Construction elevator identification and relative pose estimation method based on laser profile

Abstract

The application relates to the technical field of intelligent construction, and discloses a construction elevator identification and relative pose estimation method based on a laser profile, which comprises the following steps: performing preprocessing on two-dimensional laser scanning data, and outputting preprocessed polar coordinate laser points; converting the preprocessed polar coordinate laser points into two-dimensional plane rectangular coordinate points, and generating a complete two-dimensional point set; performing spatial clustering according to the Euclidean distance of the laser points, and taking the clustering result as the pre-input of line segment fitting; extracting the projection interval of all inner points on the direction vector to generate a standard line segment; filtering and screening in combination with the geometric structure features of the inner cavity of the construction elevator to obtain an effective construction elevator profile; transforming the coordinates in a radar coordinate system to a global coordinate system of a robot chassis to solve the original relative pose error result of an AGV relative to the construction elevator; performing time sequence smoothing and noise reduction processing on the original relative pose error to output the relative pose of the construction elevator and a pose effective flag. The application can accurately identify the construction elevator and stably output the pose.

Description

Technical Field

[0001] This application relates to the field of intelligent construction technology, and in particular to a method for construction elevator identification and relative pose estimation based on laser profiles. Background Technology

[0002] In intelligent construction scenarios within the building industry, automated guided vehicles (AGVs) often need to transport materials across floors within a construction site, and construction elevators are a crucial channel for this inter-floor transfer. Compared to conventional indoor construction elevators, construction elevators operate in environments characterized by dust, obstructions, steel structure reflections, and frequent personnel movement. The doorway contours are unstable, and the structure and placement of construction elevators vary significantly across different construction sites. Relying solely on manual marking, fixed reflectors, or QR codes for positioning is not only costly to deploy but also susceptible to dust contamination or damage at the construction site, making long-term stable use difficult. While existing visual recognition methods can detect targets on construction elevators, they are significantly affected by lighting, obstructions, and sample generalization capabilities. Methods relying solely on laser line segment extraction are easily affected by noise points, abrupt changes, and local missing parts, leading to unstable contour recognition. Summary of the Invention

[0003] In view of this, this application provides a construction elevator identification and relative pose estimation method based on laser contours. It can extract the structural contour of the construction elevator door area using two-dimensional laser radar, and combine adaptive filtering, physical corner point solving and continuous frame consistency confirmation mechanism to stably filter the target contour, and then output the relative pose between the robot and the construction elevator, providing a reliable input for the subsequent accurate entry of the AGV into the construction elevator.

[0004] This application discloses a method for construction elevator recognition and relative pose estimation based on laser contours, which includes: Step 1: The system subscribes to 2D laser scanning data, robot positioning information, and construction elevator task start signal; when the construction elevator task start signal is triggered and the robot positioning information is valid, the system switches to construction elevator laser recognition mode, performs two-level preprocessing on the 2D laser scanning data, and outputs the preprocessed polar coordinate laser points. Step 2: Convert the preprocessed polar coordinate laser points into two-dimensional Cartesian coordinate points to generate a complete two-dimensional point set; Step 3: Perform spatial clustering based on the Euclidean distance of the laser points. When the distance between adjacent laser points is less than the clustering threshold, the adjacent laser points are determined to belong to the same cluster region. The clustering result serves as the pre-input for line segment fitting. For cluster regions where the number of points reaches the minimum effective point threshold, the RANSAC algorithm or the split_merge algorithm is selected to perform line fitting. If the distance from the laser point to the fitted line is less than the inlier threshold, the laser point is determined to be an inlier of the line, and the optimal fitted line is selected based on the total number of inliers. The projection interval of all inliers on the direction vector is extracted to generate standard line segments. The direction vector is a two-dimensional vector composed of the two endpoints of a single fitted line segment, used to characterize the extension direction of the line segment to calculate the line segment tilt angle and the effective length interval of the line segment. Step 4: Using all line segments output in Step 3 as the candidate line segment set, filter and screen layer by layer based on the geometric features of the construction elevator cavity to obtain the effective construction elevator outline. Step 5: Based on the effective construction elevator cavity contour output in Step 4, extract three fitted straight lines, mathematically extend the three fitted straight lines and solve for their intersection points to obtain the coordinates of the two corner points where the two side walls and the rear wall of the construction elevator cavity intersect. Then, solve for the coordinates of the center of the construction elevator cavity in the radar coordinate system. The average of the fitted direction angles of the two inner wall segments of the construction elevator is extracted as the relative heading angle of the construction elevator. Using the robot chassis and the pre-calibrated static extrinsic parameter matrix of the lidar, the radar coordinate system is... The coordinates are transformed to the global coordinate system of the robot chassis, and the original relative pose error of the AGV relative to the construction elevator is solved. Step 6: Perform time-series smoothing and noise reduction processing on the original relative pose error, and output the relative pose and pose validity flag of the construction elevator.

