Photovoltaic cleaning robot and driving method thereof

Abstract

The present invention provides a photovoltaic cleaning robot and a driving method thereof, which relates to the technical field of photovoltaic cleaning robots, including: when the photovoltaic cleaning robot is driving on a photovoltaic panel, the positioning yaw angle in the camera coordinate system is obtained through the captured image and depth information of the depth camera; the positioning yaw angle is corrected based on the initial direction angle to obtain the heading yaw angle; the initial direction angle is the difference between the initial positioning yaw angle and the initial visual yaw angle of the photovoltaic cleaning robot; based on the heading yaw angle, the route offset distance is determined; at least based on the route offset distance, the robot is corrected to control the photovoltaic cleaning robot to travel along the straight texture of the photovoltaic panel. The present invention does not need to install any sensors, and can obtain relevant data for correction through a depth camera, and the hardware structure is simple; the data fluctuation of the positioning yaw angle in the camera coordinate system is small, and is not affected by external factors, so the robot can be better corrected to improve the cleaning effect of the photovoltaic panel.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic cleaning robots, and in particular to a photovoltaic cleaning robot and a driving method thereof. Background Art

[0002] Photovoltaic cleaning robot is an automated device specially designed for cleaning photovoltaic panels. With the popularization of photovoltaic power generation technology, how to clean photovoltaic panels has become increasingly important.

[0003] The photovoltaic cleaning robot can clean the photovoltaic panels by driving on them. The existing photovoltaic cleaning robots have three sensing modes: downward-looking camera, forward-looking camera, and forward-looking camera and sensor positioning fusion.

[0004] However, in order to correct the deviation of the photovoltaic cleaning robot, a variety of sensors are usually installed on the photovoltaic cleaning robot in addition to the forward-looking camera. The deviation is corrected by obtaining the measurement data of the sensors, which increases the complexity of the structure and circuit of the photovoltaic cleaning robot. In addition, due to external factors such as light, reflection, stains, etc., the measurement data of the photovoltaic panel fluctuates greatly, which can easily lead to poor correction effect of the photovoltaic cleaning robot, and thus worsen the cleaning effect of the photovoltaic panel. Summary of the invention

[0005] The present invention provides a photovoltaic cleaning robot and a driving method thereof, so as to solve the defects in the prior art that the installation of multiple sensors leads to increased complexity of the structure and circuit of the photovoltaic cleaning robot, and the photovoltaic panel measurement data fluctuates greatly, resulting in poor correction effect of the photovoltaic cleaning robot, so as to simplify the hardware structure, better correct the photovoltaic cleaning robot and improve the cleaning effect of the photovoltaic panel.

[0006] The present invention provides a driving method of a photovoltaic cleaning robot, wherein the photovoltaic cleaning robot comprises a depth camera arranged on a body, wherein the depth camera and the body are oriented in the same direction, and the method comprises:

[0007] When the photovoltaic cleaning robot is driving on the photovoltaic panel, the positioning yaw angle in the camera coordinate system is obtained through the image and depth information captured by the depth camera;

[0008] The positioning yaw angle is corrected based on the initial direction angle to obtain the heading yaw angle; the initial direction angle is the difference between the initial positioning yaw angle and the initial visual yaw angle of the photovoltaic cleaning robot; the initial visual yaw angle is the angle between the initial orientation of the photovoltaic cleaning robot and the straight texture of the photovoltaic panel;

[0009] Determining a route deviation distance based on the heading yaw angle;

[0010] Based at least on the route offset distance, the photovoltaic cleaning robot is corrected to control the photovoltaic cleaning robot to travel along the straight texture of the photovoltaic panel.

[0011] According to a driving method of a photovoltaic cleaning robot provided by the present invention, the determining of the route offset distance based on the heading yaw angle includes:

[0012] Determine a lateral translation distance of the fuselage based on the heading yaw angle and the distance from the depth camera to the center of the fuselage;

[0013] Determining an initial course offset distance of the fuselage based on the heading yaw angle and the course offset distance of the depth camera;

[0014] The course offset distance of the fuselage is determined based on the lateral translation distance of the fuselage, the initial course offset distance of the fuselage and the lateral offset correction value; the lateral offset correction value is the difference between the side margin and the initial side margin of the photovoltaic cleaning robot.

[0015] According to a driving method of a photovoltaic cleaning robot provided by the present invention, the method further includes:

[0016] When the forward frame cannot be monitored, the previous coordinate point is obtained to obtain the initial point of blind travel, and the farthest forward margin of the photovoltaic cleaning robot at the previous coordinate point is obtained to obtain the remaining distance to the edge; the farthest forward margin is the distance between the photovoltaic cleaning robot and the horizontal photovoltaic panel frame farthest from the photovoltaic cleaning robot in the robot's forward direction;

[0017] Calculating the distance traveled by the photovoltaic cleaning robot in the longitudinal direction relative to the initial point of the blind travel, and obtaining the relative distance between the depth camera and the front edge based on the difference between the remaining distance to the edge and the traveled distance;

[0018] When the relative distance is smaller than a set value, the photovoltaic cleaning robot is controlled to brake.

[0019] According to a driving method of a photovoltaic cleaning robot provided by the present invention, the photovoltaic cleaning robot is corrected based on at least the route offset distance, including:

[0020] inputting the route offset distance into a first proportional-integral-derivative controller;

[0021] In the case where the output parameter of the first proportional-integral-differential controller is a yaw angle parameter to be deflected, the yaw angle to be deflected output by the first proportional-integral-differential controller is superimposed and calculated with the heading yaw angle to obtain a target yaw angle; the target yaw angle is input into a second proportional-integral-differential controller to obtain a first differential throttle value, and the photovoltaic cleaning robot is corrected based on the first differential throttle value;

[0022] When the output parameter of the first proportional-integral-differential controller is a differential throttle parameter, the photovoltaic cleaning robot is corrected based on a second differential throttle value output by the first proportional-integral-differential controller.

[0023] According to a driving method of a photovoltaic cleaning robot provided by the present invention, the method further includes:

[0024] Obtain obstacle data when an obstacle is identified through a picture taken by the depth camera; the obstacle data includes the lateral distance of the left and right sides of at least one obstacle relative to the central axis of the robot, and the longitudinal distance of the at least one obstacle relative to the depth camera;

[0025] Based on the left-right width of the photovoltaic cleaning robot and the lateral distances of the left and right sides of the at least one obstacle relative to the robot's travel route, determine the obstacles within the robot's passing range;

[0026] From the obstacles within the robot's passing range, extract the obstacle with the smallest longitudinal distance as the nearest obstacle;

[0027] When the longitudinal distance to the nearest obstacle does not reach a preset braking threshold, controlling the photovoltaic cleaning robot to decelerate;

[0028] When the longitudinal distance to the nearest obstacle reaches a preset braking threshold, the photovoltaic cleaning robot is controlled to stop.

[0029] According to a driving method of a photovoltaic cleaning robot provided by the present invention, the method further includes:

[0030] The left margin and the right margin are obtained by a depth camera; the left margin is the lateral distance between the left frame of the photovoltaic panel and the photovoltaic cleaning robot, and the right margin is the lateral distance between the right frame of the photovoltaic panel and the photovoltaic cleaning robot;

[0031] If the left margin is smaller than the left width of the photovoltaic cleaning robot, it is determined that the left hanging mark is in a hanging state;

[0032] If the right margin is smaller than the right width of the photovoltaic cleaning robot, it is determined that the right suspended mark is in a suspended state;

[0033] When the left-side suspended mark is in a suspended state, or the right-side suspended mark is in a suspended state, the photovoltaic cleaning robot is controlled to perform differential turning based on the heading yaw angle.

