A laser reflector map construction method for indoor and outdoor shuttle AGV

Abstract

The application discloses a laser reflector plate map construction method for indoor and outdoor shuttle AGV, and comprises the following steps: S1, reflector plate arrangement rule: a, the shape of the reflector plate is a cylindrical shape, which can make the observable range of the reflector plate wider; b, it is ensured that the laser radar can observe at least four reflector plates at any position in the room; c, a triangle formed by any three reflector plates cannot be an equilateral triangle. The method integrates RTK positioning, laser radar positioning and the position of the cylindrical reflector plate into a map optimization problem to solve together, and the generated reflector plate map coordinate system is directly unified with the world coordinate system, thereby solving the problem of too large position jump when indoor and outdoor positioning is switched, and the constraint of RTK positioning can greatly reduce the cumulative error of the indoor reflector plate map, and improve the mapping accuracy; moreover, the method does not need manual surveying and mapping correction map in the mapping process, and the efficiency is improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of sensor positioning, and particularly relates to a laser reflector plate map construction method for indoor and outdoor shuttle AGV. BACKGROUND

[0002] The indoor and outdoor shuttle AGV uses RTK positioning outdoors and uses laser radar positioning indoors, and laser radar positioning usually needs to arrange reflector plates, and the accuracy of the reflector plate map directly affects the positioning accuracy. In the current map making method, the manual surveying and mapping method has a large workload, and the map generated by frame-by-frame matching has the problem of accumulated error; and because different positioning methods are used indoors and outdoors, the AGV needs to switch the positioning strategy when shuttling indoors and outdoors, and there may be the problem of large positioning jump when switching. SUMMARY

[0003] The application provides a laser reflector plate map construction method for indoor and outdoor shuttle AGV to solve the problems in the background.

[0004] In order to achieve the above purpose, the technical scheme adopted by the application is as follows: a laser reflector plate map construction method for indoor and outdoor shuttle AGV, specifically comprising the following steps,

[0005] S1, reflector plate arrangement rule: a, the shape of the reflector plate is a cylindrical shape, which can make the observable range of the reflector plate wider; b, ensure that the laser radar can observe at least 4 reflector plates at any position indoors; c, any triangle formed by 3 reflector plates cannot be an equilateral triangle; d, at least 3 reflector plates are arranged at each entrance from outdoor to indoor;

[0006] S2, data acquisition rule: collect the data required for mapping according to the actual running route of the AGV, including combined navigation positioning data and single-frame point cloud data of the laser radar, collect from outdoor, obtain a fixed solution by RTK, enter indoor, and finally collect from indoor to outdoor, and collect until the RTK is fixed again, and the entrance and the exit do not have to be the same;

[0007] S3, algorithm processing specific steps:

[0008] S3.1, unified coordinate system: transform the single-frame point cloud of the laser radar and the inertial navigation device into a unified vehicle body coordinate system according to the calibration parameters R and T;

[0009] S3.2, single-frame point cloud motion compensation: the data collected when the laser radar is not stationary will have motion distortion, and motion compensation needs to be performed according to the pose transformation when collecting;

[0010] S3.3, extract the reflector plate point cloud and calculate the coordinates of the center in the laser radar coordinate system;

[0011] S3.4, Use triangle matching to obtain the correspondence between the reflector in the lidar coordinate system and the reflector in the map coordinate system;

[0012] S3.5, optimization problem based on the matching relationship of the reflector and the pose construction map of the lidar;

[0013] S3.6 Find the frames with fixed solutions in RTK at the beginning and end, and add RTK pose nodes to constrain the entire graph;

[0014] S3.7 Solve the graph optimization problem to obtain the precise position of all reflectors.

[0015] A method for constructing a laser reflector map for indoor and outdoor shuttle AGVs according to claim 1, characterized in that: the coordinate transformation formula in step S3.1 is:

[0016] P C =RP L +T

[0017] Where P L P is a point cloud in the lidar coordinate system. C This is the point cloud in the vehicle coordinate system.

[0018] The compensation method in S3.2 is as follows: calculate the pose corresponding to each point based on the timestamp, and transform all points to the coordinate system corresponding to the last point of the point cloud in that frame.

[0019] S3.3 involves extracting the point cloud of the reflector and calculating the coordinates of its center in the lidar coordinate system. The specific steps are as follows:

[0020] S3.31: High reflectivity points are extracted as candidate points for the reflector based on the reflection intensity. The reflection intensity measured by different models of lidar may be different, and different thresholds need to be set according to the specific model. This patent extracts points with a reflection intensity greater than 200.

