Weld joint identifying and tracking control method and device for wall-climbing laser cleaning robot

Abstract

The invention relates to a wall-climbing laser cleaning robot weld joint identification tracking control method and device, and belongs to the technical field of robot tracking control. The maximum curvature of the arc length between the position point of the robot at the current moment and a preview point is determined, the maximum longitudinal speed corresponding to the maximum curvature is determined, and the preview point is any point on the weld joint track in the advancing direction of the robot; and determining a predicted position of the robot at the next moment according to the maximum curvature and the maximum longitudinal speed, and controlling the robot to move to the predicted position. The method solves the problem that the precision of a dynamical model and an operation control strategy of the wall-climbing robot is low because the wall surface motion control of the wall-climbing robot does not consider the influence of adsorption force and gravity.

Description

Technical Field

[0001] The present invention relates to the technical field of wall - climbing robot tracking control, and particularly to a method and device for weld seam recognition and tracking control of a wall - climbing laser cleaning robot. Background Art

[0002] With the accelerating pace of the intelligent transformation of the global manufacturing industry, the traditional manual method of rust removal / grinding of weld seams can no longer meet the urgent needs of modern industry for high - efficiency and high - quality production. Especially in the manufacturing fields of large - scale structural parts such as shipbuilding, oil and gas chemical engineering, the large - scale spliced workpieces in sectional assembly are huge in volume, complex in structure, and have irregular weld seam shapes, posing higher requirements for the rust removal / grinding operations of weld seams.

[0003] Currently, although there are semi - automated operation methods using modified elevated vehicles or remotely controlled rust removal / grinding wall - climbing robots, these methods still have certain limitations. The operation of elevated vehicles is limited by the length and flexibility of the elevated arms and is difficult to cover all weld seam areas; while remotely controlled rust removal / grinding robots can reduce the direct participation of manual labor, but the operator still needs to constantly pay attention to the pose state and rust removal / grinding situation of the robot, and cannot achieve true autonomous operation. And in the wall - surface motion control of wall - climbing robots, although a simple controller based only on the kinematic model can simplify the tracking problem of non - holonomic constraint systems, it is based on the assumption of no slip, which is inconsistent with the actual situation. In fact, the dynamic model of wheeled wall - climbing robots must consider the influence of adsorption force and gravity.

[0004] Therefore, developing more accurate dynamic models and control strategies to adapt to complex and variable operating environments is an important direction of current technological development. Summary of the Invention

[0005] In view of this, it is necessary to provide a method and device for weld seam recognition and tracking control of a wall - climbing laser cleaning robot to solve the problem in the prior art that the wall - surface motion control of the wall - climbing robot does not consider the influence of adsorption force and gravity, resulting in low accuracy of the dynamic model and operation control strategy of the wall - climbing robot.

[0006] To solve the above problems, the present invention provides a method for weld seam recognition and tracking control of a wall - climbing laser cleaning robot, including: Determine the weld seam trajectory; Determine the maximum curvature of the arc length between the position point of the robot at the current moment and the preview point, and determine the maximum longitudinal speed corresponding to the maximum curvature, where the preview point is any point on the weld seam trajectory along the forward direction of the robot; Determine the predicted position of the robot at the next moment according to the maximum curvature and the maximum longitudinal speed, and control the robot to move to the predicted position.

[0007] In a possible implementation, determining the maximum curvature of the arc length between the position point of the robot at the current moment and the preview point includes: Determine a reference point, where the reference point is the point on the weld track closest to the robot along the forward direction of the robot; Determine the preview distance according to the weld arc length between the reference point and the preview point, and determine the maximum curvature of the weld track within the preview distance.

[0008] In a possible implementation, determining the maximum longitudinal speed corresponding to the maximum curvature includes: Construct a local coordinate system with the robot as the main body; Based on the local coordinate system, determine the relationship between the maximum curvature of the robot and the longitudinal speed at any moment according to the robot kinematic model; wherein, in the robot motion model, the longitudinal speed of the robot is the average of the left and right wheel speeds, and is proportional to the difference between the yaw angular velocity and the left and right wheel speeds; According to the magnitude relationship between the maximum steering curvature of the robot and the path curvature of the weld track, and the relationship between the maximum steering curvature of the robot and the longitudinal speed at any moment, determine the maximum longitudinal speed of the robot under the maximum curvature of the weld track.

[0009] In a possible implementation, according to the maximum curvature and the maximum longitudinal speed to determine the predicted position of the robot at the next moment includes: Based on the local coordinate system, obtain the position information of the robot at the current moment and the yaw angle at the current moment; Taking the position information of the robot at the current moment and the yaw angle at the current moment as inputs, and the maximum curvature and the maximum longitudinal speed as constraint conditions, adopt a fuzzy control prediction strategy to obtain the predicted position of the robot at the next moment.

[0010] In a possible implementation, the fuzzy control prediction strategy is implemented by using a motion controller model; after determining the predicted position of the robot at the next moment, it further includes: Obtain the position of the robot at the current moment; Calculate the deviation between the position of the robot at the current moment and the predicted position, where the deviation includes the trajectory deviation angle and the deviation distance; According to the deviation, determine the magnitude of the control signal of the motion controller model; According to the control signal, adjust the motion parameters of the robot.