[0034] According to a driving method of a photovoltaic cleaning robot provided by the present invention, the initial direction angle is determined by the following method:

[0035] When the photovoltaic cleaning robot stops braking or is in the process of driving, a plurality of visual yaw angles and a plurality of positioning yaw angles within a preset time are acquired through a depth camera;

[0036] An initial direction angle is determined based on an average visual yaw angle and an average positioning yaw angle within the preset time.

[0037] According to a driving method of a photovoltaic cleaning robot provided by the present invention, the initial direction angle is updated in the following manner:

[0038] Get the side distance through the depth camera;

[0039] Comparing an average of a predetermined number of the side margins with an initial margin of the photovoltaic cleaning robot to obtain a lateral offset correction value;

[0040] Based on the initial direction angle and the lateral offset correction value, updating the initial positioning coordinate point;

[0041] Based on the direction angle of the photovoltaic cleaning robot at the updated initial positioning coordinate point, an updated initial direction angle is obtained.

[0042] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, a driving method of a photovoltaic cleaning robot as described above is implemented.

[0043] The present invention also provides a photovoltaic cleaning robot, comprising: a depth camera, a photovoltaic cleaning robot body and the electronic device as described above; wherein the depth camera is arranged on the photovoltaic cleaning robot body, and the depth camera is in the same orientation as the photovoltaic cleaning robot body.

[0044] The photovoltaic cleaning robot and the driving method thereof provided by the present invention can realize autonomous deviation correction along the straight texture of the photovoltaic panel through a depth camera, without the need to install any sensors, and the hardware structure is simple, which can reduce the complexity of the structure and circuit of the photovoltaic cleaning robot, and does not require mapping and high-precision positioning, which can reduce performance requirements, and does not require advance entry of dimensions to adapt to photovoltaic panels of any size and type; by using the positioning yaw angle for deviation correction, since the positioning yaw angle is the angle in the camera coordinate system, the data fluctuation is small, and is not affected by external factors such as light, reflection, stains, etc., which can solve the problem of large jumps in the measurement data of the photovoltaic panel; by using the initial direction angle to correct the positioning yaw angle in the camera coordinate system, long-distance straight-line driving along the texture of the photovoltaic panel can be achieved, thereby improving the cleaning effect of the photovoltaic panel. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 This is one of the flow charts of the driving method of the photovoltaic cleaning robot provided in an embodiment of the present invention.

[0047] Figure 2 It is a schematic diagram of forward braking provided by an embodiment of the present invention.

[0048] Figure 3 It is a flow chart of the dual PID control of yaw offset provided by an embodiment of the present invention.

[0049] Figure 4 It is a schematic diagram of eliminating cumulative errors provided by an embodiment of the present invention.

[0050] Figure 5 This is the second flow chart of the driving method of the photovoltaic cleaning robot provided in the embodiment of the present invention.

[0051] Figure 6 It is a schematic diagram of a state judgment process during driving provided by an embodiment of the present invention.

[0052] Figure 7 It is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] In the description of the embodiments of the present invention, it should be noted that the terms "center", "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the embodiments of the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the embodiments of the present application. In addition, the terms "first", "second", and "third" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance.

[0055] In the description of the present invention, the terms "include", "comprising" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0056] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the embodiments of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0057] The existing photovoltaic cleaning robot's perception methods include downward-looking camera, forward-looking camera, and forward-looking camera and sensor positioning fusion:

[0058] (1) Downward-looking camera. The downward-looking camera is a camera that is perpendicular to the fuselage and points downward to obtain images of the bottom of the fuselage. It uses traditional image processing algorithms or machine learning algorithms such as semantic segmentation to identify the angle between the direction of the photovoltaic cleaning robot and the texture of the photovoltaic panel (i.e., the yaw angle) to perform yaw correction when driving in a straight line. This method requires input of information such as the size of the photovoltaic panel in advance, so that the robot can identify small-scale lateral deviations based on the different textures of the photovoltaic panel.

[0059] However, this method requires advance input of information such as the size of the photovoltaic panel, and is difficult to apply to scenarios with photovoltaic panels of various sizes. It is also difficult to identify sudden offsets over longer distances. It is also impossible to use vSLAM (Simultaneous Localization and Mapping) or SLAM assisted by other sensing devices.

[0060] (2) Forward-looking camera. The forward-looking camera is a camera that faces forward, with its line of sight parallel to the direction of travel of the PV cleaning robot, so as to obtain the image directly in front of the robot. The yaw angle is obtained using traditional image processing algorithms or machine learning algorithms such as semantic segmentation. The distance to the PV panel frame is calculated using a monocular ranging algorithm or the depth value of a depth camera, and the distance that the PV cleaning robot is offset in the lateral direction from the route is converted to correct the offset when traveling straight ahead.

[0061] However, affected by external factors such as light, reflection, and stains, the measurement data of photovoltaic panels fluctuates greatly, which is not conducive to correcting the photovoltaic cleaning robot, resulting in a deterioration in the cleaning effect of the photovoltaic panels.

[0062] (3) The forward-looking camera and sensor positioning fusion is to point the camera forward and install a variety of sensing devices such as a wheel encoder, attitude sensor (Inertial Measurement Unit, IMU, also known as inertial measurement unit), anti-fall sensor, etc., and simultaneously calculate the forward photovoltaic panel ranging information and positioning information (SLAM). By fusing forward-looking vision and sensor data, the positioning accuracy is improved and the map is built.

[0063] However, this method uses a large number of perception-related devices of various types and a relatively complex structure.

[0064] Therefore, the embodiment of the present invention only needs one depth camera, and measurement, positioning, and obstacle recognition all rely on the depth camera, without the need for fusion, and the hardware structure is relatively simple; the problem of large jumps in photovoltaic panel measurement data is solved, and long-distance straight-line driving along the photovoltaic panel texture is achieved.

[0065] Figure 1 This is one of the flow charts of the driving method of the photovoltaic cleaning robot provided by the embodiment of the present invention. Figure 1The embodiment of the present invention provides a driving method of a photovoltaic cleaning robot, and the method may specifically include the following steps:

[0066] Step 101, when the photovoltaic cleaning robot is driving on the photovoltaic panel, the positioning yaw angle in the camera coordinate system is obtained through the image and depth information captured by the depth camera.

[0067] The depth camera is a camera that can capture the depth information of the scene. It can be installed at the head position of the photovoltaic cleaning robot body, and its orientation can be consistent with the orientation of the body.

[0068] In an embodiment of the present invention, a photovoltaic cleaning robot may include a photovoltaic cleaning robot body and a depth camera arranged on the body. The depth camera and the body of the photovoltaic cleaning robot may be oriented in the same direction. Thus, during the operation and driving of the photovoltaic cleaning robot, only one depth camera is used to obtain ranging data, positioning data and obstacle data for analysis, thereby achieving autonomous deviation correction (including yaw and offset) along the straight texture of the photovoltaic panel. No sensors need to be installed, and the hardware structure is simple, which can reduce the complexity of the structure and circuit of the photovoltaic cleaning robot. There is no need for mapping and high-precision positioning, which can reduce performance requirements, and there is no need to enter the size in advance to adapt to photovoltaic panels of any size and type.

[0069] It should be noted that all data acquired by the depth camera in the embodiment of the present invention may be obtained by analyzing the RGB image and depth information of the depth camera.

[0070] Specifically, when the photovoltaic cleaning robot travels on the photovoltaic panel, it can regularly obtain the red, green, and blue depth (RGBD) data of the depth camera, including RGB data (i.e., the captured image) and Depth data (i.e., depth information) used to represent color information, and process the RGB data and Depth data into photovoltaic panel ranging data, positioning data, and obstacle data.

[0071] The ranging data may include a visual yaw angle, which may refer to the angle between the orientation of the photovoltaic cleaning robot and the straight texture of the photovoltaic panel.

[0072] The positioning data may include the XYZ three-axis coordinate data, yaw angle, pitch angle, and roll angle data fed back by the visual SLAM. Among them, the positioning yaw angle may refer to the rotation angle of the robot on the XZ plane relative to the initial position in the camera coordinate system.

