AGV positioning method and system based on double reflectors

Abstract

The invention relates to an AGV positioning method and system based on double reflectors, and relates to the technical field of AGV positioning. The AGV positioning method comprises the following steps: switching a vehicle-mounted positioning mode to reflector positioning according to a preset reflector area, and calculating a single-system laser radar pose; extracting a plurality of high reflection intensity points according to a preset reflection intensity threshold in combination with the laser scanning frequency to form a high reflection point cluster; according to the diameter of the reflector, judging the distance between the head and tail high-reflection points of the high-reflection point cluster, matching a reflector model, and calculating a virtual center coordinate in a radar coordinate system; the virtual center coordinates are converted to a map coordinate system according to the single-system laser radar pose, actual center coordinates in a map are compared and matched, and a reflector coordinate set is screened; calculating line segment angles of the reflectors in different coordinate systems according to the reflector coordinate group, and aligning the different coordinate systems to obtain a rotary laser radar pose; and carrying out fusion operation on the plurality of rotary laser radar poses, and determining the running position of the AGV.

Description

Technical Field

[0001] This invention relates to the field of AGV positioning technology, and in particular to an AGV positioning method and system based on dual reflectors. Background Technology

[0002] As an important category of industrial robots, AGV forklifts integrate automatic navigation, multi-sensor control, and network interaction, enabling them to autonomously complete loading, unloading, and short-distance transportation tasks. Facing the ever-increasing demands of warehousing and logistics, unmanned AGV forklifts are gradually replacing traditional manual forklifts, demonstrating broad application prospects due to their high efficiency and flexibility.

[0003] For AGVs to complete autonomous handling tasks, accurate estimation of their own position is a prerequisite, especially in high-precision picking, placing, and stacking tasks, where positioning accuracy must be within ±10mm. Due to the complex and ever-changing warehousing and logistics environment, the movement of goods, forklifts, operators, and other dynamic objects significantly impact positioning accuracy. Visual / laser SLAM positioning, relying solely on natural environment navigation, is ill-suited to such dynamically changing scenarios. A typical scenario is drive-in racking, where AGVs need to drive into racks surrounded by stacked goods to pick up or place items. The surrounding environment is almost entirely different, making natural environment-based positioning impossible. Therefore, in dynamic scenarios, AGVs require auxiliary facilities such as reflectors and QR codes for positioning.

[0004] Application number CN202010150471.6 discloses a high-precision positioning method for AGVs that integrates the positioning of reflectors and laser features. The method includes establishing a laser map to clearly show the outline of the reflector, acquiring the reflector's position on the map, initiating AMCL positioning, using the AMCL positioning as the initial optimization value to select possible reflector positions, obtaining the distance from the reflector through laser scanning, and performing reverse calculation and optimization to obtain the precise position of the AGV. If the reflector cannot be identified, the AMCL positioning result is used as the AGV's position estimate. This invention does not rely on the strict installation of the reflector and achieves better accuracy than laser positioning, enabling industrial AGVs to operate stably in production environments for extended periods.

[0005] The existing technical solutions mentioned above have the following drawbacks: 1. Existing AGV positioning methods lack motion prediction assistance and adaptation mechanisms for the geometric features of non-standard reflectors, which can easily lead to false associations and missed identifications during the reflector matching stage. Furthermore, when relying on a single reflector for calculation, the anti-interference capability is weak, resulting in poor pose estimation stability. Summary of the Invention

[0006] To address the shortcomings of existing technologies, the present invention aims to provide an AGV positioning method and system based on dual reflectors. By combining motion prediction with geometric feature recognition of columnar reflectors through a reflector matching mechanism, the accuracy and efficiency of reflector matching in complex environments are effectively improved. At the same time, the stability of pose estimation is enhanced through multi-reflector combination calculation and angle normalization fusion strategy.

[0007] This was achieved using the following technical solutions: An AGV positioning method based on dual reflectors includes: Based on the preset reflector area, the vehicle positioning mode is switched to reflector positioning, and the pose of the single-system LiDAR is calculated. Based on the preset reflection intensity threshold and the laser scanning frequency, several high reflection intensity points are extracted to form a high reflection point cluster; The distance between the first and last high reflective points of the high reflective point cluster is determined based on the reflector diameter, and the reflector model is matched to calculate the virtual center coordinates in the radar coordinate system. The virtual center coordinates are converted to map coordinates based on the pose of the single-system lidar, and the actual center coordinates in the map are compared and matched to filter the reflector coordinate group. Calculate the line segment angles between reflectors in different coordinate systems based on the reflector coordinate set, align the different coordinate systems, and obtain the pose of the rotating lidar. The AGV vehicle's operating position is determined by fusing and calculating the poses of several rotating lidar sensors.

