Step type obstacle line pressing judgment method and device

Abstract

The invention discloses a stepped obstacle line pressing judgment method and device, and the method comprises the steps: obtaining a 2D rectangular frame, a 2D contour, a 3D rectangular frame, a 2D lane line and a 3D lane line of an obstacle in front of a vehicle according to a front view image shot by a camera of the vehicle, and obtaining a map lane line through a high-precision map in combination with the positioning information of the vehicle; according to the distance between the obstacle and the vehicle, the distance section interval where the obstacle is located is determined; according to the distance section interval where the obstacle is located, whether the obstacle presses the lane line or not is determined according to the position relation between the 2D rectangular frame and the 2D contour and the 2D lane line, the position relation between the 3D rectangular frame and the 3D lane line and the position relation between the 3D rectangular frame and the map lane line, and the position relation between the obstacle and the lane line can be more accurately and effectively judged; according to the method, the situation of line pressing misjudgment caused by inaccurate recognition of the course angle of the obstacle in the middle and long distance is remarkably reduced, and therefore the reliability and safety of auxiliary driving are improved.

Description

Technical Field

[0001] The present application relates to the technical field of line-pressing detection, and in particular to a method and device for determining line-pressing of a stepped obstacle. Background Art

[0002] At high speeds, the combined assisted driving system must accurately identify the location of obstacles on the road, particularly their relative position to lane markings, to avoid misjudgments that could trigger unnecessary cruise control slowdowns or braking, impacting driving comfort. Accurately identifying obstacles that overlap lane markings is crucial for ensuring driving safety and enhancing the driving experience, especially at medium and long distances, where even higher precision is required.

[0003] In related technologies, a lane-crossing model is used to obtain a first detection result of whether an object is crossing the lane. The model is then used to output the distance between the lane line and the object, and an overhead view is generated through inverse perspective transformation. A second detection result is further generated by judging the object and lane line model. Finally, a comprehensive judgment is made based on the first and second detection results to determine whether the obstacle ahead is crossing the lane line. However, this solution has significant flaws in actual applications: First, the accuracy of the heading angle (yaw) of the target object output by the model detection is insufficient in medium and long-range 3D space, which is a technical difficulty in the industry. Second, when the inverse perspective transformation is transformed into a overhead view, the accuracy of the lane line and the target object is greatly lost, especially for medium and long-range obstacles. This loss is more serious, resulting in inaccurate judgment of the relative position of the target object and the lane line, thereby increasing the misjudgment rate.

[0004] Therefore, how to judge whether the obstacle ahead is pressing the lane line is a technical problem that needs to be solved urgently. Summary of the Invention

[0005] The present application provides a stepped obstacle pressure line judgment method and device, which can solve the technical problem in the prior art that it is impossible to accurately judge whether the obstacle ahead is pressing the lane line.

[0006] In a first aspect, an embodiment of the present application provides a method for determining whether a stepped obstacle is pressing a line, the method comprising:

[0007] Based on the front view image captured by the vehicle's camera, the system obtains the 2D rectangular frame, 2D outline, and 3D rectangular frame of the obstacle in front of the vehicle, as well as the 2D lane lines and 3D lane lines. The system then obtains the lane lines on the map by combining the high-precision map with the vehicle's positioning information.

[0008] According to the distance between the obstacle and the vehicle, determine the distance segment in which the obstacle is located;

[0009] Based on the distance segment in which the obstacle is located, determine whether the obstacle is on the line based on the positional relationship between the 2D rectangular box and the 2D outline and the 2D lane line, between the 3D rectangular box and the 3D lane line, and between the 3D rectangular box and the map lane line.

[0010] In combination with the first aspect, in one embodiment, determining whether the obstacle is on the lane according to the distance segment in which the obstacle is located, and selecting the positional relationship between the 2D rectangular box and the 2D outline and the 2D lane line, between the 3D rectangular box and the 3D lane line, and between the 3D rectangular box and the map lane line, includes:

[0011] The distance interval segments include: a short distance segment, a medium distance segment and a long distance segment;

[0012] When the obstacle is in a close distance segment, determining whether the obstacle is pressing against a lane line according to a first judgment process, wherein the first judgment process includes: determining whether the obstacle is pressing against a lane line according to a positional relationship between the 2D rectangular frame and 2D outline frame of the obstacle and target 2D lane lines corresponding to target lane lines on both sides of the ego vehicle lane;

[0013] When the obstacle is in the middle distance segment, determining whether the obstacle is pressing the lane line according to a second judgment process, wherein the second judgment process includes: determining whether the obstacle is pressing the lane line based on a positional relationship between the 3D rectangular frame of the obstacle and the target 3D lane line corresponding to the target lane line, in combination with the first judgment process;

[0014] When the obstacle is at a long distance, a third judgment process is used to determine whether the obstacle is pressing the lane line, wherein the third judgment process includes: determining whether the obstacle is pressing the lane line based on the positional relationship between the 3D rectangular frame of the obstacle and the target map lane line corresponding to the target lane line, and the status of the vehicle positioning signal, in combination with the second judgment process.

[0015] In conjunction with the first aspect, in one embodiment, determining whether the obstacle presses the lane line based on the positional relationship between the 2D rectangular frame and 2D outline frame of the obstacle and the target 2D lane lines corresponding to the target lane lines on both sides of the ego vehicle lane includes:

[0016] Determine whether the 2D rectangular frame and the target 2D lane line intersect;

[0017] If the 2D rectangular frame and the target 2D lane line intersect, further determining whether the 2D outline and the target 2D lane line intersect; if so, determining that the obstacle is pressing the lane line; if not, determining that the obstacle is not pressing the lane line;

[0018] If the 2D rectangular frame and the target 2D lane line do not intersect, it is determined that the obstacle does not cross the lane line.