[0021] S3.32: Perform Euclidean distance clustering on the extracted candidate points, and after classification, retain all classes with 5 or more points as reflector points, and discard classes with less than 5 points;

[0022] S3.33: The center of the reflector is estimated using the trigonometric function method. Assuming the number of extracted reflector points is m and the radius of the reflector is r, the data for each point in the polar coordinate system of the lidar is represented as S. i =(ρ i ,φ i The center position of the reflector is B = (L) B ,β B Then, based on the symmetry of the circle, the central angle β can be obtained. B :

[0023]

[0024] The distance L can be obtained using trigonometric functions B :

[0025] θ i = φ i - β B

[0026]

[0027]

[0028] The S3.4, using triangle matching to obtain the corresponding relationship between the retroreflective panel in the laser radar coordinate system and the retroreflective panel in the map coordinate system, the specific steps are:

[0029] S3.41: Calculate the distance between all retroreflective panels, sort these distances from large to small, and eliminate the line segments greater than the laser radar sensing range;

[0030] S3.42: Start from the longest line segment and sequentially search for line segments with approximately equal distances in the constructed map, and the two retroreflective panels forming the line segment are used as seed retroreflective panels, and the corresponding relationship of the seed retroreflective panels in the radar coordinate system and the map coordinate system is recorded;

[0031] S3.43: The remaining other retroreflective panels and the seed retroreflective panels form a triangle, and the triangle formed in the same way in the map is matched, and if the lengths of the three sides are approximately equal, it is considered that the matching is successful;

[0032] S3.44: Use the method in b and c to traverse all line segments, and finally take the corresponding relationship with the most matching triangles as the optimal matching.

[0033] The S5, according to the matching relationship of the retroreflective panel and the laser radar pose, constructs a graph optimization problem, and the specific steps are:

[0034] The edge between adjacent laser radar pose nodes is a relative pose transformation, the edge between the retroreflective panel position node and the laser radar pose node is the distance from the laser radar point cloud of this frame to the retroreflective panel, and the optimization target is the position of the retroreflective panel and the pose of each key frame laser radar; If there is a retroreflective panel that is not found in a new frame of point cloud, it is added to the retroreflective panel map as a new retroreflective panel; If a retroreflective panel is observed by less than n key frames, it is considered to be a false target, and this node is discarded when solving the graph optimization problem, and in this example, n is set to 2.

[0035] The beneficial effects of the above technical solutions are:

[0036] The method integrates RTK positioning, laser radar positioning and the position of the cylindrical reflector into a graph optimization problem to solve together, and the generated reflector map coordinate system is directly unified with the world coordinate system, solving the problem of too large position jump when switching indoor and outdoor positioning, and the constraint of RTK positioning can greatly reduce the cumulative error of the indoor reflector map, improving the mapping accuracy; and the method does not need manual mapping and correction of the map during mapping, improving the efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0037] Fig. 1 is the algorithm flowchart of the present application;

[0038] Fig. 2 is the schematic diagram of the graph optimization structure of the present application;

[0039] Fig. 3 is the schematic diagram of the graph optimization structure - the schematic diagram of the structure after adding RTK; DETAILED DESCRIPTION

[0040] The specific embodiments of the present application will be further described in detail below with reference to the drawings, and the purpose is to help the technicians in the field to have a more complete, accurate and in-depth understanding of the concept and technical solution of the present application, and to help its implementation.

[0041] As shown in Figs. 1-3 , the present application is a laser reflector map construction method for indoor and outdoor shuttle AGV, which integrates RTK positioning, laser radar positioning and the position of the cylindrical reflector into a graph optimization problem to solve together, and the generated reflector map coordinate system is directly unified with the world coordinate system, solving the problem of too large position jump when switching indoor and outdoor positioning, and the constraint of RTK positioning can greatly reduce the cumulative error of the indoor reflector map, improving the mapping accuracy; and the method does not need manual mapping and correction of the map during mapping, improving the efficiency.

[0042] Specifically, as shown in Figs. 1-3 , it specifically includes the following steps,

[0043] S1, reflector arrangement rule: a, the shape of the reflector is cylindrical, which can make the observable range of the reflector wider; b, ensure that the laser radar can observe at least 4 reflectors at any position in the room; c, any three reflectors cannot form an equilateral triangle; d, at least 3 reflectors are arranged at each entrance from outdoor to indoor;

[0044] S2, data collection rule: collect data required for mapping according to the actual running route of AGV, including combined navigation positioning data, single-frame point cloud data of laser radar, collect from outdoor, enter indoor after RTK obtains fixed solution, finally from indoor to outdoor, collect after RTK re-fixes, and the entrance and exit do not have to be the same;

[0045] S3, algorithm processing specific steps:

[0046] S3.1, unified coordinate system: transform single-frame point cloud of laser radar and inertial navigation device into unified vehicle body coordinate system according to calibration parameters R, T;