[0005] Furthermore, the two-dimensional laser scanning data is environmental ranging data obtained by the vehicle-mounted two-dimensional laser radar periodically scanning the construction site in the scenario of AGV cross-floor transfer in an intelligent construction site; The two-dimensional laser scanning data is represented as follows:

[0006] in, This is two-dimensional laser scanning data. For the first The laser point distance value refers to the straight-line distance from the laser radar's emission center (i.e., the origin of the radar coordinate system) to the reflection point of the scanned obstacle. This represents the total number of laser points in a single scan. For the first The horizontal scanning angle corresponding to the laser beam.

[0007] Furthermore, the two-stage preprocessing of the two-dimensional laser scanning data includes: Distance range filtering: Only retain those that meet the requirements Laser points within the range are used to remove noise at very close range and invalid points at long distances; among them, For the minimum effective distance, The maximum effective distance; Adaptive continuity filtering: Calculates the distance difference between adjacent laser points :

[0008] when When the current laser point is identified as an abnormal jump point, it is removed. This is an adaptive dynamic threshold based on ranging depth.

[0009] Furthermore, a depth-based adaptive dynamic threshold is introduced. The expression is:

[0010] in, For radar angular resolution, This is the environmental dust and noise compensation constant.

[0011] Further, step 2 includes: The preprocessed polar coordinate laser points are converted into two-dimensional Cartesian coordinate points using the following formula:

[0012]

[0013] in, The horizontal x-coordinate of the obstacle reflection point in the radar coordinate system is taken along the radar X-axis. The horizontal ordinate of the obstacle reflection point in the radar coordinate system is taken along the radar Y-axis. Combine all the transformed coordinates to generate a two-dimensional point set:

[0014] in, This is a two-dimensional point set of all obstacles in the radar coordinate system, used for subsequent line segment extraction.

[0015] Furthermore, the expression for the fitted straight line is:

[0016] in, , , These represent the fixed coefficients of the equation of a standard straight line in a two-dimensional plane. , These represent the x-coordinate and y-coordinate of the laser reflection point in the radar coordinate system, respectively. The distance from the laser point to the fitted line is:

[0017] in, The vertical distance from a single laser point to a preset straight line.