[0011] In a possible implementation, after adjusting the motion parameters of the robot according to the control signal, it further includes: Based on the adjustment result of the previous robot motion parameters, the trajectory deviation angle of the robot at the current moment is weighted to obtain a weighted trajectory deviation angle; Determine the magnitude of the control signal according to the weighted deviation angle and the deviation distance.

[0012] In a possible implementation manner, the determining the weld seam trajectory includes: Obtain the weld seam plane vision image collected by the first camera device and the laser cleaning process image collected by the second camera device, where the first camera device has a top view angle relative to the robot, and the second camera device has a front view angle relative to the robot; Match the features of the co-visual relationship image key frames of the first camera device and the second camera device, the weld seam plane vision image key frames, and the laser cleaning process image key frames to determine the initial weld seam trajectory image; Extract the initial center line of the weld seam in the initial weld seam trajectory image; Perform laser cleaning on the initial center line, and determine the weld seam trajectory according to the edge information of the laser cleaning process image and the laser spot information collected by the second camera device.

[0013] In a possible implementation manner, the matching the features of the co-visual relationship image key frames of the first camera device and the second camera device, the weld seam plane vision image key frames, and the laser cleaning process image key frames to determine the initial weld seam trajectory image includes: Obtain the first feature points of the weld seam plane vision image key frames, the second feature points of the laser cleaning process image key frames, and the third feature points of the co-visual relationship image key frames, and match the first feature points, the second feature points, and the third feature points according to similarity to obtain matching features; Calibrate the internal parameters and external parameters of the first camera device, and calibrate the internal parameters and external parameters of the second camera device; Based on the calibrated internal parameters and external parameters of the first camera device and the internal parameters and external parameters of the second camera device, establish a reprojection error function according to the matching features; Use a preset optimization algorithm to determine the minimum error projection of the reprojection error function; According to the minimum error projection, determine the initial weld seam trajectory and determine the initial weld seam trajectory image.

[0014] In a possible implementation manner, the performing laser cleaning on the initial center line and determining the weld seam trajectory according to the edge information of the laser cleaning process image and the laser spot information collected by the second camera device includes: Perform laser cleaning on the initial center line, and extract the edge information of the laser spot in the initial weld seam trajectory image; Calculate the midline of the edge information of the laser spot to determine the skeleton curve; Perform a preset straight line Hough transform on the skeleton curve to determine the straight lines on both sides of the skeleton curve and determine the weld seam trajectory.

[0015] In a second aspect, the present invention also provides a weld seam recognition and tracking device for a wall-climbing laser cleaning robot, including: A trajectory determination module for determining the weld seam trajectory; A maximum longitudinal speed determination module for determining the maximum curvature of the arc length between the position point of the robot at the current moment and the preview point, and determining the maximum longitudinal speed corresponding to the maximum curvature, where the preview point is any point on the weld seam trajectory along the forward direction of the robot; A control module for determining the predicted position of the robot at the next moment according to the maximum curvature and the maximum longitudinal speed, and controlling the robot to move to the predicted position.

[0016] In a third aspect, the present invention also provides a robot motion controller, including: a processor and a memory; The memory stores a computer-readable program executable by the processor; When the processor executes the computer-readable program, it implements the steps in the above-mentioned weld seam recognition and tracking control method for the wall-climbing laser cleaning robot.

[0017] In a fourth aspect, the present invention also provides a computer-readable storage medium, which stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps in the above-mentioned weld seam recognition and tracking control method for the wall-climbing laser cleaning robot.

[0018] The beneficial effects of the present invention are: in order to ensure that the robot does not slip under the influence of adsorption force and gravity, the present invention first determines the weld seam trajectory to determine the corresponding running path of the robot during laser cleaning, and then determines the maximum curvature of the arc length between the position point of the robot at the current moment and the preview point (the preview point is any point on the weld seam trajectory along the forward direction of the robot), and determines the maximum longitudinal speed corresponding to the maximum curvature. At this time, determining the maximum curvature of the trajectory arc is to determine the maximum influence degree under the influence of adsorption force and gravity, so as to overcome this maximum influence degree through the maximum longitudinal speed to ensure that the robot does not slip; finally, taking the maximum curvature and the maximum longitudinal speed as constraint conditions, predicting the predicted position of the robot at the next moment, and thus controlling the robot to move to the predicted position, so as to perform real-time tracking control without slipping. Description of the Drawings

[0019] Figure 1Flowchart of a method according to an embodiment of the weld seam recognition and tracking control method for a wall - climbing laser cleaning robot provided by the present invention; Figure 2 Flowchart of an embodiment of step S102 in the weld seam recognition and tracking control method for a wall - climbing laser cleaning robot provided by the present invention; Figure 3 Flowchart of another embodiment of step S102 in the weld seam recognition and tracking control method for a wall - climbing laser cleaning robot provided by the present invention; Figure 4 Schematic diagram of the kinematic model of the robot in the weld seam recognition and tracking control method for a wall - climbing laser cleaning robot provided by the present invention; Figure 5 Flowchart of an embodiment of step S103 in the weld seam recognition and tracking control method for a wall - climbing laser cleaning robot provided by the present invention; Figure 6 Flowchart of a method according to another embodiment of the weld seam recognition and tracking control method for a wall - climbing laser cleaning robot provided by the present invention; Figure 7 Schematic diagram of the recognition of the weld seam position state in the weld seam recognition and tracking control method for a wall - climbing laser cleaning robot provided by the present invention; Figure 8 Schematic diagram of the deviation correction method after the robot reaches the next preview point in the weld seam recognition and tracking control method for a wall - climbing laser cleaning robot provided by the present invention; Figure 9 Schematic diagram of an embodiment of the weld seam recognition and tracking device for a wall - climbing laser cleaning robot provided by the present invention; Figure 10 Schematic diagram of the operating environment of an embodiment of the robot motion controller provided by the present invention. Detailed implementation manners

[0020] The following will specifically describe the preferred embodiments of the present invention with reference to the accompanying drawings. The accompanying drawings form a part of this application and are used together with the embodiments of the present invention to explain the principle of the present invention, and are not used to limit the scope of the present invention.