[0047] S3.2, single-frame point cloud motion compensation: data collected when laser radar is not stationary will have motion distortion, and motion compensation needs to be performed according to the pose transformation at the time of collection;

[0048] S3.3, extract retroreflective panel point cloud and calculate the coordinates of the center in the laser radar coordinate system;

[0049] S3.4, use triangle matching to obtain the corresponding relationship between the retroreflective panel in the laser radar coordinate system and the retroreflective panel in the map coordinate system;

[0050] S3.5, construct a graph optimization problem according to the matching relationship of the retroreflective panel and the pose of the laser radar;

[0051] S3.6, find the frames with fixed solution of RTK at the beginning and the end, and add the RTK pose node to constrain the entire graph;

[0052] S3.7, solve the graph optimization problem, that is, the accurate positions of all retroreflective panels can be obtained.

[0053] According to the laser retroreflective panel mapping method for indoor and outdoor shuttle AGV in claim 1, the coordinate transformation formula of S3.1 is:

[0054] P C =RP L +T

[0055] Where P L is the point cloud in the laser radar coordinate system, and P C is the point cloud in the vehicle body coordinate system.

[0056] The compensation method in S3.2 is to calculate the pose corresponding to each point according to the timestamp, and transform all points to the coordinate system corresponding to the last point of the frame point cloud.

[0057] S3.3, extract the retroreflective panel point cloud and calculate the coordinates of the center in the laser radar coordinate system, and the specific steps are:

[0058] S3.31: Extract the points with high reflectivity as the candidate points of the reflector according to the reflection intensity. The reflection intensity measured by different types of laser radars may be different, and different thresholds need to be set according to the specific type. The points with a reflection intensity greater than 200 are extracted in this patent;

[0059] S3.32: Perform Euclidean distance clustering on the extracted candidate points. After classification, all classes with 5 or more points are retained as reflector points, and classes with less than 5 points are discarded;

[0060] S3.33: Estimate the center of the reflector using the trigonometric function method. Assuming that the number of extracted reflector points is m and the radius of the reflector is r, in polar coordinates, the data of each point in the laser radar coordinate system is represented as S i =(ρ i ,φ i ). The center position of the reflector is B=(L B ,β B ). According to the symmetry of the circle, the center angle β B can be obtained:

[0061]

[0062] The distance L B can be obtained using the trigonometric function:

[0063] θ i =φ i -β B

[0064]

[0065]

[0066] S3.4, use triangle matching to obtain the correspondence between the reflector in the laser radar coordinate system and the reflector in the map coordinate system. The specific steps are as follows:

[0067] S3.41: Calculate the distance between all pairs of reflectors, sort these distances from large to small, and remove the line segments greater than the perception range of the laser radar;

[0068] S3.42: Start from the longest line segment and sequentially search for line segments with approximately equal distances in the constructed map, where the threshold is set to 0.2m. The two reflectors forming the line segment are used as seed reflectors, and the correspondence between the seed reflectors in the radar coordinate system and the map coordinate system is recorded;

[0069] S3.43: The remaining other reflectors and seed reflectors form a triangle, and the triangle formed in the same way in the map is matched. If the lengths of the three sides are approximately equal, it is considered to be a successful match;

[0070] S3.44: Use the way in b, c to traverse all line segments, finally take the corresponding relationship of the most matching successful triangle number as the optimal matching.

[0071] S5, according to the matching relationship of the retroreflective panel and the laser radar pose, a graph optimization problem is constructed, and the specific steps are as follows:

[0072] The edge between the adjacent laser radar pose nodes is the relative pose transformation, the edge between the retroreflective panel position node and the laser radar pose node is the distance from the laser radar point cloud of the frame to the retroreflective panel, the optimization target is the position of the retroreflective panel and the pose of each key frame laser radar; if there is a retroreflective panel that is not found in a new frame of point cloud, the retroreflective panel is added to the retroreflective panel map as a new retroreflective panel; if a retroreflective panel is observed by less than n key frames, the retroreflective panel is considered as a false target, and this node is discarded when solving the graph optimization problem, and in this example, n is set to 2.

[0073] The above is an exemplary description of the present application in combination with the drawings, and it is obvious that the specific implementation of the present application is not limited by the above method, as long as various non-essential improvements are made by using the method concept and technical solution of the present application; or without improvement, the above concept and technical solution of the present application is directly applied to other occasions, which is within the protection scope of the present application.