[0008] By adopting the above technical solution, feature points are extracted from the point cloud through reflection intensity threshold clustering and reflector diameter geometric verification. The virtual center coordinates in the radar coordinate system are associated with the map coordinates by combining the nearest neighbor matching algorithm. The radar pose is solved by the coordinate system alignment method based on vector angle calculation. Finally, the precise position of the AGV is output through a multi-hypothesis fusion strategy, realizing efficient, stable and interference-resistant accurate positioning in the reflector navigation area.

[0009] The present invention is further configured such that: the specific steps of switching the vehicle positioning mode to reflector positioning according to the preset reflector area and calculating the pose of the single-system LiDAR include: The AGV vehicle's operating position is self-monitored. If the AGV vehicle moves to the preset reflector area, the positioning switching logic of the vehicle positioning mode is triggered. According to the positioning switching logic, the AGV vehicle starts the lidar, automatically switches to reflector positioning, and obtains the vehicle's position coordinates, heading angle, linear velocity and angular velocity at the previous moment. The heading angle at the current moment is calculated by combining the differential driving algorithm with the heading angle at the previous moment and the sampling time difference. Based on the current heading angle and the linear and angular velocities of the previous moment, the vehicle position coordinates of the previous moment are differentially calculated to obtain the vehicle position coordinates of the current moment. Based on the vehicle's current position coordinates and heading angle, determine the pose of the single-system lidar in the map coordinate system.

[0010] By adopting the above technical solution, the positioning mode switching of the reflector is triggered by real-time monitoring of the AGV position. Based on the differential driving kinematics model, the pose and velocity information of the previous moment are used to accurately predict the pose of the current lidar in the map coordinate system through time integration and coordinate transformation. This achieves smooth switching of positioning mode and continuity of pose prediction, significantly improving matching efficiency and system stability.

[0011] The present invention is further configured such that: the specific steps of extracting several high-reflection intensity points based on a preset reflection intensity threshold and laser scanning frequency to form a high-reflection point cluster include: The reflector is scanned according to the laser scanning frequency to obtain several laser point data, and the point cloud reflection intensity is calculated. The laser point data is judged and filtered based on a preset reflection intensity threshold and the reflection intensity of the point cloud. If the reflection intensity of the point cloud is greater than the reflection intensity threshold, then target points are extracted from the current laser point data to obtain several high reflection intensity points; The high reflectivity points are clustered and grouped to generate a high reflectivity point cluster.

[0012] By adopting the above technical solution, laser point clouds are quickly screened by setting a reflection intensity threshold, and adjacent high-reflection points are grouped into high-reflection point clusters based on the continuity of the scanning sequence using a clustering algorithm. This achieves efficient and robust extraction of reflector features in complex environments, significantly improving reflector recognition efficiency and anti-interference capability.

[0013] The present invention is further configured such that: the specific steps of determining the distance between the first and last high-reflectivity points of the high-reflectivity point cluster based on the reflector diameter, matching the reflector model, and calculating the virtual center coordinates in the radar coordinate system include: The high-reflection point clusters are analyzed based on the laser scanning sequence to extract the first and last high-reflection points; Perform distance calculations on the first high inversion point and the last high inversion point to obtain the Euclidean distance of the point cluster; The Euclidean distance of the point cluster is determined based on the reflector diameter. If the Euclidean distance of the point cluster is greater than the reflector radius but less than twice the reflector diameter, then the current high-reflectivity point cluster is determined to conform to the geometric characteristics of the reflector, and the corresponding reflector model is matched. A fitting operation is performed on the cluster of high-reflection points to obtain a fitted circular arc, and the midpoint of the high-reflection point is located. The first center coordinates are determined based on the radius of curvature and direction of the fitted arc, combined with the diameter of the reflector. Based on the radar origin, the high-reflectivity intermediate point is directionally extended, and the extension distance is the radius of the reflector, to determine the coordinates of the second center. By performing a weighted operation on the first center coordinates and the second center coordinates, the virtual center coordinates of the reflector model in the radar coordinate system are obtained.

[0014] By adopting the above technical solution, the effective targets are screened by calculating the Euclidean distance between the first and last points of the high reflectivity cluster and performing geometric verification based on the reflector diameter. The curvature center is derived by combining the circular arc fitting algorithm and the radial extension method based on the radar origin to calculate the coordinates of the first and second centers respectively. Finally, the virtual center coordinates are obtained by weighted fusion. The multi-algorithm complementary mechanism effectively reduces the impact of scanning angle and point cloud noise on center positioning, and significantly improves the accuracy of reflector recognition and the robustness of the system.