[0021] In the description of the embodiments of the present invention, unless otherwise specified, "a plurality of" means two or more. "And / or" describes the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone.

[0022] In the embodiments of the present invention, the descriptions such as "first" and "second" are only for descriptive purposes and should not be construed as indicating or implying their relative importance or implicitly specifying the quantity of the indicated technical features. Therefore, the technical features defined with "first" and "second" may explicitly or implicitly include at least one such feature.

[0023] Reference to "embodiment" herein means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the present invention. The phrase appears in various places in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0024] Before presenting the embodiments, the following terms are explained first.

[0025] Weld seam trajectory: Refers to the geometric path of the weld seam in space, usually a continuous curve or broken line, which describes the shape and position of the weld seam.

[0026] Wall-climbing laser cleaning robot: It is a robot system that can crawl on vertical walls or other complex surfaces and use laser technology for cleaning operations. The wall-climbing laser cleaning robot usually has a certain degree of autonomy and can perform autonomous movement and cleaning operations according to preset paths or real-time perceived environmental information.

[0027] Recognition and tracking: It means that the robot identifies the target path (such as the weld seam trajectory) through sensors (such as vision sensors) and adjusts its own movement trajectory in real time to maintain consistency with the target path. Specifically, recognition is to obtain the image information of the weld seam through a vision sensor (such as an industrial camera) and extract the weld seam trajectory through image processing algorithms (such as edge detection, feature extraction, etc.). Tracking is that the robot adjusts its own movement parameters (such as speed, steering angular velocity, etc.) in real time according to the recognized weld seam trajectory to ensure that the laser cleaner is always aligned with the center line of the weld seam.

[0028] Maximum curvature: Refers to the curvature value of the point with the maximum curvature in a section of path or curve. Curvature is a quantitative description of the degree of path bending and is usually defined as the reciprocal of the curvature radius of a certain point on the path.

[0029] The present invention provides a method and device for controlling the recognition and tracking of the weld seam of a wall-climbing laser cleaning robot, which are described separately below.

[0030] Figure 1 It is a schematic flowchart of an embodiment of the method for controlling the recognition and tracking of the weld seam of the wall-climbing laser cleaning robot provided by the present invention. As Figure 1 shown, the method for controlling the recognition and tracking of the weld seam of the wall-climbing laser cleaning robot includes: S101. Determine the weld seam trajectory; It should be noted that the weld seam trajectory is the basis for the robot's motion control. The robot needs to perform laser cleaning operations along this trajectory. Therefore, determining the accurate extraction of the weld seam trajectory can improve the accuracy and efficiency of the robot's tracking.

[0031] In this embodiment, the method for determining the weld seam trajectory is not limited.

[0032] S102. Determine the maximum curvature of the arc length between the position point of the robot at the current moment and the preview point, and determine the maximum longitudinal speed corresponding to the maximum curvature, where the preview point is any point on the weld seam trajectory along the forward direction of the robot; It should be noted that the wall-climbing robot operates on a vertical wall surface or a complex surface, and its working environment is different from that of a ground vehicle robot. Due to the particularity of wall surface operations (such as the need for stable adsorption and avoiding falling off, etc.), the wall-climbing robot does not need to travel at high speed. And when operating on the wall surface, the pitch angle and roll angle of the robot usually do not have a significant impact on its main motion postures (such as the forward direction and steering ability). Therefore, in this embodiment, when considering the robot slip factor, the influence of the pitch angle on the parameters does not need to be considered.

[0033] Furthermore, when the robot is tracking a path, its steering ability is restricted by its kinematic characteristics. If the curvature of the path is too large and the speed of the robot is too high, the robot may not be able to accurately track the path and may even deviate from the path. Therefore, without considering factors such as the pitch angle, the speed of the robot is restricted according to the curvature of the path to ensure that its steering ability can meet the tracking requirements.

[0034] Furthermore, in this embodiment, by constraining the relationship between the maximum curvature and the maximum longitudinal speed, determining the maximum curvature of the trajectory arc at this time is to determine the maximum influence degree under the influence of the adsorption force and gravity, so as to overcome this maximum influence degree through the maximum longitudinal speed to ensure that the robot does not slip; and predicting the predicted position of the robot at the next moment with the maximum curvature and the maximum longitudinal speed as the limiting conditions, and then controlling the motion of the robot according to the predicted position at the next moment to ensure that the robot travels along the weld seam trajectory without slipping.

[0035] It should be noted that the longitudinal speed refers to the speed of the robot along its forward direction (usually the weld seam direction).

