Multi-degree-of-freedom dredging and winching operation track planning and control method

Abstract

The present application relates to a multi-degree-of-freedom dredging and obstacle removal operation trajectory planning and control method, and belongs to the technical field of dredging and obstacle removal operation, comprising: dividing an underwater image collected by a multi-beam image sonar into a plurality of sub-images, performing two-dimensional Fourier transform on each sub-image to obtain a frequency domain signal after Fourier transform and denoising to obtain a denoised underwater image; performing color compensation on the denoised underwater image to obtain a compensated underwater image and training a deep learning model to obtain a silt recognition model; and using the silt recognition model to detect silt in water at a target dredging position to complete a dredging operation. The present application trains a deep learning model using a denoised underwater image, constructs a silt recognition model for silt characteristics, and can adaptively extract deep features of silt in multiple dimensions such as texture, color and shape compared with traditional image processing methods based on threshold segmentation, edge detection and the like, so as to realize high-precision detection and segmentation of silt regions.

Description

Technical Field

[0001] This invention relates to the field of dredging and obstacle removal operation technology, specifically to a method for planning and controlling the trajectory of multi-degree-of-freedom dredging and obstacle removal operations. Background Technology

[0002] Rivers, lakes, reservoirs, and urban drainage networks are prone to blockages due to siltation, floating debris accumulation, landslides, and various solid foreign objects during long-term operation. This can lead to reduced cross-sectional area, decreased flood control capacity, and even problems such as pipe overflows, backflow, and black and odorous water bodies. To restore the normal function of hydraulic structures and drainage systems, regular or irregular dredging and clearing operations are necessary. Traditional dredging and clearing methods mainly rely on manual operations in wells or underwater, or the use of large machinery such as excavators and grab boats. These methods suffer from harsh working environments, high labor intensity, low operational precision, low efficiency, and significant personnel safety risks, making them unsuitable for the current needs of refined and routine water environment management and facility operation and maintenance.

[0003] With the development of robotics and automatic control technologies, various dredging and obstacle removal equipment suitable for underwater, submerged, or confined pipeline environments has emerged. Examples include pipeline maintenance robots, underwater operation robots, and dredging systems using tracked or wheeled mobile platforms equipped with robotic arms. These devices are typically equipped with multi-degree-of-freedom robotic arms or multi-joint mechanisms, using end effectors such as gripping tools, cutting tools, suction tools, and high-pressure water jets to break up, grasp, clear, and transport silt or obstacles. Compared to traditional methods, robots and multi-degree-of-freedom mechanisms can operate in narrow, complex, and hazardous environments, offering advantages such as high automation, improved personnel safety, and longer continuous operation time.

[0004] However, in existing technologies, the trajectory planning and control methods for multi-degree-of-freedom dredging and obstacle removal equipment are still relatively crude, often adopting general planning and control strategies from the fields of industrial robotic arms or general mobile robots, which are difficult to fully adapt to the complexity of underwater or submerged dredging conditions. On the one hand, existing systems mostly adopt operation path planning based on human experience or simple geometric rules, such as sequential scanning along the pipeline axis or regular grid, or linear reciprocating scraping, lacking comprehensive modeling and intelligent planning of environmental information such as siltation morphology, obstacle distribution, spatial constraints, and water flow disturbances. This results in redundant operation paths, insufficient coverage, or repeated cleaning of local areas, affecting overall operation efficiency. Summary of the Invention

[0005] To address the aforementioned problems, the purpose of this invention is to provide a method for planning and controlling the trajectory of multi-degree-of-freedom dredging and obstacle removal operations.