[0015] The present invention is further configured such that: the specific steps of calculating the line segment angles between the reflectors in different coordinate systems based on the reflector coordinate set, aligning the different coordinate systems, and obtaining the pose of the rotating lidar include: Based on the reflector coordinate system, the angles of line segments between different reflectors are calculated in different coordinate systems. Line segment angle in radar coordinate system Angle of line segment in map coordinate system ; in, Number the reflectors. For virtual center coordinates, The actual center coordinates; The difference between the line segment angles in the radar coordinate system and the line segment angles in the map coordinate system is calculated to obtain the deviation conversion angle. ; The deviation conversion angle is normalized according to a preset angle range to obtain the standard conversion angle; The radar coordinate system and the map coordinate system are rotated and aligned according to the standard transformation angle to obtain a rotated coordinate system, and the composite center coordinates corresponding to the reflector are calculated. : ; Based on the composite center coordinates and the actual center coordinates of any reflector, coordinate transformation is performed on the lidar to obtain the corresponding rotating lidar pose. .

[0016] By adopting the above technical solution, the deviation angle is obtained by calculating the vector angle between the reflector and the line connecting the radar and map coordinate systems. An angle normalization algorithm is used to eliminate the circular jump problem. Then, the precise pose of the radar is derived by coordinate system rotation transformation and composite center coordinates. The global pose is directly solved by geometric constraints, avoiding the computational complexity caused by iterative optimization, and significantly improving the real-time performance of pose estimation and the robustness of cross-coordinate system matching.

[0017] The present invention is further configured such that the specific steps for fusing and calculating several of the rotating lidar poses to determine the running position of the AGV vehicle include: Match n reflectors to each other to obtain n(n-1) / 2 pairs of reflectors, and calculate n(n-1) / 2 standard conversion angles and n(n-1) rotating lidar coordinates; If the standard conversion angle is in the negative angle range, then the standard conversion angle is converted to a full angle to obtain the corresponding positive conversion angle; The average conversion angle is calculated by averaging all the positive conversion angles, and then compared with the straight angle for judgment. If the average conversion angle is greater than the horizontal angle, the difference between the average conversion angle and the circumference angle is calculated to obtain the final conversion angle and determine the heading angle of the AGV vehicle. The median value of all the rotating lidar coordinates is calculated to obtain the median coordinates of the lidar; The coordinates of the rotating lidar are calculated by performing a difference operation on the median coordinates of the radar to obtain a coordinate distance value, which is then compared with a preset coordinate difference threshold. If the coordinate distance value is within the coordinate difference range, then the current rotating lidar coordinates are determined to be compliant. The average value of all compliant rotating lidar coordinates is calculated to obtain the final rotating lidar coordinates, which determines the driving position of the AGV vehicle.

[0018] By adopting the above technical solutions, the negative angle and circumferential angle jump problems are handled by the angle normalization algorithm, and the abnormal pose values ​​are eliminated by median screening and threshold judgment. Finally, the optimal positioning result is obtained by weighted fusion. The multi-coordinate fusion strategy significantly improves the robustness and real-time performance of the AGV positioning system in scenarios where the reflectors are sparsely distributed or partially obscured.

[0019] Secondly, the present invention also provides an AGV positioning system based on dual reflectors, which adopts the following technical solution: An AGV positioning system based on dual reflectors includes: The pose determination module is used to switch the vehicle positioning mode to reflector positioning based on the preset reflector area and calculate the pose of the single-system LiDAR. The point cluster construction module is used to extract several high-reflection intensity points based on a preset reflection intensity threshold and laser scanning frequency to form a high-reflection point cluster. The virtual calculation module is used to determine the distance between the first and last high reflective points of the high reflective point cluster based on the reflector diameter, match the reflector model, and calculate the virtual center coordinates. The coordinate filtering module is used to convert the virtual center coordinates to the map coordinate system based on the pose of the single-system lidar, compare and match the actual center coordinates, and filter the reflector coordinate group. The coordinate alignment module is used to calculate the line segment angles between reflectors in different coordinate systems based on the reflector coordinate group, and align the different coordinate systems to obtain the pose of the rotating lidar. The vehicle positioning module is used to fuse and calculate the poses of several rotating lidar sensors to determine the running position of the AGV vehicle.

[0020] By adopting the above technical solution, the pose determination module realizes the switching of reflector positioning mode and radar pose prediction based on differential driving algorithm and kinematic model; the point cluster construction module extracts high reflectivity point clusters by clustering using reflection intensity threshold and scanning frequency; the virtual calculation module calculates the virtual center coordinates by geometric verification and arc fitting weighted calculation; the coordinate filtering module filters reflector coordinate groups by coordinate transformation and nearest neighbor matching; the coordinate alignment module realizes coordinate system rotation alignment by vector angle calculation and angle normalization; and the vehicle positioning module processes multi-pose data by angle circumference transformation, median filtering and weighted fusion to finally determine the precise position of AGV. Through the collaborative work of multiple module algorithm chains, the real-time performance, accuracy and robustness to complex environments of the positioning system are significantly improved.