[0036] S103. Determine the predicted position of the robot at the next moment according to the maximum curvature and the maximum longitudinal speed, and control the robot to move to the predicted position.

[0037] It should be noted that, with the maximum curvature and the maximum longitudinal speed as the constraint conditions, determining the maximum curvature of the trajectory arc is to determine the maximum influence degree under the influence of the adsorption force and the gravity, and then overcome this maximum influence degree through the maximum longitudinal speed to ensure that the robot does not slip, so as to obtain the predicted position at the next moment, and ensure that the robot always travels along the weld track.

[0038] Furthermore, the prediction method is not limited herein.

[0039] In summary, for the weld recognition and tracking control method of the wall-climbing laser cleaning robot provided by the embodiment of the present invention, in order to ensure that the robot does not slip under the influence of the adsorption force and the gravity, the present invention first determines the weld track to determine the corresponding running path of the robot during laser cleaning, and then determines the maximum curvature of the arc length between the position point of the robot at the current moment and the preview point (the preview point is any point on the weld track along the advancing direction of the robot), and determines the maximum longitudinal speed corresponding to the maximum curvature. At this time, determining the maximum curvature of the trajectory arc is to determine the maximum influence degree under the influence of the adsorption force and the gravity, and then overcome this maximum influence degree through the maximum longitudinal speed to ensure that the robot does not slip; finally, with the maximum curvature and the maximum longitudinal speed as the constraint conditions, predict the predicted position of the robot at the next moment, so as to control the robot to move to the predicted position, so as to perform real-time tracking control without slipping.

[0040] In some embodiments of the present invention, as Figure 2 shown, step S102 may include: S201. Determine a reference point, where the reference point is the point on the weld track along the advancing direction of the robot that is closest to the robot; S202. Determine the preview distance according to the weld arc length between the reference point and the preview point, and determine the maximum curvature of the weld track within the preview distance.

[0041] It should be noted that the selection of the reference point on the weld track is random, but the closer the reference point is to the robot, the smaller the error of its predicted position. Therefore, in this embodiment, the reference point is the point closest to the robot.

[0042] In this embodiment, by calculating the maximum curvature and the corresponding upper limit of the longitudinal speed in real time, the motion control of the robot under different path conditions is made more flexible and adaptable. Especially in the area where the path curvature changes greatly, the robot can automatically reduce the speed to improve the tracking ability, so as to adapt to various complex weld shapes.

[0043] Furthermore, there is an inverse relationship between the curvature of the weld track and the speed of the robot. In order to establish the relationship between the maximum curvature and the maximum longitudinal speed, asFigure 3 As shown, step S102 further includes: S301. Construct a local coordinate system with the robot as the main body; It should be noted that when establishing a local coordinate system with the robot as the core, the motion state of the robot itself, such as speed, acceleration, yaw angular velocity, etc., is more intuitive and simple, and there is no need to consider the complex transformation in the global coordinate system.

[0044] In a specific embodiment, the origin of the coordinate system xoy of the local coordinate system is the center of the connection line between the two wheels of the robot, and a reference coordinate system xtotyt for the robot's motion at time t is established. Since the laser cleaner is fixedly connected to the robot, that is, the running trajectory of the laser cleaner and the emitted laser is consistent with the robot's motion trajectory, and the fixed distance between the laser cleaner and the origin of the local coordinate system is l.

[0045] S302. Based on the local coordinate system, determine the relationship between the maximum curvature of the robot and the longitudinal velocity at any moment according to the robot kinematic model; wherein, the robot motion model includes that the longitudinal velocity of the robot is the average value of the left and right wheel velocities, and the yaw angular velocity is proportional to the difference between the left and right wheel velocities; It should be noted that the kinematic model is used to describe the motion characteristics of the robot in space, including the changes in position, velocity, and attitude. It does not involve dynamics but focuses on the geometric and time characteristics of motion. As Figure 4 shown, through the kinematic model, the path of the robot from the starting point to the end point can be planned, and the required motion parameters can be calculated.

[0046] According to the robot motion model, the following relationship holds among the longitudinal velocity, yaw angular velocity, and linear velocities of the left and right driving wheels of the robot:

[0047] In the formula, V x is the longitudinal velocity of the robot, V 1, V 2 are the linear velocities of the left and right driving wheels, is the yaw angular velocity of the robot, B is the distance between the left and right driving wheels.

[0048] Since there is a maximum longitudinal velocity, and the maximum linear velocity of a single driving wheel is equal to the maximum longitudinal velocity, when the longitudinal velocity is known, the maximum yaw angular velocity of the robot satisfies:

[0049] In the formula, is the maximum longitudinal velocity, is the maximum yaw angular velocity.

[0050] Therefore, when the longitudinal speed is known, the maximum turning curvature of the robot c max can be expressed as:

[0051] When the curvature of the weld path c ref is known and the robot has the path tracking ability, the turning ability of the robot satisfies:

[0052] Therefore, it can be obtained that:

[0053] That is, the longitudinal speed must satisfy:

[0054] It can be seen from this that when the curvature of the reference path is known, if the ability of the robot to track the reference path is to be guaranteed, there is an upper limit to its longitudinal speed.

[0055]

[0056] In the formula, is the upper limit of the feasible longitudinal speed of the robot under the premise of ensuring the path tracking ability.

