Intelligent identification and pose adjustment method for operation and maintenance robot

Abstract

The application discloses an intelligent identification and pose adjustment method for an operation and maintenance robot. The method comprises the following steps: the operation and maintenance robot drives to the vicinity of a photovoltaic panel, a mechanical arm is stretched and drives the cleaning mechanism to be placed above the photovoltaic panel; a plurality of distance measuring laser sensors respectively measure a plurality of measurement distance values of the photovoltaic panel, and the operation and maintenance robot adjusts the cleaning surface of the cleaning mechanism to be parallel to the surface of the photovoltaic panel based on the calculation of the plurality of measurement distance values; a laser scanning radar scans the airspace where the photovoltaic panel is located, obtains panel three-dimensional point cloud data, and the operation and maintenance robot identifies the plane and the edge line of the photovoltaic panel by using the three-dimensional point cloud data, adjusts and controls the spatial position of the cleaning mechanism, and cleans close to the surface of the photovoltaic panel. The method can accurately identify the photovoltaic panel, intelligently adjust the pose according to the structure and layout characteristics of the photovoltaic panel, and realize accurate, safe and reliable cleaning work.

Description

Technical Field

[0001] This invention relates to the field of intelligent robot technology, and in particular to an intelligent recognition and posture adjustment method for maintenance robots. Background Technology

[0002] Independent maintenance robots struggle to achieve automated positioning and intelligent operation control when cleaning photovoltaic panels, making them unsuitable for diverse application scenarios involving photovoltaic panel installations. Summary of the Invention

[0003] The main technical problem solved by this invention is to provide an intelligent identification and posture adjustment method for maintenance robots, which enables maintenance robots to accurately identify photovoltaic panels and intelligently adjust their posture according to the structure and layout characteristics of the photovoltaic panels, thereby achieving accurate, safe and reliable cleaning work.

[0004] To address the aforementioned technical problems, one technical solution adopted by this invention is to provide an intelligent recognition and pose adjustment method for an operation and maintenance robot. The operation and maintenance robot includes a laser scanning radar mounted at the end of a robotic arm, and a cleaning mechanism mounted at the end of the robotic arm, the cleaning mechanism being equipped with multiple ranging laser sensors. The method includes the following steps:

[0005] The maintenance robot moves to a location near the photovoltaic panel, and the robotic arm extends, bringing the cleaning mechanism to a position above the photovoltaic panel.

[0006] Multiple ranging laser sensors respectively measure multiple distance values ​​of the photovoltaic panel. Based on the calculation of multiple distance values, the maintenance robot adjusts and controls the cleaning surface of the cleaning mechanism to be parallel to the surface of the photovoltaic panel.

[0007] The laser scanning radar scans the airspace where the photovoltaic panel is located to obtain three-dimensional point cloud data of the panel. The maintenance robot uses the three-dimensional point cloud data to identify the plane and edge lines of the photovoltaic panel, adjusts the spatial position of the cleaning mechanism, and moves close to the surface of the photovoltaic panel to prepare for cleaning.

[0008] Optionally, before the maintenance robot travels to a location near the photovoltaic panel, the maintenance robot uses RTK positioning, odometer, and IMU inertial navigation system to measure and calculate the location of the maintenance robot.

[0009] Optionally, the method for RTK positioning of the maintenance robot includes:

[0010] RTK base stations are installed around the photovoltaic power plant. Differential locators are mounted on the mobile chassis of the maintenance robot, and these differential locators wirelessly measure distances with each RTK base station. The RTK base stations are... , , , Differential positioner is Select the two base stations closest to the differential locator. , The distance between the differential locator and the base station is measured using RTK. With base station The distance is With base station The distance is Base station , The distance is Then there is a first included angle. Second angle They respectively satisfy: , The positioning coordinates of the maintenance robot are also based on If we express it as: , Therefore, the starting point is calculated to be the base station. The dynamic position of the maintenance robot at its current location. .

[0011] Optionally, there are four ranging laser sensors, namely M1, M2, M3, and M4, which are fixedly mounted on the cleaning mechanism. The four ranging laser sensors measure the vertical distance to the photovoltaic panel, which corresponds to the vertical distance to four vertical projection points T1, T2, T3, and T4 on the photovoltaic panel. The end effector M0 of the robotic arm is fixed relative to the cleaning mechanism and serves as a reference point. Four projection planes are formed by any three of the four vertical projection points T1, T2, T3, and T4, namely the first projection plane T1T2T3, the second projection plane T1T2T4, the third projection plane T2T3T4, and the fourth projection plane T1T3T4. The projection distance from the end effector M0 of the robotic arm to these four projection planes is calculated, and then the average of these four projection distances is calculated to obtain the distance from the end effector M0 of the robotic arm to the photovoltaic panel.

