Obstacle detection method, device and equipment, medium and vehicle

Abstract

The invention discloses an obstacle detection method, device and equipment, a medium and a vehicle, and the method comprises the steps: determining the distance between a plurality of radars and a target obstacle based on obtained multi-frame echo signals; constructing a corresponding target ellipse according to each distance, and determining a plurality of obstacle line segments of the target obstacle by using tangent lines of the target ellipse; and fusing the plurality of obstacle line segments to obtain a detection result of the target obstacle. The sensing precision of obstacles in different shapes can be improved, and consumption of computing resources is effectively reduced.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, specifically to an obstacle detection method, device, equipment, medium, and vehicle. Background Technology

[0002] Ultrasonic radar plays a crucial role in parking systems. In current mainstream parking solutions, ultrasonic radar primarily handles short-range perception and can use its output obstacle information to correct the recognition results of visual sensors. Especially in complex scenarios involving various irregularly shaped obstacles, ultrasonic radar often serves as a last line of defense, ensuring the vehicle avoids collisions.

[0003] However, ultrasonic radar still has limitations in outputting obstacle contour information for irregularly shaped obstacles. Existing technologies mostly generate obstacle points based on distance echo signals, and then generate obstacle line segments through clustering and fitting to characterize their contours. This method performs well in the recognition of obstacles with regular boundaries, but the recognition accuracy drops significantly in scenarios with irregularly shaped obstacles where the point cloud distribution is irregular, and the clustering operation has high computational resource requirements.

[0004] Other methods output obstacle outlines directly as points: First, a point cloud is generated using ultrasonic radar echoes; then, the point cloud is validated, and points deemed valid obstacles are output. However, in practical applications, this method faces two problems: First, the point cloud is massive, resulting in high processing overhead; second, for large obstacles, the generated points have certain measurement errors, leading to a loss of perception accuracy. Summary of the Invention

[0005] This application provides an obstacle detection method, apparatus, device, medium, and vehicle, which improves the perception accuracy of obstacles of different shapes and effectively reduces the consumption of computing resources.

[0006] Firstly, this application provides an obstacle detection method, the method comprising:

[0007] The distances between multiple radars and target obstacles are determined based on the acquired multi-frame echo signals;

[0008] Construct the corresponding target ellipse based on each distance, and use the tangent of the target ellipse to determine multiple obstacle segments of the target obstacle;

[0009] Multiple obstacle segments are fused together to obtain the detection results of the target obstacle.

[0010] In some examples of this application, the radar includes a first radar and a second radar;

[0011] Based on the acquired multi-frame echo signals, the distances between multiple radars and target obstacles are determined, including:

[0012] The first distance between the first radar and the target obstacle is determined based on each first echo signal. The first echo signal is the echo signal received by the first radar after the detection signal emitted by the first radar is reflected by the target obstacle.

[0013] The second distance from the first radar to the target obstacle and then reflected back to the second radar is determined based on each second echo signal. The second echo signal is the echo signal received by the second radar after being emitted by the first radar and reflected by the target obstacle.

[0014] Construct the corresponding target ellipse based on each distance, including:

[0015] Using the first and second radars as focal points and the second distance as the length of the major axis of the target ellipse, a corresponding target ellipse is constructed.

[0016] Multiple obstacle segments of the target obstacle are determined using the tangent to the target ellipse, including:

[0017] Using the first distance as the normal distance of the obstacle line segment relative to the first radar, a tangent that satisfies the normal distance constraint is determined on the target ellipse, and the obstacle line segment is obtained based on the determined tangent.

[0018] In some examples of this application, multiple obstacle segments are fused to obtain the detection results of the target obstacle, including:

[0019] Update the endpoints of obstacle segments based on the intersections between them;

[0020] The updated line segments are projected according to a preset angle, the projected line segments are merged, and the contour information of the target obstacle is determined based on the merged line segments. The contour information is then used as the detection result.

[0021] In some examples of this application, the updated line segments are projected according to a preset angle, and the projected line segments are then blended, including:

[0022] If two updated line segments do not intersect and the included angle is less than the angle threshold, the preset angle is obtained by weighted calculation based on the length, relative angle and relative distance of the two updated line segments.

[0023] Using a preset angle as the directional reference, obtain the average angle straight line passing through the center points of the two updated line segments respectively;

[0024] Project the endpoints of the two updated line segments onto their respective average angle lines to obtain the projected line segments. Then move and splice the two projected line segments to obtain the merged line segments.

[0025] In some examples of this application, the two updated line segments are the first line segment and the second line segment, respectively;

[0026] Angle weighting is performed based on the lengths, relative angles, and relative distances of the two updated line segments to obtain the preset angle, including:

[0027] Obtain the first length and first angle corresponding to the first line segment, and the second length and second angle corresponding to the second line segment;

[0028] Length weights are obtained based on the first length and the second length; angle weights are obtained based on the relative angles between the first line segment and the second line segment; and distance weights are obtained based on the relative distance between the first line segment and the second line segment.

[0029] The first angle and the second angle are weighted and summed according to the length weight, angle weight and distance weight to obtain the preset angle.

[0030] In some examples of this application, the endpoints of obstacle segments are updated based on the intersections between obstacle segments, including:

[0031] When there is an intersection between two obstacle line segments, compare the distance between the intersection point and the center point of the two obstacle line segments, and determine the line segment cutting direction based on the distance comparison result;

[0032] Update the endpoints of the two obstacle line segments based on the line segment cut direction.