[0057] S303. Determine the maximum longitudinal speed of the robot at the maximum curvature of the weld path according to the magnitude relationship between the maximum turning curvature of the robot and the path curvature of the weld path, and the relationship between the maximum turning curvature of the robot and the longitudinal speed at any moment.

[0058] Since the maximum value of the longitudinal speed of the robot is a fixed value, the upper limit of the feasible longitudinal speed of the robot under the premise of ensuring the path tracking ability is inversely proportional to the curvature of the reference path. Taking the point on the weld path closest to the robot as the reference point, a preview point is taken on the weld path along the forward direction of the robot. The arc length of the weld path between this point and the reference point is the preview distance. Then, the maximum value of the curvature of the reference path within the preview distance is taken as the basis for calculating the upper limit of the feasible longitudinal speed. Based on the above formula, the upper limit of the feasible longitudinal speed within the preview distance can be calculated.

[0059] In some embodiments of the present invention, as Figure 5 shown, step S103 includes: S501. Based on the local coordinate system, obtain the position information of the robot at the current moment and the yaw angle at the current moment; S502. Using the robot's current position information and current yaw angle as inputs, and with the maximum curvature and the maximum longitudinal speed as constraint conditions, a fuzzy control prediction strategy is adopted to obtain the predicted position of the robot at the next moment.

[0060] It should be noted that in this embodiment, the fuzzy control prediction strategy is implemented using a motion controller model. Specifically, the motion controller model divides motion control into three steps: prediction, correction, and update. First, the state vector of the robot is obtained through wheel encoders and IMUs, which is used to measure and estimate the pose information of the robot during motion, including the moving distance and the yaw angle. In other embodiments, other control methods can also be used.

[0061] In this embodiment, the smaller value between the upper limit of the feasible longitudinal speed within the preview distance and the artificially set value of the longitudinal speed is taken as the final expected value of the longitudinal speed.

[0062]

[0063] In the formula, is the final expected value of the longitudinal speed, is the artificially set value of the longitudinal speed, is the upper limit of the feasible longitudinal speed within the preview distance.

[0064] The deviation e between the current longitudinal speed and the final expected value of the longitudinal speed can be expressed as:

[0065] Considering that this deviation is a time-varying quantity, a control law can be designed:

[0066] In the formula, K is a coefficient, and the output speed value is further obtained as:

[0067] It should be noted that the control law is a method for regulating the magnitude of the control signal in the motion controller model.

[0068] In a specific embodiment, the wall-climbing robot obtains the rotational speed of the wheels through the Hall encoders carried by the drive wheels, and establishes a kinematic state equation of the robot with the center point of the two wheels as the reference point:

[0069] Among them, the state quantity is: [x, y, θ]T, and the control quantity is: [v, w]T. At this time, there is a coupling relationship between the angle and the position. For the convenience of logical calculation and subsequent result calculation, a first-order Taylor expansion is used for approximate linearization processing at the expected position point:

[0070] Among them

[0071] Obtained by subtracting the above formulas:

[0072] Among them: , , ,

[0073] After linearization, a continuous linear kinematic model is formed. In order to achieve it within the control period, the Euler forward difference method is used to discretize the model:

[0074] Transformed to get:

[0075] Among them, , , both are time-varying matrices. Expanding, we can get:

[0076]

[0077] In order to predict the state relationship and trend between the next moment and the current moment, the current state quantity and the control quantity of the previous moment are combined as the input of the new state quantity to form a prediction equation.

[0078]

[0079] According to the discretized linear expression, a linear homogeneous matrix expression can be obtained:

[0080] Among them, , , m is the number of control quantities, and n is the number of state quantities.

[0081] According to the discretized prediction equation, combined with the prediction time domain and the control time domain, combined with the above matrix expression, we get:

[0082]

[0083]

[0084] According to the recurrence rule, it can be written in vector form as follows:

[0085] Among them, , Np is the prediction horizon. The other parameter matrices are as follows:

[0086] , where NC is the control horizon.

[0087] In some embodiments of the present invention, referring to Figure 6 , after determining the predicted position of the robot at the next moment, it further includes: S601. Obtain the position of the robot at the current moment; S602. Calculate the deviation between the position of the robot at the current moment and the predicted position, where the deviation includes a trajectory deviation angle and a deviation distance; S603. Determine the magnitude of the control signal of the motion controller model according to the deviation; S604. Adjust the motion parameters of the robot according to the control signal.

[0088] It should be noted that the present invention combines real-time feedback, predictive control, and optimization strategies, and can effectively improve the adaptability and accuracy of the robot in complex path tracking tasks.

[0089] In a specific embodiment, updating the predicted position by using a motion controller model includes: Re-decompose the state quantity and the control quantity, and let: , where , n is the number of state quantities, and the coefficient matrix can be updated as follows.

[0090] ,

[0091] Construct a cost function based on the principle of smooth velocity change as follows.

[0092]

[0093] Convert the objective function into the form of quadratic programming, and substitute the prediction equation into the cost function, then the standard quadratic programming form of the matrix equation is:

[0094] Among them, , , , .

[0095] Taking the point on the weld path closest to the wall-climbing robot as the reference point, a preview point is taken on the weld path along the advancing direction of the robot. The arc length of the reference path between this point and the reference point is the preview distance. Then, the maximum value of the curvature of the reference path within the preview distance is used as the basis for calculating the upper limit of the feasible longitudinal speed.

