A trajectory self-planning system and method for a large tank body inspection robot

Abstract

The application discloses a kind of large tank inspection robot trajectory self-planning system and method, belong to the technical field of inspection robot, system includes large tank inspection robot, it includes main body device, slave device, electromagnetic absorption device and connecting device, main body device and slave device are connected by connecting device, electromagnetic absorption device is located at the bottom of main body device and slave device;Further comprising: drive module, it is located at the two sides of main body device and slave device;Image acquisition module, it is located at the front of main body device;Detection module, it is located at the top of slave device rear end;Control module, it is located in main body device;The application can realize two kinds of trajectory self-planning mode, and can be self-adaptive on curved surface magnetic wall surface, can flexibly turn and move, the invention has strong load capacity, high inspection accuracy, good flexibility, high work efficiency;Solve the problem that existing inspection robot control is complex and positioning accuracy is low.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of inspection robots, and particularly relates to a track self-planning system and method of a large tank inspection robot. BACKGROUND

[0002] Large tank containers and pipelines are widely used in industry and are important in the fields of chemical industry, pipeline transportation, smelting, etc. In the long-term use, the tank is inevitably damaged. For example, for the smelting loss weighing tank, large pressure container, and corrosive material container, when the tank is subjected to a large pressure, corrosive liquid and gas, and due to the defect in a certain place not being detected and repaired, a major safety accident such as explosion or leakage of toxic substances is easily caused.

[0003] The development of the tank will be more and more large-scale. Artificial detection not only has a large labor intensity and a poor working environment, but also cannot guarantee the detection accuracy. For narrow or high space, the worker cannot complete the work. The application of nondestructive testing and the cooperation of the inspection robot can make the vertical tank detection intelligent. At present, for the track planning inspection robot, the detection environment needs to be scanned or preset, and then a two-dimensional map is generated for track planning. The control is complex and there is a certain error in the positioning accuracy. In the case of positioning error, the robot cannot continue to navigate, so the track planning should be avoided by positioning. SUMMARY

[0004] The present application aims to provide a track self-planning system and method of a large tank inspection robot to solve the problems in the background.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical scheme:

[0006] A track self-planning system of a large tank inspection robot, the system comprises a large tank inspection robot, and the large tank inspection robot comprises:

[0007] a main body device;

[0008] a slave device;

[0009] an electromagnetic adsorption device, which is installed at the bottom of the main body device and the slave device;

[0010] a connecting device, which is located between the main body device and the slave device and is used for connecting the main body device and the slave device;

[0011] The system further comprises:

[0012] a driving module, which adopts differential steering and is installed on both sides of the main body device and the slave device;

[0013] An image acquisition module is installed at the front of the main device for acquiring weld images;

[0014] A detection module is installed at the top of the rear end of the slave device, carrying an eddy current sensor and an ultrasonic sensor, for detecting weld defects and tank defects;

[0015] A control module is installed inside the main device for receiving, processing information of each module and controlling the movement of the robot.

[0016] As a further aspect of the application: the electromagnetic adsorption device includes a plurality of electromagnetic adsorption units, which are fixedly connected to the main device or the slave device by bolts, the magnet in the electromagnetic adsorption unit adopts an electromagnet, and the electromagnetic adsorption unit includes an attracting coil and a core, which generates a magnetic field after being energized.

[0017] As a further aspect of the application: the connecting device includes a first member, a second member, a passive joint one, a passive joint two and a passive joint three, the second member is connected to the slave device through the passive joint three, the first member and the second member are connected through the passive joint two, the first member is connected to the main device through the passive joint one, the passive joint one can rotate in the x-y plane, the passive joint two can rotate in the x-z plane, and the passive joint three can rotate in the y-z plane.

[0018] As a further aspect of the application: the driving module is a wheeled mobile mechanism, which is internally provided with an encoder, there are eight driving modules in total, four of which are installed on the two sides of the main device and the slave device, the rear two wheels of the main device and the slave device are driving wheels, and the driving modules are connected to the main device and the slave device through passive joints and can rotate in the y-z plane.

[0019] As a further aspect of the application: the image acquisition module includes two linear laser emitters and a ccd camera, which are installed inside the image acquisition module in a straight-shooting and oblique-receiving manner.