[0018] Further, step 4 includes: Abnormal short line segments and long wall line segments are removed. Abnormal short line segments are invalid short line segments whose projected length is less than the minimum side wall length threshold of the construction elevator door frame and are generated by fitting dust noise, partial pedestrian obstruction, and steel structure reflective noise. Long wall line segments are excessively long line segments whose projected length is greater than the maximum width threshold of the construction elevator cavity and correspond to large-area continuous walls, supporting steel beams, and other non-construction elevator outlines on the construction site. Match the target location corresponding to the current construction elevator task and remove irrelevant line segments whose center distance from the target location exceeds a preset distance threshold; the target location is the theoretical stopping coordinate point of the construction elevator specified by the current construction elevator task. Based on the internal geometry of the construction elevator, three geometric constraints of parallelism, perpendicularity, and width are set to select three target line segments that constitute the outline of the construction elevator entrance. Parallel constraints, i.e., constraints on both sides of the construction elevator cavity walls: Select two line segments of the construction elevator cavity side walls that satisfy... ; , These are the fitted direction angles for the two inner wall segments of the construction elevator cavity, respectively. Vertical constraint, specifically the vertical constraint between the rear wall and side walls of the construction elevator cavity: Select a line segment of the rear wall of the construction elevator cavity that is approximately perpendicular to both side walls, satisfying... ; The fitting direction angle for the line segment of the rear wall inside the construction elevator cavity; Width constraint, specifically the width constraint of the construction elevator cavity entrance: the vertical distance between the two sidewall segments of the construction elevator cavity matches the preset standard width of the construction elevator cavity, satisfying... ; in, The permissible deviation threshold for the parallel angle. The allowable deviation threshold for the vertical angle. The allowable deviation threshold for the width of the construction elevator; The frame is determined to have detected a compliant construction elevator outline only if it meets all the constraint conditions of parallel constraint, vertical constraint and width constraint. Two-dimensional lidar periodically scans the construction environment. A single complete scan outputs one frame of lidar point cloud data, which is recorded as the current frame. Multiple consecutive periodic scans output multiple sets of point clouds, which are called continuous sampling frames. Perform consistency verification on the contour recognition results of consecutive sampling frames: when N consecutive frames meet the constraint judgment conditions, the contour is marked as a valid construction elevator contour.

[0019] Furthermore, in step 5, the geometric midpoints of Pcorner1 and Pcorner2 are defined as the relative center position of the construction elevator using the following formula:

[0020]

[0021] in, Let x be the x-coordinate of the center of the construction elevator cavity in the radar coordinate system. The coordinates are the longitudinal coordinates of the center of the construction elevator cavity in the radar coordinate system.

[0022] Further, step 6 includes: The relative pose error results are subjected to temporal noise reduction using the following formula:

[0023] in, This represents the relative pose of the construction elevator after filtering and smoothing at the current time t, i.e., the relative pose of the construction elevator. These are the filter weight coefficients. To obtain the unsmoothed original relative pose error results for a single frame, This is the smoothed relative pose of the construction elevator output after low-pass filtering at the previous moment.

[0024] Furthermore, the pose validity flag is a Boolean output flag, and its generation logic is as follows: First condition: All N consecutive frames satisfy the constraint judgment condition to detect the valid internal contour of the construction elevator. Second condition: The construction elevator's internal cavity width, center coordinates, and relative heading angle corresponding to its relative pose all fall within the preset physical range; When both the first and second conditions are met, the pose validity flag is set to valid; if either condition is not met, the pose validity flag is set to invalid.

[0025] Due to the adoption of the above technical solution, this application has the following advantages: (1) Simple deployment and good adaptability. This invention is based on two-dimensional laser contour for construction elevator identification, without relying on manual markings such as QR codes, April Tags or fixed reflectors, which can reduce the deployment cost on the construction site and improve the system adaptability.

[0026] (2) Strong resistance to dust and high-frequency jump interference. This invention adds a laser continuity filtering mechanism with depth adaptive dynamic threshold to the traditional distance filtering, which can effectively suppress the problem of frequent laser point jump caused by steel structure edge reflection, dust interference and dynamic personnel movement at the construction site, and improve the stability of line segment extraction.

[0027] (3) Strong resistance to false recognition. Based on the structural prior features of the construction elevator and combined with the continuous frame consistency confirmation mechanism, this invention effectively suppresses single-frame misjudgment, reduces the false detection rate, and improves the stability and reliability of the construction elevator contour output.

[0028] (4) Stable output, facilitating subsequent control. This invention smooths the relative pose by solving for physical corner points and using low-pass filtering, which can reduce the impact of local occlusion and point cloud jitter on the results, providing a more stable input for the AGV to enter the construction elevator control module. Attached Figure Description

[0029] 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 recorded in the embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings.

[0030] Figure 1 This is a flowchart illustrating a laser profile-based method for construction elevator identification and relative pose estimation, according to an embodiment of this application. Detailed Implementation

[0031] The present application will be further described in conjunction with the accompanying drawings and embodiments. The described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments obtained by those skilled in the art should fall within the protection scope of the embodiments of the present application.