[0033] In some examples of this application, after determining multiple obstacle segments of the target obstacle using the tangent of the target ellipse, the method further includes:

[0034] Based on historical obstacle segments, multiple obstacle segments are filtered according to a preset association radius to obtain valid segments;

[0035] Merging multiple obstacle segments, including:

[0036] Merge valid line segments.

[0037] Secondly, this application also provides an obstacle detection device, the device comprising:

[0038] The distance acquisition module is used to determine the distances between multiple radars and target obstacles based on the acquired multi-frame echo signals.

[0039] The line segment determination module is used to construct the corresponding target ellipse based on each distance, and to determine multiple obstacle line segments of the target obstacle using the tangent of the target ellipse;

[0040] The line segment fusion module is used to fuse multiple obstacle line segments to obtain the detection results of the target obstacle.

[0041] Thirdly, this application also provides an electronic device, including a memory and a processor; the memory stores a computer program, and the processor is used to run the computer program in the memory to perform the operations in the obstacle detection method provided in the first aspect.

[0042] Fourthly, this application also provides a storage medium storing a plurality of instructions adapted for loading by a processor to execute the steps in the obstacle detection method provided in the first aspect.

[0043] Fifthly, this application also provides a computer program product, including a computer program or instructions, which, when executed by a processor, implement the steps in the obstacle detection method provided in the first aspect.

[0044] Sixthly, this application also provides a vehicle equipped with at least one of the obstacle detection device provided in the second aspect, the electronic device provided in the third aspect, the storage medium provided in the fourth aspect, and the computer program product provided in the fifth aspect.

[0045] Through one or more embodiments of the above embodiments in this application, at least the following technical effects can be achieved:

[0046] The obstacle detection method, apparatus, equipment, medium, and vehicle provided in this application include a method that uses multi-frame echo signals to determine the distances between multiple radars and target obstacles, constructs corresponding target ellipses based on these distances, and further generates obstacle line segments from multiple different orientations using the tangents of the target ellipses. This process omits the intermediate steps of point cloud generation and line fitting in traditional methods, reducing computational complexity and error propagation. Subsequently, the obstacle line segments generated from different orientations are fused to comprehensively reflect the overall contour, size distribution, and spatial orientation of the obstacle. The obstacle detection method provided in this application can quickly obtain obstacle line segment information after acquiring echo signals, significantly improving the real-time performance of data processing and reducing computational load. Simultaneously, by fusing the contour information corresponding to obstacle line segments from multiple orientations, the shape characteristics of obstacles of different shapes, such as long linear obstacles, irregularly shaped obstacles, and large-volume obstacles, can be more accurately characterized, improving perception accuracy. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 This is one of the flowcharts illustrating an obstacle detection method provided in an embodiment of this application;

[0049] Figure 2 This is one of the schematic diagrams of obstacle line segments provided in the embodiments of this application;

[0050] Figure 3 This is the second schematic diagram of obstacle line segments provided in the embodiments of this application;

[0051] Figure 4 This is a schematic diagram of line segment fusion provided in an embodiment of this application;

[0052] Figure 5 This is a schematic diagram of obstacle segment endpoint update provided in an embodiment of this application;

[0053] Figure 6 This is a second schematic flowchart of the obstacle detection method provided in the embodiments of this application;

[0054] Figure 7 This is a schematic diagram of the obstacle detection device provided in the embodiments of this application;

[0055] Figure 8 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0056] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0057] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined as "first" or "second" may explicitly or implicitly include one or more features. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0058] "A and / or B" includes the following three combinations: A only, B only, and a combination of A and B.

[0059] The use of "applies to" or "configured to" in this application implies open and inclusive language, which does not preclude applicability to or configuration to devices performing additional tasks or steps. Furthermore, the use of "based on" implies openness and inclusivity, because processes, steps, calculations, or other actions "based on" one or more conditions or values ​​may in practice be based on additional conditions or values ​​beyond those conditions.

[0060] In this application, the term "exemplary" is used to mean "used as an example, illustration, or description." Any embodiment described as "exemplary" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use this application. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that this application can be made without using these specific details. In other instances, well-known structures and processes are not described in detail to avoid obscuring the description of this application with unnecessary detail. Therefore, this application is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0061] Current ultrasonic radar sensing technology still has several shortcomings. First, the traditional line segment fitting method represents the contour information of obstacles by fitting straight line segments. This simplified approach based on straight lines is difficult to adapt to targets with complex shapes, especially during parking. The contour depiction of irregularly shaped obstacles is often not accurate enough, thus reducing parking efficiency and success rate.

[0062] Secondly, while the method of representing obstacle contours in the form of point clouds can achieve high accuracy when the obstacle is small, the processing difficulty increases significantly when the number of points is too large, and the consumption of computing resources is large; while when the obstacle is large, the contour accuracy may decrease due to measurement errors.

[0063] To address the aforementioned issues, embodiments of this application provide an obstacle detection method, apparatus, device, medium, and vehicle. The obstacle detection method, apparatus, device, medium, and vehicle provided in this application are described below with reference to the accompanying drawings.

[0064] like Figure 1 As shown in the figure, this application provides an obstacle detection method, which includes the following steps:

[0065] S101 determines the distance between multiple radars and target obstacles based on the acquired multi-frame echo signals.

[0066] In this context, multi-frame echo signals refer to echo data synchronously acquired by multiple ultrasonic radars within the same sampling period. A target obstacle is an object within the radar's detection range that can effectively reflect ultrasonic signals; this can be a long linear obstacle, a short linear obstacle, or an irregularly shaped obstacle. Each frame of echo signal determines the distance between a radar and the target obstacle at a given time; while multiple consecutive frames of echo signals provide the distances between the same radar and the same target obstacle at different times.