[0021] Thirdly, the present invention also provides an electronic device, comprising: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement any of the methods described in the above scheme.

[0022] Fourthly, the present invention also provides a storage medium storing at least one instruction, at least one program, code set, or instruction set, wherein the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by a processor to implement the AGV positioning method based on dual reflectors as described above.

[0023] In summary, the beneficial technical effects of the present invention are as follows: 1. By combining motion prediction with geometric feature recognition of columnar reflectors, the reflector matching mechanism effectively improves the accuracy and efficiency of reflector matching in complex environments. At the same time, through multi-reflector combination calculation and angle normalization fusion strategy, the stability of pose estimation and the fault tolerance of the system are significantly enhanced. 2. The global pose is directly calculated through the geometric relationship of the dual reflectors, avoiding the real-time problems caused by iterative calculations. At the same time, the angle normalization and median filtering mechanisms in multi-pose fusion effectively solve the problems of critical angle jumps and abnormal coordinate point interference, greatly improving the robustness and reliability of the positioning output. Attached Figure Description

[0024] Figure 1 This is a flowchart illustrating an AGV positioning method according to one embodiment of the present invention.

[0025] Figure 2 This is a schematic diagram illustrating the calculation of the center position of the reflector according to one embodiment of the present invention; Figure 3 This is a schematic diagram of the coordinates of a reflector according to one embodiment of the present invention; Figure 4 This is a schematic diagram of the structure of an AGV positioning system according to one embodiment of the present invention. Detailed Implementation

[0026] The present invention will be further described in detail below with reference to the accompanying drawings.

[0027] Reference Figure 1 The present invention discloses an AGV positioning method based on dual reflectors, comprising: S1: Based on the preset reflector area, the vehicle positioning mode is switched to reflector positioning, and the pose of the single-system LiDAR is calculated; S2: Based on the preset reflection intensity threshold and the laser scanning frequency, extract several high reflection intensity points to form a high reflection point cluster; S3: Determine the distance between the first and last high reflective points of the high reflective point cluster based on the reflector diameter, match the reflector model, and calculate the virtual center coordinates in the radar coordinate system; S4: Based on the pose of the single-system lidar, convert the virtual center coordinates to the map coordinate system, compare and match the actual center coordinates in the map, and filter the reflector coordinate group; S5: Calculate the line segment angles between the reflectors in different coordinate systems based on the reflector coordinate set, align the different coordinate systems, and obtain the pose of the rotating lidar; S6: Perform fusion calculations on the poses of several rotating lidar sensors to determine the running position of the AGV vehicle.

[0028] The implementation principle of this embodiment is as follows: the reflector positioning mode is automatically triggered by real-time monitoring of the AGV position, and the initial radar pose is estimated based on the differential driving model; high reflectivity point clusters are extracted by clustering using reflection intensity threshold and scanning frequency, and the virtual center coordinates of the reflector are calculated by geometric feature verification and weighted fusion algorithm; a reflector coordinate group is established by coordinate transformation and nearest neighbor matching, and the coordinate system is rotated and aligned by vector angle calculation and angle normalization; finally, the circumferential angle jump and coordinate anomaly are handled by multi-hypothesis fusion strategy, and the accurate pose of the AGV is finally output, realizing efficient, stable and anti-interference accurate positioning in the navigation area.

[0029] Step S1 includes: The AGV vehicle's operating position is self-monitored. If the AGV vehicle moves to the preset reflector area, the positioning switching logic of the vehicle positioning mode is triggered. According to the positioning switching logic, the AGV vehicle starts the lidar, automatically switches to reflector positioning, and obtains the vehicle's position coordinates, heading angle, linear velocity and angular velocity at the previous moment. The heading angle at the current moment is calculated by combining the differential driving algorithm with the heading angle at the previous moment and the sampling time difference. Based on the current heading angle and the linear and angular velocities of the previous moment, the vehicle position coordinates of the previous moment are differentially calculated to obtain the vehicle position coordinates of the current moment. Based on the vehicle's current position coordinates and heading angle, determine the pose of the single-system lidar in the map coordinate system.

[0030] Step S2 includes: The reflector is scanned according to the laser scanning frequency to obtain several laser point data, and the point cloud reflection intensity is calculated. The laser point data is judged and filtered based on a preset reflection intensity threshold and the reflection intensity of the point cloud. If the reflection intensity of the point cloud is greater than the reflection intensity threshold, then target points are extracted from the current laser point data to obtain several high reflection intensity points; The high reflectivity points are clustered and grouped to generate a high reflectivity point cluster.