[0020] As a further aspect of the application: the control module includes a main control module and a slave control module, the main control module processes information fed back by the sensor and sends a delay signal to the slave control module, the main control module controls the main driving module, and the slave control module controls the slave driving module according to the delay signal.

[0021] Another object of the embodiment of the application is to provide a trajectory self-planning method of a large tank inspection robot.

[0022] S1, system initialization, select trajectory self-planning inspection mode;

[0023] S2, select the inspection mode one, the trajectory self-planning steps are as follows:

[0024] a1, image acquisition, the image acquisition module collects the weld image in front of the robot movement, and transmits the image to the control module;

[0025] a2, image preprocessing, the control module pre-processes the received weld image;

[0026] a3, center line extraction, using eight neighborhood template to detect all pixel points in the pre-processed image, if the detected pixel point is 8simple, and the point is not isolated, if the point is a boundary point, delete the point, and iterate several times to finally obtain the center line;

[0027] a4, feature point extraction, all effective pixel points in the center line image plane are converted to the parameter plane, the straight line parameters in the Hough fitting straight line are determined, and two straight lines can be fitted in the parameter plane according to the point coordinates in the image and the straight line parameters. The intersection of the straight line equations of the two straight lines is the feature point;

[0028] a5, planning motion trajectory, converting the feature point coordinates obtained from multiple images into world coordinate system, and using least square method fitting to realize weld trajectory planning;

[0029] a6, inspection movement, the control module controls the motor movement according to the trajectory, and uses the encoder to position and adjust the robot pose;

[0030] a7, upload defect information, in the process of inspection movement, the eddy current sensor works continuously, when detecting defects, the control module obtains the positioning information through the encoder, and transmits the positioning information to the upper computer;

[0031] S3, select the inspection mode two, the trajectory self-planning steps are as follows:

[0032] a1, specify the starting point coordinate, adjust the angle between the side of the robot and the horizontal plane to zero degree;

[0033] a2, linear motion, the robot moves linearly on the tank, adjusts the pose through the encoder and keeps uniform speed;

[0034] The method for the robot to move linearly around the tank bottom is as follows:

[0035] Calculate the driving wheel movement distance Δd L , Δd R , calculate the pulse number N L , N R of the left and right wheels in a sampling period, and calculate the movement distance of the two wheels in a sampling period;

[0036] Δd L = 2πRN L1 ;

[0037] Δd R = 2πRN R1 ;

[0038] In the formula, R represents the radius of the wheel, Δd L the movement distance, Δd L the movement distance

[0039] The computer calculates the movement distance and the offset angle ΔD and Δθ of the robot,

[0040] ΔD = (Δd L + Δd R ) / 2;

[0041] Δθ = (Δd L - Δd R ) / α;

[0042] In the formula, α represents the distance between the two wheels;

[0043] The distance ΔD traveled by the robot can be known as the position of the robot relative to the initial position;

[0044] a3, static steering, the control module knows through the encoder that the robot has made a round around the bottom of the tank, stops moving, and rotates in place to make the side of the robot and the horizontal plane have an angle of

[0045] The method for the robot to obtain a round around the bottom of the tank is as follows:

[0046] When ΔD = the circumference S of the bottom of the tank, the robot completes a round movement around the bottom of the tank;

[0047] The rotation degree is obtained as follows:

[0048] The rotation angle calculation method is established as follows:

[0049]

[0050] In the formula, P represents the axial distance of two adjacent trajectories in the same plane;

[0051] Input the width L of the robot and the diameter d of the tank body, and the robot rotation angle and the axial distance P of two adjacent trajectories can be calculated;

[0052] a4, spiral movement, the control module knows through the angle sensor that the robot has completed rotation, at this time the side of the robot and the horizontal plane have an angle of Start the spiral movement around the tank body;

[0053] The positioning method for the robot's spiral motion around the tank is as follows:

[0054] The time t for completing one spiral motion and the distance D traveled in one revolution can be calculated using the following equations.

[0055]

[0056] In the formula, D represents the distance the robot travels in one spiral motion, and V represents the speed at which the robot moves at a constant speed.

[0057] The robot's exact position at a given moment is determined by calculating the ratio of its running time T to the time t required to complete one revolution of the spiral.

[0058]

[0059] In the formula, N is an integer representing the quotient of T and t, and M represents the remainder of T and t.