[0073] In the embodiment of the present invention, when the photovoltaic cleaning robot is driving on the photovoltaic panel, the positioning yaw angle in the camera coordinate system can be obtained by the depth camera at regular intervals, so as to perform deviation correction based on the positioning yaw angle and the initial direction angle. Since the positioning yaw angle is the angle in the camera coordinate system, the data fluctuation is small and will not be affected by external factors such as light, reflection, stains, etc., which is conducive to better correction of the photovoltaic cleaning robot and improve the cleaning effect of the photovoltaic panel.

[0074] In addition, the embodiment of the present invention obtains the positioning yaw angle through the depth camera at regular intervals and corrects the current positioning yaw angle, so as to correct the driving direction of the photovoltaic cleaning robot based on the corrected yaw angle, thereby realizing uninterrupted correction during the robot's driving process, which can avoid the robot from having increasingly larger offsets / yaws and is beneficial to improving the cleaning effect of the photovoltaic panels.

[0075] In some embodiments, the depth camera can be replaced by other sensing devices. For example, in the measurement part, a forward-looking monocular camera can be used to obtain the yaw angle and the frame distance, and the recognition and monocular ranging algorithm can be used to obtain the obstacle information. The downward-looking camera can be used to obtain the yaw angle, and the cameras on both sides can be used to obtain the yaw angle and the frame distance. The forward multi-line laser radar, millimeter wave array, etc. can be used to extract the frame point cloud to calculate the yaw angle, frame distance and obstacles. For another example, real-time kinematic positioning (RTK), global positioning system (GPS), etc. can be used to obtain longitude and latitude for positioning; ultra-wideband (UWB) can be used to obtain relative position for positioning; 360-degree multi-line laser radar can be used to obtain omnidirectional point cloud and then calculate SLAM.

[0076] Step 102, correcting the positioning yaw angle based on the initial direction angle to obtain the heading yaw angle; the initial direction angle is the difference between the initial positioning yaw angle of the photovoltaic cleaning robot and the initial visual yaw angle; the initial visual yaw angle is the angle between the initial orientation of the photovoltaic cleaning robot and the straight texture of the photovoltaic panel.

[0077] Before the photovoltaic cleaning robot starts, the initial positioning yaw angle and the initial visual yaw angle of the robot at the initial position can be obtained through the depth camera, and the difference between the initial positioning yaw angle and the initial visual yaw angle can be calculated to obtain the initial direction angle.

[0078] The route yaw angle may refer to the angle between the forward direction of the photovoltaic cleaning robot and the route to be traveled. During the driving process of the photovoltaic cleaning robot, the current positioning yaw angle may be corrected by the initial direction angle to obtain the heading yaw angle, that is, the corrected positioning yaw angle. The specific correction method may be as follows:

[0079] (1);

[0080] Among them, yaw i Indicates the initial direction angle, yaw s Indicates the positioning yaw angle, yaw now Indicates the current course yaw angle.

[0081] In some embodiments, since external interference may cause cumulative errors in the positioning yaw angle, the embodiments of the present invention can update the positioning coordinate information before each start, which is beneficial to reducing the cumulative error and improving the driving correction effect.

[0082] Specifically, after the photovoltaic cleaning robot travels a certain distance and brakes to stop, before the photovoltaic cleaning robot starts, the current starting position can be used as the initial position, and the robot's positioning yaw angle and visual yaw angle can be re-acquired through the depth camera to re-determine a new initial direction angle, so as to determine the heading yaw angle based on the current positioning yaw angle and the re-determined initial direction angle after starting.

[0083] Step 103: Determine the route deviation distance based on the heading yaw angle.

[0084] In the embodiment of the present invention, the heading yaw angle may be converted to determine the heading offset distance.

[0085] Step 104, correcting the photovoltaic cleaning robot based at least on the route offset distance, so as to control the photovoltaic cleaning robot to travel along the straight texture of the photovoltaic panel.

[0086] The route offset distance may refer to the offset distance between the body of the photovoltaic cleaning robot and the route that needs to be traveled.

[0087] Specifically, the yaw angle that needs to be deflected can be determined based on the route offset distance; the target yaw angle can be determined based on the yaw angle that needs to be deflected and the heading yaw angle; based on the target yaw angle, the photovoltaic cleaning robot is corrected to control the photovoltaic cleaning robot to travel along the straight texture of the photovoltaic panel.

[0088] The present invention can realize autonomous deviation correction along the straight texture of photovoltaic panels with only one depth camera, without the need to install any sensors, and has a simple hardware structure, which can reduce the complexity of the structure and circuit of the photovoltaic cleaning robot, and does not require mapping and high-precision positioning, which can reduce performance requirements, and does not require advance entry of dimensions to adapt to photovoltaic panels of any size and type; by using the positioning yaw angle for deviation correction, since the positioning yaw angle is the angle in the camera coordinate system, the data fluctuation is small, and is not affected by external factors such as light, reflection, and stains, the problem of large jumps in the measurement data of the photovoltaic panel can be solved; by using the initial direction angle to correct the positioning yaw angle in the camera coordinate system, long-distance straight-line driving along the texture of the photovoltaic panel can be achieved, thereby improving the cleaning effect of the photovoltaic panel.

[0089] Based on any of the above embodiments, determining the route offset distance based on the heading yaw angle may specifically include:

[0090] Based on the heading yaw angle and the distance from the depth camera to the center of the fuselage, the lateral translation distance of the fuselage is determined; based on the heading yaw angle and the course offset distance of the depth camera, the initial course offset distance of the fuselage is determined; based on the lateral translation distance of the fuselage, the initial course offset distance of the fuselage and the lateral offset correction value, the course offset distance of the fuselage is determined; the lateral offset correction value is the difference between the side margin and the initial side margin of the photovoltaic cleaning robot.

[0091] In the embodiment of the present invention, the coordinate system of the robot's running plane can be converted into a relative coordinate system with the initial coordinate point as the origin and the initial direction angle as zero degrees. In the relative coordinate system, the Z' axis is the distance the photovoltaic cleaning robot travels forward along the route, and the X' axis is the lateral offset distance of the photovoltaic cleaning robot relative to the route.

[0092] Since the positioning data feeds back the position of the depth camera, when the course yaw angle is not zero, the rotation center of the fuselage is offset relative to the camera, that is, the offset distance of the depth camera and the offset distance of the midpoint of the photovoltaic cleaning robot body will deviate. Therefore, the lateral translation distance of the fuselage can be calculated based on the distance from the depth camera to the center of the fuselage and the heading yaw angle using the Pythagorean theorem, and the initial course offset distance of the fuselage can be calculated based on the heading yaw angle and the course offset distance of the depth camera using the Pythagorean theorem. Finally, the lateral translation distance of the fuselage, the initial course offset distance of the fuselage, and the lateral offset correction value can be superimposed to obtain the course offset distance of the fuselage at this time.

[0093] The embodiment of the present invention takes into account the situation where the rotation center point of the fuselage is offset relative to the camera. The lateral translation distance of the fuselage is calculated through the distance from the depth camera to the center of the fuselage and the heading yaw angle, and the course offset distance of the depth camera is converted into the course offset distance of the fuselage. The course offset distance of the fuselage is calculated based on the lateral translation distance of the fuselage, the initial course offset distance of the fuselage and the lateral offset correction value. The accuracy of the calculated course offset distance of the fuselage can be improved, which is conducive to better correcting the photovoltaic cleaning robot.