[0067] S102, construct the corresponding target ellipse according to each distance, and use the tangent of the target ellipse to determine multiple obstacle segments of the target obstacle.

[0068] Among them, multiple obstacle line segments correspond to the line segment representation of the target obstacle under different detection angles, which are used to depict the contour features of the obstacle in different directions.

[0069] Schematic, by taking the positions of any two radars as the two foci of an ellipse and the sum of their measured distances as the length of the ellipse's major axis, a target ellipse can be uniquely determined, and the target obstacle lies on the trajectory of this ellipse. When the obstacle is a linear structure, its contact point with the ellipse can be considered as the tangent point of the ellipse, and the direction of the tangent at this point is consistent with the actual extension direction of the obstacle. Therefore, this tangent can serve as the geometric expression of the obstacle line segment. By determining multiple target ellipses through different radar combinations, obstacle tangents at different angles can be obtained, thereby generating multiple obstacle line segments, each depicting the contour features of the target obstacle in different orientations.

[0070] S103 fuses multiple obstacle line segments to obtain the detection result of the target obstacle. The fusion result comprehensively reflects the overall shape information of the target obstacle, including its outer contour, size distribution, and orientation features. By fusing line segment information from different angles, not only are the limitations of single-direction detection eliminated, but the multi-dimensional contour of the target obstacle can also be reconstructed more comprehensively. This is particularly suitable for the accurate detection of irregularly shaped or large-volume obstacles, thus significantly improving detection accuracy and robustness.

[0071] The obstacle detection method provided in this application utilizes multi-frame echo signals to determine the distances between multiple radars and target obstacles, and constructs corresponding target ellipses based on these distances. Furthermore, it uses the tangents of the target ellipses to generate obstacle line segments from multiple different orientations. This process omits the intermediate steps of point cloud generation and line fitting in traditional methods, reducing computational complexity and error propagation from the outset. Subsequently, the obstacle line segments generated from different orientations are fused to comprehensively reflect the overall contour, size distribution, and spatial orientation of the obstacle. The obstacle detection method provided in this application can quickly obtain obstacle line segment information after acquiring echo signals, significantly improving the real-time performance of data processing. Simultaneously, by fusing the contour information corresponding to obstacle line segments from multiple orientations, it can more accurately characterize the shape features of obstacles of different shapes, such as long linear obstacles, irregularly shaped obstacles, and large-volume obstacles, improving the accuracy and reliability of ultrasonic radar environmental perception.

[0072] In some embodiments of this application, the radar includes a first radar and a second radar.

[0073] Based on the acquired multi-frame echo signals, the distances between multiple radars and target obstacles are determined, including:

[0074] The first distance between the first radar and the target obstacle is determined based on each first echo signal. The first echo signal is the echo signal received by the first radar after the detection signal emitted by the first radar is reflected by the target obstacle. The first echo signal reflects half of the two-way propagation distance of the first radar to the surface of the target obstacle in the vertical direction.

[0075] The second distance from the first radar to the target obstacle and then reflected back to the second radar is determined based on each second echo signal. The second echo signal is the echo signal received by the second radar after being emitted by the first radar and reflected by the target obstacle. This second echo signal reflects the total propagation distance from the first radar to the reflection point of the target obstacle and back to the second radar. It can be understood that both the first echo signal and the second echo signal originate from the detection signal emitted by the first radar. The difference is that the first echo signal is a self-emitted and self-received echo emitted by the first radar and received by itself; while the second echo signal is the echo emitted by the first radar and reflected by the target obstacle, received by the second radar.

[0076] Construct the corresponding target ellipse based on each distance, including:

[0077] Using the first and second radars as focal points and the second distance as the length of the major axis of the target ellipse, a corresponding target ellipse is constructed.

[0078] Multiple obstacle segments of the target obstacle are determined using the tangent to the target ellipse, including:

[0079] Using the first distance as the normal distance of the obstacle line segment relative to the first radar, a tangent that satisfies the normal distance constraint is determined on the target ellipse, and the obstacle line segment is obtained based on the determined tangent.

[0080] It is understandable that within a local area of ​​the target obstacle, the surface tangential direction at its reflection point coincides with the tangential direction of the target ellipse at that point, and this tangential direction simultaneously satisfies the fixed-range and geometric constraints formed by the ranging of the first and second radars. Therefore, by locating the tangent point on the target ellipse using the known normal distance and extracting the corresponding tangential direction, the obstacle line segment consistent with the local surface orientation of the obstacle can be directly obtained. This process does not require point cloud generation and line fitting, and can obtain high-precision geometric information with lower computational complexity, thereby achieving rapid and accurate depiction of the obstacle contour.

[0081] In other examples, the number of radars can be three, four, or even more. Based on any combination of first and second radars acquiring echo signals at the same time, an obstacle segment can be calculated using the method described above. As the number of radars increases, different radar combinations can generate multiple sets of ranging data, thus generating multiple obstacle segments in different orientations. By fusing the obstacle segments corresponding to the echo signals obtained from multiple radar combinations, not only can the outline features of obstacles be depicted from more angles, but the completeness and accuracy of detecting large-volume obstacles, irregularly shaped obstacles, and partially obscured obstacles can also be improved, significantly enhancing the overall environmental perception capability and reliability.

[0082] The following illustration uses radar A and radar B as examples.