[0060] If there is no remainder when T and t are divided, then the robot is at the position axially distance NP from its initial position.

[0061] If there is a remainder when T and t are divided, substitute the remainder M into the equation. The time t in the equation can be used to determine the position of the robot at a certain point on the circumference β, which is a distance NP+h from its initial position along the axial direction.

[0062] Substitute the remainder M In the equation t, we can determine that the robot is on a straight line a with a lateral distance l from its initial position;

[0063] The robot's exact location is at the intersection of line a and circle β;

[0064] a5. Stationary turning: The control module determines the position by positioning. When the axial distance between the robot and the starting point is equal to the tank height H minus the vehicle length L, the robot stops moving and rotates in place so that the angle between the side of the robot and the horizontal plane is zero.

[0065] a6. Linear motion: The robot adheres to the tank and moves in a straight line for real-time positioning. It stops moving after circling the top of the tank once.

[0066] a7. Upload defect information. During the inspection process, the ultrasonic sensor works continuously. When a defect is detected, the control module transmits the positioning information to the host computer.

[0067] As a further aspect of the present invention: the preprocessing steps of the control module in the first inspection mode for the received weld image include: image grayscale conversion, extracting the RGB components of each point in the image, and performing operations on the three components using the average value method R = G = B = 1 / 3(R + G + B) to achieve image grayscale conversion; image filtering, using a Gaussian normal distribution curve... The image is filtered by a Gaussian filter template established by statistically analyzing the gray values ​​of each pixel in the neighborhood and assigning corresponding weight coefficients. For image segmentation, the original gray image is f(x,y). A value T is selected as the segmentation threshold according to a certain rule. The gray values ​​of each pixel are compared with the value of T to transform the gray image f(x,y) into a binary image g(x,y).

[0068] As a further aspect of the present invention: the feature point coordinates (x, y) in the first inspection mode are converted to the world coordinate system (X, Y, Z), and the transformation relationship S is as follows:

[0069]

[0070] In the formula, M is the intrinsic parameter matrix of the CCD camera, R is the rotation matrix of the CCD camera, and T is the translation vector.

[0071] As a further aspect of the present invention: in inspection mode one and inspection mode two, the posture is adjusted by an encoder, and the speed v of the left and right wheels is measured by a wheel-type encoder. r and v l Based on the differential motion model, the motion state model is derived as follows:

[0072]

[0073] Integrating the robot's motion state variables yields the robot's pose information model:

[0074]

[0075] In the formula, θ is the heading angle, ω is the angular velocity of the rotation axis, l is the width between the left and right wheels, and v r v is the rotational speed of the right wheel. l This represents the rotational speed of the left wheel.

[0076] As a further aspect of the present invention: In the second inspection mode, the method for adjusting the robot's angle while stationary is as follows: The error between the robot's angle value and the commanded angle value is obtained through an angle sensor. The angle error value and the rate of change of the angle error value are used as inputs, and the robot's steering is controlled by a fuzzy PID controller. The calculation process is as follows:

[0077]

[0078] In the formula, u(t) represents the output control quantity, and θ(t) represents the angle error value.

[0079] Compared with the prior art, the beneficial effects of the present invention are: the present invention enables the robot to work flexibly on curved surfaces of different radii, the eight drive mechanisms can achieve rotation at different angles, and can achieve self-adaptation of curved surfaces in spiral upward motion. Compared with single inspection robots, it can adapt to a wider range of curved surface radii, has a greater adsorption capacity, stronger adaptability, better flexibility, stronger load capacity, and longer battery life.

[0080] This invention provides a self-planning trajectory method for a tank inspection robot, comprising two inspection modes. One mode utilizes a laser sensor to detect weld seams, generates a motion trajectory through a control module, and completes weld seam tracking inspection, using encoder mileage for positioning. The other mode involves spiral motion inspection of the tank, using encoder feedback to the control module of the robot's movement time, distance, and initial position for positioning. By inputting the tank's diameter and the robot's width, the control module can calculate the robot's rotation angle at the start of the spiral motion, completing the inspection and positioning work. Compared to methods that require scanning or pre-setting the inspection environment and then generating a two-dimensional map for trajectory planning, this invention eliminates the need for map pre-setting. Furthermore, by utilizing the encoder feedback time and distance combined with the invention's calculation formula, positioning accuracy is improved, preventing positioning issues that could hinder robot navigation, reducing workload, lowering control difficulty, and increasing work efficiency. Attached Figure Description

[0081] Figure 1 This is a three-dimensional structural diagram of a large tank inspection robot according to the present invention.