[0012] Optionally, the posture of the cleaning mechanism is adjusted based on the vertical distances measured by the four ranging laser sensors to the photovoltaic panel. Then, these four vertical distances are measured multiple times, and the posture of the cleaning mechanism is adjusted multiple times accordingly, until the four vertical distances are nearly equal. In this case, the cleaning surface of the cleaning mechanism is parallel to the surface of the photovoltaic panel.

[0013] Optionally, the method for obtaining the three-dimensional point cloud data of the panel includes: after obtaining the three-dimensional point cloud data of the panel, preprocessing the three-dimensional point cloud data of the panel, and by setting a spatial threshold between adjacent point clouds, excluding point clouds that are significantly larger than the threshold, thereby obtaining the overall point cloud set PM of the photovoltaic panel.

[0014] Optionally, the method for identifying the plane of the photovoltaic panel includes:

[0015] First, a first planar subset PM(1) is randomly selected from the overall point cloud set PM of the photovoltaic panel. Then, point clouds from the overall point cloud set PM of the photovoltaic panel that are less than a set threshold from the first planar subset PM(1) are selected. These point clouds are considered valid and are combined with the point clouds in the first planar subset PM(1) to obtain a corrected planar subset. The point clouds within the range of the corrected planar subset are averaged to fit a second planar subset. If the number of point clouds in the second planar subset is greater than the number of point clouds in the first planar subset PM(1), the point clouds in the first planar subset PM(1) are updated to the point clouds in the second planar subset.

[0016] The second planar subset PM(2) is randomly selected from the overall point cloud set PM of the photovoltaic panel for the second time, and the second planar subset PM(2) is corrected.

[0017] If the number of point clouds in the second planar subset PM(2) is equal to or less than the number of point clouds in the first planar subset PM(1), then the first planar subset PM(1) is taken as the final planar point cloud set of the photovoltaic panel.

[0018] If the number of point clouds in the second planar subset PM(2) is greater than the number of point clouds in the first planar subset PM(1), then the third planar subset PM(3) is randomly selected from the overall point cloud set PM of the photovoltaic panel and corrected until the number of point clouds in the nth planar subset PM(n) no longer increases. Then the nth planar subset PM(n) is taken as the final planar point cloud set of the photovoltaic panel, where n is greater than or equal to 3.

[0019] Optionally, the method for identifying the edge lines of the photovoltaic panel includes:

[0020] First, a first fitted edge subset PX(1) is randomly selected from the overall point cloud set PM of the photovoltaic panel. Then, point clouds whose distance from the first edge subset PX(1) is less than a set threshold are selected from the overall point cloud set of the photovoltaic panel. These point clouds are considered valid and are combined with the point clouds in the first edge subset PX(1) to obtain a corrected edge subset. The point clouds within the range of the corrected edge subset are averaged to fit a second edge subset. If the number of point clouds in the second edge subset is greater than the number of point clouds in the first edge subset PX(1), the point clouds in the first edge subset PX(1) are updated to the point clouds in the second edge subset.

[0021] The second fitted edge subset PX(2) is randomly selected from the overall point cloud set PM of the photovoltaic panel and the second edge subset PX(2) is corrected.

[0022] If the number of point clouds in the second edge subset PX(2) is equal to or less than the number of point clouds in the first edge subset PX(1), then the first edge subset PX(1) is taken as the final edge point cloud set of the photovoltaic panel.

[0023] If the number of point clouds in the second edge subset PX(2) is greater than the number of point clouds in the first edge subset PX(1), then the third fitted edge subset PX(3) is randomly selected from the overall point cloud set PM of the photovoltaic panel for the third time and corrected until the number of point clouds in the nth edge subset PX(n) no longer increases. Then the nth edge subset PX(n) is taken as the final edge point cloud set of the photovoltaic panel, where n is greater than or equal to 3.

[0024] Optionally, the method for adjusting the spatial position of the cleaning mechanism includes:

[0025] In a three-dimensional coordinate system, the X-axis represents the up-down direction, the Y-axis represents the left-right direction, and the Z-axis represents the forward-backward direction, which is also the direction in which the maintenance robot moves forward and backward during the cleaning process. During the cleaning process, the swaying along the Y-axis is adaptively adjusted by the left-right sway support shaft installed between the robotic arm and the cleaning mechanism, as well as the support rollers that move on the panel. The swaying along the Z-axis is adaptively adjusted by the forward-backward sway support shaft installed between the robotic arm and the cleaning mechanism, as well as the support rollers that move on the panel.

[0026] Optionally, for the end point of the robotic arm The distance from the photovoltaic panel is established on a two-dimensional plane along the X and Y axes, with the end point... The motion model in this two-dimensional plane is used to monitor the feedback closed-loop control in real time: that is, through the base rocker arm, the end of the base rocker arm, and the end point. Adjustments are made based on the joint motion analysis of the connected virtual arms, including:

[0027] The length of the base rocker arm is The end of the base swing arm and the end point of the robotic arm The length of the virtual arm of the connecting line is The angle between the base rocker arm and the Y-axis is... The angle between the base swing arm and the virtual arm is The beginning of the base swing arm to The angle between the line connecting the two axes and the Y-axis is... And there are .