[0006] Multi-degree-of-freedom dredging and obstacle removal operation trajectory planning and control methods, including:

[0007] Step 1: Use multibeam imaging sonar to acquire underwater images of the dredging location;

[0008] Step 2: Divide the underwater image into multiple sub-images, and perform a two-dimensional Fourier transform on each sub-image to obtain the frequency domain signal after Fourier transform;

[0009] Step 3: Construct a corresponding denoising threshold for each sub-image and remove noise from the frequency domain signal to obtain the denoised frequency domain signal;

[0010] Step 4: Reconstruct the denoised frequency domain signal to obtain the denoised underwater image;

[0011] Step 5: Perform color compensation on the denoised underwater image to obtain the compensated underwater image;

[0012] Step 6: Train a deep learning model using the compensated underwater images to obtain a silt recognition model;

[0013] Step 7: Use a silt identification model to detect silt in the water at the target dredging location;

[0014] Step 8: When the dredging robot approaches the silt, use the preset automatic operation instructions to complete the dredging operation.

[0015] Preferably, in step 2, the formula is used:

[0016]

[0017] A two-dimensional Fourier transform is performed on each sub-image to obtain the frequency domain signal after the Fourier transform; where, Indicates sub-image in Pixel value at that location, Indicates the length of the sub-image. Indicates the width of the sub-image. This represents the frequency domain signal after Fourier transform. Represents frequency coordinates.

[0018] Preferably, step 3: constructing a corresponding denoising threshold based on each sub-image and removing noise from the frequency domain signal to obtain the denoised frequency domain signal includes:

[0019] Step 3.1: Obtain the Fourier spectrum from the frequency domain signal after Fourier transform;

[0020] Step 3.2: Perform a two-dimensional discrete sine transform on the Fourier spectrum to obtain the transform coefficients; the two-dimensional discrete sine transform process is as follows:

[0021] ;

[0022] in, Represents the transformation coefficients. Represents frequency coordinates. Indicates the Fourier spectrum in The range at the point;

[0023] Step 3.3: Determine the denoising threshold for each sub-image using the transform coefficients;

[0024] Step 3.4: Set the transform coefficients that are less than the denoising threshold to 0, and construct the denoised frequency domain signal based on the remaining transform coefficients.

[0025] Preferably, step 3.3 includes:

[0026] The transform coefficients are sorted in ascending order to form an ascending sequence, and a denoising threshold is calculated based on the ascending sequence; wherein the formula for calculating the denoising threshold is:

[0027]

[0028]

[0029] in, Indicates the noise reduction threshold. This represents the k-th value in the ascending sequence. This represents the k-th value in the ascending sequence. Represents the sorting function. This represents the k-th value in the ascending sequence. This indicates the number of elements in the ascending sequence. This represents the mean of all transformation coefficients.

[0030] Preferably, step 5: performing color compensation on the denoised underwater image to obtain a compensated underwater image, includes:

[0031] Step 5.1: Map the denoised underwater image to a pseudo-color image and calculate the mean of each color channel;

[0032] Step 5.2: Sort the mean values ​​of each color channel and filter out the maximum, minimum, and median values;

[0033] Step 5.3: Calculate the weighting coefficients using the maximum, minimum, and median values;

[0034] Step 5.4: Use weighting coefficients to compensate the pixel values ​​of the target color channel to obtain the compensated underwater image.

[0035] Preferably, in step 5.4, weighting coefficients are used to compensate the color channels corresponding to the median and the minimum values ​​to obtain the compensated underwater image; wherein the compensation process is as follows:

[0036]

[0037] in, This represents the first weighting coefficient. This represents the second weighting coefficient. This indicates the color channel corresponding to the median value. This represents the compensation value for the color channel corresponding to the minimum value. This represents the compensation value for the color channel corresponding to the maximum value. This represents the minimum value. This represents the median.

[0038] Preferably, step 6: training a deep learning model using the compensated underwater image to obtain a silt recognition model includes:

[0039] The compensated underwater image is input into the YOLOv5 network for training to obtain a silt recognition model; wherein, during the training of the YOLOv5 network, the following formula is used:

[0040] )

[0041] )

[0042]

[0043]

[0044]

[0045] The model parameters of the YOLOv5 network are updated; among them, Indicates the current time step. Indicates the previous time step, For time steps First-order moment estimation at time, For time steps Second-order moment estimation at time, For gradient, As the first hyperparameter, This is the second hyperparameter. For learning rate, It is a constant. For time steps First-order moment estimation at time, For time steps Second-order moment estimation at time, For regularization parameters, This is the bias correction value for the first moment estimate at time step t. For time step Model parameters at time, For time step Model parameters at that time.