[0019] In combination with the first aspect, in one embodiment, determining whether the 2D rectangular frame and the target 2D lane line intersect includes:

[0020] Determine whether the 2D rectangular frame intersects the target 2D lane line, and whether a first ratio between a first area of ​​the 2D rectangular frame crossing the target 2D lane line and entering the vehicle lane and an area of ​​the 2D rectangular frame is greater than a preset first ratio threshold;

[0021] If so, determining that the 2D rectangular box and the target 2D lane line have an intersection;

[0022] If not, it is determined that the 2D rectangular frame and the target 2D lane line do not intersect.

[0023] In combination with the first aspect, in one embodiment, determining whether the 2D contour and the target 2D lane line intersect includes:

[0024] Determining whether the 2D contour intersects the target 2D lane line, and whether a second ratio between a second area of ​​the 2D contour where the 2D contour crosses the target 2D lane line and enters the vehicle lane and an area of ​​the 2D contour is greater than a preset second ratio threshold;

[0025] If yes, determining that the 2D contour and the target 2D lane line have an intersection;

[0026] If not, it is determined that the 2D contour does not intersect the target 2D lane line.

[0027] In combination with the first aspect, in one embodiment, determining whether the 2D contour intersects the target 2D lane line includes:

[0028] determining a bounding box of the 2D contour based on coordinates of each vertex of the 2D contour in the vehicle image coordinate system, wherein the bounding box is constructed based on a maximum horizontal coordinate, a minimum horizontal coordinate, a maximum vertical coordinate, and a minimum vertical coordinate of all vertices of the 2D contour;

[0029] Determining in sequence whether each straight line forming the 2D contour intersects the target 2D lane line within the bounding box;

[0030] If any two straight lines forming the 2D contour and the target 2D lane line have an intersection within the bounding box, it is determined that the 2D contour intersects the target 2D lane line; otherwise, it is determined whether the 2D contour intersects the target 2D lane line.

[0031] In conjunction with the first aspect, in one embodiment, determining whether the obstacle is pressing the lane based on the positional relationship between the 3D rectangular frame of the obstacle and the target 3D lane line corresponding to the target lane line, in combination with the first judgment process, includes:

[0032] Determining whether the 3D rectangular frame intersects the target 3D lane line, and whether a third ratio between a third area of ​​the 3D rectangular frame where the 3D rectangular frame crosses the target 3D lane line and enters the vehicle lane and the area of ​​the 3D rectangular frame is greater than a preset third ratio threshold;

[0033] If yes, then determining that the obstacle is pressing the lane line;

[0034] Otherwise, the first judgment process is executed to determine whether the obstacle is pressing the line.

[0035] In conjunction with the first aspect, in one embodiment, determining whether the obstacle presses the lane line based on the positional relationship between the 3D rectangular frame of the obstacle and the target map lane line corresponding to the target lane line, the state of the ego-vehicle positioning signal, and the second judgment process includes:

[0036] determining whether the 3D rectangular frame intersects the target map lane line, whether a fourth ratio between a fourth area of ​​the 3D rectangular frame that crosses the target map lane line and enters the ego vehicle lane and the area of ​​the 3D rectangular frame is greater than a preset fourth ratio threshold, and whether the ego vehicle positioning signal is in a valid state;

[0037] If yes, then determining that the obstacle is pressing the lane line;

[0038] Otherwise, executing the second judgment process to determine whether the obstacle is pressing the line;

[0039] The values ​​of the first ratio threshold, the second ratio threshold, the third ratio threshold and the fourth ratio threshold increase in sequence.

[0040] In conjunction with the first aspect, in one embodiment, the method further includes:

[0041] If the obstacle is in the left lane adjacent to the own vehicle lane, the left lane marking of the own vehicle lane is determined as the target lane marking;

[0042] If the obstacle is in the right lane adjacent to the own vehicle lane, the right lane marking of the own vehicle lane is determined as the target lane marking;

[0043] If the obstacle is in the own vehicle lane, the left lane marking and the right lane marking of the own vehicle lane are determined as the target lane markings.

[0044] In a second aspect, an embodiment of the present application provides a stepped obstacle line-pressing judgment device, the stepped obstacle line-pressing judgment device comprising:

[0045] The acquisition module is used to obtain the 2D rectangular frame, 2D outline, and 3D rectangular frame of the obstacle in front of the vehicle based on the front view image captured by the vehicle's camera, as well as the 2D lane lines and 3D lane lines. It also obtains the map lane lines by combining the high-precision map with the vehicle's positioning information.

[0046] A first determination module is used to determine the distance segment in which the obstacle is located based on the distance between the obstacle and the vehicle;

[0047] The second determination module is used to determine whether the obstacle is on the line based on the positional relationship between the 2D rectangular box and the 2D outline and the 2D lane line, between the 3D rectangular box and the 3D lane line, and between the 3D rectangular box and the map lane line according to the distance segment in which the obstacle is located.