[0028] Furthermore, the following structural positional relationship is satisfied:

[0029] The corresponding ones are:

[0030] And, and , .

[0031] Based on the above structural positional relationship, the end-effector pose of the maintenance robot can be adjusted to maintain the attitude matching between the cleaning structure and the photovoltaic panel.

[0032] The beneficial effects of this invention are as follows: This invention discloses an intelligent identification and posture adjustment method for a maintenance robot. The method includes the maintenance robot moving to a nearby photovoltaic panel, the robotic arm extending and positioning the cleaning mechanism above the photovoltaic panel; multiple ranging laser sensors measuring multiple distance values ​​of the photovoltaic panel; the maintenance robot calculating and adjusting the cleaning surface of the cleaning mechanism to be parallel to the surface of the photovoltaic panel based on these multiple distance values; a laser scanning radar scanning the airspace where the photovoltaic panel is located to obtain three-dimensional point cloud data of the panel; the maintenance robot using the three-dimensional point cloud data to identify the plane and edges of the photovoltaic panel, and adjusting the spatial position of the cleaning mechanism to clean close to the surface of the photovoltaic panel. This method can accurately identify photovoltaic panels and intelligently adjust their posture according to the structure and layout characteristics of the photovoltaic panels, achieving accurate, safe, and reliable cleaning work. Attached Figure Description

[0033] Figure 1 This is a flowchart according to an embodiment of the present invention;

[0034] Figure 2 This is a schematic diagram of RTK positioning of a maintenance robot according to an embodiment of the present invention;

[0035] Figure 3 This is a schematic diagram of the positioning of the motion odometer of an operation and maintenance robot according to an embodiment of the present invention;

[0036] Figure 4 This is a schematic diagram of the laser ranging adjustment of the cleaning mechanism according to an embodiment of the present invention;

[0037] Figure 5 This is a three-dimensional point cloud schematic diagram of a photovoltaic panel according to an embodiment of the present invention;

[0038] Figure 6 This is a schematic diagram of a planar point cloud of a photovoltaic panel according to an embodiment of the present invention;

[0039] Figure 7 This is a schematic diagram of the correction selection of a planar point cloud of a photovoltaic panel according to an embodiment of the present invention;

[0040] Figure 8 This is a pose control motion model of an operation and maintenance robot according to an embodiment of the present invention;

[0041] Figure 9 This is a schematic diagram of the composition of an operation and maintenance robot according to an embodiment of the present invention. Detailed Implementation

[0042] To facilitate understanding of the present invention, a more detailed description is provided below with reference to the accompanying drawings and specific embodiments. Preferred embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a thorough and complete understanding of the disclosure of the invention.

[0043] It should be noted that, unless otherwise defined, all technical and scientific terms used in this specification have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. The term "and / or" as used in this specification includes any and all combinations of one or more of the associated listed items.

[0044] Figure 1 An embodiment of the intelligent recognition and pose adjustment method for the maintenance robot of the present invention is shown. The maintenance robot in this embodiment includes a laser scanning radar installed at the end of a robotic arm, and a cleaning mechanism installed at the end of the robotic arm. The cleaning mechanism is equipped with multiple ranging laser sensors. The method includes the following steps:

[0045] Step S1: The maintenance robot moves to a nearby photovoltaic panel, and the robotic arm extends and drives the cleaning mechanism to be positioned above the photovoltaic panel;

[0046] Step S2: The multiple ranging laser sensors respectively measure multiple distance values ​​of the photovoltaic panel. Based on the multiple measured distance values, the maintenance robot adjusts and controls the cleaning surface of the cleaning mechanism to be parallel to the surface of the photovoltaic panel.

[0047] Step S3: The laser scanning radar scans the airspace where the photovoltaic panel is located to obtain three-dimensional point cloud data of the panel. The maintenance robot uses the three-dimensional point cloud data to identify the plane and edge lines of the photovoltaic panel, and adjusts the spatial position of the cleaning mechanism to clean the surface of the photovoltaic panel close to it.

[0048] As can be seen, through the above operational steps, the maintenance robot can autonomously move to the location of the photovoltaic panel. This type of maintenance robot has strong autonomous movement flexibility, enhancing its working area. Based on the control of the robotic arm, the maintenance robot also has greater freedom of spatial adjustment, allowing the cleaning mechanism to adapt to photovoltaic panels of various tilt angles and sizes. Furthermore, precise distance measurement and 3D point cloud data acquisition and processing via a ranging laser sensor enhance the accuracy and safety of the integration between the cleaning mechanism and the photovoltaic panel.

[0049] Optionally, before step S1, in order for the maintenance robot to travel to the nearest photovoltaic panel, in addition to the satellite positioning coordinates of the photovoltaic panel as usual, it also needs to provide more accurate location positioning and tracking. This includes using carrier phase differential technology RTK (Real-time kinematic) for differential positioning and updating the maintenance robot's position in the actual space in a timely manner.