[0094] Based on any of the above embodiments, the method may further include:

[0095] When the forward frame cannot be monitored, the previous coordinate point is obtained to obtain the initial point of the blind travel, and the farthest forward margin of the photovoltaic cleaning robot at the previous coordinate point is obtained to obtain the remaining distance to the edge; the farthest forward margin is the distance between the photovoltaic cleaning robot and the horizontal photovoltaic panel frame farthest from the photovoltaic cleaning robot in the forward direction of the robot; the longitudinal distance traveled by the photovoltaic cleaning robot relative to the initial point of the blind travel is calculated, and based on the difference between the remaining distance to the edge and the traveled distance, the relative distance between the depth camera and the forward edge is obtained; when the relative distance is less than a set value, the photovoltaic cleaning robot is controlled to brake.

[0096] In some embodiments, the farthest forward margin and the left and right suspended states can be obtained periodically through a depth camera; if the farthest forward margin is less than a preset deceleration threshold, the photovoltaic cleaning robot can be controlled to decelerate; when decelerating to a coordinate point where the forward border cannot be monitored, and the left suspended state or the right suspended state is both suspended, the previous coordinate point can be used as the initial point of blind travel, and the farthest forward margin of the photovoltaic cleaning robot at the previous coordinate point can be obtained to obtain the remaining distance to the edge.

[0097] The distance measurement data of the depth camera can also include the forward margin, the left hanging state and the right hanging state. The forward margin can refer to the distance between the photovoltaic cleaning robot and the horizontal frames of each photovoltaic panel in its driving direction. The left hanging state can be used to characterize whether there is a photovoltaic panel on the leftmost side of the depth camera's field of view, that is, whether the left side of the robot is hanging. The right hanging state can be used to characterize whether there is a photovoltaic panel on the rightmost side of the depth camera's field of view, that is, whether the right side of the robot is hanging.

[0098] During the driving process, all the acquired forward margin data can be traversed, and the one farthest from the photovoltaic cleaning robot can be extracted as the farthest forward margin.

[0099] When the photovoltaic cleaning robot is about to reach the front edge, it needs to slow down in advance and brake after pushing the front cleaning mechanism out of the edge of the photovoltaic panel. However, there is a visual blind spot in the forward vision from the front cleaning mechanism to the depth camera, and the relative positions of the front cleaning mechanism, the photovoltaic cleaning robot body and the front edge cannot be "seen" through the image data of the depth camera.

[0100] Therefore, the embodiment of the present invention performs forward-to-side braking by combining distance measurement data such as the forward margin and left and right suspended markers, as well as positioning data such as the longitudinal distance traveled by the robot relative to the initial point of the blind travel.

[0101] Reference Figure 2 , Figure 2 It is a schematic diagram of forward braking provided by an embodiment of the present invention. If the farthest forward margin is less than the preset deceleration threshold, counting starts, and the photovoltaic cleaning robot is controlled to decelerate after the count exceeds a preset number of times (for example, 3 times), thereby avoiding the problem of false deceleration due to failure to identify a farther photovoltaic panel. While decelerating, the coordinate point and the farthest forward margin obtained at the coordinate point can be recorded regularly. If the farthest forward margin is detected to be greater than the preset deceleration threshold during the deceleration process, it means that a farther photovoltaic panel has been identified and normal driving can be resumed.

[0102] If the forward border cannot be monitored during deceleration, and the left and right suspended marks in the distance measurement data are both in a suspended state, the coordinate point recorded at the last moment can be used as the blind travel initial point, and the farthest forward margin obtained at the blind travel initial point can be used as the remaining distance to the edge.

[0103] In the subsequent deceleration process, the distance traveled by the current photovoltaic cleaning robot and the initial point of the blind travel in the Z axis, i.e. the longitudinal direction, can be calculated regularly, and the remaining distance to the edge minus the traveled distance can be obtained to obtain the relative distance between the depth camera position on the photovoltaic cleaning robot and the front edge (i.e. the front edge distance). When the relative distance is less than the set value, the photovoltaic cleaning robot can be controlled to brake, so that the front cleaning mechanism can be pushed out of the edge and the robot body has not reached the edge.

[0104] In some embodiments, the preset deceleration threshold may be composed of a set braking distance constant and a response lag distance, and the response lag distance may be obtained by multiplying the current straight-line speed of the robot by the response lag time of the entire system.

[0105] In some embodiments, the set value can be set according to the driving speed of the photovoltaic cleaning robot. The greater the driving speed, the greater the set value can be.

[0106] In the related art, due to the limitation of Field of View (FOV), there is a blind spot directly below the forward-looking camera, and there is a blind spot at close distances, so the robot can only brake at a distance far from the forward edge, or can only rely on the anti-fall sensor to detect the forward edge and brake; the downward-looking camera can only recognize the boundary after reaching the edge, and the body where the camera is located needs to be extended out of the photovoltaic panel. The embodiment of the present invention can achieve the goal of pushing the front cleaning mechanism out of the edge of the photovoltaic panel while the body does not extend out of the edge of the photovoltaic panel, thereby improving the cleaning effect and ensuring the safety of the robot operation.

[0107] Based on any of the above embodiments, the correcting of the photovoltaic cleaning robot based on the route offset distance and the heading yaw angle may specifically include:

[0108] The route offset distance is input into a first proportional-integral-differential controller; when the output parameter of the first proportional-integral-differential controller is a yaw angle parameter to be deflected, the yaw angle to be deflected output by the first proportional-integral-differential controller is superimposed with the heading yaw angle to obtain a target yaw angle; the target yaw angle is input into a second proportional-integral-differential controller to obtain a first differential throttle value, and the photovoltaic cleaning robot is corrected based on the first differential throttle value; when the output parameter of the first proportional-integral-differential controller is a differential throttle parameter, the photovoltaic cleaning robot is corrected based on the second differential throttle value output by the first proportional-integral-differential controller.

[0109] In the embodiment of the present invention, when the lateral distance between the photovoltaic cleaning robot and the route is greater than a threshold, differential turning is required to control the photovoltaic cleaning robot to gradually approach and return to the route. The embodiment of the present invention can adopt a variety of methods of gradually approaching and returning to the route through differential turning, and a better method can be selected according to various conditions such as motor power, drive lag, track slippage, etc.

[0110] In some embodiments, the yaw angle can be obtained by linear adjustment. Specifically, the yaw angle to be deflected to the route can be directly converted into the required number of degrees of deflection to the route according to the route offset distance of the fuselage through a linear relationship, and then added to the heading yaw angle for calculation to obtain the target yaw angle.

[0111] Figure 3 Schematic diagram of the flow of the yaw offset dual PID control provided by the embodiment of the present invention. Figure 3 , the offset and yaw can be processed by a dual proportional-integral-differential (PID) control method.

[0112] In some embodiments, when the output parameter of the yaw PID (i.e., the first PID controller) is set as the parameter of the yaw angle to be deflected, the yaw PID can output the yaw angle. The course offset distance can be used as input, and a position type or incremental PID can be used to output the yaw angle to be deflected, which is superimposed on the heading yaw angle for calculation to obtain the target yaw angle.

[0113] In some embodiments, when the output parameter of the yaw PID is set as a differential throttle parameter, the yaw PID can output a differential throttle value. The second differential throttle value can be directly output using a position or incremental PID with the course offset distance as input.

[0114] In the embodiment of the present invention, when the yaw angle formed by the robot's driving direction and the route is greater than a threshold, differential turning is required to control the photovoltaic cleaning robot's driving direction to be parallel to the route. The embodiment of the present invention can adopt a variety of methods of making the driving direction parallel to the route through differential turning, and a better method can be selected according to various conditions such as motor power, drive lag, track slippage, etc.

[0115] In some embodiments, the differential throttle value can be outputted piecewise linearly. For intervals with smaller yaw angles, a linear formula with a larger slope can be used to calculate the differential throttle value to increase the correction sensitivity; for intervals with larger yaw angles, a linear formula with a smaller slope can be used to calculate the differential throttle value to avoid jitter.

[0116] In some embodiments, the offset PID (ie, the second PID) may output a differential throttle value. The target yaw angle may be used as input, and a position or incremental PID may be used to directly output a first differential throttle value.