[0096] Then, in front of this reference point, a series of points are selected as the reference path point sequence within this control period. The number of them is at least equal to the prediction steps, and the arc length between every two points in the point sequence is

[0097] The information vector form of the reference path point sequence is:

[0098] where the prediction steps are P and the control steps are C.

[0099] Furthermore, in order to obtain a more accurate weld path, the information of the reference path point sequence is substituted into the cost function to obtain the optimization objective function:

[0100] where Q and R are weight coefficients, represents the i-th state prediction information at time t, represents the i-th future control signal at time t. When it is zero, it is the actual control signal output by the controller in the previous control period. T is the iteration period, is the relaxation factor parameter.

[0101] Then the system constraints are: , ,

[0102] where is the yaw angular velocity.

[0103] In a specific embodiment, such as Figure 7As shown in the figure, when the robotic cleaner reaches the weld position obtained visually, parameters such as the robot speed v, angular velocity ω, and control period T fed back by the motion controller are used to solve and correct the estimated distance deviation between the robot motion trajectory and the weld centerline trajectory, establish the correspondence between the actual robot motion trajectory and the weld centerline trajectory, and estimate the estimated angle θr between the robot and the weld and the deviation distance d of the weld centerline at the next moment, where Δθt is the angle increment within the sampling time. The predicted angle between the robot and the weld at time t + 1 is weighted and corrected by calculating the offset of the angle γ at the previous moment between the image weld trajectory and the actual robot motion trajectory. According to the twice-corrected γ value, the position of the robot in the global coordinate system is normalized to obtain the corrected angle. The angle is corrected through multiple iterations. In each correction step, an error prediction mechanism is added, that is, the current correction strategy is adjusted according to the correction result at the previous moment to reduce the deviation introduced by system delay or sensor error. Finally, the motion controller executes the motion parameter input at the next moment.

[0104] In some embodiments of the invention, in order to obtain the weld trajectory in step S101, the wall-climbing robot is provided with industrial cameras with two viewing angles, namely, a top-mounted industrial camera and a horizontally-mounted industrial camera. Please refer to Figure 8 and the placement positions of the two cameras are related, that is, the position relationship of the viewing coordinate systems has been determined. Among them, the functions of the two cameras for image acquisition are different. The role of the top-mounted industrial camera is to acquire images of the weld in the planar field of view. The camera viewing angle is horizontal relative to the plane where the weld is located, and it acquires and provides the state information of the weld plane. The other horizontally-mounted viewing industrial camera can observe the pre-aiming position point of the target weld in front and acquire and provide the weld image during the laser cleaning process of the weld (for providing the defect characteristics after cleaning).

[0105] Furthermore, the two viewing-angle cameras have a co-view relationship, that is, there is an overlapping part in the viewing field. By adjusting the fill light intensity, the light brightness in the observation environment is maintained at a certain level, thereby avoiding the problem of overexposure of the brightness during the laser cleaning process. By re-feature matching between the key frames of the images of the two industrial cameras and multiple co-view key frames, a new weld position coordinate can be generated by fusion enhancement. Combining the internal and external parameter calibration of the industrial camera, through jointly minimizing the point-line reprojection cost function, higher-precision three-dimensional point-line feature information is obtained, and the three-dimensional coordinates of the weld in space are calculated.

[0106] In a specific embodiment, the host computer preprocesses the collected weld image, including gray-scale transformation, target area extraction, filtering to process image noise, removing longitudinal interference, threshold segmentation, area domain filtering, morphological processing, etc. When the weld shape is irregular or there are discontinuities on the weld surface, first, extract the weld edge line as the initial guiding line, cut the image into several parts in the direction perpendicular to the guiding line, and scan each part of the image one by one starting from the first part to find the upper and lower boundaries of each connected region. Calculate the midpoints of the upper and lower boundaries of each connected region as the reference points. Use these reference points to fit a line as the guiding line for the weld direction center line. Secondly, correct the position of the guiding line based on the gray value histogram, that is, take a region with a certain width centered on the midpoint of each guiding line. Then calculate the gray value histogram of the pixels in this region, and then sort the histogram to find the gray value with the most occurrences, and take the pixel position corresponding to this gray value as the new corrected position of the guiding line. Then, for the possible broken center line in this direction, repeat the above histogram statistics process to correct the broken center line and remove the deviated points. If the distance between the new guiding line and the original guiding line is too far, the correction is abandoned. For the new guiding line, remove the points that are far from the center line to smooth the center line, and filter the points on the boundary to reduce the influence of noise.

[0107] Similarly, for the weld images that appear perpendicular to the original weld direction during the weld scanning process, that is, T-shaped or cross-shaped welds, scan each new weld one by one according to the above idea. Find the left and right boundaries of each connected region, and calculate the midpoint as the guiding line for the new weld direction. Perform histogram statistics on the pixels on the guiding line to find the gray value with the most occurrences as the center line of the new weld.

[0108] At the same time, for the key-frame weld images collected by the flat industrial camera, when the laser cleaner emits laser on the weld surface, extract the edge of the laser spot from the edge information and take its center line as the skeleton to obtain a smooth single-pixel-width skeleton curve. Finally, perform the Hough transformation of the straight line on the skeleton curve to identify the two straight lines on both sides of the arc and mark the two endpoints of the arc. By finding the arc part, detect the position of the weld in the straight-line laser beam and this frame of image. By combining the two position images with visual feedback and using the feature matching algorithm between consecutive frames to ensure the continuity of the weld path, obtain the real-time updated weld center line trajectory. At the same time, the characteristics of the cleaned weld become more obvious and can be used for defect detection.