[0050] like Figure 2 As shown, RTK base stations are installed around the photovoltaic power plant. Differential locators are mounted on the mobile chassis of the maintenance robot, enabling wireless ranging between the RTK base stations. RTK base stations are provided. , , , Differential positioner is Select the two base stations with the closest distance between the differential locators. , The distance between the differential locator and the base station is measured and fed back via RTK: Differential locator With base station The distance is With base station The distance is Base station , The distance is Then there is a first included angle. Satisfying the second included angle They respectively satisfy: , The positioning coordinates of the maintenance robot are also based on If we express it as: , Therefore, the starting point can be calculated as the base station. The dynamic position of the maintenance robot at its current location. .

[0051] Optionally, this also includes controlling the speed of the robot's wheels to obtain the chassis's odometer readings and control the chassis's position. For example... Figure 3 As shown, the distance from the wheel to the center of the chassis... The radius of the circular motion at the center of the chassis Chassis centerline speed Chassis center angular velocity The linear velocities of the left and right wheels are respectively: , If we establish a two-dimensional coordinate system with the ground as the plane, then the differential chassis system has three degrees of freedom. ,in, The horizontal coordinates of the chassis The vertical coordinate is... It is a corner.

[0052] Optionally, the linear velocities of the left and right wheels can be read and calculated using an encoder installed on the chassis differential drive servo motor and the motion time: , ;

[0053] During differential motion, the angular velocities of the two wheels Same, that is: The radius of motion of the central arc of the chassis can be calculated from this. Chassis centerline speed .

[0054] After the initial movement of the chassis, the cumulative mileage is calculated to obtain the actual chassis position, thus determining the initial coordinates of the chassis center. Then we have: ,

[0055] In every tiny moment Inside, the distance the chassis travels is , If the change in the angle of the chassis movement is Δθ, then: , , By integrating, the real-time coordinates of the chassis during its dynamic movement can be calculated based on the initial position. : ; ; .

[0056] Optionally, the actual position of the chassis can be dynamically measured using an IMU (Inertial Measurement Unit) mounted on the mobile chassis. Furthermore, by integrating RTK positioning, odometer calculation, and IMU inertial navigation, the position of the maintenance robot determined by the mobile chassis through RTK positioning, odometer calculation, and IMU inertial navigation is comprehensively calculated and compared. , , An alarm will sound when the error exceeds the set threshold (default 200mm).

[0057] Therefore, based on the above-mentioned operation control of the maintenance robot, it can be ensured that the maintenance robot can accurately reach the location of the photovoltaic panel that needs to be cleaned. That is, before the maintenance robot moves to the vicinity of the photovoltaic panel, the maintenance robot identifies the three-dimensional spatial position of the photovoltaic panel, and the robotic arm controls the cleaning mechanism to move to the starting position of the photovoltaic panel.

[0058] In step S1, the maintenance robot moves to a position near the photovoltaic panel, and the robotic arm extends, causing the cleaning mechanism to be positioned above the photovoltaic panel. Optionally, since the tilt angle of the photovoltaic panel can be adjusted, or the photovoltaic panel has multiple tilt angle settings, the cleaning mechanism needs to have the same tilt angle to be parallel to the surface of the photovoltaic panel. Maintaining this requires that the angle of the cleaning mechanism can be adjusted according to the relative position between the cleaning mechanism and the surface of the photovoltaic panel.

[0059] For step S2, further as follows Figure 4 As shown, there are four ranging laser sensors, namely M1, M2, M3, and M4, which are fixedly mounted on the cleaning mechanism. For example, if the cleaning mechanism is rectangular, the sensors are distributed at the four corners; if it is circular, they can be evenly distributed around the circumference. Correspondingly, in Figure 4In this process, four ranging laser sensors measure the vertical distance to the photovoltaic panel, corresponding to the vertical distance to four vertical projection points T1, T2, T3, and T4 on the photovoltaic panel. Since the robotic arm end effector M0 is fixed relative to the cleaning mechanism, it can be used as a reference point. Any three of the four vertical projection points T1, T2, T3, and T4 can form four planes: the first projection plane T1T2T3, the second projection plane T1T2T4, the third projection plane T2T3T4, and the fourth projection plane T1T3T4. Based on the existing spatial position data of the robotic arm end effector M0 and the cleaning mechanism, and the measured distances, the projected distances from the robotic arm end effector M0 to these four projection planes are calculated. Then, the average of these four projected distances is taken to obtain the distance from the robotic arm end effector M0 to the photovoltaic panel.

[0060] Furthermore, based on the vertical distances measured by the four ranging laser sensors to the photovoltaic panel, the posture of the cleaning mechanism is adjusted. Then, these four vertical distances are measured again, and the adjustment is repeated multiple times until these four vertical distances tend to be close to equal. This indicates that the cleaning surface of the cleaning mechanism is parallel to the surface of the photovoltaic panel.