[0048] An embodiment of the present application provides a stepped obstacle crossing line judgment method and device, which obtains the 2D rectangular frame, 2D outline and 3D rectangular frame of the obstacle in front of the own vehicle based on the front view image taken by the own vehicle camera, as well as the 2D lane line and 3D lane line, and obtains the map lane line through the combination of high-precision map and own vehicle positioning information; determines the distance segment interval in which the obstacle is located according to the distance between the obstacle and the own vehicle; and determines whether the obstacle crosses the line based on the positional relationship between the 2D rectangular frame and 2D outline and the 2D lane line, between the 3D rectangular frame and the 3D lane line, and between the 3D rectangular frame and the map lane line according to the distance segment interval in which the obstacle is located. This method can more accurately and effectively judge the positional relationship between the obstacle and the lane line, thereby accurately judging whether the obstacle crosses the line. This method significantly reduces the misjudgment of crossing the line caused by inaccurate obstacle heading angle recognition at medium and long distances, thereby improving the reliability and safety of assisted driving. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a flow chart of an embodiment of the step-by-step obstacle line-crossing determination method of the present application;

[0050] Figure 2 A schematic diagram of the target lane line corresponding to the obstacle in this application;

[0051] Figure 3 This is a schematic diagram of the specific process of the step-by-step obstacle pressure line judgment method of this application;

[0052] Figure 4 Schematic diagram of the intersection of a 2D rectangular box and a 2D lane line;

[0053] Figure 5Schematic diagram of the intersection of 2D contour and 2D lane line;

[0054] Figure 6 This is a functional module diagram of an embodiment of a stepped obstacle line-crossing judgment device of the present application. DETAILED DESCRIPTION

[0055] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 this application.

[0056] In order to make the objectives, technical solutions and advantages of this application clearer, the implementation methods of this application will be further described in detail below with reference to the accompanying drawings.

[0057] In a first aspect, an embodiment of the present application provides a method for determining whether a stepped obstacle is pressing a line.

[0058] In one embodiment, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the step-by-step obstacle line-pressing judgment method of this application. Figure 1 As shown, the method for judging whether a stepped obstacle is pressing the line includes:

[0059] Step S1: Based on the front view image captured by the vehicle's camera, obtain the 2D rectangular frame, 2D outline, and 3D rectangular frame of the obstacle in front of the vehicle, as well as the 2D lane lines and 3D lane lines, and obtain the map lane lines by combining the high-precision map with the vehicle's positioning information.

[0060] For example, the forward-view image captured by the ego vehicle's camera is first input into a pre-set perception model, which then generates 2D and 3D image spatial attributes. 2D image spatial attributes include 2D rectangles, 2D outlines, and 2D lane lines; 3D image spatial attributes include 3D rectangles and 3D lane lines. Simultaneously, map lane lines are obtained by combining a high-precision map with the ego vehicle's positioning information. The obtained 2D rectangles, 2D outlines, 2D lane lines, 3D rectangles, 3D lane lines, and map lane lines are then converted to the ego vehicle's image coordinate system.

[0061] It's worth noting that the perception model in this embodiment is a trained algorithmic model, typically based on deep learning technology. It analyzes and understands input images, identifying various objects and features within them. The perception model is trained using a large amount of annotated data, enabling it to learn how to extract and output obstacle and lane information from images.

[0062] A 2D rectangular box is a simple representation of an obstacle in a front-view image. It is a rectangular area on a two-dimensional plane. The coordinates of the four corner points of the rectangular box determine the position and size of the obstacle in the image.

[0063] The 2D contour is a more accurate description of the edge of an obstacle. It is a description of the edge of an object composed of a series of continuous two-dimensional points, which are connected to form a contour shape. It reflects the shape characteristics of the object more accurately than a 2D rectangular box.

[0064] 2D lane lines are two-dimensional lines that represent lane lines in an image. The perception model can detect the position and shape of these lane lines by analyzing the image and output them in the form of two-dimensional lines.

[0065] A 3D rectangle represents the position and size of an object in three-dimensional space. It is a three-dimensional rectangle that contains the object's position, size, and orientation in the vehicle coordinate system. Compared to a 2D rectangle, a 3D rectangle provides richer spatial information about the object.

[0066] 3D lane lines are a representation of lane lines in three-dimensional space. They not only include the two-dimensional position information of the lane lines in the image, but also take into account the three-dimensional spatial position of the lane lines on the road.

[0067] Step S2: Determine the distance segment in which the obstacle is located based on the distance between the obstacle and the vehicle.

[0068] Exemplarily, the distance segment of the obstacle is determined based on the distance between the obstacle and the ego vehicle in 3D space output by the perception model. In this embodiment, when the distance between the obstacle and the ego vehicle is greater than or equal to 0 and less than 20 meters, that is, [0, 20m), the obstacle is determined to be in the close distance segment; when the distance between the obstacle and the ego vehicle is greater than or equal to 20 and less than 80 meters, [20, 80m), the obstacle is determined to be in the medium distance segment; and when the distance between the obstacle and the ego vehicle is greater than or equal to 80 meters, [80, Max), the obstacle is determined to be in the long distance segment.

[0069] It's worth noting that the range of the distance interval can be dynamically adjusted based on the performance of the perception model. Because the upper limit of Max perception and obstacle recognition is generally between 100 and 200 meters, affected by lens resolution and algorithm performance, the value of Max does not affect the judgment logic of this application.

[0070] Step S3: Based on the distance segment in which the obstacle is located, determine whether the obstacle is on the lane based on the positional relationship between the 2D rectangular box and the 2D outline and the 2D lane line, between the 3D rectangular box and the 3D lane line, and between the 3D rectangular box and the map lane line.

[0071] Preferably, before determining whether the obstacle is crossing the lane, a target lane line is first determined. The method for determining the target lane line includes: if the obstacle is in the left lane adjacent to the vehicle's lane, determining the left lane line of the vehicle's lane as the target lane line; if the obstacle is in the right lane adjacent to the vehicle's lane, determining the right lane line of the vehicle's lane as the target lane line; if the obstacle is in the vehicle's lane, determining both the left and right lane lines of the vehicle's lane as the target lane lines.