[0117] In some embodiments, the photovoltaic cleaning robot is corrected based on the differential throttle value, which may specifically include:

[0118] When the output is the adjusted target yaw angle, the yaw PID can continue to process. When the output is the differential throttle value, the yaw PID will not continue to process.

[0119] When the output is the differential throttle value, if the differential throttle value is less than the driving throttle, the track on the side where the robot's front is biased is controlled to keep the driving throttle unchanged, and the track on the other side is decelerated to the difference between the driving throttle and the differential throttle; if the differential throttle value is greater than the driving throttle, the track on the side where the front is biased is controlled to accelerate to the differential throttle value, and the track on the other side is braked, thereby correcting the deviation of the photovoltaic cleaning robot and making the photovoltaic cleaning robot travel along the straight texture of the photovoltaic panel.

[0120] The embodiment of the present invention processes the offset and yaw through a dual PID control method, so as to more accurately and stably control the differential turning of the photovoltaic cleaning robot, which is conducive to better correcting the photovoltaic cleaning robot.

[0121] Based on any of the above embodiments, the method further includes:

[0122] When the presence of an obstacle is recognized by the depth camera, obstacle data is obtained; the obstacle data includes the lateral distance of the left and right sides of at least one obstacle relative to the central axis of the robot, and the longitudinal distance of the at least one obstacle relative to the depth camera; based on the left and right widths of the photovoltaic cleaning robot, and the lateral distances of the left and right sides of the at least one obstacle relative to the robot's driving route, the obstacles within the robot's passing range are determined; from the obstacles within the robot's passing range, the obstacle with the smallest longitudinal distance is extracted as the nearest obstacle; when the longitudinal distance of the nearest obstacle does not reach a preset braking threshold, the photovoltaic cleaning robot is controlled to decelerate; when the longitudinal distance of the nearest obstacle reaches a preset braking threshold, the photovoltaic cleaning robot is controlled to stop.

[0123] When an obstacle is identified by the depth camera, the obstacle data can be obtained by the depth camera. All obstacle data can be traversed, and the lateral distances of the left and right sides of all obstacles from the central axis of the photovoltaic cleaning robot are first compared with the left and right widths of the photovoltaic cleaning robot to determine whether the obstacle is within the robot's passing range.

[0124] Specifically, if the left side and the right side of the obstacle are respectively located on both sides of the photovoltaic cleaning robot, then the obstacle is within the passing range of the robot.

[0125] If all obstacles are located on the left side of the photovoltaic cleaning robot, the lateral distance between the right side of the obstacle and the robot's central axis can be compared with the left width of the robot. If the lateral distance is less than the left width of the robot, it can be said that the obstacle is within the robot's passing range.

[0126] If all obstacles are located on the right side of the photovoltaic cleaning robot, the lateral distance between the left side of the obstacle and the robot's central axis can be compared with the right width of the robot. If the lateral distance is less than the right width of the robot, it can be said that the obstacle is within the robot's passing range.

[0127] After screening out multiple obstacles within the robot's passing range, an obstacle with the smallest longitudinal distance relative to the robot may be extracted as the nearest obstacle.

[0128] In an embodiment of the present invention, when the photovoltaic cleaning robot approaches an obstacle, it can decelerate in advance and brake after reaching a set braking distance. Specifically, when an obstacle on the robot's route is identified but the longitudinal distance of the nearest obstacle does not reach the set braking threshold, the photovoltaic cleaning robot can be controlled to decelerate. If the longitudinal distance of the nearest obstacle reaches the set braking threshold, the photovoltaic cleaning robot is controlled to stop.

[0129] The embodiment of the present invention can obtain obstacle data through a depth camera and identify obstacles within the robot's passing range, so that when the photovoltaic cleaning robot approaches an obstacle, it can slow down in time and brake after reaching the set braking distance, thereby avoiding collision between the robot and the obstacle and ensuring the safety of the cleaning process.

[0130] Based on any of the above embodiments, the method further includes:

[0131] The left and right margins are obtained through a depth camera; the left margin is the lateral distance between the left side frame of the photovoltaic panel and the photovoltaic cleaning robot, and the right margin is the lateral distance between the right side frame of the photovoltaic panel and the photovoltaic cleaning robot; if the left margin is smaller than the left width of the photovoltaic cleaning robot, the left suspended mark is determined to be in a suspended state; if the right margin is smaller than the right width of the photovoltaic cleaning robot, the right suspended mark is determined to be in a suspended state; when the left suspended mark is in a suspended state, or the right suspended mark is in a suspended state, the photovoltaic cleaning robot is controlled to perform a differential turn based on the heading yaw angle.

[0132] In the embodiment of the present invention, when the photovoltaic cleaning robot approaches the edge of the photovoltaic panel laterally, it needs to make a differential turn to move away from the edge of the photovoltaic panel laterally to avoid falling.

[0133] During the robot's driving process, the side distance can be recorded regularly by the depth camera, and the side distance can include the left distance and the right distance. When it is recognized that the left distance in the ranging data is smaller than the left width of the robot, the left suspended mark of the ranging data can be set to a suspended state; when it is recognized that the right distance in the ranging data is smaller than the right width of the robot, the right suspended mark of the ranging data can be set to a suspended state.

[0134] At this time, if the heading yaw angle is toward the direction of the suspension or close to zero, it can be considered that there is a risk of falling sideways. Therefore, the embodiment of the present invention can control the photovoltaic cleaning robot to perform differential turning based on the size of the heading yaw angle, thereby avoiding falling sideways.

[0135] Specifically, if the heading yaw angle is small, the track on the suspended side is controlled to decelerate, and the track on the other side is braked until the heading yaw angle reaches the threshold in the opposite direction, so that the robot is away from the edge of the photovoltaic panel to avoid falling; if the heading yaw angle is large, the track on the suspended side is controlled to decelerate, and the track on the other side is braked first and then reversed at a low speed, which provides commutation buffer for the motor and reduces the turning radius at the same time, so that the robot is away from the edge of the photovoltaic panel to avoid falling.

[0136] Based on any of the above embodiments, the initial direction angle can be determined by:

[0137] When the photovoltaic cleaning robot brakes to a stop or is driving, a plurality of visual yaw angles and a plurality of positioning yaw angles within a preset time are acquired through a depth camera; an initial direction angle is determined based on an average value of the visual yaw angle and an average value of the positioning yaw angle within the preset time.

[0138] In the embodiment of the present invention, when the photovoltaic cleaning robot is stopped by braking, or when the photovoltaic cleaning robot is driving, the depth camera can be used to regularly obtain visual data, and the distance measurement data, positioning data and obstacle data within a preset time can be stored. When calculating the initial direction angle, the average value of the distance measurement data and the average value of the positioning data within the preset time can be taken.

[0139] The visual yaw angle in the ranging data can reflect the angle between the photovoltaic cleaning robot and the straight texture of the photovoltaic panel, and the data fluctuates greatly; the positioning yaw angle in the positioning data can reflect the position of the robot that has rotated from the horizontal plane at the initial moment, and the data fluctuates less, but there will be accumulated errors. The calculation of the initial direction angle requires combining the visual yaw angle with the positioning yaw angle to obtain the angle value of the positioning yaw angle when the photovoltaic cleaning robot is assumed to be parallel to the straight texture of the photovoltaic panel. The calculation of the initial direction angle can be as follows:

[0140] (2);

[0141] Among them, yaw i Can represent the initial direction angle, yaw s Can represent the positioning yaw angle, yaw c It can represent the visual yaw angle, and n can represent the number of data obtained.