[0061] The distance-measuring laser sensor installed on the cleaning mechanism provides timely feedback on the distance to the photovoltaic panel. When the distance deviates from the set threshold (such as 5mm), the robotic arm is dynamically adjusted in time to ensure that the cleaning machine can contact the photovoltaic panel without damaging it.

[0062] Optionally, the laser scanning radar scans the airspace where the photovoltaic panel is located to obtain three-dimensional point cloud data of the panel. The maintenance robot uses the three-dimensional point cloud data to identify the plane and edge lines of the photovoltaic panel, adjusts the spatial position of the cleaning mechanism, and prepares to clean the surface of the photovoltaic panel.

[0063] For step S3, the method for obtaining the three-dimensional point cloud data of the panel includes: after the laser scanning radar scans the airspace where the photovoltaic panel is located and obtains the three-dimensional point cloud data of the panel, the three-dimensional point cloud data of the panel is preprocessed. By setting a spatial threshold between adjacent point clouds, for example, a default of 8mm, point clouds that are significantly larger than the threshold are excluded, and unnecessary point clouds such as ground, other panels, and supports are removed, thereby reducing the amount of calculation and enhancing the quality of the point cloud.

[0064] Furthermore, based on the known structural dimensions of the photovoltaic panel, a threshold range for the overall structure of the photovoltaic panel can be set. The overall structure of the photovoltaic panel is approximately a cuboid, and this range includes thresholds for thickness, width, and vertical length. For example... Figure 5The diagram shows a schematic of a photovoltaic panel. The thickness threshold F1 can range from ±8mm, the walking width threshold F2 from ±1000mm, and the vertical length threshold F3 from ±3000mm. By using these overall structural threshold ranges for the photovoltaic panel, the overall outline of the point cloud data can be defined. The point cloud data within the resulting cuboid is considered valid and serves as the overall point cloud set (PM) for the photovoltaic panel, excluding point cloud data that is clearly outside this range.

[0065] Optionally, a further method for identifying the plane where the photovoltaic panel is located includes: randomly extracting a first fitted plane subset PM from the overall point cloud set PM of the photovoltaic panel (1), such as... Figure 6 As shown; then, as Figure 7 As shown, point clouds selected from the overall point cloud set of the photovoltaic panel whose distance to the first planar subset PM(1) is less than a set threshold are considered valid and combined with the point clouds in the first planar subset PM(1) to obtain a corrected planar subset; further, the point clouds within the range of the corrected planar subset are averaged to fit a second planar subset. If the number of point clouds in the second planar subset is greater than the number of point clouds in the first planar subset PM(1), the point clouds in the first planar subset PM(1) are updated to the point clouds in the second planar subset.

[0066] Optionally, a further method for identifying the plane where the photovoltaic panel is located includes: randomly extracting a first fitted plane subset PM from the overall point cloud set PM of the photovoltaic panel (1), such as... Figure 6 As shown; then, as Figure 7 As shown, point clouds selected from the overall point cloud set of the photovoltaic panel whose distance to the first planar subset PM(1) is less than a set threshold QM are considered valid and combined with the point clouds in the first planar subset PM(1) to obtain a corrected planar subset; further, the point clouds within the range of the corrected planar subset are averaged to fit a second planar subset QN. If the number of point clouds in the second planar subset is greater than the number of point clouds in the first planar subset PM(1), the point clouds in the first planar subset PM(1) are updated to the point clouds in the second planar subset.

[0067] Then, a second fitted planar subset PM(2) is randomly selected from the overall point cloud set PM of the photovoltaic panel for the second time, and the second planar subset PM(2) is corrected in the same way. If the number of point clouds in the second planar subset PM(2) is equal to or less than the number of point clouds in the first planar subset PM(1), then the first planar subset PM(1) is taken as the final planar point cloud set of the photovoltaic panel. If the number of point clouds in the second planar subset PM(2) is greater than the number of point clouds in the first planar subset PM(1), then a third fitted planar subset PM(3) is randomly selected from the overall point cloud set PM of the photovoltaic panel for the third time and corrected, until the number of point clouds in the nth planar subset PM(n) no longer increases, then the nth planar subset PM(n) is taken as the final planar point cloud set of the photovoltaic panel.

[0068] Optionally, after obtaining the planar point cloud set of the photovoltaic panel, it can be used as a timely feedback object relative to the pose of the robotic arm to dynamically control the movement of the robotic arm's end effector.