[0082] Figure 2 This is a bottom view of a large tank inspection robot according to the present invention.

[0083] Figure 3 This is a flowchart of a trajectory self-planning method for a large tank inspection robot according to the present invention.

[0084] Figure 4 This is a flowchart of a trajectory self-planning method for a large tank inspection robot according to the present invention.

[0085] Figure 5 The flowchart of the trajectory self-planning method of a large tank inspection robot of the present invention is shown in the second mode.

[0086] Figure 6 A schematic diagram of the spiral trajectory positioning of a large tank inspection robot according to the present invention.

[0087] Figure reference numerals: 1-Main body device; 2-Slave body device; 3-Detection module; 4-Passive joint three; 5-Second component; 6-Connecting device; 7-Passive joint two; 8-First component; 9-Passive joint one; 10-Drive module; 11-Acquisition module; 12-Passive joint; 13-Electromagnetic adsorption device. Detailed Implementation

[0088] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention.

[0089] In the description of this invention, it should be understood that the terms "bottom", "front", "top", "rear", "middle", "inner", etc., which indicate orientation or positional relationship, are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the components or elements referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as limiting this invention.

[0090] Example 1.

[0091] like Figure 1 and 2 The diagram illustrates the trajectory self-planning system of the large tank inspection robot of this embodiment. The system includes a large tank inspection robot comprising: a main body device 1, a slave device 2, an electromagnetic adsorption device 13, and a connecting device 6. The system also includes: a drive module 10, an image acquisition module 11, a detection module 3, and a control module (not shown in the diagram). The electromagnetic adsorption device 13 is installed at the bottom of the main body device 1 and the slave device 2. The connecting device 6 is located between the main body device 1 and the slave device 2, connecting them. The drive module 10 employs differential steering and is installed on both sides of the main body device 1 and the slave device 2. The image acquisition module 11 is installed at the front of the main body device 1, for acquiring weld images. The detection module 3 is installed at the top rear end of the slave device 2, equipped with an eddy current sensor and an ultrasonic sensor, for detecting weld defects and tank defects. The control module is installed inside the main body device 1, for receiving and processing information from each module and controlling the robot's movement.

[0092] The electromagnetic adsorption device 13 includes several electromagnetic adsorption units. The electromagnetic adsorption units are fixedly connected to the main body device 1 or the slave device 2 by bolts. The magnet in the electromagnetic adsorption unit is an electromagnet. The electromagnetic adsorption unit includes an attraction coil and an iron core. When energized, the attraction coil generates a magnetic field, so that the robot will not slip or fall when moving on the tank.

[0093] The connecting device 6 includes a first component 8, a second component 5, a passive joint 1 9, a passive joint 2 7, and a passive joint 3 4. The second component 5 is connected to the slave frame 2 through the passive joint 3 4. The first component 8 and the second component 5 are connected through the passive joint 2. The first component 8 is connected to the main device 1 through the passive joint 1. The passive joint 1 can rotate in the xy plane, the passive joint 2 7 can rotate in the xz plane, and the passive joint 3 4 can rotate in the yz plane, so that the robot can rotate freely in the xy plane, the xz plane, and the yz plane, can adapt to a certain range of curved tanks, and can turn freely.

[0094] The drive module 10 is a wheeled movement mechanism with an encoder inside. There are eight drive modules 10 in total, with four installed on each side of the main body device 1 and the slave device 2. The two rear wheels of the main body device 1 and the two rear wheels of the slave device 2 are drive wheels. The drive module is connected to the main body device and the slave device through the passive joint 12 and can rotate in the yz plane. This allows the robot to better adapt to curved surfaces when the radii of the tanks it contacts on both sides are different or when it is in spiral motion, ensuring the fit with the curved surface and improving the robot's adsorption capacity and friction.

[0095] The image acquisition module 11 includes two linear laser emitters and a CCD camera, which are installed inside the image acquisition module 11 using a direct-to-oblique receiving method. The image information acquired by the image acquisition module is used to fit a motion trajectory in the control module.