[0109] In summary, the present invention utilizes the co-visual relationship and feature matching of two industrial cameras, combines internal and external parameter calibration, and obtains high-precision three-dimensional point-line feature information by jointly minimizing the point-line reprojection cost function, thereby obtaining the spatial three-dimensional coordinates of the weld seam. Through the guiding line fitting technology, even in the case of irregular weld seam shapes or discontinuous surfaces, the weld seam edge line can be accurately extracted, improving the stability and accuracy of tracking. Further, based on the design speed adjustment and the prediction of the distance deviation of the tracking weld seam trajectory, the deviation correction with real-time feedback is realized, achieving the precise correspondence between the robot tracking trajectory and the weld seam center line trajectory. Based on the control strategy of prediction and correction, the tracking ability of the robot for the weld seam trajectory is improved, especially in the area where the weld seam curvature changes greatly, in the form of preview points, step by step. Furthermore, the weld seam features after being tracked and laser cleaned by the wall-climbing robot of the present invention become more obvious. On this basis, the present invention can utilize these smooth features to complete the secondary calibration of the route and features, making it easier to perform defect detection and tracking, and improving the monitoring ability of the weld seam trajectory quality.

[0110] Based on the above-mentioned weld seam recognition and tracking control method for a wall-climbing laser cleaning robot, an embodiment of the present invention further provides a weld seam recognition and tracking control device for a wall-climbing laser cleaning robot. Please refer to Figure 9 , including: A trajectory determination module 910, configured to determine the weld seam trajectory; A maximum longitudinal speed determination module 920, configured to determine the maximum curvature of the arc length between the position point of the robot at the current moment and the preview point, and determine the maximum longitudinal speed corresponding to the maximum curvature, where the preview point is any point on the weld seam trajectory along the advancing direction of the robot; A control module 930, configured to determine the predicted position of the robot at the next moment according to the maximum curvature and the maximum longitudinal speed, and control the robot to move to the predicted position.

[0111] As Figure 10 shown, based on the above-mentioned weld seam recognition and tracking control method for a wall-climbing laser cleaning robot, the present invention also correspondingly provides a robot motion controller.

[0112] The robot motion controller at least includes a processor 1010, a memory 1020, and a display 1030. Figure 10 Only some components of the electronic device are shown, but it should be understood that it is not required to implement all the shown components, and more or fewer components can be alternatively implemented.

[0113] The memory 1020 may be an internal storage unit of the electronic device in some embodiments, such as the hard disk or memory of the electronic device. The memory 1020 may also be an external storage electronic device of the electronic device in some other embodiments, such as a plug-in hard disk equipped on the electronic device, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. Further, the memory 1020 may also include both the internal storage unit of the electronic device and the external storage electronic device. The memory 1020 is used to store application software installed on the electronic device and various types of data, such as program codes installed on the electronic device. The memory 1020 may also be used to temporarily store data that has been output or will be output. In one embodiment, a weld recognition and tracking program 1040 for a wall-climbing laser cleaning robot is stored on the memory 1020, and the weld recognition and tracking program 1040 for the wall-climbing laser cleaning robot can be executed by the processor 1010, so as to implement the weld recognition and tracking control method for the wall-climbing laser cleaning robot in various embodiments of the present application.

[0114] The processor 1010 may be a central processing unit (CPU), a microprocessor or other data processing chips in some embodiments, and is used to run the program codes stored in the memory 1020 or process data, such as executing the weld recognition and tracking control method for the wall-climbing laser cleaning robot.

[0115] The display 1030 may be an LED display, a liquid crystal display, a touch liquid crystal display, and an OLED (Organic Light-Emitting Diode) toucher, etc. in some embodiments. The display 1030 is used to display information of the wall-climbing laser cleaning robot weld recognition and tracking electronic device and to display a visual user interface. The components 1010-1030 of the electronic device communicate with each other through a system bus.

[0116] The present invention also correspondingly provides a robot, including a robot body and a robot motion controller, wherein the robot body is equipped with a laser cleaner, and the robot body is controlled by the robot motion controller to identify and track the weld, so as to enable the robot to move along the weld track without slipping on the wall surface.

[0117] Those skilled in the art can understand that all or part of the processes of implementing the methods in the above embodiments can be completed by instructing relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. Among them, the computer-readable storage medium is a disk, an optical disc, a read-only memory or a random access memory, etc.

[0118] As described above, it is only the preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A weld seam recognition and tracking control method for a wall-climbing laser cleaning robot, characterized in that: include: Determine the weld trajectory; Determine the maximum curvature of the arc length between the current position of the robot and the preview point, and determine the maximum longitudinal speed corresponding to the maximum curvature, wherein the preview point is any point on the weld trajectory along the robot's forward direction; The predicted position of the robot at the next moment is determined according to the maximum curvature and the maximum longitudinal speed, and the robot is controlled to move to the predicted position.