[0083] like Figure 2 As shown, assume there are two radars, A and B, to detect obstacle CD. At time t, radar A acquires a direct echo (the first echo signal) via self-transmission and self-reception. The measured distance is twice the round-trip distance from point A to the obstacle's reflection point E (A→E→A). Radar B does not transmit a detection signal during this process; it only receives the echo signal (the second echo signal) from the detection wave emitted by radar A, which is transmitted to radar B via the obstacle's reflection point F. The measured distance is the total propagation path length from A→F→B. Since ultrasonic sensors can only acquire distance information from echoes, the measured distance for the receiving radar precisely satisfies the property of an elliptic distance with the transmitting radar A and the receiving radar B as foci and this distance as the length of the major axis. Point O is the center of the ellipse. This elliptical relationship is used to characterize the correspondence between the ranging result and the geometric position of the obstacle.

[0084] For obstacle CD, if the distance signal measured by the first radar A and the distance signal measured by the second radar B are mistakenly considered to originate from the same obstacle reflection point, then the obstacle position calculated based on this assumption (e.g.) Figure 2 The position shown by the pentagram will deviate significantly from the actual obstacle position, thus introducing positioning error.

[0085] To avoid the aforementioned errors, in this embodiment, for obstacle CD, the first distance measured by the first radar A (self-transmitting and self-receiving) corresponds to the distance from point A to point E, the perpendicular foot of obstacle line segment CD. The echo signal received by the second radar B comes from any obstacle reflection point F that satisfies the condition that "points A and B are the foci of an ellipse, and the reflection point is located on the elliptical trajectory." Based on this, the position of reflection point F can be represented on the ellipse in the form of unknown parameters, and the equation of the tangent line at point F can be obtained; the perpendicular distance from point A to this tangent line is the first distance, and the positions of obstacle endpoints E and F can be determined by solving the corresponding equation. Using this method, the obstacle line segment EF can be directly output, avoiding the positioning errors caused by inaccurate assumptions in point cloud clustering or triangulation in traditional methods.

[0086] The obstacle detection method provided in this application constructs a target ellipse with two radars as focal points and a second distance as the major axis length, which strictly satisfies the geometric constraints of dual-radar ranging. Based on this, the tangent point on the ellipse is located using the first distance, and the tangent line is calculated, obtaining the line segment direction consistent with the obstacle surface orientation, thereby improving the geometric accuracy of the obstacle contour. Directly constructing the ellipse and calculating the tangent line using echo distance eliminates the multi-step processing steps of point cloud generation, clustering, and line fitting in traditional ultrasonic sensing; the computational link is shorter, and the data processing volume is significantly reduced. Traditional triangulation methods are prone to positioning errors in the case of long obstacles or inconsistent reflection points, while the obstacle detection method provided in this application locates the tangent point under the dual conditions of geometric constraints and normal distance, effectively avoiding systematic errors caused by inconsistent measurement points; the tangent direction is highly consistent with the actual orientation of the obstacle, thus reducing error accumulation in subsequent line segment fusion and contour reconstruction.

[0087] In some embodiments of this application, multiple obstacle segments are fused to obtain the detection result of the target obstacle, including:

[0088] Update the endpoints of obstacle segments based on the intersections between them.

[0089] Understandably, obstacle segments may intersect or partially overlap during the calculation process from different perspectives or with different radar combinations. Direct fusion could lead to excessively long contours or distorted shapes. By truncating at intersections and updating endpoints, redundancy and inaccuracies can be eliminated, ensuring that the spatial extent of each segment matches the true boundary of the obstacle.

[0090] The updated line segments are projected according to a preset angle, the projected line segments are merged, and the contour information of the target obstacle is determined based on the merged line segments. The contour information is then used as the detection result.

[0091] The preset angle refers to the direction of the obstacle line segments, the main extension direction of the obstacle as a whole, or the average direction calculated by weighting multiple line segments, which is used as a unified projection reference.

[0092] Projection fusion maps line segments belonging to the same obstacle but located at different positions and in different directions onto a straight line in a unified direction. The projected segments are then merged based on their position and length, thus eliminating minor deviations in line segment direction caused by measurement noise or differences in local reflections. This method yields contour lines with better continuity and smoother boundaries, avoiding jagged or discontinuous contours and improving the integrity and accuracy of the overall contour information.

[0093] Finally, the outer contour information of the target obstacle is determined based on the fused line segment set, and this contour information is output as the detection result.

[0094] In some examples, such as Figure 3 As shown, for short linear obstacles, the obstacle points are generated in the same way as for long linear obstacles. That is, the obstacle is assumed to be a long linear obstacle during calculation, and the obstacle segment GH is obtained according to the aforementioned method for obstacle segment generation. Although there is a certain deviation between the calculated GH and the actual obstacle segment EF, this deviation can be effectively eliminated through subsequent segment fusion processing. Specifically, segment fusion utilizes segment information obtained from multiple perspectives and at multiple times. Through steps such as intersection point updating, direction projection, and overlap elimination, GH is comprehensively corrected with other observed segment data, ensuring that the final generated obstacle contour is highly consistent with the actual EF. Therefore, segment fusion not only improves the positioning accuracy of short linear obstacles but also enhances the stability and robustness of the overall detection results.

[0095] In some embodiments of this application, the updated line segment is projected according to a preset angle, and the projected line segment is then fused, including:

[0096] When two updated line segments do not intersect and their included angle is less than the angle threshold α, a preset angle is obtained by weighting the angles based on the lengths, relative angles, and relative distances of the two updated line segments. The longer the line segment, the closer the relative angles, and the smaller the relative distance, the greater the contribution to the weighted result, ensuring that the preset angle more accurately reflects the overall extension direction of the two line segments.