[0032] See Figure 1 This application provides an embodiment of a construction elevator identification and relative pose estimation method based on laser contours, which includes: Step 1: The system subscribes to 2D laser scanning data, robot positioning information, and construction elevator task start signal; when the construction elevator task start signal is triggered and the robot positioning information is valid, the system switches to construction elevator laser recognition mode, performs two-level preprocessing on the 2D laser scanning data, and outputs the preprocessed polar coordinate laser points. Step 2: Convert the preprocessed polar coordinate laser points into two-dimensional Cartesian coordinate points to generate a complete two-dimensional point set; Step 3: Perform spatial clustering based on the Euclidean distance of the laser points. When the distance between adjacent laser points is less than the clustering threshold, the adjacent laser points are determined to belong to the same cluster region. The clustering result serves as the pre-input for line segment fitting. For cluster regions where the number of points reaches the minimum effective point threshold, the RANSAC algorithm or the split_merge algorithm is selected to perform line fitting. If the distance from the laser point to the fitted line is less than the inlier threshold, the laser point is determined to be an inlier of the line, and the optimal fitted line is selected based on the total number of inliers. The projection interval of all inliers on the direction vector is extracted to generate standard line segments. The direction vector is a two-dimensional vector composed of the two endpoints of a single fitted line segment, used to characterize the extension direction of the line segment to calculate the line segment tilt angle and the effective length interval of the line segment. Step 4: Using all line segments output in Step 3 as the candidate line segment set, filter and screen layer by layer based on the geometric features of the construction elevator cavity to obtain the effective construction elevator outline. Step 5: Based on the effective construction elevator cavity contour output in Step 4, extract three fitted straight lines, mathematically extend the three fitted straight lines and solve for their intersection points to obtain the coordinates of the two corner points where the two side walls and the rear wall of the construction elevator cavity intersect. Then, solve for the coordinates of the center of the construction elevator cavity in the radar coordinate system. The average of the fitted direction angles of the two inner wall segments of the construction elevator is extracted as the relative heading angle of the construction elevator. Using the robot chassis and the pre-calibrated static extrinsic parameter matrix of the lidar, the radar coordinate system is... The coordinates are transformed to the global coordinate system of the robot chassis, and the original relative pose error of the AGV relative to the construction elevator is solved. Step 6: Perform time-series smoothing and noise reduction processing on the original relative pose error, and output the relative pose and pose validity flag of the construction elevator.

[0033] Optionally, the construction elevator task is the target operation process of AGV autonomously entering the construction elevator to complete the cross-floor material transfer; The three fitted lines are the fitted line of the left side wall of the construction elevator cavity, the fitted line of the right side wall of the construction elevator cavity, and the fitted line of the rear wall of the construction elevator cavity.

[0034] Optionally, the two-dimensional laser scanning data is environmental ranging data obtained by the vehicle-mounted two-dimensional laser radar periodically scanning the construction site in the scenario of AGV cross-floor transfer in a smart construction site. The two-dimensional laser scanning data is represented as follows:

[0035] in, This is two-dimensional laser scanning data. For the first The laser point distance value refers to the straight-line distance from the laser radar's emission center (i.e., the origin of the radar coordinate system) to the reflection point of the scanned obstacle. This represents the total number of laser points in a single scan. For the first The horizontal scanning angle corresponding to the laser beam.

[0036] Optionally, the two-stage preprocessing of the two-dimensional laser scanning data includes: Distance range filtering: Only retain those that meet the requirements Laser points within the range are used to remove noise at very close range (noise below a preset distance) and invalid points at long distance (points beyond a specified distance are invalid); among them, For the minimum effective distance, The maximum effective distance; Adaptive continuity filtering: Calculates the distance difference between adjacent laser points :

[0037] when When the current laser point is identified as an abnormal jump point, it is removed. An adaptive dynamic threshold based on ranging depth is used to reduce the discrete noise caused by edge reflections and local occlusions of steel structures at the construction site, and avoids the problem of false filtering at distant locations caused by a fixed threshold.