[0072] For example, the lane where the obstacle is located is determined based on its lateral coordinates in the vehicle's spatial coordinate system. Figure 2 As shown, the obstacles are target vehicles 1, 2, and 3. AD vehicles represent the ego vehicle with combined assisted driving. Target vehicle 1 is in the left adjacent lane, target vehicle 2 is the vehicle directly ahead, i.e., in the ego vehicle lane, and target vehicle 3 is in the right adjacent lane. To determine whether target vehicle 1 has crossed the lane, lane line l2 is used as the target lane line, and a determination is made as to whether target vehicle 1 and lane line l2 intersect. To determine whether target vehicle 3 has crossed the lane, lane line l3 is used as the target lane line, and a determination is made as to whether target vehicle 3 and lane line l3 intersect. For target vehicle 2, lane lines l2 and l3 are used as the target lane lines, and a determination is made as to whether target vehicle 2 intersects with lane lines l2 and l3. The determination rules are logically consistent with those for target vehicle 1 or target vehicle 3.

[0073] In one embodiment, if Figure 3 As shown, step S3 specifically includes the following steps:

[0074] Step S301: When the obstacle is in a close-range segment, determine whether the obstacle is pressing against the lane line according to a first judgment process, wherein the first judgment process includes: determining whether the obstacle is pressing against the lane line based on the positional relationship between the obstacle's 2D rectangular frame and 2D contour frame and the target 2D lane lines corresponding to the target lane lines on both sides of the vehicle's lane.

[0075] Specifically, the first judgment process: based on the positional relationship between the obstacle's 2D rectangular frame and 2D outline frame and the target 2D lane lines corresponding to the target lane lines on both sides of the vehicle's lane, determines whether the obstacle is pressing the lane line, specifically including:

[0076] Step S3011: Determine whether the 2D rectangular frame and the target 2D lane line have an intersection.

[0077] Step S3012: If the 2D rectangular frame and the target 2D lane line intersect, further determine whether the 2D outline and the target 2D lane line intersect; if the 2D outline and the target 2D lane line intersect, determine that the obstacle is pressing the lane line; if not, determine that the obstacle is not pressing the lane line.

[0078] Step S3013: If the 2D rectangular frame and the target 2D lane line do not intersect, it is determined that the obstacle does not press the lane line.

[0079] Among them, judging whether the 2D rectangular box and the target 2D lane line have an intersection includes: judging whether the 2D rectangular box and the target 2D lane line intersect, and whether a first ratio between a first area of ​​the 2D rectangular box crossing the target 2D lane line and entering the own vehicle lane and an area of ​​the 2D rectangular box is greater than a preset first ratio threshold; if so, determining that the 2D rectangular box and the target 2D lane line have an intersection; if not, determining that the 2D rectangular box and the target 2D lane line do not have an intersection.

[0080] Exemplary, such as Figure 4 As shown, take the example of judging whether the target vehicle 3 is crossing the lane line l3. The 2D rectangular frame of the obstacle is a polygon ABCD , whose area is S ABCD , polygon ABCD Intersecting with the target 2D lane line (red curve) at points E and F, the polygon ABCD The first area of ​​the vehicle that crosses the target 2D lane line and enters the vehicle lane is S BEF Then the first ratio between the first area and the area of ​​the 2D rectangular frame is d1=S BEF / S ABCD .

[0081] In calculating the first area S BEF When , the curve EF can be approximated as a straight line (green straight line) between the two points EF for calculation. In this embodiment, the first ratio threshold is set to 0.2, and the first ratio threshold can be dynamically adjusted according to the actual scenario. Therefore, when the 2D rectangular box intersects with the target 2D lane line and the first ratio d1>0.2, it is determined that the two have an intersection, otherwise there is no intersection.

[0082] Among them, judging whether the 2D contour and the target 2D lane line have an intersection includes: judging whether the 2D contour and the target 2D lane line intersect, and whether a second ratio between a second area of ​​the 2D contour crossing the target 2D lane line and entering the own vehicle lane and the area of ​​the 2D contour is greater than a preset second ratio threshold; if so, determining that the 2D contour and the target 2D lane line have an intersection; if not, determining that the 2D contour and the target 2D lane line do not have an intersection.

[0083] Exemplary, such as Figure 5 As shown, take the example of judging whether the target vehicle 3 is crossing the lane line l3. The 2D outline of the obstacle is a polygon BCDEG , whose area is S BCDEG , polygon BCDEGIntersecting with the target 2D lane line (red curve) at points A and F, the polygon BCDEG The second area of ​​the lane that crosses the target 2D lane line and enters the vehicle lane is S ABCF Then the first ratio between the second area and the area of ​​the 2D rectangular frame is d2=S ABCF / S BCDEG .

[0084] In calculating the second area S ABCF When the curve EF is approximated as the straight line (green straight line) between the two points AF, the calculation is performed. In this embodiment, the second ratio threshold is set to 0.05, and the second ratio threshold can be dynamically adjusted according to the actual scenario. When the 2D contour intersects the target 2D lane line and the second ratio d2>0.2, it is determined that the two have an intersection; otherwise, there is no intersection.

[0085] In one embodiment, determining whether the 2D contour intersects with the target 2D lane line includes the following steps: determining a bounding box of the 2D contour based on the coordinates of each vertex of the 2D contour in the vehicle image coordinate system, wherein the bounding box is constructed based on the maximum lateral coordinate, the minimum lateral coordinate, the maximum longitudinal coordinate, and the minimum longitudinal coordinate of all vertices of the 2D contour; sequentially determining whether each straight line forming the 2D contour and the target 2D lane line have an intersection within the bounding box; if any two straight lines forming the 2D contour and the target 2D lane line have an intersection within the bounding box, determining whether the 2D contour intersects with the target 2D lane line; otherwise, determining whether the 2D contour intersects with the target 2D lane line.