[0046] This invention also provides a multi-degree-of-freedom dredging and obstacle removal robotic arm trajectory planning and control system, comprising:

[0047] The data acquisition module is used to acquire underwater images of the dredging location using multibeam imaging sonar;

[0048] The frequency domain transformation module is used to divide the underwater image into multiple sub-images and perform a two-dimensional Fourier transform on each sub-image to obtain the frequency domain signal after Fourier transform.

[0049] The denoising module is used to construct a corresponding denoising threshold based on each sub-image and remove noise from the frequency domain signal to obtain the denoised frequency domain signal.

[0050] The reconstruction module is used to reconstruct the denoised frequency domain signal to obtain the denoised underwater image;

[0051] The color compensation module is used to perform color compensation on the denoised underwater image to obtain the compensated underwater image.

[0052] The training module is used to train a deep learning model using compensated underwater images to obtain a silt recognition model;

[0053] The silt detection module is used to detect silt in the water at the target dredging location using a silt identification model;

[0054] The control module is used to complete the dredging operation using preset automatic operation instructions when the dredging robot approaches the silt.

[0055] The present invention also provides an electronic device, including a bus, a transceiver, a memory, a processor, and a computer program stored in the memory and executable on the processor. The transceiver, the memory, and the processor are connected via the bus. The computer program, when executed by the processor, implements the steps in the above-described multi-degree-of-freedom dredging and obstacle removal operation trajectory planning and control method.

[0056] The present invention also provides a computer-readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it implements the steps in the above-described multi-degree-of-freedom dredging and obstacle removal operation trajectory planning and control method.

[0057] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0058] This invention relates to a method for trajectory planning and control of multi-degree-of-freedom dredging and obstacle removal operations. Compared with existing technologies, this invention utilizes denoised underwater images to train a deep learning model, constructing a silt recognition model tailored to silt characteristics. Compared to traditional image processing methods based on threshold segmentation and edge detection, this model can adaptively extract deep features of silt in multiple dimensions such as texture, color, and shape, achieving high-precision detection and segmentation of silt areas. This silt recognition model maintains high accuracy and recall even under complex backgrounds, low contrast, and partial occlusion conditions, thereby improving the reliability of silt detection at target dredging locations.

[0059] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0060] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0061] Figure 1 Flowchart of the multi-degree-of-freedom dredging and obstacle removal operation trajectory planning and control method provided by the present invention;

[0062] Figure 2 This is a diagram illustrating the YOLOv5 network training process provided by the present invention.

[0063] Figure 3 The diagram shows the multi-degree-of-freedom dredging and obstacle removal operation interface provided by this invention. Detailed Implementation

[0064] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," and "counterclockwise," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0065] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0066] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0067] Please see Figure 1 Multi-degree-of-freedom dredging and obstacle removal operation trajectory planning and control methods, including:

[0068] Step 1: Use multibeam imaging sonar to acquire underwater images of the dredging location;

[0069] Step 2: Divide the underwater image into multiple sub-images, and perform a two-dimensional Fourier transform on each sub-image to obtain the frequency domain signal after Fourier transform;

[0070] In step 2, the formula is used:

[0071]

[0072] A two-dimensional Fourier transform is performed on each sub-image to obtain the frequency domain signal after the Fourier transform; where, Indicates sub-image in Pixel value at that location, Indicates the length of the sub-image. Indicates the width of the sub-image. This represents the frequency domain signal after Fourier transform. Represents frequency coordinates.