[0038] Optionally, considering the physical phenomenon that the distance between points naturally increases due to the laser divergence characteristics, a depth-based adaptive dynamic threshold is introduced here. Its expression is:

[0039] in, For radar angular resolution, This is the environmental dust and noise compensation constant.

[0040] Optionally, step 2 includes: The preprocessed polar coordinate laser points are converted into two-dimensional Cartesian coordinate points using the following formula:

[0041]

[0042] in, The horizontal x-coordinate of the obstacle reflection point in the radar coordinate system is taken along the radar X-axis. The horizontal ordinate of the obstacle reflection point in the radar coordinate system is taken along the radar Y-axis. Combine all the transformed coordinates to generate a two-dimensional point set:

[0043] in, This is a two-dimensional point set of all obstacles in the radar coordinate system, used for subsequent line segment extraction.

[0044] Optionally, the expression for the fitted straight line is:

[0045] in, , , These represent the fixed coefficients of the equation of a standard straight line in a two-dimensional plane. , These represent the x-coordinate and y-coordinate of the laser reflection point in the radar coordinate system, respectively. The distance from the laser point to the fitted line is:

[0046] in, The vertical distance from a single laser point to a preset straight line.

[0047] Optionally, step 4 includes: Abnormal short line segments and long wall line segments are removed. Abnormal short line segments are invalid short line segments whose projected length is less than the minimum side wall length threshold of the construction elevator door frame and are generated by fitting dust noise, partial pedestrian obstruction, and steel structure reflective noise. Long wall line segments are excessively long line segments whose projected length is greater than the maximum width threshold of the construction elevator cavity and correspond to large-area continuous walls, supporting steel beams, and other non-construction elevator outlines on the construction site. Match the target location corresponding to the current construction elevator task and remove irrelevant line segments whose center distance from the target location exceeds a preset distance threshold; the target location is the theoretical stopping coordinate point of the construction elevator specified by the current construction elevator task. Based on the internal geometry of the construction elevator, three geometric constraints of parallelism, perpendicularity, and width are set to select three target line segments that constitute the outline of the construction elevator entrance. Parallel constraints, i.e., constraints on both sides of the construction elevator cavity walls: Select two line segments of the construction elevator cavity side walls that satisfy... ; , These are the fitted direction angles for the two inner wall segments of the construction elevator cavity, respectively. Vertical constraint, specifically the vertical constraint between the rear wall and side walls of the construction elevator cavity: Select a line segment of the rear wall of the construction elevator cavity that is approximately perpendicular to both side walls, satisfying... ; The fitting direction angle for the line segment of the rear wall inside the construction elevator cavity; Width constraint, specifically the width constraint of the construction elevator cavity entrance: the vertical distance between the two sidewall segments of the construction elevator cavity matches the preset standard width of the construction elevator cavity, satisfying... ; in, The permissible deviation threshold for the parallel angle. The allowable deviation threshold for the vertical angle. The allowable deviation threshold for the width of the construction elevator; The frame is determined to have detected a compliant construction elevator outline only if it meets all the constraint conditions of parallel constraint, vertical constraint and width constraint. Two-dimensional lidar periodically scans the construction environment. A single complete scan outputs one frame of lidar point cloud data, which is recorded as the current frame. Multiple consecutive periodic scans output multiple sets of point clouds, which are called continuous sampling frames. To further reduce instantaneous false detections caused by dynamic personnel movement, short-term occlusion, and steel structure reflections at the construction site, a consistency check is performed on the contour recognition results of continuous sampling frames: when N consecutive frames meet the constraint judgment conditions, the contour is marked as a valid construction elevator contour.

[0048] Optionally, in step 5, the geometric midpoints of Pcorner1 and Pcorner2 are defined as the relative center position of the construction elevator using the following formula:

[0049]

[0050] in, Let x be the x-coordinate of the center of the construction elevator cavity in the radar coordinate system. The coordinates are the longitudinal coordinates of the center of the construction elevator cavity in the radar coordinate system.