[0097] Using a preset angle as the direction reference, obtain the average angle straight line passing through the center point of the two updated line segments respectively, so that the average angle straight line is consistent with the preset angle.

[0098] The endpoints of the two updated line segments are projected onto their corresponding average angle lines to obtain projected line segments. These two projected line segments are then moved and spliced ​​together to obtain a fused line segment. In other words, the two projected line segments are first aligned according to their relative positions on the average angle line, then appropriately moved along the axis perpendicular to that average angle line to create a continuous connection in space; finally, they are spliced ​​together into a fused line segment. This method eliminates breaks and misalignments in the original line segments caused by minor deviations in angle or position, resulting in obstacle contour lines with consistent direction and good continuity.

[0099] Indicatively, such as Figure 4 As shown, if two updated line segments do not intersect and the included angle is less than the angle threshold α, then the preset angle is obtained by first performing angle weighted calculation based on the length of each of the two updated line segments, the angle of their respective line segments AB between the first radar A and the second radar B (i.e., relative angle), and the length of the major axis of the target ellipse (i.e., relative distance).

[0100] Then, using the center points of the two updated line segments as references, an average angle line passing through the center points is drawn at a preset angle. The two updated line segments are projected onto their respective average angle lines to obtain corresponding new line segments, which are the projected line segments. Finally, according to the length of the projected line segments, they are appropriately moved and aligned in the direction perpendicular to the line segments to obtain the merged line segments.

[0101] Understandably, in the detection of irregularly shaped or short obstacles, the original calculations, assuming the obstacle is a long line, may generate multiple line segments that are close to each other and have similar directions, sometimes even exceeding the actual size of the obstacle. Therefore, line segment fusion can merge these redundant or repetitive segments into a single line segment, effectively reducing redundant detection results and preventing the obstacle outline from being overextended. For obstacles with irregular shapes or local unevenness, fusion of adjacent line segments generates a more realistic overall outline. Even when long obstacles are divided into multiple segments, they can be restored to complete, extended line segments, improving the completeness and accuracy of the detection results.

[0102] In some embodiments of this application, the two updated line segments are the first line segment and the second line segment, respectively.

[0103] Angle weighting is performed based on the lengths, relative angles, and relative distances of the two updated line segments to obtain the preset angle, including:

[0104] Obtain the first length L1 and first angle θ1 corresponding to the first line segment, and the second length L2 and second angle θ2 corresponding to the second line segment. It can be understood that the first angle θ1 and the second angle θ2 are the angles of the first and second line segments in the global coordinate system, respectively.

[0105] The length weight is obtained based on the first length and the second length, the angle weight is obtained based on the relative angle between the first line segment and the second line segment, and the distance weight is obtained based on the relative distance between the first line segment and the second line segment.

[0106] The length weight is determined based on the ratio of the first length L1 to the second length L2. The longer the line segment, the higher the weight, in order to reflect its representativeness in the overall extension direction.

[0107] The angle weight is calculated based on the relative angle between the first angle and the second angle. The smaller the relative angle, the more consistent the directions of the two line segments are, and the higher the weight.

[0108] The distance weight is determined based on the relative distance between the first and second line segments. The smaller the relative distance, the closer the two line segments are in space, and the higher the weight.

[0109] The first angle and the second angle are weighted and summed according to length weight, angle weight, and distance weight to obtain a preset angle. That is, the length weight, angle weight, and distance weight are normalized and used as weighting coefficients to sum the first angle θ1 and the second angle θ2 to obtain a preset angle for the fusion direction. This preset angle can comprehensively reflect the comprehensive characteristics of the two line segments in terms of length, directional consistency, and spatial proximity, and is used as the directional reference for subsequent projection fusion.

[0110] In some examples, the length weight of the first line segment The length weight of the second line segment

[0111] The angle weight corresponding to the relative angle β1 of the first line segment The angle weight corresponding to the relative angle β2 of the second line segment Wherein, β1 and β2 are the relative angles between the first line segment and the second line segment and the major axis in the elliptical coordinate system, respectively. The smaller the relative angle, the higher the corresponding angle weight.

[0112] Distance weight of the first line segment Distance weight of the second line segment Here, len1 and len2 represent the sum of the distances from the first line segment, the second line segment, and the tangent point of the ellipse to the ellipse, respectively. In other words, they are the distances relative to the ellipse. The smaller the relative distance, the higher the corresponding distance weight.

[0113] Based on the aforementioned determination of the length weight, angle weight, and distance weight of the first and second line segments, the total weight w1 of the first line segment and the total weight w2 of the second line segment are obtained respectively:

[0114]

[0115] Based on the total weight w1 of the first line segment and the total weight w2 of the second line segment, the first angle θ1 and the second angle θ2 can be weighted and summed to obtain the preset angle θ:

[0116] θ = θ1*w1 + θ2*w2.

[0117] In some embodiments of this application, updating the endpoints of obstacle segments based on the intersections between obstacle segments includes:

[0118] When there is an intersection between two obstacle line segments, compare the distance between the intersection point and the center point of the two obstacle line segments, and determine the direction of line segment cutting based on the distance comparison result.

[0119] Indicatively, such as Figure 5 As shown, if the distance from the intersection point to the center point of the obstacle segment is less than the distance from the other end of the segment to the center point, it indicates that the intersection point is located on the side of the obstacle segment closer to the center point. In this case, the part between the intersection point and the center point should be retained, and the other end should be cut off.