Claims

1. A method for constructing a laser reflector map for indoor and outdoor shuttle AGVs, characterized in that: Specifically comprising the following steps, S1, the arrangement rule of the reflector plate: a, the reflector plate shape adopts a cylindrical shape, which can make the observable range of the reflector plate wider; b, ensure that the lidar can observe at least 4 reflector plates at any position in the room; c, any three reflector plates cannot form an equilateral triangle; d, at least 3 reflector plates are arranged at each entrance from outdoor to indoor; S2, the data acquisition rule: collect the data required for mapping according to the actual running route of the AGV, including combined navigation positioning data and single-frame point cloud data of the lidar, start from outdoor, obtain a fixed solution by RTK, enter indoor, and finally from indoor to outdoor, re-fix by RTK, and collect data until the end, the entrance and the exit do not have to be the same; S3, the specific steps of algorithm processing: S3.1, unified coordinate system: transform the single-frame point cloud of the lidar and the inertial navigation device into a unified vehicle body coordinate system according to the calibration parameters R and T; S3.2, single-frame point cloud motion compensation: the data collected when the lidar is not stationary will have motion distortion, which needs to be compensated according to the pose transformation at the time of collection; S3.3, extract the reflector plate point cloud and calculate the coordinates of the center in the lidar coordinate system; S3.4, use triangle matching to obtain the corresponding relationship between the reflector plates in the lidar coordinate system and the reflector plates in the map coordinate system; S3.5, construct a graph optimization problem according to the matching relationship of the reflector plates and the pose of the lidar; S3.6, find the frames with fixed solutions at the beginning and the end of RTK, add the RTK pose nodes to constrain the entire graph; S3.7, solve the graph optimization problem, that is, the accurate positions of all reflector plates can be obtained. 2.The laser retro-reflective plate map construction method for an indoor-outdoor shuttling AGV according to claim 1, characterized in that: The coordinate transformation formula of S3.1 is: ， wherein is a point cloud in the laser radar coordinate system, is a point cloud in the vehicle body coordinate system. 3.The laser retro-reflective plate map construction method for an indoor-outdoor shuttling AGV according to claim 1, characterized in that: The compensation method in S3.2 is: calculate the pose corresponding to each point according to the timestamp, and transform all points to the coordinate system corresponding to the last point of the frame point cloud. 4.The laser retro-reflective plate map construction method for an indoor-outdoor shuttling AGV according to claim 1, characterized in that: S3.3, extract the reflector plate point cloud and calculate the coordinates of the center in the lidar coordinate system, the specific steps are: S3.31: extract high-reflectivity points as candidate points according to the reflection intensity, different models of lidar may have different reflection intensities, so different thresholds need to be set according to the specific model, and points with reflection intensity greater than 200 in the lidar scanning data are extracted as candidate reflector plate points; S3.32: perform Euclidean distance clustering on the extracted candidate points, and retain all classes with more than or equal to 5 points as reflector plate points after classification, and discard classes with less than 5 points; S3.33: Estimate the center of the retro-reflective plate using the trigonometric method, assuming that the number of extracted retro-reflective plate points is m, and the radius of the retro-reflective plate is r. In polar coordinates, the data of each point in the laser radar coordinate system is represented as , the center position of the retro-reflective plate , the center angle of the circle can be obtained according to the symmetry of the circle : ； The distance can be found using trigonometry : ； ； 。 5.The laser retro-reflective plate map construction method for an indoor-outdoor shuttling AGV according to claim 1, characterized in that: S3.4, use triangle matching to obtain the corresponding relationship between the reflector plates in the lidar coordinate system and the reflector plates in the map coordinate system, the specific steps are: S3.41: calculate the distance between all pairs of reflector plates, sort these distances from large to small, and remove the line segments greater than the perception range of the lidar; S3.42: start from the longest line segment and sequentially search for line segments with approximately equal distances in the constructed map, and the two reflector plates forming the line segment are used as seed reflector plates, and the corresponding relationship of the seed reflector plates in the radar coordinate system and the map coordinate system is recorded. S3.43: The remaining other retro-reflective panels and the seed retro-reflective panel form a triangle, which is matched with the triangle formed in the same way in the map. If the lengths of the three sides are approximately equal, it is considered that the matching is successful; S3.44: Use the method in b and c to traverse all line segments, and finally take the corresponding relationship of the group with the most matching successful triangles as the optimal matching. 6.The laser retro-reflective plate map construction method for an indoor-outdoor shuttling AGV according to claim 1, characterized in that: The step S3.5 constructs a map optimization problem according to the matching relationship of the retro-reflective panel and the pose of the laser radar, and the specific steps are as follows: The edge between adjacent laser radar pose nodes is the relative pose transformation, the edge between the retro-reflective panel position node and the laser radar pose node is the distance from the laser radar point cloud of the frame to the retro-reflective panel, and the optimization target is the position of the retro-reflective panel and the pose of each key frame laser radar; If there is a retro-reflective panel that is not found in a new frame of point cloud, it is added to the retro-reflective panel map as a new retro-reflective panel; If a retro-reflective panel is observed by less than n key frames, it is considered to be a false target, and this node is discarded when solving the map optimization problem, and n is set to 2.

Images