[0069] Optionally, the method for identifying the edge lines of the photovoltaic panel includes: randomly selecting a first fitted edge line subset PX(1) from the overall point cloud set PM of the photovoltaic panel; then, selecting point clouds from the overall point cloud set of the photovoltaic panel whose distance to the first edge line subset PX(1) is less than a set threshold, and considering these point clouds as valid, combining them with the point clouds in the first edge line subset PX(1) to obtain a corrected edge line subset; further averaging the point clouds within the range of the corrected edge line subset to fit a second edge line subset; if the number of point clouds in the second edge line subset is greater than the number of point clouds in the first edge line subset PX(1), then updating the point clouds in the first edge line subset PX(1) to the point clouds in the second edge line subset.

[0070] Then, a second fitted edge subset PX(2) is randomly selected from the overall point cloud set PM of the photovoltaic panel for the second time, and the second edge subset PX(2) is corrected in the same way. If the number of point clouds in the second edge subset PX(2) is equal to or less than the number of point clouds in the first edge subset PX(1), then the first edge subset PX(1) is taken as the final edge point cloud set of the photovoltaic panel. If the number of point clouds in the second edge subset PX(2) is greater than the number of point clouds in the first edge subset PX(1), then a third fitted edge subset PX(3) is randomly selected from the overall point cloud set PM of the photovoltaic panel for the third time and corrected until the number of point clouds in the nth edge subset PX(n) no longer increases. Then the nth edge subset PX(n) is taken as the final edge point cloud set of the photovoltaic panel.

[0071] Optionally, after obtaining the edge point cloud set of the photovoltaic panel, it can be used as a timely feedback object relative to the pose of the robotic arm to dynamically control the movement of the robotic arm's end effector.

[0072] During the initial positioning process, the planar pose and the straight line pose of the bounding box are extracted from the panel point cloud scanned by the laser scanning radar as the starting positioning of the maintenance robot. Then, the pose of the end effector M0 coordinate is calculated by 6 degrees of freedom to control the movement of the robot arm.

[0073] The method for adjusting the spatial position of the cleaning mechanism includes:

[0074] like Figure 8 As shown, in the three-dimensional coordinate system, the X-axis represents the up-down direction, the Y-axis represents the left-right direction, and the Z-axis (perpendicular to the paper) represents the front-back direction, which is also the direction in which the robot moves forward and backward during the cleaning process.

[0075] During normal cleaning operation, the main deviations come from uneven ground, changes in the dimensions of the photovoltaic panels, and operational errors. Specifically, the swaying along the Y-axis can be adjusted using the left-right sway support shaft installed between the robotic arm and the cleaning mechanism, as well as the support rollers that travel on the panel; the swaying along the Z-axis can be adjusted using the forward-backward sway support shaft installed between the robotic arm and the cleaning mechanism, as well as the support rollers that travel on the panel.

[0076] Additionally, the end effector of the robotic arm is The distance from the photovoltaic panel (PV) needs to be established on a two-dimensional plane along the X and Y axes, with the end point... The motion model in this two-dimensional plane is used to monitor the feedback closed-loop control in real time: that is, through the base rocker arm JZB, the end of the base rocker arm JZB and the end point. Adjustments are made using joint kinematic analysis of the connected virtual arm XNB. Specifically:

[0077] Establish a 2-DOF kinematic model for the pose feedback adjustment of the end effector of the maintenance robot, including: Figure 8 The end point of the robotic arm is The length of the base rocker arm is The end of the base swing arm and The length of the virtual arm of the connecting line is The angle between the base rocker arm and the Y-axis is... The angle between the base swing arm and the virtual arm is The beginning of the base swing arm to The angle between the line connecting the two axes and the Y-axis is... And there are .

[0078] Furthermore, the following structural positional relationship is satisfied:

[0079] The corresponding ones are:

[0080] And, and , .

[0081] Based on the above structural positional relationship, the end-effector pose of the maintenance robot can be adjusted to maintain the attitude matching between the cleaning structure and the photovoltaic panel.

[0082] Optionally, a force touch sensor is also installed on the panel of the cleaning mechanism to provide timely feedback on the pressure between the cleaning mechanism and the photovoltaic panel. When the pressure exceeds the set threshold, the robotic arm is dynamically adjusted to lift, ensuring that the cleaning mechanism does not damage the photovoltaic panel.

[0083] like Figure 9 As shown, the maintenance robot includes a walking platform 1, a robotic arm 2, and a cleaning mechanism 3. The robotic arm 2 includes a first sub-arm (i.e., a base swing arm), a second sub-arm, and a third sub-arm connected in sequence, with the cleaning mechanism 3 installed at the end of the third sub-arm.

[0084] Therefore, this invention discloses an intelligent identification and posture adjustment method for a maintenance robot. The method includes the maintenance robot moving to a nearby photovoltaic panel, the robotic arm extending and positioning the cleaning mechanism above the photovoltaic panel; multiple ranging laser sensors measuring multiple distance values ​​of the photovoltaic panel; the maintenance robot calculating and adjusting the cleaning surface of the cleaning mechanism to be parallel to the surface of the photovoltaic panel based on these multiple distance values; a laser scanning radar scanning the airspace where the photovoltaic panel is located to obtain three-dimensional point cloud data of the panel; the maintenance robot using the three-dimensional point cloud data to identify the plane and edges of the photovoltaic panel, and adjusting the spatial position of the cleaning mechanism to clean close to the surface of the photovoltaic panel. This method can accurately identify photovoltaic panels and intelligently adjust their posture according to the structure and layout characteristics of the photovoltaic panels, achieving accurate, safe, and reliable cleaning work.