[0120] If the distance from the intersection point to the center point of the obstacle segment is large, it indicates that the intersection point is located on the side of the obstacle segment away from the center point. In this case, the part from the center point to the other end should be retained, and the part on one side of the intersection point should be cut off.

[0121] The endpoints of two obstacle line segments are updated based on the truncation direction. That is, the truncation direction of each line segment is determined based on the comparison results above, and the intersection point is used as the new endpoint to replace the cut-off end, thereby updating the endpoint positions of the two obstacle line segments. This method can effectively remove redundant extensions outside the intersection point, ensuring that the truncated line segment endpoints accurately conform to the actual boundary of the obstacle.

[0122] In some embodiments of this application, considering that some obstacle segments may be mistakenly generated due to interference signals, abnormal instantaneous echoes, etc., it is necessary to verify the validity of the generated obstacle segments before performing segment fusion.

[0123] Specifically, after determining multiple obstacle segments of the target obstacle using the tangent of the target ellipse, the method further includes:

[0124] Based on historical obstacle segments, multiple obstacle segments are filtered according to a preset association radius to obtain valid segments.

[0125] Merging multiple obstacle segments, including:

[0126] Merge valid line segments.

[0127] In some examples, the obstacle segments generated in the most recent K time steps are first obtained as a historical segment set Segs, where all segments are confirmed historical detection results. Then, the multiple obstacle segments obtained at the current time step are filtered according to the set minimum association radius dis (defined as the minimum spatial distance between two segments) and the minimum number of associated segments n.

[0128] During the screening process, the set of obstacle segments generated at the current moment is traversed. For any given current segment, it is checked whether it satisfies the minimum correlation radius dis with historical segments in the historical set Segs, and the number of historical segments that meet this condition is recorded. If the number of historical segments is greater than or equal to the preset minimum number of correlated segments n, then the current segment is determined to have high temporal continuity and spatial stability with the historical segments generated at the aforementioned K moments, and is thus considered a valid segment. Otherwise, the current segment is determined to be a false target caused by occasional false detection or noise interference, and is discarded, not included in subsequent segment fusion.

[0129] In the subsequent line segment fusion process, only the aforementioned valid line segments are fused. Since invalid line segments have been removed, the fusion result can more accurately reflect the true contour of the target obstacle, and can reduce the interference of falsely detected line segments on the final detection result, thereby improving the robustness and accuracy of the overall detection.

[0130] like Figure 6 As shown in the figure, this application embodiment also provides an obstacle detection method, which includes the following steps:

[0131] S601: Based on multi-frame echo signals, it obtains the distance between multiple radars and target obstacles.

[0132] S602: Construct a target ellipse using each set of distances, and determine the tangent line on the ellipse that satisfies the constraints. Then, determine the endpoints of the obstacle line segments based on the tangent line, thereby obtaining the obstacle line segments. Repeat the above calculation for multiple sets of distances to obtain multiple obstacle line segments.

[0133] Specifically, a target ellipse is constructed by taking the spatial positions of the first and second radars as the two foci of an ellipse and the second distance as the length of the major axis of the ellipse. The first distance is the perpendicular distance from the first radar to the target obstacle, and the second distance is the total propagation distance of the signal transmitted by the first radar after reflection from the target obstacle to the second radar. The first distance is used as the normal distance of the obstacle line segment relative to the first radar. A tangent point satisfying this normal distance constraint is located on the target ellipse, and the tangent line corresponding to this tangent point is obtained. Based on the direction and position of this tangent line, the coordinates of the endpoints of the obstacle line segment are determined, thus obtaining the obstacle line segment.

[0134] S603: Traverse all generated obstacle segments, compare each obstacle segment with the segments in the historical segment set Segs, and count the number of historical segments that meet the preset condition. The preset condition is: the minimum distance between the obstacle segment and the historical segment is less than the preset minimum association radius dis. If the count is greater than the minimum number of associated segments n, the current obstacle segment is determined to be a valid segment; otherwise, it is determined to be an invalid segment and skipped, and the next obstacle segment is processed.

[0135] S604: Traverse the valid line segments and determine whether the current valid line segment intersects with other valid line segments. If they intersect, update the endpoint positions of the two intersecting line segments based on the intersection point, thereby removing redundant extensions outside the intersection point; if they do not intersect, continue to determine whether the next valid line segment has an intersection point.

[0136] S605: Traverse the updated line segments and determine whether any two updated line segments meet the fusion condition. If they do, merge the two line segments; otherwise, continue to the next set of updated line segments. The fusion condition includes: the two line segments do not intersect, and their included angle is less than a preset angle threshold α.

[0137] Specifically, during the fusion process, firstly, a weighted calculation is performed based on the length, relative angle, and relative distance of the two updated line segments to obtain a preset angle for the fusion direction; then, using this preset angle as the direction reference, the average angle straight lines passing through the center points of the two updated line segments are obtained respectively; next, the endpoints of the two updated line segments are projected onto their respective corresponding average angle straight lines to obtain projected line segments with the same direction; finally, the two projected line segments are aligned and translated in a direction perpendicular to the preset angle, and then spliced ​​together to form a fused line segment.

[0138] Through the above steps, the detection results of the target obstacle can be obtained, which can accurately reflect the overall outline, spatial distribution and directional characteristics of the obstacle.