[0073] Step 3: Construct a corresponding denoising threshold for each sub-image and remove noise from the frequency domain signal to obtain the denoised frequency domain signal;

[0074] Furthermore, step 3 includes:

[0075] Step 3.1: Obtain the Fourier spectrum from the frequency domain signal after Fourier transform;

[0076] Step 3.2: Perform a two-dimensional discrete sine transform on the Fourier spectrum to obtain the transform coefficients; the two-dimensional discrete sine transform process is as follows:

[0077] ;

[0078] in, Represents the transformation coefficients. Represents frequency coordinates. Indicates the Fourier spectrum in The range at the point;

[0079] Step 3.3: Determine the denoising threshold for each sub-image using the transform coefficients;

[0080] In step 3.3, the transform coefficients are sorted in ascending order to form an ascending sequence, and a denoising threshold is calculated based on the ascending sequence; wherein, the formula for calculating the denoising threshold is:

[0081]

[0082]

[0083] in, Indicates the noise reduction threshold. This represents the k-th value in the ascending sequence. This represents the k-th value in the ascending sequence. Represents the sorting function. This represents the k-th value in the ascending sequence. This indicates the number of elements in the ascending sequence. This represents the mean of all transformation coefficients.

[0084] Step 3.4: Set the transform coefficients that are less than the denoising threshold to 0, and construct the denoised frequency domain signal based on the remaining transform coefficients.

[0085] Images typically contain noise from various sources (such as sensor noise and uneven lighting), which manifests as high-frequency components in the frequency domain. By denoising the frequency domain signal after Fourier transform, these unwanted noise components can be effectively reduced, resulting in a clearer image.

[0086] This invention provides a method for denoising underwater images by dividing them into sub-images, which can adapt to different characteristics of local areas of the image. For example, underwater images often have differences in lighting in different areas, blurring, and the movement of underwater objects. By dividing the image into multiple sub-images and processing them independently, an appropriate denoising strategy can be selected based on the specific characteristics of each sub-image, resulting in more accurate denoising.

[0087] Step 4: Reconstruct the denoised frequency domain signal to obtain the denoised underwater image;

[0088] Step 5: Perform color compensation on the denoised underwater image to obtain the compensated underwater image;

[0089] Furthermore, step 5 includes:

[0090] Step 5.1: Map the denoised underwater image to a pseudo-color image and calculate the mean of each color channel;

[0091] Step 5.2: Sort the mean values ​​of each color channel and filter out the maximum, minimum, and median values;

[0092] Step 5.3: Calculate the weighting coefficients using the maximum, minimum, and median values;

[0093] Step 5.4: Use weighting coefficients to compensate the pixel values ​​of the target color channel to obtain the compensated underwater image.

[0094] In step 5, weighting coefficients are used to compensate the color channels corresponding to the median and the minimum values ​​to obtain the compensated underwater image; the compensation process is as follows:

[0095]

[0096] in, This represents the first weighting coefficient. This represents the second weighting coefficient. This indicates the color channel corresponding to the median value. This represents the compensation value for the color channel corresponding to the minimum value. This represents the compensation value for the color channel corresponding to the maximum value. This represents the minimum value. This represents the median.

[0097] Due to the absorption and scattering of light by water, the grayscale distribution of the original image is often "concentrated in a narrow range and unevenly distributed," resulting in an overall grayish or dark appearance, with some areas being too bright or too dark. The large difference between the maximum and minimum values ​​of the pixel histogram means that a few grayscale levels accumulate a large number of pixels, while many grayscale levels are almost unused, leading to low image contrast and obscured details. The compensated underwater image tends to be more balanced across the entire effective dynamic range, with each grayscale level achieving a certain pixel proportion, thus significantly improving the visibility and contrast of the underwater image.

[0098] Step 6: Train a deep learning model using underwater images to obtain a silt recognition model;

[0099] like Figure 2As shown, the compensated underwater image is input into the YOLOv5 network for training to obtain a silt recognition model; wherein, during the training of the YOLOv5 network, the following formula is used:

[0100] )

[0101] )

[0102]

[0103]

[0104]

[0105] The model parameters of the YOLOv5 network are updated; among them, Indicates the current time step. Indicates the previous time step, For time steps First-order moment estimation at time, For time steps Second-order moment estimation at time, For gradient, As the first hyperparameter, This is the second hyperparameter. For learning rate, It is a constant. For time steps First-order moment estimation at time, For time steps Second-order moment estimation at time, For regularization parameters, This is the bias correction value for the first moment estimate at time step t. For time steps Model parameters at time, For time steps Model parameters at that time.