[0096] The detection module 3 can be equipped with eddy current sensors and ultrasonic sensors. During the robot's inspection process, the sensors are always working and will feed back to the control module when a defect is detected. The control module will then transmit the location information to the host computer.

[0097] The control module includes a main control module and a slave control module. The main control module processes information from sensors and encoders and transmits positioning information to the host computer. During linear motion, the main control module and the slave control module synchronously control the main drive module and the slave drive module, respectively. When turning, the main control module controls the main drive module to complete the turn first, and then sends the turn signal to the slave control module, thereby controlling the slave drive module to turn.

[0098] Example 2

[0099] like Figures 3-5 The diagram illustrates the trajectory self-planning method for the large tank inspection robot in this embodiment. The specific steps are as follows:

[0100] S1. System initialization selects trajectory self-planning inspection mode;

[0101] S2. Select Inspection Mode 1. The trajectory self-planning steps are as follows:

[0102] a1. Image acquisition: The image acquisition module acquires images of the weld seam in front of the robot's movement and transmits the images to the control module;

[0103] a2. Image preprocessing: The control module preprocesses the received weld image. The image preprocessing steps include: image grayscale conversion, extracting the RGB components of each point in the image, and using the averaging method R = G = B = 1 / 3(R + G + B) to operate on the three components to achieve image grayscale conversion; image filtering, using a Gaussian normal distribution curve... The image is filtered by a Gaussian filter template established by statistically analyzing the gray values ​​of each pixel in the neighborhood and assigning corresponding weight coefficients. For image segmentation, the original gray image is f(x,y). A value T is selected as the segmentation threshold according to a certain rule. The gray values ​​of each pixel are compared with the value of T to transform the gray image f(x,y) into a binary image g(x,y).

[0104] a3. Centerline extraction: Use an eight-neighbor template to detect all pixels in the preprocessed image. If the detected pixel is an 8-simple and not isolated, or if it is a boundary point, delete the point. Repeat this process multiple times to finally obtain the centerline.

[0105] a4. Feature point extraction: All effective pixels in the centerline image plane are transformed to the parameter plane. The line parameters in the Hough fitting line are determined. Based on the coordinates of the points in the image, two lines can be fitted in the parameter plane by combining the line parameters. The intersection of the two lines can be obtained through the line equations of the two lines, which are the feature points.

[0106] a5. Plan the motion trajectory and convert the feature point coordinates obtained from multiple frames of images into the world coordinate system. In the first inspection mode, the feature point coordinates (x, y) are converted into the world coordinate system (X, Y, Z). The transformation relationship S is as follows:

[0107]

[0108] In the formula, M is the intrinsic parameter matrix of the CCD camera, R is the rotation matrix of the CCD camera, and T is the translation vector.

[0109] We then used the least squares method to fit and plan the weld trajectory.

[0110] a6. Inspection Movement: The control module controls the motor movement according to the trajectory, and uses encoders to position and adjust the robot's posture. Wheel encoders measure the speeds v of the left and right wheels. r and v l Based on the differential motion model, the motion state model is derived as follows:

[0111]

[0112] Integrating the robot's motion state variables yields the robot's pose information model:

[0113]

[0114] In the formula, θ is the heading angle, ω is the angular velocity of the rotation axis, l is the width between the left and right wheels, and v r v is the rotational speed of the right wheel. l This refers to the rotational speed of the left wheel;

[0115] a7. Upload defect information. During the inspection process, the eddy current sensor works continuously. When a defect is detected, the control module obtains the positioning information through the encoder and transmits the positioning information to the host computer.

[0116] S3. Select inspection mode two. The trajectory self-planning steps are as follows:

[0117] a1. Specify the starting point coordinates and adjust the angle between the robot's side and the horizontal plane to zero degrees;

[0118] a2. Linear motion: The robot adheres to the tank and moves in a straight line, adjusting its posture and maintaining a constant speed through an encoder.

[0119] The method for positioning a robot in a straight line around the bottom of a can is as follows:

[0120] Calculate the distance Δd traveled by the drive wheel L , Δd R Calculate the number of pulses N between the left and right wheels within a sampling period. L N R And use this to calculate the distance traveled by the two wheels within one sampling period;

[0121] Δd L =2πRN L1 ;

[0122] Δd R =2πRN R1 ;

[0123] In the formula, R represents the wheel radius, and Δd L Distance of movement, Δd L distance of movement

[0124] Calculate the robot's movement distance and offset angles ΔD and Δθ.