[0031] Step S3 includes: The high-reflection point clusters are analyzed based on the laser scanning sequence to extract the first and last high-reflection points; Perform distance calculations on the first high inversion point and the last high inversion point to obtain the Euclidean distance of the point cluster; The Euclidean distance of the point cluster is determined based on the reflector diameter. If the Euclidean distance of the point cluster is greater than the reflector radius but less than twice the reflector diameter, then the current high-reflectivity point cluster is determined to conform to the geometric characteristics of the reflector, and the corresponding reflector model is matched. A fitting operation is performed on the cluster of high-reflection points to obtain a fitted circular arc, and the midpoint of the high-reflection point is located. The first center coordinates are determined based on the radius of curvature and direction of the fitted arc, combined with the diameter of the reflector. Based on the radar origin, the high-reflectivity intermediate point is directionally extended, and the extension distance is the radius of the reflector, to determine the coordinates of the second center. By performing a weighted operation on the first center coordinates and the second center coordinates, the virtual center coordinates of the reflector model in the radar coordinate system are obtained.

[0032] Step S4 includes: The virtual center coordinates in the radar coordinate system are transformed based on the pose of the single-system lidar to obtain the global center coordinates. The global center coordinates are matched with the actual center coordinates on the map, the coordinate difference is calculated, and a judgment is made against a preset matching threshold. If the coordinate difference is less than the matching threshold, the current reflector is determined to be successfully matched, and the virtual center coordinates and the actual center coordinates are aggregated to construct a reflector coordinate group.

[0033] Step S5 includes: Based on the reflector coordinate system, the angles of line segments between different reflectors are calculated in different coordinate systems. Line segment angle in radar coordinate system Angle of line segment in map coordinate system ; in, Number the reflectors. For virtual center coordinates, The actual center coordinates; The difference between the line segment angles in the radar coordinate system and the line segment angles in the map coordinate system is calculated to obtain the deviation conversion angle. ; The deviation conversion angle is normalized according to a preset angle range to obtain the standard conversion angle; The radar coordinate system and the map coordinate system are rotated and aligned according to the standard transformation angle to obtain a rotated coordinate system, and the composite center coordinates corresponding to the reflector are calculated. : ; Based on the composite center coordinates and the actual center coordinates of any reflector, coordinate transformation is performed on the lidar to obtain the corresponding rotating lidar pose. .

[0034] Step S6 includes: Match n reflectors to each other to obtain n(n-1) / 2 pairs of reflectors, and calculate n(n-1) / 2 standard conversion angles and n(n-1) rotating lidar coordinates; If the standard conversion angle is in the negative angle range, then the standard conversion angle is converted to a full angle to obtain the corresponding positive conversion angle; The average conversion angle is calculated by averaging all the positive conversion angles, and then compared with the straight angle for judgment. If the average conversion angle is greater than the horizontal angle, the difference between the average conversion angle and the circumference angle is calculated to obtain the final conversion angle and determine the heading angle of the AGV vehicle. The median value of all the rotating lidar coordinates is calculated to obtain the median coordinates of the lidar; The coordinates of the rotating lidar are calculated by performing a difference operation on the median coordinates of the radar to obtain a coordinate distance value, which is then compared with a preset coordinate difference threshold. If the coordinate distance value is within the coordinate difference range, then the current rotating lidar coordinates are deemed compliant. The average value of all compliant rotating lidar coordinates is calculated to obtain the final rotating lidar coordinates, which determines the driving position of the AGV vehicle.

[0035] The implementation principle of this embodiment is as follows: the reflector positioning mode is switched by triggering the AGV position self-monitoring; the initial radar pose is estimated by fusing historical pose and motion state based on the differential driving model; high reflectivity point clusters are generated by using reflection intensity threshold screening and point cloud clustering; virtual coordinates are calculated by geometric feature verification and dual-algorithm weighted fusion (circular arc fitting center and radial extension center); reflector coordinate group is constructed by using coordinate transformation and nearest neighbor matching; coordinate system rotation and alignment are achieved by using vector angle calculation and angle normalization; finally, the circumferential angle jump problem is handled by a multi-hypothesis fusion strategy; abnormal poses are eliminated by combining median screening and threshold judgment; and the final AGV pose is output by weighted average, forming a complete closed loop from environmental perception to positioning decision, which significantly improves the accuracy and robustness of the system in complex scenarios. Example

[0036] When the AGV enters the reflector area, it switches from other positioning methods (such as laser SLAM positioning, QR code positioning, etc.) to reflector positioning. Based on the position and velocity information of the previous moment, the current pose of the navigation radar is estimated; Extract continuous high-reflectivity points from the current laser scanning frame. If the reflection intensity exceeds the preset value T0, multiple high-reflectivity point clusters are formed. Assume the reflector is cylindrical with a diameter of D. For each cluster of high reflectivity points, calculate the distance L between the first and last high reflectivity points. If L > D / 2 and L < 2D, then the cluster of high reflectivity points corresponds to one reflector.