[0106] Step 7: Use a silt identification model to detect silt in the water at the target dredging location;

[0107] Step 8: When the dredging robot approaches the silt, use the preset automatic operation instructions to complete the dredging operation.

[0108] like Figure 3 As shown, in practical applications, the operator can manually control the angle and speed of any joint of the robotic arm through an Android host computer. After receiving the instruction through the network interface, the robot controls the joint to move to the corresponding position, and works in conjunction with the slurry pump, crushing motor and chassis movement to complete the dredging operation.

[0109] The operator sends a semi-automatic (automatic arm swinging) command to the robot via an Android host computer. After receiving the command, the robot controls the three joints of the robotic arm to automatically complete the automatic arm swinging action, which, in conjunction with the slurry pump, crushing motor, and chassis movement, completes the dredging operation.

[0110] The operator sends automatic operation instructions to the robot via an Android host computer. After receiving the instructions, the robot controls the three joints of the robotic arm to automatically complete the automatic swinging motion of the corresponding pipe diameter. In addition, after the specified swinging cycle, the robot automatically controls the chassis to perform stepping motion. The stepping time and distance are adjustable. Together with the slurry pump and crushing motor, the robot completes the dredging operation.

[0111] For example: performing an automated operation on a 600mm pipe:

[0112] The operator starts the automatic operation mode on the Android host computer.

[0113] In the pop-up dialog box, select 600mm pipe diameter, step time, and swing arm cycle parameters, and start automatic operation.

[0114] After receiving the instruction, the robot selects the 600mm diameter automatic swing arm and performs the automatic swing arm cycle for the specified period. Then, according to the required time, it controls the robot to complete the forward movement and starts the slurry pump and crushing motor to complete the automatic dredging operation.

[0115] The dredging robot in this invention also has a fault diagnosis function:

[0116] The fault diagnosis system adopts a two-tier design, generating unified fault codes that are invoked by the diagnostic service. The upper layer is responsible for centrally processing fault codes from various sub-components in the lower layer, providing a unified fault code and interface to the diagnostic service. The lower layer consists of sub-components, each identifying its own faults based on real-time operating conditions and generating fault codes and device codes. Each sub-component's device code is unique.

[0117] ii. The diagnostic service provides a shell diagnostic command interface. It supports viewing the current highest-level faults of the system, faults of individual sub-devices, and historical faults.

[0118] iii. The application layer will check the system fault status in real time and perform protection / recovery actions such as emergency shutdown and fault recovery according to the fault situation.

[0119] iv. For example: IMU fault identification and diagnosis process

[0120] v. When the IMU task is initialized, the IMU device address is set in its fault codes.

[0121] vi. During the task cycle, read IMU data. If five consecutive data reads fail, it indicates an IMU communication error. In this case, an IMU connection failure fault is generated in the IMU task module, with the fault priority being urgent.

[0122] vii. The upper-level fault diagnosis task periodically retrieves fault codes from each sub-component and checks for changes. If a change occurs, the latest fault code is saved, a timestamp is added to the fault code to record the time of the fault, and a historical estimate is written.

[0123] The viii.Shell human-machine interface allows users to view the system's current highest priority faults, historical estimates, and current faults of all devices using the dtc now|history|devices command.

[0124] This invention utilizes compensated underwater images to train a deep learning model, constructing a silt recognition model tailored to silt characteristics. Compared to traditional image processing methods based on threshold segmentation and edge detection, this model can adaptively extract deep features of silt across multiple dimensions, including texture, color, and shape, achieving high-precision detection and segmentation of silt regions. This silt recognition model maintains high accuracy and recall even in complex backgrounds, low contrast conditions, and partial occlusion, thereby improving the reliability of silt detection at target dredging locations.

[0125] This invention also provides a multi-degree-of-freedom dredging and obstacle removal robotic arm trajectory planning and control system, comprising:

[0126] The data acquisition module is used to acquire underwater images of the dredging location using multibeam imaging sonar;

[0127] The frequency domain transformation module is used to divide the underwater image into multiple sub-images and perform a two-dimensional Fourier transform on each sub-image to obtain the frequency domain signal after Fourier transform.