[0051] Optionally, step 6 includes: Due to dynamic interference at the construction site, direct output of pose is prone to jitter. Therefore, a first-order hysteresis low-pass filter (weighted exponential smoothing) is used for timing noise reduction.

[0052] in, This represents the relative pose of the construction elevator after filtering and smoothing at the current time t, i.e., the relative pose of the construction elevator. These are the filter weight coefficients. To obtain the unsmoothed original relative pose error results for a single frame, This is the smoothed relative pose of the construction elevator output after low-pass filtering at the previous moment.

[0053] Optionally, the pose validity flag is a Boolean output flag, and the generation logic is as follows: First condition: All N consecutive frames satisfy the constraint judgment condition to detect the valid internal contour of the construction elevator. Second condition: The construction elevator's internal cavity width, center coordinates, and relative heading angle corresponding to its relative pose all fall within the preset physical range; When both the first and second conditions are met, the pose validity flag is set to valid; if either condition is not met, the pose validity flag is set to invalid.

[0054] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application and not to limit them. Although this application has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of this application. Any modifications or equivalent substitutions that do not depart from the spirit and scope of this application should be covered within the protection scope of the claims of this application.

Claims

1. A method for construction elevator identification and relative pose estimation based on laser profiles, characterized in that, include: Step 1: The system subscribes to 2D laser scanning data, robot positioning information, and construction elevator task start signal; when the construction elevator task start signal is triggered and the robot positioning information is valid, the system switches to construction elevator laser recognition mode, performs two-level preprocessing on the 2D laser scanning data, and outputs the preprocessed polar coordinate laser points. Step 2: Convert the preprocessed polar coordinate laser points into two-dimensional Cartesian coordinate points to generate a complete two-dimensional point set; Step 3: Perform spatial clustering based on the Euclidean distance between laser points. When the distance between adjacent laser points is less than the clustering threshold, the adjacent laser points are determined to belong to the same cluster region. The clustering result serves as the pre-input for line segment fitting. For cluster regions where the number of points reaches the minimum effective point threshold, the RANSAC algorithm or the split_merge algorithm is selected to perform line fitting. If the distance from a laser point to the fitted line is less than the inlier threshold, the laser point is determined to be an inlier of the line, and the optimal fitted line is selected based on the total number of inliers. The projection intervals of all inliers on the direction vector are extracted to generate standard line segments. Step 4: Using all line segments output in Step 3 as the candidate line segment set, filter and screen layer by layer based on the geometric features of the construction elevator cavity to obtain the effective construction elevator outline. Step 5: Based on the effective construction elevator cavity contour output in Step 4, extract three fitted straight lines, mathematically extend the three fitted straight lines and solve for their intersection points to obtain the coordinates of the two corner points where the two side walls and the rear wall of the construction elevator cavity intersect. Then, solve for the coordinates of the center of the construction elevator cavity in the radar coordinate system. The average of the fitted direction angles of the two inner wall segments of the construction elevator is extracted as the relative heading angle of the construction elevator. Using the robot chassis and the pre-calibrated static extrinsic parameter matrix of the lidar, the radar coordinate system is... The coordinates are transformed to the global coordinate system of the robot chassis, and the original relative pose error of the AGV relative to the construction elevator is solved. Step 6: Perform time-series smoothing and noise reduction processing on the original relative pose error, and output the relative pose and pose validity flag of the construction elevator.

2. The method according to claim 1, characterized in that, The two-dimensional laser scanning data refers to the environmental ranging data obtained by the vehicle-mounted two-dimensional laser radar periodically scanning the construction site in the scenario of AGV cross-floor transfer in an intelligent construction site. The two-dimensional laser scanning data is represented as follows: in, This is two-dimensional laser scanning data. For the first Distance value of each laser point, This represents the total number of laser points in a single scan. For the first The horizontal scanning angle corresponding to the laser beam.

3. The method according to claim 2, characterized in that, The two-stage preprocessing of the two-dimensional laser scanning data includes: Distance range filtering: Only retain those that meet the requirements Laser points within the range, of which, For the minimum effective distance, The maximum effective distance; Adaptive continuity filtering: Calculates the distance difference between adjacent laser points : when When the current laser point is identified as an abnormal jump point, it is removed. This is an adaptive dynamic threshold based on ranging depth.