[0037] Calculating the reflector center: To ensure the reflector position is the same when scanned from different angles, the center of the reflector needs to be further determined. For cylindrical reflectors, the center position can be estimated from the arc contour of the high-reflectivity point cluster. A rough calculation can be performed by searching for the midpoint of the high-reflectivity point cluster and extending it by D / 2 as the reflector center position, such as... Figure 3 As shown.

[0038] Reflector matching: Based on the radar's current pose, the reflector is transformed into the global map coordinate system and its coordinates are compared one by one with those of reflectors on the map. If the coordinate difference is less than a set threshold, the reflector is considered to have matched successfully. This results in multiple sets of successfully matched reflector coordinates, each set containing both the radar coordinates and the reflector coordinates on the map, such as... Figure 4 As shown.

[0039] LiDAR pose calculation: Angle: Calculate the angle of the line segment from reflector 1 to reflector 2 in both the radar coordinate system and the map coordinate system. The difference is the angle of the radar coordinate system in the map coordinate system. The specific calculation formula is as follows: Line segment angles in radar coordinate system: ; Line segment angles in map coordinate system: ; Radar angle in map coordinate system: Normalize the angle to the range of [-180°, 180°] or [0, 360°].

[0040] Location: The location of a radar in the map coordinate system can be calculated for each reflector. The specific calculation formula is as follows: Align the rotating radar coordinate system with the map coordinate system, and calculate the coordinates of reflector 1 in the rotating coordinate system: ; Calculate the radar's position coordinates in the map coordinate system: ; Similarly, the radar's coordinates on the map can also be obtained from reflector 2. ; In summary, based on the two reflectors, one angle and two positions of the radar in the map coordinate system can be calculated.

[0041] If there are n matching reflectors, pairwise combinations can yield n(n-1) / 2 pairs of reflectors, from which n(n-1) / 2 angles and n(n-1) positions can be calculated. To obtain more stable calculation results, multiple poses need to be fused.

[0042] Angle blending: Assume an angle range of [-180°, 180°]. Special attention needs to be paid to angles close to ±180° during angle blending. For example, averaging two angles, 180.1° and -179.9°, yields 0.2°, which is clearly incorrect.

[0043] One feasible approach is to add 360° to all angles in the range [-180°, 0°], shifting them to the range [180°, 360°]. At this point, all angles are positive and within the range [0°, 360°]. Then, average or take the median of the n(n-1) / 2 angles. If the result exceeds 180°, subtract 360° to obtain the final angle after merging.

[0044] Location fusion: The fusion of x and y coordinates can be achieved using the median elimination method. First, the median of the coordinates is calculated, and several data points closest to the median are retained. Then, the average is taken. The final fused position is obtained.

[0045] The implementation principle of this embodiment is as follows: When the AGV enters the reflector area, the system first estimates the current radar pose based on the kinematic model, then extracts high-reflectivity point clusters through reflection intensity threshold clustering and verifies the reflector identity using columnar geometric features, and calculates the virtual center coordinates by combining arc contour calculation and radial extension; then, the radar coordinates are transformed to the map coordinates through coordinate transformation and matched with the preset reflector, the radar yaw angle is calculated using the vector angle method and the circumferential angle jump problem is handled by angle normalization; finally, multiple pose hypotheses are generated based on the combination of multiple reflectors, and the final pose output is optimized through angle transformation fusion and coordinate median elimination strategy, forming a complete high-precision and robust reflector positioning solution.

[0046] An AGV positioning system based on dual reflectors, applied to the aforementioned AGV positioning method, includes: The pose determination module is used to switch the vehicle positioning mode to reflector positioning based on the preset reflector area and calculate the pose of the single-system LiDAR. The point cluster construction module is used to extract several high-reflection intensity points based on a preset reflection intensity threshold and laser scanning frequency to form a high-reflection point cluster. The virtual calculation module is used to determine the distance between the first and last high reflective points of the high reflective point cluster based on the reflector diameter, match the reflector model, and calculate the virtual center coordinates. The coordinate filtering module is used to convert the virtual center coordinates to the map coordinate system based on the pose of the single-system lidar, compare and match the actual center coordinates, and filter the reflector coordinate group. The coordinate alignment module is used to calculate the line segment angles between reflectors in different coordinate systems based on the reflector coordinate group, and align the different coordinate systems to obtain the pose of the rotating lidar. The vehicle positioning module is used to fuse and calculate the poses of several rotating lidar sensors to determine the running position of the AGV vehicle.

