: The vehicle collision risk data information includes first risk index data information, second risk index data information, third risk index data information, and no collision risk data information.

7. An autonomous driving collision risk detection system based on lane information, comprising a computer device, characterized in that, The computer device is programmed or configured to perform the steps of the lane information-based autonomous driving collision risk detection method as described in any one of claims 1 to 6.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that is programmed or configured to perform the lane information-based autonomous driving collision risk detection method according to any one of claims 1 to 6.

Images