[0125] ΔD=(Δd L +Δd R ) / 2;

[0126] Δθ=(Δd L -Δd R ) / α;

[0127] In the formula, α represents the distance between the two wheels;

[0128] The robot's position relative to its initial position can be determined by the distance ΔD that the robot has traveled.

[0129] a3. Stationary Turning: The control module, upon detecting via the encoder that the robot has completed one revolution around the bottom of the tank, stops moving and rotates in place until the robot's side forms an angle with the horizontal plane.

[0130] The robot obtains the signal by circling the bottom of the can as follows:

[0131] When ΔD = the circumference S of the tank bottom, the robot completes one revolution around the tank bottom;

[0132] Obtain rotation The method is as follows:

[0133] Establish a method for calculating rotation angles:

[0134]

[0135] In the formula, P represents the axial distance between two adjacent trajectories in the same plane;

[0136] By inputting the robot's width L and the tank's diameter d, the robot's rotation angle can be calculated. And the axial distance P between two adjacent trajectories;

[0137] a4. Spiral motion: The control module detects that the robot has completed rotation via an angle sensor. At this point, the robot's side forms an angle with the horizontal plane. It begins spiral motion around the tank;

[0138] like Figure 6 As shown, the positioning method for the robot's spiral motion around the tank is as follows:

[0139] The time t for completing one spiral motion and the distance D traveled in one revolution can be calculated using the following equations.

[0140]

[0141] In the formula, D represents the distance the robot travels in one spiral motion, and V represents the speed at which the robot moves at a constant speed.

[0142] The robot's exact position at a given moment is determined by calculating the ratio of its running time T to the time t required to complete one revolution of the spiral.

[0143]

[0144] In the formula, N is an integer representing the quotient of T and t, and M represents the remainder of T and t.

[0145] If there is no remainder when T and t are divided, then the robot is at the position axially distance NP from its initial position.

[0146] If there is a remainder when T and t are divided, substitute the remainder M into the equation. The time t in the equation can be used to determine the position of the robot at a certain point on the circumference β, which is a distance NP+h from its initial position along the axial direction.

[0147] Substitute the remainder M In the equation t, we can determine that the robot is on a straight line a with a lateral distance l from its initial position;

[0148] The robot's exact location is at the intersection of line a and circle β;

[0149] a5. Stationary turning: The control module determines the position by positioning. When the axial distance between the robot and the starting point is equal to the tank height H minus the vehicle length L, the robot stops moving and rotates in place so that the angle between the side of the robot and the horizontal plane is zero.

[0150] a6. Linear motion: The robot adheres to the tank and moves in a straight line for real-time positioning. It stops moving after circling the top of the tank once.

[0151] a7. Upload defect information. During the inspection process, the ultrasonic sensor works continuously. When a defect is detected, the control module transmits the positioning information to the host computer.

[0152] The method for adjusting the robot's angle while stationary is as follows: The error between the robot's angle value and the commanded angle value is obtained through an angle sensor. The angle error value and its rate of change are used as inputs, and fuzzy PID control is employed to regulate the robot's steering. The calculation process is as follows:

[0153]

[0154] In the formula, u(t) represents the output control quantity, and θ(t) represents the angle error value.