2. The weld seam recognition and tracking control method of the wall-climbing laser cleaning robot according to claim 1 is characterized in that: The step of determining the maximum curvature of the arc length between the robot's current position point and the preview point includes: Determine a reference point, where the reference point is a point on the weld track along the robot's forward direction that is closest to the robot; The preview distance is determined according to the arc length of the weld between the reference point and the preview point, and the maximum curvature of the weld trajectory within the preview distance is determined.

3. The weld seam recognition and tracking control method of the wall-climbing laser cleaning robot according to claim 1 is characterized in that: Determining the maximum longitudinal speed corresponding to the maximum curvature includes: Construct a local coordinate system with the robot as the main body; Based on the local coordinate system, the relationship between the maximum curvature of the robot and the longitudinal speed at any time is determined according to the robot kinematic model; wherein the longitudinal speed of the robot in the robot kinematic model is the average value of the left and right wheel speeds, and the yaw angular velocity is proportional to the difference between the left and right wheel speeds; According to the relationship between the maximum turning curvature of the robot and the path curvature of the weld track, and the relationship between the maximum turning curvature of the robot and the longitudinal speed at any moment, the maximum longitudinal speed of the robot at the maximum curvature of the weld track is determined.

4. The weld seam recognition and tracking control method of the wall-climbing laser cleaning robot according to claim 3 is characterized in that: Determining the predicted position of the robot at the next moment according to the maximum curvature and the maximum longitudinal speed includes: Based on the local coordinate system, obtain the current position information and the current yaw angle of the robot; The robot's current position information and the current yaw angle are used as inputs, the maximum curvature and the maximum longitudinal speed are used as constraints, and a fuzzy control prediction strategy is adopted to obtain the predicted position of the robot at the next moment.

5. The weld seam recognition and tracking control method of the wall-climbing laser cleaning robot according to claim 4 is characterized in that: The fuzzy control prediction strategy is implemented using a motion controller model; after determining the predicted position of the robot at the next moment, it also includes: Get the current position of the robot; Calculate the deviation between the robot's current position and the predicted position, wherein the deviation includes a trajectory deviation angle and a deviation distance; Determining the control signal size of the motion controller model according to the deviation; According to the control signal, the robot motion parameters are adjusted.

6. The weld seam recognition and tracking control method of the wall-climbing laser cleaning robot according to claim 5 is characterized in that: After adjusting the robot motion parameters according to the control signal, the method further includes: According to the adjustment result of the robot motion parameters last time, the trajectory deviation angle of the robot at the current moment is weighted to obtain the weighted trajectory deviation angle; The control signal size is determined according to the weighted deviation angle and the deviation distance.

7. The weld seam recognition and tracking control method of a wall-climbing laser cleaning robot according to claim 1, characterized in that: Determining the weld trajectory includes: Acquire a weld plane field image acquired by a first camera device, and acquire a laser cleaning process image acquired by a second camera device, wherein the first camera device has a top view angle relative to the robot, and the second camera device has a horizontal view angle relative to the robot; Matching features of the common-view relationship image key frames of the first camera device and the second camera device, the weld plane field of view image key frames, and the laser cleaning process image key frames to determine an initial weld trajectory image; Extracting an initial centerline of the weld in the initial weld trajectory image; The initial center line is laser cleaned, and the weld trajectory is determined based on the laser cleaning process image and edge information of the laser spot information collected by the second camera device.

8. The weld seam recognition and tracking control method of the wall-climbing laser cleaning robot according to claim 7, characterized in that: The matching of features of the common view relationship image key frame of the first camera device and the second camera device, the weld plane field of view image key frame and the laser cleaning process image key frame to determine the initial weld trajectory image includes: Acquire the first feature point of the key frame of the weld plane field of view image, the second feature point of the key frame of the laser cleaning process image, and the third feature point of the key frame of the common view relationship image, and match the first feature point, the second feature point, and the third feature point according to similarity to obtain matching features; Calibrate the internal and external parameters of the first camera device, and calibrate the internal and external parameters of the second camera device; Based on the calibrated internal parameters and external parameters of the first camera device and the internal parameters and external parameters of the second camera device, a reprojection error function is established according to the matching features; Determine the minimum error projection of the reprojection error function using a preset optimization algorithm; According to the minimum error projection, the initial weld trajectory is determined, and the initial weld trajectory image is determined.

9. The weld seam recognition and tracking control method of a wall-climbing laser cleaning robot according to claim 7, characterized in that: The laser cleaning of the initial center line and determining the weld trajectory according to the laser cleaning process image and edge information of the laser spot information collected by the second camera device include: Laser cleaning is performed on the initial center line to extract edge information of the laser spot in the initial weld trajectory image; Calculating the center line of the edge information of the laser spot to determine the skeleton curve; Performing a preset straight line Hough transform on the skeleton curve, determining straight lines on both sides of the skeleton curve, and determining a weld trajectory.

10. A weld seam recognition and tracking control device for a wall-climbing laser cleaning robot, characterized in that: include: A trajectory determination module, used to determine the weld trajectory; A maximum longitudinal speed determination module is used to determine the maximum curvature of the arc length between the current position point of the robot and the preview point, and determine the maximum longitudinal speed corresponding to the maximum curvature, wherein the preview point is any point on the weld track along the robot's forward direction; The control module is used to determine the predicted position of the robot at the next moment according to the maximum curvature and the maximum longitudinal speed, and control the robot to move to the predicted position.

Images