[0047] The implementation principle of this embodiment is as follows: the pose determination module realizes the automatic switching of the positioning mode after the AGV enters the reflector area, and predicts the initial radar pose based on the differential driving model; the point cluster construction module extracts high-reflectivity point clusters from the laser point cloud using the reflection intensity threshold and scanning continuity; the virtual calculation module calculates the virtual center coordinates of the reflector through geometric verification and weighted fusion of dual algorithms; the coordinate filtering module establishes the coordinate correspondence of the reflector using coordinate transformation and nearest neighbor matching; the coordinate alignment module realizes the coordinate system rotation alignment by using vector angle calculation and angle normalization; the vehicle positioning module handles circumferential angle jumps and coordinate anomalies through a multi-hypothesis fusion strategy, and finally outputs the accurate pose of the AGV, forming a complete closed-loop system from environmental perception to positioning decision.

[0048] An electronic device, comprising: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement any of the methods described in the above scheme.

[0049] A storage medium storing at least one instruction, at least one program, a code set, or an instruction set, wherein the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by a processor to implement the AGV positioning method based on dual reflectors as described above. The embodiments described herein are preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape, and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. An AGV positioning method based on dual reflectors, characterized in that, include: Based on the preset reflector area, the vehicle positioning mode is switched to reflector positioning, and the pose of the single-system LiDAR is calculated. Based on the preset reflection intensity threshold and the laser scanning frequency, several high reflection intensity points are extracted to form a high reflection point cluster; The distance between the first and last high reflective points of the high reflective point cluster is determined based on the reflector diameter, and the reflector model is matched to calculate the virtual center coordinates in the radar coordinate system. The virtual center coordinates are converted to map coordinates based on the pose of the single-system lidar, and the actual center coordinates in the map are compared and matched to filter the reflector coordinate group. Calculate the line segment angles between reflectors in different coordinate systems based on the reflector coordinate set, align the different coordinate systems, and obtain the pose of the rotating lidar. The AGV vehicle's operating position is determined by fusing and calculating the poses of several rotating lidar sensors.

2. The AGV positioning method based on dual reflectors according to claim 1, characterized in that, The specific steps for switching the vehicle positioning mode to reflector positioning based on the preset reflector area and calculating the pose of the single-system LiDAR include: The AGV vehicle's operating position is self-monitored. If the AGV vehicle moves to the preset reflector area, the positioning switching logic of the vehicle positioning mode is triggered. According to the positioning switching logic, the AGV vehicle starts the lidar, automatically switches to reflector positioning, and obtains the vehicle's position coordinates, heading angle, linear velocity and angular velocity at the previous moment. The heading angle at the current moment is calculated by combining the differential driving algorithm with the heading angle at the previous moment and the sampling time difference. Based on the current heading angle and the linear and angular velocities of the previous moment, the vehicle position coordinates of the previous moment are differentially calculated to obtain the vehicle position coordinates of the current moment. Based on the vehicle's current position coordinates and heading angle, determine the pose of the single-system lidar in the map coordinate system.

3. The AGV positioning method based on dual reflectors according to claim 1, characterized in that, The specific steps for extracting several high-reflection intensity points and forming a high-reflection point cluster based on a preset reflection intensity threshold and laser scanning frequency include: The reflector is scanned according to the laser scanning frequency to obtain several laser point data, and the point cloud reflection intensity is calculated. The laser point data is judged and filtered based on a preset reflection intensity threshold and the reflection intensity of the point cloud. If the reflection intensity of the point cloud is greater than the reflection intensity threshold, then target points are extracted from the current laser point data to obtain several high reflection intensity points; The high reflectivity points are clustered and grouped to generate a high reflectivity point cluster.

4. The AGV positioning method based on dual reflectors according to claim 1, characterized in that, The specific steps for determining the distance between the first and last high-reflectivity points of the high-reflectivity point cluster based on the reflector diameter, matching the reflector model, and calculating the virtual center coordinates in the radar coordinate system include: The high-reflection point clusters are analyzed based on the laser scanning sequence to extract the first and last high-reflection points; Perform distance calculations on the first high inversion point and the last high inversion point to obtain the Euclidean distance of the point cluster; The Euclidean distance of the point cluster is determined based on the reflector diameter. If the Euclidean distance of the point cluster is greater than the reflector radius but less than twice the reflector diameter, then the current high-reflectivity point cluster is determined to conform to the geometric characteristics of the reflector, and the corresponding reflector model is matched. A fitting operation is performed on the cluster of high-reflection points to obtain a fitted circular arc, and the midpoint of the high-reflection point is located. The first center coordinates are determined based on the radius of curvature and direction of the fitted arc, combined with the diameter of the reflector. Based on the radar origin, the high-reflectivity intermediate point is directionally extended, and the extension distance is the radius of the reflector, to determine the coordinates of the second center. By performing a weighted operation on the first center coordinates and the second center coordinates, the virtual center coordinates of the reflector model in the radar coordinate system are obtained.