[0155] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0156] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A trajectory self-planning system for a large tank inspection robot, characterized in that, The system includes a large tank inspection robot, which comprises: Main unit; Slave device; An electromagnetic adsorption device is installed at the bottom of the main body device and the slave device; A connecting device, located between the main body device and the slave device, for connecting the main body device and the slave device; The system also includes: The drive module, which employs differential steering, is installed on both sides of the main body and the slave body. An image acquisition module is installed at the front of the main body device and is used to acquire weld images; The detection module is installed at the top rear end of the slave device and is equipped with an eddy current sensor and an ultrasonic sensor for detecting weld defects and tank defects. The control module, installed inside the main unit, receives and processes information from various modules and controls the robot's movement. The trajectory self-planning method of the large tank inspection robot's trajectory self-planning system includes the following steps: S1. System initialization selects trajectory self-planning inspection mode; S2. Select Inspection Mode 1. The trajectory self-planning steps are as follows: a1. Image acquisition: Acquire images of the weld seam in front of the robot's movement and transmit the images to the control module; a2. Image preprocessing: The control module preprocesses the received weld images. a3. Centerline extraction: Use an eight-neighbor template to detect all pixels in the preprocessed image. If the detected pixel is an 8-simple and not isolated, or if it is a boundary point, delete the point. Repeat this process multiple times to finally obtain the centerline. a4. Feature point extraction: Transform all effective pixels in the centerline image plane to the parameter plane, determine the line parameters in the Hough fitting line, and fit two lines in the parameter plane based on the point coordinates in the image and the line parameters. The intersection of the two lines is obtained through the line equations of the two lines, which are the feature points. a5. Plan the motion trajectory, convert the feature point coordinates obtained from multiple frames of images into the world coordinate system, and use the least squares method to fit the weld seam trajectory. a6. Inspection movement: The control module controls the motor movement according to the trajectory and uses the encoder to position and adjust the robot's posture. a7. Upload defect information. During the inspection process, the eddy current sensor works continuously. When a defect is detected, the control module obtains the positioning information through the encoder and transmits the positioning information to the host computer. S3. Select inspection mode two. The trajectory self-planning steps are as follows: a1. Specify the starting point coordinates and adjust the angle between the robot's side and the horizontal plane to zero degrees; a2. Linear motion: The robot adheres to the tank and moves in a straight line, adjusting its posture and maintaining a constant speed through an encoder. The method for positioning a robot in a straight line around the bottom of a can is as follows: Calculate the distance Δd traveled by the drive wheel L , Δd R Calculate the number of pulses N between the left and right wheels within a sampling period. L N R And use this to calculate the distance traveled by the two wheels within one sampling period; Δd L =2πRN L1 Δd R =2πRN R1 In the formula, R represents the wheel radius, and Δd L Distance of movement, Δd L distance of movement Calculate the robot's movement distance and offset angles ΔD and Δθ. ΔD=(Δd L +Δd R ) / 2 Δθ=(Δd L -Δd R ) / a In the formula, α represents the distance between the two wheels; The robot's position relative to its initial position can be determined by the distance ΔD that the robot has traveled. a3. Stationary Turning: The control module, upon detecting via the encoder that the robot has completed one revolution around the bottom of the tank, stops moving and rotates in place until the robot's side forms an angle with the horizontal plane. The robot obtains the signal by circling the bottom of the can as follows: When ΔD = the circumference S of the tank bottom, the robot completes one revolution around the tank bottom; Obtain rotation The method is as follows: Establish a method for calculating rotation angles: In the formula, P represents the axial distance between two adjacent trajectories in the same plane; By inputting the robot's width L and the tank's diameter d, the robot's rotation angle can be calculated. And the axial distance P between two adjacent trajectories; a4. Spiral motion: The control module detects that the robot has completed rotation via an angle sensor. At this point, the robot's side forms an angle with the horizontal plane. It begins spiral motion around the tank; The positioning method for the robot's spiral motion around the tank is as follows: The time t for completing one spiral motion and the distance D traveled in one revolution can be calculated using the following equations. In the formula, D represents the distance the robot travels in one spiral motion, and V represents the speed at which the robot moves at a constant speed. The robot's exact position at a given moment is determined by calculating the ratio of its running time T to the time t required to complete one revolution of the spiral. In the formula, N is an integer representing the quotient of T and t, and M represents the remainder of T and t. If there is no remainder when T and t are divided, then the robot is at the position axially distance NP from its initial position. If there is a remainder when T and t are divided, substitute the remainder M into the equation. In time t, we obtain the position of the robot at a certain point on the circumference β, which is a distance NP+h from the initial position along the axial direction. Substitute the remainder M In the equation t, we obtain the position of the robot on a straight line a with a lateral distance l from its initial position; The robot's exact location is at the intersection of line a and circle β; a5. Stationary turning: The control module determines the position by positioning. When the axial distance between the robot and the starting point is equal to the tank height H minus the vehicle length L, the robot stops moving and rotates in place so that the angle between the side of the robot and the horizontal plane is zero. a6. Linear motion: The robot adheres to the tank and moves in a straight line for real-time positioning. It stops moving after circling the top of the tank once. a7. Upload defect information. During the inspection process, the ultrasonic sensor works continuously. When a defect is detected, the control module transmits the positioning information to the host computer.