[0139] The obstacle detection method provided in this application determines the distance between the radar and the target obstacle using the echo signal acquired by the ultrasonic radar. It then constructs a target ellipse using a second distance and the positions of the two radars, and locates the tangent point on the ellipse using the normal constraint of the first distance. The tangent line is then calculated, and the obstacle line segment is directly generated. This process skips the traditional point cloud generation and line fitting process, allowing the obstacle contour generation to rely solely on the currently acquired distance information. This results in a shorter computational link, significantly reduced computational load, and reduced error propagation from intermediate processing steps. Furthermore, the obstacle detection method also performs validity screening on multiple obstacle line segments based on historical line segments from the most recent K time points. This effectively eliminates occasional false line segments caused by transient interference or noise, significantly reducing the false detection rate and improving the reliability of subsequent contour fusion. In addition, the obstacle detection method also updates the endpoints of intersecting valid line segments, truncating and removing redundant extensions based on the intersection point, thereby avoiding excessively long line segments or boundary extensions caused by multi-view calculations. During line segment fusion, a preset angle is calculated using a weighted average of length, relative angle, and relative distance to make the fusion direction more consistent with the main extension direction of the obstacle. By projecting the line segments onto a straight line in a unified direction and then aligning and stitching them together, slight directional deviations caused by measurement noise or local reflection differences can be eliminated, resulting in smoother and more continuous contour lines.

[0140] like Figure 7 As shown in the figure, this application embodiment also provides an obstacle detection device, which corresponds to the aforementioned obstacle detection method. The device includes a distance acquisition module 701, a line segment determination module 702, and a line segment fusion module 703.

[0141] The distance acquisition module 701 is used to determine the distance between multiple radars and target obstacles based on the acquired multi-frame echo signals.

[0142] The line segment determination module 702 is used to construct the corresponding target ellipse based on each distance, and to determine multiple obstacle line segments of the target obstacle using the tangent of the target ellipse.

[0143] The line segment fusion module 703 is used to fuse multiple obstacle line segments to obtain the detection results of the target obstacle.

[0144] In some embodiments of this application, the radar includes a first radar and a second radar;

[0145] The distance acquisition module 701 is specifically used for:

[0146] The first distance between the first radar and the target obstacle is determined based on each first echo signal. The first echo signal is the echo signal received by the first radar after the detection signal emitted by the first radar is reflected by the target obstacle.

[0147] The second distance from the first radar to the target obstacle and then reflected back to the second radar is determined based on each second echo signal. The second echo signal is the echo signal received by the second radar after being emitted by the first radar and reflected by the target obstacle.

[0148] The line segment determination module 702 includes an ellipse construction unit and a tangent determination unit, wherein the ellipse construction unit is specifically used for:

[0149] Using the first and second radars as focal points and the second distance as the length of the major axis of the target ellipse, a corresponding target ellipse is constructed.

[0150] The tangent determination unit is specifically used for:

[0151] Using the first distance as the normal distance of the obstacle line segment relative to the first radar, a tangent that satisfies the normal distance constraint is determined on the target ellipse, and the obstacle line segment is obtained based on the determined tangent.

[0152] In some embodiments of this application, the line segment fusion module 703 includes an endpoint update unit and a fusion unit. The endpoint update unit is used for:

[0153] Update the endpoints of obstacle segments based on the intersections between them;

[0154] The fusion unit is used for:

[0155] The updated line segments are projected according to a preset angle, the projected line segments are merged, and the contour information of the target obstacle is determined based on the merged line segments. The contour information is then used as the detection result.

[0156] In some embodiments of this application, the fusion unit is specifically used for:

[0157] If two updated line segments do not intersect and the included angle is less than the angle threshold, the preset angle is obtained by weighted calculation based on the length, relative angle and relative distance of the two updated line segments.

[0158] Using a preset angle as the directional reference, obtain the average angle straight line passing through the center points of the two updated line segments respectively;

[0159] Project the endpoints of the two updated line segments onto their respective average angle lines to obtain the projected line segments. Then move and splice the two projected line segments to obtain the merged line segments.

[0160] In some embodiments of this application, the two updated line segments are a first line segment and a second line segment, respectively;

[0161] The fusion unit is specifically used for:

[0162] Obtain the first length and first angle corresponding to the first line segment, and the second length and second angle corresponding to the second line segment;

[0163] Length weights are obtained based on the first length and the second length; angle weights are obtained based on the relative angles between the first line segment and the second line segment; and distance weights are obtained based on the relative distance between the first line segment and the second line segment.

[0164] The first angle and the second angle are weighted and summed according to the length weight, angle weight and distance weight to obtain the preset angle.

[0165] In some embodiments of this application, the endpoint update unit is specifically used for:

[0166] When there is an intersection between two obstacle line segments, compare the distance between the intersection point and the center point of the two obstacle line segments, and determine the line segment cutting direction based on the distance comparison result;

[0167] Update the endpoints of the two obstacle line segments based on the line segment cut direction.

[0168] In some embodiments of this application, the obstacle detection device further includes a validity determination module, used for:

[0169] Based on historical obstacle segments, multiple obstacle segments are filtered according to a preset association radius to obtain valid segments;

[0170] Accordingly, the line segment fusion module is specifically used for:

[0171] Merge valid line segments.

[0172] Based on any of the above embodiments, another embodiment of this application also provides an electronic device, such as... Figure 8 As shown, the electronic device may include a processor 810, a communication interface 820, a memory 830, and a communication bus 840, wherein the processor 810, the communication interface 820, and the memory 830 communicate with each other through the communication bus 840. The processor 810 can call logical instructions in the memory 830 to execute the obstacle detection method described above.