[0086] For example, the polygon corresponding to the 2D contour BCDEG In the ego-vehicle image coordinate system, the bounding box of the 2D contour is constructed based on the maximum horizontal coordinate, minimum horizontal coordinate, maximum vertical coordinate and minimum vertical coordinate of the five points BCDEG: Sk={[x min ,x max ],[y min ,y max ]}.in:

[0087] x min =min(x B ,x C ,x D ,x E ,x G )

[0088] x max =max(x B ,x C ,x D ,x E ,x G)

[0089] y min =min(y B ,y C ,y D ,y E ,y G )

[0090] y max =max(y,y C ,y D ,y E ,y G )

[0091] Among them, Sk is the bounding box of the 2D contour, x max is the maximum horizontal coordinate, x min is the minimum horizontal coordinate, y max is the maximum vertical coordinate, y min If the target 2D lane line intersects the 2D contour, the target lane line must be within the bounding box.

[0092] The target 2D lane line is a cubic polynomial:

[0093] y=a0x 3 +b0x 2 +c0x+d

[0094] Each pair of points in the 2D contour polygon BCDEG can generate a line: y = kx + b. Each line is determined to intersect the target 2D lane line and to determine if it is within the bounding box. If not, the intersection does not meet the requirements. If it does, the intersection is retained. If two intersection points are retained, such as points A and F, the 2D contour is determined to intersect the target 2D lane line.

[0095] In an optional implementation, the shoelace formula can be used to calculate the area of ​​a polygon. Taking the area of ​​polygon BCDEG as an example, its vertices are sorted clockwise, and the area calculation formula is as follows:

[0096]

[0097] Among them, the coordinates (x i ,y i ) represent the vertex coordinates, and n is the number of vertices.

[0098] Regarding the intersection of the lane line and the polygon, such as points A and F shown above, we can approximate points A and F as a straight line and use the shoelace formula to calculate the polygon area S. ABCF .

[0099] It is worth noting that in this embodiment, at close distances, whether an obstacle is on the lane is determined by the positional relationship between the obstacle's 2D polygonal outline and the target 2D lane line. This not only provides a more refined geometric feature analysis based on the 2D rectangular frame, but also can more accurately identify and determine the relative position of the obstacle and the lane line, thereby significantly improving the accuracy of the judgment and the reliability of the results.

[0100] Step S302: When the obstacle is in the middle distance segment, determine whether the obstacle is pressing the lane line according to a second judgment process, wherein the second judgment process includes: determining whether the obstacle is pressing the lane line based on the positional relationship between the 3D rectangular frame of the obstacle and the target 3D lane line corresponding to the target lane line, in combination with the first judgment process.

[0101] Specifically, the second step of the judgment process includes the following steps:

[0102] Step S3021: Determine whether the 3D rectangular frame intersects the target 3D lane line, and whether a third ratio between a third area of ​​the 3D rectangular frame that crosses the target 3D lane line and enters the vehicle lane and the area of ​​the 3D rectangular frame is greater than a preset third ratio threshold.

[0103] Step S3022: If yes, determine that the obstacle is pressing the lane line.

[0104] Step S3023: Otherwise, execute the first judgment process to determine whether the obstacle is pressing the line.

[0105] It is worth noting that, in order to facilitate the calculation of the third ratio, this embodiment projects the 3D rectangular box onto a 2D box under the perspective of a top view, and calculates the intersection ratio of the 2D box and the lane line under the perspective of BEV (Bird's Eye View). The calculation method is the same as the calculation method of the first ratio, except that the perspective is different (projection under BEV three-dimensional space). Because the stability of the 3D rectangular box and the 3D lane line in the distance segment output by the perception model is better, the third ratio threshold is set to 0.1 here, and the third ratio threshold can be dynamically adjusted according to the actual scenario. If the third ratio d3 is greater than the third ratio threshold of 0.1, it is determined that the obstacle is pressing the lane line, otherwise, enter the first judgment process and continue to judge according to steps S3011 to S3013.

[0106] Step S303: Determine whether the obstacle is pressing against the lane line based on the positional relationship between the 3D rectangular frame of the obstacle and the target lane line corresponding to the target lane line, the state of the vehicle positioning signal, and the second judgment process.

[0107] Specifically, the third judgment process includes the following steps:

[0108] Step S3031: Determine whether the 3D rectangular frame intersects with the target map lane line, whether a fourth ratio between a fourth area of ​​the 3D rectangular frame that crosses the target map lane line and enters the vehicle lane and the area of ​​the 3D rectangular frame is greater than a preset fourth ratio threshold, and whether the vehicle positioning signal is in a valid state.

[0109] Step S3032: If yes, determine that the obstacle is pressing the lane line.

[0110] Step S3033: Otherwise, execute the second judgment process to determine whether the obstacle is pressing the line;

[0111] The values ​​of the first ratio threshold, the second ratio threshold, the third ratio threshold and the fourth ratio threshold increase in sequence.

[0112] It's worth noting that if the obstacle is too far from the vehicle's longitudinal distance, the vehicle's camera's ability to identify lane markings may be limited to the obstacle's longitudinal distance, and accurate 3D lane markings may not be obtained. In this case, the third judgment process is used to determine if the obstacle is crossing the lane. If the vehicle's positioning signal is normal, the 3D rectangles representing the lane markings on the map and the obstacle are used to determine if the obstacle is crossing the lane. If the vehicle's positioning signal fails, accurate lane markings on the map cannot be obtained, and the second judgment process is used to determine if the obstacle is crossing the lane.