2. The trajectory self-planning system for a large tank inspection robot according to claim 1, characterized in that, The electromagnetic adsorption device includes several electromagnetic adsorption units, which are fixedly connected to the main body or the slave body by bolts. The magnet in the electromagnetic adsorption unit is an electromagnet, and the electromagnetic adsorption unit includes an attraction coil and an iron core. When energized, the attraction coil generates a magnetic field.

3. The trajectory self-planning system for a large tank inspection robot according to claim 1, characterized in that, The connecting device includes a first component, a second component, a passive joint one, a passive joint two, and a passive joint three. The second component is connected to the slave device through the passive joint three. The first component and the second component are connected through the passive joint two. The first component is connected to the main device through the passive joint one. The passive joint one rotates in the xy plane, the passive joint two rotates in the xz plane, and the passive joint three rotates in the yz plane.

4. The trajectory self-planning system for a large tank inspection robot according to claim 1, characterized in that, The drive module is a wheeled moving mechanism with an encoder inside. There are a total of eight drive modules, with four installed on each side of the main body and the slave body. The two rear wheels of the main body and the two rear wheels of the slave body are drive wheels. The drive modules are connected to the main body and the slave body through passive joints and rotate in the yz plane.

5. The trajectory self-planning system for a large tank inspection robot according to claim 1, characterized in that, The image acquisition module includes two linear laser emitters and a CCD camera, which are installed inside the image acquisition module using a direct-to-oblique receiving method.

6. The trajectory self-planning system for a large tank inspection robot according to claim 1, characterized in that, The control module includes a main control module and a slave control module. The main control module processes the information fed back by the sensor and sends a delay signal to the slave control module. The main control module controls the main drive module, and the slave control module controls the slave drive module according to the delay signal.

7. The trajectory self-planning system for a large tank inspection robot according to claim 1, characterized in that, The trajectory self-planning method of the trajectory self-planning system of the large tank inspection robot, in inspection mode one, involves the control module performing preprocessing on the received weld seam image, including: image grayscale conversion, extracting the RGB components of each point in the image, and performing image grayscale conversion by averaging the three components using the method R = G = B = 1 / 3(R + G + B); image filtering, using a Gaussian normal distribution curve... The image is filtered by a Gaussian filter template established by statistically analyzing the gray values ​​of each pixel in the neighborhood and assigning corresponding weight coefficients. For image segmentation, the original gray image is f(x,y). A value T is selected as the segmentation threshold according to a certain rule. The gray values ​​of each pixel are compared with the value of T to transform the gray image f(x,y) into a binary image g(x,y).

8. The trajectory self-planning system for a large tank inspection robot according to claim 1, characterized in that, In the inspection mode one of the trajectory self-planning method of the trajectory self-planning system of the large tank inspection robot, the feature point coordinates (x, y) are converted to the world coordinate system (X, Y, Z), and the transformation relationship S is as follows: In the formula, M is the intrinsic parameter matrix of the CCD camera, R is the rotation matrix of the CCD camera, and T is the translation vector.

9. The trajectory self-planning system for a large tank inspection robot according to claim 1, characterized in that, In inspection modes one and two of the trajectory self-planning method of the trajectory self-planning system of the large tank inspection robot, the posture is adjusted by an encoder, and the speed v of the left and right wheels is measured by a wheel encoder. r and v l Based on the differential motion model, the motion state model is derived as follows: Integrating the robot's motion state variables yields the robot's pose information model: In the formula, θ is the heading angle, ω is the angular velocity of the rotation axis, l is the width between the left and right wheels, and v r v is the rotational speed of the right wheel. l This represents the rotational speed of the left wheel.

10. The trajectory self-planning system for a large tank inspection robot according to claim 1, characterized in that, In the inspection mode two of the trajectory self-planning method of the trajectory self-planning system of the large tank inspection robot, the method for adjusting the robot's angle while stationary is as follows: The error between the robot's angle value and the commanded angle value is obtained through an angle sensor. The angle error value and the rate of change of the angle error value are used as inputs, and the robot's turning is controlled by fuzzy PID control. The calculation process is as follows: In the formula, u(t) represents the output control quantity, and θ(t) represents the angle error value.

Images