[0173] Furthermore, the logical instructions in the aforementioned memory 830 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0174] Furthermore, when the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0175] On the other hand, embodiments of this application also provide a storage medium storing a plurality of instructions adapted for loading by a processor to execute the obstacle detection method provided in the above embodiments.

[0176] On the other hand, embodiments of this application also provide a computer program product, including a computer program that, when executed by a processor, implements the above-described obstacle detection method.

[0177] On the other hand, embodiments of this application also provide a vehicle equipped with at least one of the aforementioned obstacle detection device, electronic device, storage medium, and computer program product.

[0178] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0179] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0180] The foregoing has provided a detailed description of an obstacle detection method, apparatus, device, medium, and vehicle provided in the embodiments of this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. An obstacle detection method, characterized in that, The method includes: The distances between multiple radars and target obstacles are determined based on the acquired multi-frame echo signals; Construct corresponding target ellipses based on the distances, and use the tangents of the target ellipses to determine multiple obstacle segments of the target obstacle; The multiple obstacle segments are fused together to obtain the detection result of the target obstacle.

2. The obstacle detection method according to claim 1, characterized in that, The radar includes a first radar and a second radar; The method of determining the distance between multiple radars and target obstacles based on the acquired multi-frame echo signals includes: A first distance between the first radar and the target obstacle is determined based on each first echo signal. The first echo signal is the echo signal received by the first radar after the detection signal emitted by the first radar is reflected by the target obstacle. A second distance is determined from the first radar to the target obstacle and then reflected back to the second radar based on each second echo signal, wherein the second echo signal is the echo signal received by the second radar after being emitted by the first radar and reflected by the target obstacle; The construction of the corresponding target ellipse based on each of the distances includes: Using the first radar and the second radar as focal points, and the second distance as the length of the major axis of the target ellipse, the corresponding target ellipse is constructed to obtain the target ellipse. The step of determining multiple obstacle segments of the target obstacle using the tangent of the target ellipse includes: Using the first distance as the normal distance of the obstacle line segment relative to the first radar, a tangent line satisfying the normal distance constraint is determined on the target ellipse, and the obstacle line segment is obtained based on the determined tangent line.

3. The obstacle detection method according to claim 1 or 2, characterized in that, The process of fusing the multiple obstacle segments to obtain the detection result of the target obstacle includes: Update the endpoints of the obstacle segments based on the intersections between them; The updated line segments are projected according to a preset angle, the projected line segments are merged, and the contour information of the target obstacle is determined based on the merged line segments. The contour information is then used as the detection result.

4. The obstacle detection method according to claim 3, characterized in that, The process of projecting the updated line segments according to a preset angle and then merging the projected line segments includes: When the two updated line segments do not intersect and the included angle is less than the angle threshold, the preset angle is obtained by performing angle weighted calculation based on the length, relative angle and relative distance of the two updated line segments. Using the preset angle as the direction reference, obtain the average angle straight line passing through the center points of the two updated line segments respectively; The endpoints of the two updated line segments are projected onto their respective average angle lines to obtain the projected line segments. The two projected line segments are then moved and spliced ​​together to obtain the fused line segment.

5. The obstacle detection method according to claim 4, characterized in that, The two updated line segments are the first line segment and the second line segment, respectively; The step of calculating the preset angle by weighting the angles based on the lengths, relative angles, and relative distances of the two updated line segments includes: Obtain the first length and first angle corresponding to the first line segment, and the second length and second angle corresponding to the second line segment; Length weights are obtained based on the first length and the second length; angle weights are obtained based on the relative angles between the first line segment and the second line segment; and distance weights are obtained based on the relative distance between the first line segment and the second line segment. The first angle and the second angle are weighted and summed according to the length weight, the angle weight, and the distance weight to obtain the preset angle.

6. The obstacle detection method according to claim 3, characterized in that, The step of updating the endpoints of the obstacle segments based on the intersections between the obstacle segments includes: When there is an intersection between two obstacle line segments, the distance between the intersection and the center point of the two obstacle line segments is compared, and the line segment cutting direction is determined based on the distance comparison result. The endpoints of the two obstacle segments are updated based on the truncation direction of the line segments.

7. The obstacle detection method according to any one of claims 1 to 6, characterized in that, After determining multiple obstacle segments of the target obstacle using the tangent of the target ellipse, the method further includes: Based on historical obstacle segments, the multiple obstacle segments are filtered according to a preset association radius to obtain valid segments; The fusion of the multiple obstacle segments includes: The effective line segments are merged.

8. An obstacle detection device, characterized in that, The device includes: The distance acquisition module is used to determine the distances between multiple radars and target obstacles based on the acquired multi-frame echo signals. A line segment determination module is used to construct a corresponding target ellipse based on each of the distances, and to determine multiple obstacle line segments of the target obstacle using the tangent of the target ellipse; The line segment fusion module is used to fuse the multiple obstacle line segments to obtain the detection result of the target obstacle.

9. An electronic device, characterized in that, It includes a memory and a processor; the memory stores a computer program, and the processor is used to run the computer program in the memory to perform the operations in the obstacle detection method according to any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium stores a plurality of instructions adapted for loading by a processor to execute the steps of the obstacle detection method according to any one of claims 1 to 7.

11. A vehicle, characterized in that, The vehicle is equipped with at least one of the obstacle detection device of claim 8, the electronic device of claim 9, and the storage medium of claim 10.

Images