5. The AGV positioning method based on dual reflectors according to claim 1, characterized in that, The specific steps of converting the virtual center coordinates to the map coordinate system based on the pose of the single-system lidar, comparing and matching the actual center coordinates in the map, and filtering the reflector coordinate group include: The virtual center coordinates in the radar coordinate system are transformed based on the pose of the single-system lidar to obtain the global center coordinates. The global center coordinates are matched with the actual center coordinates on the map, the coordinate difference is calculated, and a judgment is made against a preset matching threshold. If the coordinate difference is less than the matching threshold, the current reflector is determined to be successfully matched, and the virtual center coordinates and the actual center coordinates are aggregated to construct a reflector coordinate group.

6. The AGV positioning method based on dual reflectors according to claim 1, characterized in that, The specific steps for calculating the line segment angles between reflectors in different coordinate systems based on the reflector coordinate set, aligning the different coordinate systems, and obtaining the pose of the rotating lidar include: Based on the reflector coordinate system, the angles of line segments between different reflectors are calculated in different coordinate systems. Line segment angle in radar coordinate system Angle of line segment in map coordinate system ; in, Number the reflectors. For virtual center coordinates, The actual center coordinates; The difference between the line segment angles in the radar coordinate system and the line segment angles in the map coordinate system is calculated to obtain the deviation conversion angle. ; The deviation conversion angle is normalized according to a preset angle range to obtain the standard conversion angle; The radar coordinate system and the map coordinate system are rotated and aligned according to the standard transformation angle to obtain a rotated coordinate system, and the composite center coordinates corresponding to the reflector are calculated. : ; Based on the composite center coordinates and the actual center coordinates of any reflector, coordinate transformation is performed on the lidar to obtain the corresponding rotating lidar pose. .

7. The AGV positioning method based on dual reflectors according to claim 1, characterized in that, The specific steps for fusing and calculating the poses of several rotating lidar sensors to determine the running position of the AGV vehicle include: Match n reflectors to each other to obtain n(n-1) / 2 pairs of reflectors, and calculate n(n-1) / 2 standard conversion angles and n(n-1) rotating lidar coordinates; If the standard conversion angle is in the negative angle range, then the standard conversion angle is converted to a full angle to obtain the corresponding positive conversion angle; The average conversion angle is calculated by averaging all the positive conversion angles, and then compared with the straight angle for judgment. If the average conversion angle is greater than the horizontal angle, the difference between the average conversion angle and the circumference angle is calculated to obtain the final conversion angle and determine the heading angle of the AGV vehicle. The median value of all the rotating lidar coordinates is calculated to obtain the median coordinates of the lidar; The coordinates of the rotating lidar are calculated by performing a difference operation on the median coordinates of the radar to obtain a coordinate distance value, which is then compared with a preset coordinate difference threshold. If the coordinate distance value is within the coordinate difference range, then the current rotating lidar coordinates are deemed compliant. The average value of all compliant rotating lidar coordinates is calculated to obtain the final rotating lidar coordinates, which determines the driving position of the AGV vehicle.

8. An AGV positioning system based on dual reflectors applied to the method of any one of claims 1 to 7, characterized in that, include: The pose determination module is used to switch the vehicle positioning mode to reflector positioning based on the preset reflector area and calculate the pose of the single-system LiDAR. The point cluster construction module is used to extract several high-reflection intensity points based on a preset reflection intensity threshold and laser scanning frequency to form a high-reflection point cluster. The virtual calculation module is used to determine the distance between the first and last high reflective points of the high reflective point cluster based on the reflector diameter, match the reflector model, and calculate the virtual center coordinates. The coordinate filtering module is used to convert the virtual center coordinates to the map coordinate system based on the pose of the single-system lidar, compare and match the actual center coordinates, and filter the reflector coordinate group. The coordinate alignment module is used to calculate the line segment angles between reflectors in different coordinate systems based on the reflector coordinate group, and align the different coordinate systems to obtain the pose of the rotating lidar. The vehicle positioning module is used to fuse and calculate the poses of several rotating lidar sensors to determine the running position of the AGV vehicle.

9. An electronic device, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-7.

10. A storage medium storing at least one instruction, at least one program, a code set, or an instruction set, wherein the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by a processor to implement the AGV positioning method based on dual reflectors as described in any one of claims 1 to 7.

Images