[0113] In the third judgment process, the calculation method of the fourth ratio is the same as the calculation method of the third ratio, which is to project the 3D rectangular box onto the 2D box under the top view perspective, and calculate the intersection ratio of the 2D box and the lane line under the BEV (Bird's Eye View) perspective. Because the perception model involves inaccurate obstacle heading angles at long distances, the fourth threshold is set to 0.3 here, and the fourth ratio threshold can be dynamically adjusted according to the actual scenario. If the fourth ratio d4 is greater than the fourth ratio threshold of 0.3, it is determined that the obstacle is pressing the lane line; otherwise, enter the second judgment process and continue to judge according to steps S3021 to S3023.

[0114] An embodiment of the present application provides a stepped obstacle crossing line judgment method, which divides obstacles into different distance segments according to the distance between the obstacle and the vehicle, and adopts a different judgment process for each distance segment. In the short-distance segment, by combining the positional relationship between the 2D rectangular box and the 2D contour box and the target 2D lane line, the relative position of the obstacle and the lane line can be more accurately identified and judged; in the medium-distance and long-distance segments, the positional relationship between the 3D rectangular box and the 3D lane line or the map lane line is used for judgment, and combined with the status of the vehicle positioning signal, it effectively reduces the misjudgment caused by inaccurate obstacle heading angle recognition, especially in the medium and long-distance segments, and significantly improves the reliability of judgment.

[0115] In a second aspect, an embodiment of the present application also provides a stepped obstacle line-pressing judgment device.

[0116] In one embodiment, referring to Figure 6 , Figure 6 This is a functional module diagram of an embodiment of the step-type obstacle pressure line judgment device of this application. Figure 6 As shown, the stepped obstacle line-pressing judgment device includes:

[0117] The acquisition module is used to obtain the 2D rectangular frame, 2D outline, and 3D rectangular frame of the obstacle in front of the vehicle based on the front view image captured by the vehicle's camera, as well as the 2D lane lines and 3D lane lines. It also obtains the map lane lines by combining the high-precision map with the vehicle's positioning information.

[0118] A first determination module is used to determine the distance segment in which the obstacle is located based on the distance between the obstacle and the vehicle;

[0119] The second determination module is used to determine whether the obstacle is on the line based on the positional relationship between the 2D rectangular box and the 2D outline and the 2D lane line, between the 3D rectangular box and the 3D lane line, and between the 3D rectangular box and the map lane line according to the distance segment in which the obstacle is located.

[0120] Furthermore, in one embodiment, the second determining module is further configured to:

[0121] The distance interval segments include: a short distance segment, a medium distance segment and a long distance segment;

[0122] When the obstacle is in a close distance segment, determining whether the obstacle is pressing against a lane line according to a first judgment process, wherein the first judgment process includes: determining whether the obstacle is pressing against a lane line according to a positional relationship between the 2D rectangular frame and 2D outline frame of the obstacle and target 2D lane lines corresponding to target lane lines on both sides of the ego vehicle lane;

[0123] When the obstacle is in the middle distance segment, determining whether the obstacle is pressing the lane line according to a second judgment process, wherein the second judgment process includes: determining whether the obstacle is pressing the lane line based on a positional relationship between the 3D rectangular frame of the obstacle and the target 3D lane line corresponding to the target lane line, in combination with the first judgment process;

[0124] When the obstacle is at a long distance, a third judgment process is used to determine whether the obstacle is pressing the lane line, wherein the third judgment process includes: determining whether the obstacle is pressing the lane line based on the positional relationship between the 3D rectangular frame of the obstacle and the target map lane line corresponding to the target lane line, and the status of the vehicle positioning signal, in combination with the second judgment process.

[0125] Furthermore, in one embodiment, the second determining module is further configured to:

[0126] Determine whether the 2D rectangular frame and the target 2D lane line intersect;

[0127] If the 2D rectangular frame and the target 2D lane line intersect, further determining whether the 2D outline and the target 2D lane line intersect; if so, determining that the obstacle is pressing the lane line; if not, determining that the obstacle is not pressing the lane line;

[0128] If the 2D rectangular frame and the target 2D lane line do not intersect, it is determined that the obstacle does not cross the lane line.

[0129] Furthermore, in one embodiment, the second determining module is further configured to:

[0130] Determine whether the 2D rectangular frame intersects the target 2D lane line, and whether a first ratio between a first area of ​​the 2D rectangular frame crossing the target 2D lane line and entering the vehicle lane and an area of ​​the 2D rectangular frame is greater than a preset first ratio threshold;

[0131] If so, determining that the 2D rectangular box and the target 2D lane line have an intersection;

[0132] If not, it is determined that the 2D rectangular frame and the target 2D lane line do not intersect.

[0133] Furthermore, in one embodiment, the second determining module is further configured to:

[0134] Determining whether the 2D contour intersects the target 2D lane line, and whether a second ratio between a second area of ​​the 2D contour where the 2D contour crosses the target 2D lane line and enters the vehicle lane and an area of ​​the 2D contour is greater than a preset second ratio threshold;

[0135] If yes, determining that the 2D contour and the target 2D lane line have an intersection;

[0136] If not, it is determined that the 2D contour does not intersect the target 2D lane line.

[0137] Furthermore, in one embodiment, the second determining module is further configured to:

[0138] determining a bounding box of the 2D contour based on coordinates of each vertex of the 2D contour in the vehicle image coordinate system, wherein the bounding box is constructed based on a maximum horizontal coordinate, a minimum horizontal coordinate, a maximum vertical coordinate, and a minimum vertical coordinate of all vertices of the 2D contour;

[0139] Determining in sequence whether each straight line forming the 2D contour intersects the target 2D lane line within the bounding box;

[0140] If any two straight lines forming the 2D contour and the target 2D lane line have an intersection within the bounding box, it is determined that the 2D contour intersects the target 2D lane line; otherwise, it is determined whether the 2D contour intersects the target 2D lane line.