[0085] The above are merely embodiments of the present invention and do not limit the patent scope of the present invention. Any equivalent structural transformations made based on the content of the present invention specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. An intelligent identification and pose adjustment method for an operation and maintenance robot, characterized in that, The operation and maintenance robot comprises a laser scanning radar arranged at the end of a mechanical arm, and a cleaning mechanism arranged at the end of the mechanical arm, wherein the cleaning mechanism is provided with a plurality of ranging laser sensors, and comprises the following steps: The operation and maintenance robot drives to the vicinity of a photovoltaic panel, the mechanical arm is stretched and drives the cleaning mechanism to be arranged above the photovoltaic panel; The plurality of ranging laser sensors respectively measure a plurality of measurement distance values of the photovoltaic panel, and the operation and maintenance robot adjusts the cleaning surface of the cleaning mechanism to be parallel to the surface of the photovoltaic panel based on the calculation of the plurality of measurement distance values; The laser scanning radar scans the airspace where the photovoltaic panel is located to obtain three-dimensional point cloud data of the photovoltaic panel, and the operation and maintenance robot identifies the plane and the edge line of the photovoltaic panel by using the three-dimensional point cloud data, and controls the spatial position of the cleaning mechanism to clean the surface of the photovoltaic panel; The method for obtaining the three-dimensional point cloud data of the photovoltaic panel comprises the following steps: after the three-dimensional point cloud data of the photovoltaic panel is obtained, the three-dimensional point cloud data is preprocessed, the point cloud obviously larger than the threshold value is excluded by setting the spatial threshold value between adjacent point clouds, and the overall point cloud set PM of the photovoltaic panel is obtained; The method comprises the following steps: based on the known structural size of the photovoltaic panel, the overall structural threshold range of the photovoltaic panel is set, the overall structure of the photovoltaic panel is close to a cuboid, and the overall structural threshold range comprises a thickness threshold value, a walking width threshold value and an up-down length threshold value; the overall contour of the point cloud data is set by using the overall structural threshold range of the photovoltaic panel, the point cloud in the cuboid is an effective point cloud value, and the overall point cloud set PM of the photovoltaic panel is obtained; The method for identifying the edge line of the photovoltaic panel comprises the following steps: A first edge line subset PX(1) is randomly taken out from the overall point cloud set PM of the photovoltaic panel for the first time, then the point cloud with a distance less than a set threshold value from the first edge line subset PX(1) is selected from the overall point cloud set PM of the photovoltaic panel, the point cloud is considered to be effective, and the point cloud is combined with the point cloud in the first edge line subset PX(1) to obtain a corrected edge line subset; the point cloud in the range of the corrected edge line subset is further averaged to fit a second edge line subset, and if the number of point clouds in the second edge line subset is greater than the number of point clouds in the first edge line subset PX(1), the point clouds in the first edge line subset PX(1) are updated to the point clouds in the second edge line subset; A second edge line subset PX(2) is randomly taken out from the overall point cloud set PM of the photovoltaic panel for the second time, and the second edge line subset PX(2) is corrected; If the number of point clouds in the second edge line subset PX(2) is equal to or less than the number of point clouds in the first edge line subset PX(1), the first edge line subset PX(1) is taken as the final edge line point cloud set of the photovoltaic panel. If the number of point clouds in the second edge line subset PX(2) is greater than the number of point clouds in the first edge line subset PX(1), a third edge line subset PX(3) is randomly taken out from the whole point cloud set PM of the photovoltaic panel for the third time, and correction is performed until the number of point clouds in the nth edge line subset PX(n) is no longer increased, and the nth edge line subset PX(n) is taken as the final edge line point cloud set of the photovoltaic panel, and n is greater than or equal to 3.