4. The method according to claim 3, characterized in that, Depth-based adaptive dynamic thresholding The expression is: in, For radar angular resolution, This is the environmental dust and noise compensation constant.

5. The method according to claim 1, characterized in that, Step 2 includes: The preprocessed polar coordinate laser points are converted into two-dimensional Cartesian coordinate points using the following formula: in, The horizontal x-coordinate of the obstacle reflection point in the radar coordinate system is taken along the radar X-axis. The horizontal ordinate of the obstacle reflection point in the radar coordinate system is taken along the radar Y-axis. Combine all the transformed coordinates to generate a two-dimensional point set: in, This is a two-dimensional point set of all obstacles in the radar coordinate system, used for subsequent line segment extraction.

6. The method according to claim 1, characterized in that, The expression for the fitted straight line is: in, , , These represent the fixed coefficients of the equation of a standard straight line in a two-dimensional plane. , These represent the x-coordinate and y-coordinate of the laser reflection point in the radar coordinate system, respectively. The distance from the laser point to the fitted line is: in, The vertical distance from a single laser point to a preset straight line.

7. The method according to claim 1, characterized in that, Step 4 includes: Remove abnormal short line segments and long wall line segments; match the target position corresponding to the current construction elevator task and remove irrelevant line segments whose center distance from the target position exceeds the preset distance threshold; Based on the internal geometry of the construction elevator, three geometric constraints of parallelism, perpendicularity, and width are set to select three target line segments that constitute the outline of the construction elevator entrance. Parallel constraints, i.e., constraints on both sides of the construction elevator cavity walls: Select two line segments of the construction elevator cavity side walls that satisfy... ; , These are the fitted direction angles for the two construction elevator inner cavity side wall segments, respectively. Vertical constraint, specifically the vertical constraint between the rear wall and side walls of the construction elevator cavity: Select a line segment of the rear wall of the construction elevator cavity that is approximately perpendicular to both side walls, satisfying... ; The fitting direction angle for the line segment of the rear wall inside the construction elevator cavity; Width constraint, specifically the width constraint of the construction elevator cavity entrance: the vertical distance between the two sidewall segments of the construction elevator cavity matches the preset standard width of the construction elevator cavity, satisfying... ; in, The permissible deviation threshold for the parallel angle. The allowable deviation threshold for the vertical angle. The allowable deviation threshold for the width of the construction elevator; The frame is determined to have detected a compliant construction elevator outline only if it meets all the constraint conditions of parallel constraint, vertical constraint and width constraint. Perform consistency verification on the contour recognition results of consecutive sampling frames: when N consecutive frames meet the constraint judgment conditions, the contour is marked as a valid construction elevator contour.

8. The method according to claim 1, characterized in that, In step 5, the coordinates of the construction elevator cavity center in the radar coordinate system are solved by using the geometric midpoints of the two corner points: in, Let x be the x-coordinate of the center of the construction elevator cavity in the radar coordinate system. The coordinates are the longitudinal coordinates of the center of the construction elevator cavity in the radar coordinate system.

9. The method according to claim 1, characterized in that, Step 6 includes: The relative pose error results are subjected to temporal noise reduction using the following formula: in, This represents the relative pose of the construction elevator after filtering and smoothing at the current time t, i.e., the relative pose of the construction elevator. These are the filter weight coefficients. To obtain the unsmoothed original relative pose error results for a single frame, This is the smoothed relative pose of the construction elevator output after low-pass filtering at the previous moment.

10. The method according to claim 1, characterized in that, The pose validity flag is a Boolean output identifier, and its generation logic is as follows: First condition: All N consecutive frames satisfy the constraint judgment condition to detect the valid internal contour of the construction elevator. Second condition: The construction elevator's internal cavity width, center coordinates, and relative heading angle corresponding to its relative pose all fall within the preset physical range; When both the first and second conditions are met, the pose validity flag is set to valid; if either condition is not met, the pose validity flag is set to invalid.

Images