[0128] The denoising module is used to construct a corresponding denoising threshold based on each sub-image and remove noise from the frequency domain signal to obtain the denoised frequency domain signal.

[0129] The reconstruction module is used to reconstruct the denoised frequency domain signal to obtain the denoised underwater image;

[0130] The color compensation module is used to perform color compensation on the denoised underwater image to obtain the compensated underwater image.

[0131] The training module is used to train a deep learning model using compensated underwater images to obtain a silt recognition model;

[0132] The silt detection module is used to detect silt in the water at the target dredging location using a silt identification model;

[0133] The control module is used to complete the dredging operation using preset automatic operation instructions when the dredging robot approaches the silt.

[0134] Compared with the prior art, the beneficial effects of the multi-degree-of-freedom dredging and obstacle removal robotic arm trajectory planning and control system provided by the present invention are the same as the beneficial effects of the multi-degree-of-freedom dredging and obstacle removal robotic arm trajectory planning and control method described above, and will not be repeated here.

[0135] The present invention also provides an electronic device, including a bus, a transceiver, a memory, a processor, and a computer program stored in the memory and executable on the processor. The transceiver, the memory, and the processor are connected via the bus. The computer program, when executed by the processor, implements the steps in the above-described multi-degree-of-freedom dredging and obstacle removal operation trajectory planning and control method. Compared with the prior art, the beneficial effects of the electronic device provided by the present invention are the same as those of the above-described multi-degree-of-freedom dredging and obstacle removal robotic arm operation trajectory planning and control method, and will not be elaborated here.

[0136] The present invention also provides a computer-readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it implements the steps in the above-described multi-degree-of-freedom dredging and obstacle removal operation trajectory planning and control method. Compared with the prior art, the beneficial effects of the electronic device provided by the present invention are the same as the beneficial effects of the above-described multi-degree-of-freedom dredging and obstacle removal robotic arm operation trajectory planning and control method, and will not be elaborated here.

[0137] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for trajectory planning and control of multi-degree-of-freedom dredging and obstacle removal operations, characterized in that, include: Step 1: Use multibeam imaging sonar to acquire underwater images of the dredging location; Step 2: Divide the underwater image into multiple sub-images, and perform a two-dimensional Fourier transform on each sub-image to obtain the frequency domain signal after Fourier transform; Step 3: Construct a corresponding denoising threshold for each sub-image and remove noise from the frequency domain signal to obtain the denoised frequency domain signal; Step 3 involves constructing a corresponding denoising threshold for each sub-image and removing noise from the frequency domain signal to obtain the denoised frequency domain signal, including: Step 3.1: Obtain the Fourier spectrum from the frequency domain signal after Fourier transform; Step 3.2: Perform a two-dimensional discrete sine transform on the Fourier spectrum to obtain the transform coefficients; the two-dimensional discrete sine transform process is as follows: ； in, Represents the transformation coefficients. Represents frequency coordinates. Indicates the Fourier spectrum in The range at the point; Step 3.3: Determine the denoising threshold for each sub-image using the transform coefficients; Step 3.3 includes: The transform coefficients are sorted in ascending order to form an ascending sequence, and a denoising threshold is calculated based on the ascending sequence; wherein the formula for calculating the denoising threshold is: in, Indicates the noise reduction threshold. This represents the k-th value in the ascending sequence. This represents the k-th value in the ascending sequence. Represents the sorting function. This represents the k-th value in the ascending sequence. This indicates the number of elements in the ascending sequence. This represents the mean of all transformation coefficients; Step 3.4: Set the transform coefficients that are less than the denoising threshold to 0, and construct the denoised frequency domain signal based on the remaining transform coefficients; Step 4: Reconstruct the denoised frequency domain signal to obtain the denoised underwater image; Step 5: Perform color compensation on the denoised underwater image to obtain the compensated underwater image; Step 6: Train a deep learning model using the compensated underwater images to obtain a silt recognition model; Step 7: Use a silt identification model to detect silt in the water at the target dredging location; Step 8: When the dredging robot approaches the silt, use the preset automatic operation instructions to complete the dredging operation.