[0142] The embodiments of the present invention can avoid the problem of frame loss of image data acquired during driving, thereby affecting the measurement accuracy by measuring the initial direction angle before starting or measuring the initial direction angle during driving; by using the average value of the visual yaw angle and the average value of the positioning yaw angle to calculate the initial direction angle, the error caused by the data fluctuation of the visual yaw angle and the positioning yaw angle can be reduced, and the accuracy of the calculated initial direction angle can be improved, which is conducive to better correction of the photovoltaic cleaning robot.

[0143] Based on any of the above embodiments, the initial direction angle is updated in the following manner:

[0144] The side margins are obtained by a depth camera; an average of a predetermined number of the side margins is compared with the initial margin of the photovoltaic cleaning robot to obtain a lateral offset correction value; based on the initial direction angle and the lateral offset correction value, an initial positioning coordinate point is updated; based on the direction angle of the photovoltaic cleaning robot at the updated initial positioning coordinate point, an updated initial direction angle is obtained.

[0145] If only positioning data is used during driving, then when driving long distances, slight deviations in positioning yaw angle and photovoltaic panel problems will produce more obvious lateral deviations, and external interference may also cause accumulated errors in positioning yaw angle. The embodiment of the present invention can regularly update the positioning coordinate information based on the accumulated visual yaw angle and the left and right side margins, and update the initial direction angle based on the updated positioning coordinate information.

[0146] During the robot's driving process, since the positioning yaw angle obtained by the depth camera is the direction of the photovoltaic cleaning robot, it is necessary to continuously calculate the visual yaw angle of the robot's route parallel to the straight texture of the photovoltaic panel, so as to eliminate the accumulated error of the positioning data during driving. Therefore, it is necessary to calculate the difference between the positioning yaw angle and the visual yaw angle at the same time, and accumulate and store the positioning yaw angle error of the route.

[0147] (3);

[0148] Among them, yaw diff Can represent the cumulative error, yaw s Can represent the positioning yaw angle, yaw c Can represent the visual yaw angle.

[0149] Figure 4 Schematic diagram of eliminating cumulative errors provided by an embodiment of the present invention. Figure 4 ,When the photovoltaic cleaning robot has been driving for a certain time, the initial information can be updated once.

[0150] Specifically, according to the data accumulated when calculating the heading yaw angle, the 10 most recent differences between the positioning yaw angle and the visual yaw angle can be selected, and the average value of the 10 differences can be calculated to obtain the updated initial direction angle.

[0151] (4);

[0152] yaw diff It can be expressed as the cumulative error, It can represent the updated initial direction angle.

[0153] In some embodiments, for the multiple sets of left margin data and multiple sets of right margin data recorded, the data that is too large or too small can be deleted to reduce errors, and the left margin average value and the right margin average value are calculated respectively, and the side with the smaller margin average value is used as the correction reference side. For example, if the left margin average value is greater than the right margin average value, the left side is used as the correction reference side.

[0154] Based on the correction reference side, multiple side distance data stored during driving can be extracted, and the side distance difference can be calculated according to the weight. Among them, the side distance with an earlier recording time can have a smaller weight. For example, the latest 3 side distance data can be extracted, and the side distance average can be calculated according to the weight of 5:3:2.

[0155] In an embodiment of the present invention, the mean of the measured margins can be compared with the initial margins to obtain the distance of the lateral offset compared with the initial state. Specifically, the difference between the mean of the measured margins and the initial margins measured before the photovoltaic cleaning robot starts can be calculated to obtain the lateral offset correction value.

[0156] In an embodiment of the present invention, the initial positioning coordinate point can be updated according to the updated initial direction angle and the lateral offset correction value. Specifically, according to the current coordinate point, the updated initial direction angle can be used as the yaw angle, and the current pitch angle and roll angle can be used to perform coordinate translation and three-axis rotation to form a robot coordinate system with the current point as the origin and the direction indicated by the heading angle as the Z axis. Based on the lateral offset correction value, the point where the X-axis, Y-axis and Z-axis are zero can be determined as the updated starting point. After reverse three-axis rotation and translation, the coordinate point of the updated starting point in the positioning coordinate system is obtained to update the initial direction angle based on the coordinate point of the updated starting point.

[0157] In order to better illustrate the driving method of the photovoltaic cleaning robot provided by the present invention, a specific example is given below for further explanation:

[0158] Prerequisite: Process the RGBD data of the depth camera, that is, the RGB image data and Depth data, into photovoltaic panel ranging, positioning and obstacle data.

[0159] The ranging data may include the visual yaw angle between the center line of sight and the photovoltaic panel, the lateral distance between the left and right side frames of the photovoltaic panel in the middle of the field of view and the robot, i.e. the left and right distances, the distance between each lateral frame in the direction of travel of the robot, i.e. the forward distance, and whether there are photovoltaic panels on the left and right sides of the field of view, i.e. left hanging and right hanging.

[0160] The positioning data may include the XYZ three-axis coordinate data fed back by the visual SLAM, and the three-directional angle data of yaw, pitch, and roll.

[0161] The obstacle data may include the leftmost and rightmost lateral distances of each obstacle above the photovoltaic panel plane within the field of view relative to the robot's travel route, the longitudinal distance relative to the camera position, and the height of the obstacle itself.

[0162] Reference Figure 5 , Figure 5 This is the second flow chart of the driving method of the photovoltaic cleaning robot provided in the embodiment of the present invention.

[0163] Step 1. Measure before starting

[0164] Step 1-1, brake and stop to obtain visual data, store the distance measurement, positioning and obstacle data within a certain period of time, among which the distance measurement data and positioning data can be directly averaged, and the obstacle data can be obtained by selecting the obstacle whose lateral position intersects with the robot's lateral width and is closest to the robot.

[0165] Step 1-2, calculate the initial direction angle. The visual yaw angle in the ranging data can reflect the angle between the robot and the straight texture of the photovoltaic panel, and the data fluctuates greatly; the positioning yaw angle in the positioning data can reflect the position of the robot that has rotated on the horizontal plane at the initial moment, and the data fluctuates less, but there will be accumulated errors. The initial direction angle needs to be calculated by combining the visual yaw angle and the positioning yaw angle to obtain the angle value of the positioning yaw angle when assuming that the robot is parallel to the photovoltaic panel texture.

[0166] Step 1-3, calculate the initial margin. The left and right margins in the distance measurement data can reflect the relative position of the robot and the photovoltaic panel frame. It is necessary to eliminate the error values ​​that are too large or too small, calculate the average value, and take the smaller side as the correction reference side.

[0167] Step 1-4, record the initial coordinate point and the other two angles. The three-axis coordinate point of the averaged positioning data is used as the starting coordinate [X i , Y i , Z i ] Record the average pitch angle Pitch i and roll angle Roll i .

[0168] Step 2: Data processing during driving

[0169] Step 2-1, calculate the route yaw angle. Calculate the difference between the positioning yaw angle and the initial direction angle at this time, and use it as the route yaw angle of the robot and the route to be driven.

[0170] Step 2-2, calculate the route offset distance. Convert the coordinate system of the robot's running plane into a relative coordinate system with the initial coordinate point as the origin and the initial direction angle as zero degrees. In this coordinate system, Z' is the distance traveled forward along the route, and X' is the lateral offset distance relative to the route. Since the positioning data feeds back the position of the camera, when the route yaw angle is not zero, the offset distance of the camera and the offset distance of the midpoint of the fuselage will deviate. Therefore, it is necessary to calculate the lateral translation distance of the fuselage based on the distance from the camera to the center of the fuselage, and use this to convert the camera's route offset distance into the fuselage's route offset distance. Add the lateral offset correction value to get the route offset distance at this time.

[0171] Step 2-3, identify the farthest forward margin. Traverse all the acquired forward edge distance data and extract the farthest one as the farthest forward margin.

[0172] Step 2-4, identify the nearest obstacle. Traverse all obstacle data, first eliminate obstacles that are not within the robot's passing range based on the lateral distance between the left and right sides of the obstacle and the robot's central axis and the left and right width of the robot, then extract the obstacle with the smallest longitudinal distance to the robot as the nearest obstacle, and use the data of this obstacle as the nearest obstacle.