2. The intelligent identification and pose adjustment method of the operation and maintenance robot according to claim 1, characterized in that, Before the operation and maintenance robot drives to the photovoltaic panel, the operation and maintenance robot measures and calculates the position of the operation and maintenance robot through RTK positioning, a walking odometer and an IMU inertial instrument. 3.The intelligent identification and pose adjustment method of the operation and maintenance robot according to claim 2, characterized in that, The method for the operation and maintenance robot to perform RTK positioning comprises the following steps: RTK base stations are installed around the photovoltaic power plant. Differential locators are mounted on the mobile chassis of the maintenance robot, and these differential locators wirelessly measure distances with each RTK base station. The RTK base stations are... , , , Differential positioner is Select the two base stations closest to the differential locator. , The distance between the differential locator and the RTK base station is fed back through RTK measurement: Differential locator With base station The distance is With base station The distance is Base station , The distance between them is Then there is a first included angle. Second angle They respectively satisfy: , The positioning coordinates of the maintenance robot are also based on If we express it as: , Therefore, the starting point is calculated to be the base station. The dynamic position of the maintenance robot at its current location. . 4.The method of Claim 1, wherein The four ranging laser sensors, namely M1, M2, M3 and M4, are fixedly arranged on the cleaning mechanism, and the four ranging laser sensors measure the vertical distances to the photovoltaic panel, which correspond to the vertical distances to the four vertical projection points T1, T2, T3 and T4 on the photovoltaic panel; the mechanical arm end M0 is fixed relative to the cleaning mechanism and serves as a reference point, and any three of the four vertical projection points T1, T2, T3 and T4 form four projection planes, namely a first projection plane T1T2T3, a second projection plane T1T2T4, a third projection plane T2T3T4 and a fourth projection plane T1T3T4; the projection distances from the mechanical arm end M0 to the four projection planes are calculated respectively, and then the average of the four projection distances is calculated to obtain the distance from the mechanical arm end M0 to the photovoltaic panel. 5.The intelligent identification and pose adjustment method of the operation and maintenance robot according to claim 4, characterized in that, The posture of the cleaning mechanism is adjusted according to the vertical distances measured by the four ranging laser sensors, and then the four vertical distances are measured again, and the posture of the cleaning mechanism is adjusted again, until the four vertical distances are close to each other, and then the cleaning surface of the cleaning mechanism is parallel to the surface of the photovoltaic panel. 6.The intelligent identification and pose adjustment method of the operation and maintenance robot according to claim 1, characterized in that, The method for identifying the plane of the photovoltaic panel comprises the following steps: A first plane subset PM(1) is randomly taken out from the whole point cloud set PM of the photovoltaic panel; then, point clouds with a distance to the first plane subset PM(1) less than a set threshold value are selected from the whole point cloud set PM of the photovoltaic panel, and these point clouds are considered to be effective and combined with the point clouds in the first plane subset PM(1) to obtain a corrected plane subset; the point clouds in the corrected plane subset are further averaged to fit a second plane subset, and if the number of point clouds in the second plane subset is greater than the number of point clouds in the first plane subset PM(1), the point clouds in the first plane subset PM(1) are updated to the point clouds in the second plane subset; A second plane subset PM(2) is randomly taken out from the whole point cloud set PM of the photovoltaic panel, and the second plane subset PM(2) is corrected; If the number of point clouds in the second plane subset PM(2) is equal to or less than the number of point clouds in the first plane subset PM(1), the first plane subset PM(1) is taken as the final plane point cloud set of the photovoltaic panel. If the number of point clouds in the second plane subset PM(2) is greater than the number of point clouds in the first plane subset PM(1), a third plane subset PM(3) is randomly selected from the entire point cloud set PM of the photovoltaic panel, and the third plane subset PM(3) is fitted and corrected, until the number of point clouds in the nth plane subset PM(n) obtained no longer increases, and the nth plane subset PM(n) is taken as the final plane point cloud set of the photovoltaic panel, and n is greater than or equal to 3. 7.The intelligent identification and pose adjustment method of the operation and maintenance robot according to claim 1, characterized in that, The method for regulating the spatial position of the cleaning mechanism comprises: In the three-dimensional coordinate axis, the X-axis represents the up-down direction, the Y-axis represents the left-right direction, and the Z-axis represents the front-back direction, which is also the direction of the operation and maintenance robot running forward and backward during cleaning; during the walking and cleaning process of the operation and maintenance robot, the swing along the Y-axis is adaptively adjusted through the left-right swing support shaft installed between the mechanical arm and the cleaning mechanism and the support rollers walking on the panel; the swing along the Z-axis is adaptively adjusted through the front-back swing support shaft installed between the mechanical arm and the cleaning mechanism and the support rollers walking on the panel. 8.The method of claim 7, wherein, For the end point of the robot arm The distance from the photovoltaic panel is established on a two-dimensional plane of the X and Y axes, the end point of the robot arm The motion model on this two-dimensional plane is used to monitor the feedback closed-loop control in real time: that is, by adjusting the base swing arm, the base swing arm end, and the end point of the robot arm The joint motion analysis of the virtual arm of the connecting line is used to adjust, including: The length of the base rocker arm is The end of the base swing arm and the end point of the robotic arm The length of the virtual arm of the connecting line is The angle between the base rocker arm and the Y-axis is... The angle between the base swing arm and the virtual arm is The beginning of the base swing arm to The angle between the line connecting the two axes and the Y-axis is... And there are ; Further, the following structural position relationship is satisfied: corresponding to: and also , ; According to the above structural position relationship, the end pose of the mechanical arm of the operation and maintenance robot is feedback adjusted to maintain the attitude matching between the cleaning structure and the photovoltaic panel.

Images