[0141] Furthermore, in one embodiment, the second determining module is further configured to:

[0142] Determining whether the 3D rectangular frame intersects the target 3D lane line, and whether a third ratio between a third area of ​​the 3D rectangular frame where the 3D rectangular frame crosses the target 3D lane line and enters the vehicle lane and the area of ​​the 3D rectangular frame is greater than a preset third ratio threshold;

[0143] If yes, then determining that the obstacle is pressing the lane line;

[0144] Otherwise, the first judgment process is executed to determine whether the obstacle is pressing the line.

[0145] Furthermore, in one embodiment, the second determining module is further configured to:

[0146] determining whether the 3D rectangular frame intersects the target map lane line, whether a fourth ratio between a fourth area of ​​the 3D rectangular frame that crosses the target map lane line and enters the ego vehicle lane and the area of ​​the 3D rectangular frame is greater than a preset fourth ratio threshold, and whether the ego vehicle positioning signal is in a valid state;

[0147] If yes, then determining that the obstacle is pressing the lane line;

[0148] Otherwise, executing the second judgment process to determine whether the obstacle is pressing the line;

[0149] The values ​​of the first ratio threshold, the second ratio threshold, the third ratio threshold and the fourth ratio threshold increase in sequence.

[0150] Furthermore, in one embodiment, the first determining module is further configured to:

[0151] If the obstacle is in the left lane adjacent to the own vehicle lane, the left lane marking of the own vehicle lane is determined as the target lane marking;

[0152] If the obstacle is in the right lane adjacent to the own vehicle lane, the right lane marking of the own vehicle lane is determined as the target lane marking;

[0153] If the obstacle is in the own vehicle lane, the left lane marking and the right lane marking of the own vehicle lane are determined as the target lane markings.

[0154] Among them, the functional implementation of each module in the above-mentioned stepped obstacle pressure line judgment device corresponds to the various steps in the above-mentioned stepped obstacle pressure line judgment method embodiment, and its functions and implementation processes are no longer repeated here.

[0155] It should be noted that the serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0156] The terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices. The terms "first", "second" and "third" are used to distinguish different objects, etc., and do not represent a sequence, nor do they limit the "first", "second" and "third" to different types.

[0157] In the description of the embodiments of this application, the words "exemplary," "for example," or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary," "for example," or "for example" in the embodiments of this application should not be construed as being preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary," "for example," or "for example" is intended to present the relevant concepts in a concrete manner.

[0158] In the description of the embodiments of the present application, unless otherwise specified, “ / ” means or, for example, A / B can mean A or B; “and / or” in the text is merely a description of the association relationship of associated objects, indicating that three relationships may exist, for example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the description of the embodiments of the present application, “multiple” refers to two or more than two.

[0159] In some processes described in the embodiments of the present application, multiple operations or steps are included that appear in a specific order. However, it should be understood that these operations or steps may not be performed in the order in which they appear in the embodiments of the present application or may be performed in parallel. The sequence numbers of the operations are only used to distinguish between different operations, and the sequence numbers themselves do not represent any order of execution. In addition, these processes may include more or fewer operations, and these operations or steps may be performed in sequence or in parallel, and these operations or steps may be combined.

[0160] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform, of course, it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for enabling a terminal device to execute the methods described in each embodiment of the present application.

[0161] The above are only preferred embodiments of the present application and do not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for determining if a step-by-step obstacle is pressing a line, characterized in that: The step-type obstacle pressure line judgment method includes: Based on the front view image captured by the vehicle's camera, the system obtains the 2D rectangular frame, 2D outline, and 3D rectangular frame of the obstacle in front of the vehicle, as well as the 2D lane lines and 3D lane lines. The system then obtains the lane lines on the map by combining the high-precision map with the vehicle's positioning information. According to the distance between the obstacle and the vehicle, determine the distance segment in which the obstacle is located; Based on the distance segment in which the obstacle is located, determine whether the obstacle is on the line based on the positional relationship between the 2D rectangular box and the 2D outline and the 2D lane line, between the 3D rectangular box and the 3D lane line, and between the 3D rectangular box and the map lane line.

2. The method for determining if a stepped obstacle is pressing a line according to claim 1, wherein: The determining whether the obstacle is on the lane according to the positional relationship between the 2D rectangular box and the 2D outline and the 2D lane line, between the 3D rectangular box and the 3D lane line, and between the 3D rectangular box and the map lane line, based on the distance segment in which the obstacle is located, includes: The distance interval segments include: a short distance segment, a medium distance segment and a long distance segment; When the obstacle is in a close distance segment, determining whether the obstacle is pressing against a lane line according to a first judgment process, wherein the first judgment process includes: determining whether the obstacle is pressing against a lane line according to a positional relationship between the 2D rectangular frame and 2D outline frame of the obstacle and target 2D lane lines corresponding to target lane lines on both sides of the ego vehicle lane; When the obstacle is in the middle distance segment, determining whether the obstacle is pressing the lane line according to a second judgment process, wherein the second judgment process includes: determining whether the obstacle is pressing the lane line based on a positional relationship between the 3D rectangular frame of the obstacle and the target 3D lane line corresponding to the target lane line, in combination with the first judgment process; When the obstacle is at a long distance, a third judgment process is used to determine whether the obstacle is pressing the lane line, wherein the third judgment process includes: determining whether the obstacle is pressing the lane line based on the positional relationship between the 3D rectangular frame of the obstacle and the target map lane line corresponding to the target lane line, and the status of the vehicle positioning signal, in combination with the second judgment process.