[0173] Step 2-5, record the yaw angle difference. Since the positioning yaw angle at this time is the direction of the robot, it is necessary to continuously calculate the positioning yaw angle of the route parallel to the photovoltaic panel texture to eliminate the accumulated error of the positioning data in the process. Therefore, it is necessary to calculate the difference between the positioning yaw angle and the visual yaw angle at the same time, and accumulate and store the positioning yaw angle of the route.

[0174] Step 2-6, record the side distance. Directly store the recognized left / right side distance.

[0175] Step 3: Driving process state machine

[0176] According to the data calculated in step 2, determine whether it is necessary to enter a certain state in step 3.

[0177] Reference Figure 6 , Figure 6 It is a schematic diagram of a state judgment process during driving provided by an embodiment of the present invention.

[0178] Step 3-1, forward to the edge

[0179] In this state, when the robot is close to reaching the front edge, it slows down in advance and pushes the front cleaning mechanism out of the edge of the photovoltaic panel before braking. The difficulty lies in the fact that the forward vision has a visual blind spot from the front cleaning mechanism to the camera, and it is impossible to directly "see" the relative positions of the front cleaning mechanism, the robot body, and the front edge. It requires the combination of distance measurement and positioning data.

[0180] To reach the edge, you need to calculate the deceleration threshold first. This threshold is composed of the set braking distance constant and the response lag distance. The response lag distance is the product of the current straight-line speed and the response lag time of the entire system.

[0181] If the farthest forward margin is less than the deceleration threshold, the counting starts. In order to avoid false deceleration due to failure to identify the photovoltaic panel far away, it is necessary to start deceleration after counting more than 3 times, and start recording the farthest forward distance and coordinate point at this time. If the farthest forward margin is found to be greater than the deceleration threshold during deceleration, normal driving will be resumed.

[0182] If the farthest forward distance cannot be identified, and both the left and right hanging marks in the distance measurement data are reported as hanging, the coordinate point recorded last time is used as the blind initial point and the remaining distance to the edge. After that, the distance traveled by the robot in the Z axis, i.e. the longitudinal direction, between the blind initial point is calculated each time, and the relative distance between the robot's camera position and the forward edge can be calculated by subtracting the traveled distance from the remaining distance, so that the robot can brake accurately at the position where the front cleaning mechanism pushes out the edge but the robot body has not reached the edge.

[0183] Step 3-2: Encountering obstacles

[0184] In this state, when the robot approaches an obstacle, it decelerates in advance and brakes after reaching the set braking distance.

[0185] When an obstacle on the route is identified, but the longitudinal distance of the nearest obstacle is not up to the set braking threshold, the robot slows down. If the threshold is reached, the robot brakes.

[0186] Step 3-3, sideways suspension

[0187] In this state, when the robot approaches the edge sideways, it makes a differential turn to move away from the lateral edge.

[0188] When it is recognized that the left / right side frame distance in the ranging data is less than the left / right side width of the robot, the left / right side hanging mark feedback of the ranging data is hanging, and the heading yaw angle is in the hanging direction or close to zero, it is considered that there is a risk of sideways falling.

[0189] If the heading yaw angle is small, the track on the suspended side will slow down and the track on the other side will brake until the heading yaw angle reaches the threshold in the opposite direction; if the heading yaw angle is large, the track on the suspended side will slow down and the track on the other side will brake first and then reverse at a low speed to provide a buffer for the motor commutation while reducing the turning radius.

[0190] Step 3-4, offset

[0191] In this state, when the lateral distance between the robot and the route is greater than the threshold, differential turning is required to gradually approach and return to the route. The present invention adopts three different methods, and selects the better method according to various conditions such as motor power, drive lag, track slippage, etc. Some of the methods process the offset and yaw through the dual PID control method, while the rest of the methods use the ordinary method.

[0192] The first method is to adjust the yaw angle linearly. According to the offset distance in Step 2-2, the yaw angle that needs to be deflected to the route is directly converted through a linear relationship, and then added to the heading yaw angle in Step 2-1 to enter Step 3-5 for calculation.

[0193] The second method is to use PID to output the yaw angle. With the offset distance as input, position or incremental PID is used to output the yaw angle that needs to be deflected, which is added to the heading yaw angle in Step 2-1 and then entered into Step 3-5 for calculation.

[0194] The third type is PID output differential throttle. With the offset distance as input, position or incremental PID is used to directly output the throttle value required for differential turning.

[0195] Step 3-5, Yaw

[0196] In this state, when the yaw angle formed by the robot's driving direction and the route is greater than the threshold, differential turning is required to make the driving direction parallel to the route. The present invention adopts two different methods, and the better method is selected according to various conditions such as motor power, drive lag, track slippage, etc.

[0197] The first is the segmented linear output differential throttle. In the range of small yaw angle, a linear formula with a larger slope is used to increase the correction sensitivity; in the range of large yaw angle, a linear formula with a smaller slope is used to avoid jitter.

[0198] The second type is PID output differential throttle. With the yaw angle as input, position or incremental PID is used to directly output the throttle value required for differential turning.

[0199] If the first and second methods are used in Step 3-4 offset, the output is the adjusted yaw angle, and it is necessary to enter Step 3-5 yaw to continue processing. However, the third method of Step 3-4 offset outputs the throttle value, and does not enter Step 3-5 yaw.

[0200] If the differential throttle output by the third method of offset in Step 3-4 and the two methods of yaw in Step 3-5 is less than the driving throttle, the track on the side where the front of the vehicle is deviated from keeps the driving throttle unchanged, and the track on the other side is decelerated by the difference between the driving throttle and the differential throttle; if the differential throttle is greater than the driving throttle, the track on the side where the front of the vehicle is deviated from is accelerated to the differential throttle, and the track on the other side is braked.

[0201] Step 4: Update the coordinates during driving

[0202] If only positioning data is used, then when driving long distances, slight deviations in positioning yaw angle and photovoltaic panel problems will result in more obvious lateral deviations. At the same time, external interference may also cause cumulative errors in positioning yaw angle. Therefore, it is necessary to regularly update the positioning coordinate information based on the accumulated visual yaw angle and left and right side margins. Figure 5 shown.

[0203] After driving for a certain period of time, if it is determined in Step 3 that the vehicle does not enter any state and keeps going straight, the initial information is updated once.

[0204] Step 4-1, heading angle update

[0205] According to the records accumulated when B1 calculates the heading yaw angle, the differences between the latest 10 positioning yaw angles and the visual yaw angle are selected, and the average is taken to calculate the new initial direction angle.

[0206] Step 4-2, update the horizontal offset correction value of the coordinate point

[0207] According to the reference left / right side determined by A3, extract the latest three accumulated records of this side stored in Step 2-6, and calculate the average value according to the weight of 5:3:2. Compare this value with the initial margin distance of Step 1-3 to obtain the lateral offset distance compared with the initial state.

[0208] Step 4-3, coordinate point update

[0209] Get the latest positioning coordinate point and update the coordinate point according to the heading angle in Step 4-1 and the offset correction value in Step 4-2.

[0210] According to the current coordinate point, use the heading angle of Step 4-1 as the yaw angle, and use the current pitch angle and roll angle to perform coordinate translation and three-axis rotation to form a robot coordinate system with the current point as the origin and the direction of the heading angle as the Z axis. Then use the offset correction value of Step 4-2 as the point where the X-axis, Y-axis and Z-axis are zero as the updated starting point, and after reverse three-axis rotation and translation, obtain the coordinate point of the updated starting point in the positioning coordinate system.

[0211] Figure 7A schematic diagram of the physical structure of an electronic device is shown as an example. Figure 7 , the electronic device may include: a processor 710, a communication interface 720, a memory 730 and a communication bus 740, wherein the processor 710, the communication interface 720 and the memory 730 communicate with each other through the communication bus 740. The processor 710 may call the logic instructions in the memory 730 to execute the driving method of the photovoltaic cleaning robot, the method comprising: when the photovoltaic cleaning robot is driving on the photovoltaic panel, the positioning yaw angle in the camera coordinate system is obtained through the captured image and depth information of the depth camera; the positioning yaw angle is corrected based on the initial direction angle to obtain the heading yaw angle; the initial direction angle is the difference between the initial positioning yaw angle and the initial visual yaw angle of the photovoltaic cleaning robot; the initial visual yaw angle is the angle between the initial orientation of the photovoltaic cleaning robot and the straight line texture of the photovoltaic panel; based on the heading yaw angle, the route offset distance is determined; at least based on the route offset distance, the photovoltaic cleaning robot is corrected to control the photovoltaic cleaning robot to travel along the straight line texture of the photovoltaic panel.

[0212] In addition, the logic instructions in the above-mentioned memory 730 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0213] The present invention also provides a photovoltaic cleaning robot, comprising: a depth camera, a photovoltaic cleaning robot body and the above-mentioned electronic device; wherein the depth camera is arranged on the photovoltaic cleaning robot body, and the depth camera is in the same orientation as the photovoltaic cleaning robot body.

[0214] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0215] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0216] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A driving method of a photovoltaic cleaning robot, characterized in that: The photovoltaic cleaning robot includes a depth camera disposed on a body, the depth camera and the body are oriented in the same direction, and the method includes: When the photovoltaic cleaning robot is driving on the photovoltaic panel, the positioning yaw angle in the camera coordinate system is obtained through the image and depth information captured by the depth camera; The positioning yaw angle is corrected based on the initial direction angle to obtain the heading yaw angle; the initial direction angle is the difference between the initial positioning yaw angle and the initial visual yaw angle of the photovoltaic cleaning robot; the initial visual yaw angle is the angle between the initial orientation of the photovoltaic cleaning robot and the straight texture of the photovoltaic panel; Determining a route deviation distance based on the heading yaw angle; Based at least on the route offset distance, the photovoltaic cleaning robot is corrected to control the photovoltaic cleaning robot to travel along the straight texture of the photovoltaic panel; Wherein, determining the route deviation distance based on the heading yaw angle includes: The lateral translation distance of the fuselage is calculated based on the heading yaw angle and the distance from the depth camera to the center of the fuselage using the Pythagorean theorem; The initial course offset distance of the fuselage is calculated based on the heading yaw angle and the course offset distance of the depth camera using the Pythagorean theorem; wherein the course offset distance of the depth camera is the difference between the offset distance of the depth camera and the offset distance of the midpoint of the fuselage when the rotation center point of the fuselage is offset relative to the depth camera; The lateral translation distance of the fuselage, the initial course offset distance of the fuselage and the lateral offset correction value are superimposed to obtain the course offset distance of the fuselage; the lateral offset correction value is the difference between the side margin of the photovoltaic cleaning robot and the initial side margin; The step of correcting the photovoltaic cleaning robot based at least on the route offset distance includes: Determining a yaw angle to be deflected based on the route offset distance; Determining a target yaw angle based on the yaw angle to be deflected and the heading yaw angle; Based on the target yaw angle, the photovoltaic cleaning robot is corrected.

2. The driving method of the photovoltaic cleaning robot according to claim 1, characterized in that: The method further comprises: When the forward frame cannot be monitored, the previous coordinate point is obtained to obtain the initial point of blind travel, and the farthest forward margin of the photovoltaic cleaning robot at the previous coordinate point is obtained to obtain the remaining distance to the edge; the farthest forward margin is the distance between the photovoltaic cleaning robot and the horizontal photovoltaic panel frame farthest from the photovoltaic cleaning robot in the robot's forward direction; Calculating the distance traveled by the photovoltaic cleaning robot in the longitudinal direction relative to the initial point of the blind travel, and obtaining the relative distance between the depth camera and the front edge based on the difference between the remaining distance to the edge and the traveled distance; When the relative distance is smaller than a set value, the photovoltaic cleaning robot is controlled to brake.

3. The driving method of the photovoltaic cleaning robot according to claim 1, characterized in that: The step of correcting the photovoltaic cleaning robot based at least on the route offset distance includes: inputting the route offset distance into a first proportional-integral-derivative controller; In the case where the output parameter of the first proportional-integral-differential controller is a yaw angle parameter to be deflected, the yaw angle to be deflected output by the first proportional-integral-differential controller is superimposed and calculated with the heading yaw angle to obtain a target yaw angle; the target yaw angle is input into a second proportional-integral-differential controller to obtain a first differential throttle value, and the photovoltaic cleaning robot is corrected based on the first differential throttle value; When the output parameter of the first proportional-integral-differential controller is a differential throttle parameter, the photovoltaic cleaning robot is corrected based on a second differential throttle value output by the first proportional-integral-differential controller.

4. The driving method of the photovoltaic cleaning robot according to claim 1, characterized in that: The method further comprises: When an obstacle is identified by the depth camera, obstacle data is obtained; the obstacle data includes the lateral distance of the left and right sides of at least one obstacle relative to the central axis of the robot, and the longitudinal distance of the at least one obstacle relative to the depth camera; Based on the left-right width of the photovoltaic cleaning robot and the lateral distances of the left and right sides of the at least one obstacle relative to the robot's travel route, determine the obstacles within the robot's passing range; From the obstacles within the robot's passing range, extract the obstacle with the smallest longitudinal distance as the nearest obstacle; When the longitudinal distance to the nearest obstacle does not reach a preset braking threshold, controlling the photovoltaic cleaning robot to decelerate; When the longitudinal distance to the nearest obstacle reaches a preset braking threshold, the photovoltaic cleaning robot is controlled to stop.

5. The driving method of the photovoltaic cleaning robot according to claim 1, characterized in that: The method further comprises: The left margin and the right margin are obtained by a depth camera; the left margin is the lateral distance between the left frame of the photovoltaic panel and the photovoltaic cleaning robot, and the right margin is the lateral distance between the right frame of the photovoltaic panel and the photovoltaic cleaning robot; If the left margin is smaller than the left width of the photovoltaic cleaning robot, it is determined that the left hanging mark is in a hanging state; If the right margin is smaller than the right width of the photovoltaic cleaning robot, it is determined that the right suspended mark is in a suspended state; When the left-side suspended mark is in a suspended state, or the right-side suspended mark is in a suspended state, the photovoltaic cleaning robot is controlled to perform differential turning based on the heading yaw angle.

6. The driving method of the photovoltaic cleaning robot according to claim 1, characterized in that: The initial direction angle is determined by: When the photovoltaic cleaning robot stops braking or is in the process of driving, a plurality of visual yaw angles and a plurality of positioning yaw angles within a preset time are acquired through a depth camera; An initial direction angle is determined based on an average visual yaw angle and an average positioning yaw angle within the preset time.

7. The driving method of the photovoltaic cleaning robot according to claim 6, characterized in that: The initial direction angle is updated in the following way: Get the side distance through the depth camera; Comparing an average of a predetermined number of the side margins with an initial margin of the photovoltaic cleaning robot to obtain a lateral offset correction value; Based on the initial direction angle and the lateral offset correction value, updating the initial positioning coordinate point; Based on the direction angle of the photovoltaic cleaning robot at the updated initial positioning coordinate point, an updated initial direction angle is obtained.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the driving method of the photovoltaic cleaning robot as described in any one of claims 1 to 7 is implemented.

9. A photovoltaic cleaning robot, characterized in that: include: A depth camera, a photovoltaic cleaning robot body and an electronic device as claimed in claim 8; Wherein, the depth camera is arranged on the body of the photovoltaic cleaning robot, and the depth camera is in the same direction as the body of the photovoltaic cleaning robot.

Images