3. The method for determining if a stepped obstacle is pressing a line according to claim 2, wherein: The determining whether the obstacle presses the lane line according to the positional relationship between the 2D rectangular frame and the 2D contour frame of the obstacle and the target 2D lane lines corresponding to the target lane lines on both sides of the ego vehicle lane includes: Determine whether the 2D rectangular frame and the target 2D lane line intersect; If the 2D rectangular frame and the target 2D lane line intersect, further determining whether the 2D outline and the target 2D lane line intersect; if so, determining that the obstacle is pressing the lane line; if not, determining that the obstacle is not pressing the lane line; If the 2D rectangular frame and the target 2D lane line do not intersect, it is determined that the obstacle does not press the lane line.

4. The method for determining if a stepped obstacle is pressing a line according to claim 3, wherein: The determining whether the 2D rectangular frame and the target 2D lane line intersect includes: Determine whether the 2D rectangular frame intersects the target 2D lane line, and whether a first ratio between a first area of ​​the 2D rectangular frame crossing the target 2D lane line and entering the vehicle lane and an area of ​​the 2D rectangular frame is greater than a preset first ratio threshold; If so, determining that the 2D rectangular box and the target 2D lane line have an intersection; If not, it is determined that the 2D rectangular frame and the target 2D lane line do not intersect.

5. The method for determining if a stepped obstacle is crossing a line according to claim 4, wherein: Determining whether the 2D contour and the target 2D lane line intersect includes: Determining whether the 2D contour intersects the target 2D lane line, and whether a second ratio between a second area of ​​the 2D contour where the 2D contour crosses the target 2D lane line and enters the vehicle lane and an area of ​​the 2D contour is greater than a preset second ratio threshold; If yes, determining that the 2D contour and the target 2D lane line have an intersection; If not, it is determined that the 2D contour does not intersect the target 2D lane line.

6. The method for determining if a stepped obstacle is crossing a line according to claim 5, wherein: The determining whether the 2D contour intersects the target 2D lane line includes: determining a bounding box of the 2D contour based on coordinates of each vertex of the 2D contour in the vehicle image coordinate system, wherein the bounding box is constructed based on a maximum horizontal coordinate, a minimum horizontal coordinate, a maximum vertical coordinate, and a minimum vertical coordinate of all vertices of the 2D contour; Determining in sequence whether each straight line forming the 2D contour intersects the target 2D lane line within the bounding box; If any two straight lines forming the 2D contour and the target 2D lane line have an intersection within the bounding box, it is determined that the 2D contour intersects the target 2D lane line; otherwise, it is determined whether the 2D contour intersects the target 2D lane line.

7. The method for determining if a stepped obstacle is crossing a line according to claim 5, wherein: The determining whether the obstacle is pressing the lane based on the positional relationship between the 3D rectangular frame of the obstacle and the target 3D lane line corresponding to the target lane line in combination with the first judgment process includes: Determining whether the 3D rectangular frame intersects the target 3D lane line, and whether a third ratio between a third area of ​​the 3D rectangular frame where the 3D rectangular frame crosses the target 3D lane line and enters the vehicle lane and the area of ​​the 3D rectangular frame is greater than a preset third ratio threshold; If yes, then determining that the obstacle is pressing the lane line; Otherwise, the first judgment process is executed to determine whether the obstacle is pressing the line.

8. The method for determining if a stepped obstacle is crossing a line according to claim 7, wherein: The determining whether the obstacle is pressing the lane line based on the positional relationship between the 3D rectangular frame of the obstacle and the target map lane line corresponding to the target lane line, the state of the vehicle positioning signal, and the second judgment process includes: determining whether the 3D rectangular frame intersects the target map lane line, whether a fourth ratio between a fourth area of ​​the 3D rectangular frame that crosses the target map lane line and enters the ego vehicle lane and the area of ​​the 3D rectangular frame is greater than a preset fourth ratio threshold, and whether the ego vehicle positioning signal is in a valid state; If yes, then determining that the obstacle is pressing the lane line; Otherwise, executing the second judgment process to determine whether the obstacle is pressing the line; The values ​​of the first ratio threshold, the second ratio threshold, the third ratio threshold and the fourth ratio threshold increase in sequence.

9. The method for determining if a stepped obstacle is crossing a line according to claim 2, wherein: The method further includes: If the obstacle is in the left lane adjacent to the own vehicle lane, the left lane marking of the own vehicle lane is determined as the target lane marking; If the obstacle is in the right lane adjacent to the own vehicle lane, the right lane marking of the own vehicle lane is determined as the target lane marking; If the obstacle is in the own vehicle lane, the left lane marking and the right lane marking of the own vehicle lane are determined as the target lane markings.

10. A stepped obstacle line-pressing judgment device, characterized in that: The step-type obstacle pressure line judgment device includes: The acquisition module is used to obtain the 2D rectangular frame, 2D outline, and 3D rectangular frame of the obstacle in front of the vehicle based on the front view image captured by the vehicle's camera, as well as the 2D lane lines and 3D lane lines. It also obtains the map lane lines by combining the high-precision map with the vehicle's positioning information. A first determination module is used to determine the distance segment in which the obstacle is located based on the distance between the obstacle and the vehicle; The second determination module is used to determine whether the obstacle is on the line based on the positional relationship between the 2D rectangular box and the 2D outline and the 2D lane line, between the 3D rectangular box and the 3D lane line, and between the 3D rectangular box and the map lane line according to the distance segment in which the obstacle is located.

Images