2. The multi-degree-of-freedom dredging and obstacle removal operation trajectory planning and control method according to claim 1, characterized in that, In step 2, the formula is used: A two-dimensional Fourier transform is performed on each sub-image to obtain the frequency domain signal after the Fourier transform; where, Indicates sub-image in Pixel value at that location, Indicates the length of the sub-image. Indicates the width of the sub-image. This represents the frequency domain signal after Fourier transform. Represents frequency coordinates.

3. The multi-degree-of-freedom dredging and obstacle removal operation trajectory planning and control method according to claim 1, characterized in that, Step 5: Perform color compensation on the denoised underwater image to obtain the compensated underwater image, including: Step 5.1: Map the denoised underwater image to a pseudo-color image and calculate the mean of each color channel; Step 5.2: Sort the mean values ​​of each color channel and filter out the maximum, minimum, and median values; Step 5.3: Calculate the weighting coefficients using the maximum, minimum, and median values; Step 5.4: Use weighting coefficients to compensate the pixel values ​​of the target color channel to obtain the compensated underwater image.

4. The multi-degree-of-freedom dredging and obstacle removal operation trajectory planning and control method according to claim 3, characterized in that, In step 5.4, weighting coefficients are used to compensate the color channels corresponding to the median and the minimum values ​​to obtain the compensated underwater image; the compensation process is as follows: in, This represents the first weighting coefficient. This represents the second weighting coefficient. This indicates the color channel corresponding to the median value. This represents the compensation value for the color channel corresponding to the minimum value. This represents the compensation value for the color channel corresponding to the maximum value. This represents the minimum value. This represents the median.

5. The multi-degree-of-freedom dredging and obstacle removal operation trajectory planning and control method according to claim 1, characterized in that, Step 6: Using the compensated underwater images to train a deep learning model to obtain a silt recognition model, including: The compensated underwater image is input into the YOLOv5 network for training to obtain a silt recognition model; wherein, during the training of the YOLOv5 network, the following formula is used: ) ) The model parameters of the YOLOv5 network are updated; among them, Indicates the current time step. Indicates the previous time step, For time steps First-order moment estimation at time, For time steps Second-order moment estimation at time, For gradient, As the first hyperparameter, This is the second hyperparameter. For learning rate, It is a constant. For time steps First-order moment estimation at time, For time steps Second-order moment estimation at time, For regularization parameters, This is the bias correction value for the first moment estimate at time step t. For time steps Model parameters at time, For time steps Model parameters at that time.

6. A multi-degree-of-freedom dredging and obstacle removal robotic arm trajectory planning and control system, characterized in that, include: The data acquisition module is used to acquire underwater images of the dredging location using multibeam imaging sonar; The frequency domain transformation module is used to divide the underwater image into multiple sub-images and perform a two-dimensional Fourier transform on each sub-image to obtain the frequency domain signal after Fourier transform. The denoising module is used to construct a corresponding denoising threshold based on each sub-image and remove noise from the frequency domain signal to obtain the denoised frequency domain signal. Specifically, a corresponding denoising threshold is constructed based on each sub-image, and noise is removed from the frequency domain signal to obtain the denoised frequency domain signal, including: Step 3.1: Obtain the Fourier spectrum from the frequency domain signal after Fourier transform; Step 3.2: Perform a two-dimensional discrete sine transform on the Fourier spectrum to obtain the transform coefficients; the two-dimensional discrete sine transform process is as follows: ； in, Represents the transformation coefficients. Represents frequency coordinates. Indicates the Fourier spectrum in The range at the point; Step 3.3: Determine the denoising threshold for each sub-image using the transform coefficients; Step 3.3 includes: The transform coefficients are sorted in ascending order to form an ascending sequence, and a denoising threshold is calculated based on the ascending sequence; wherein the formula for calculating the denoising threshold is: in, Indicates the noise reduction threshold. This represents the k-th value in the ascending sequence. This represents the k-th value in the ascending sequence. Represents the sorting function. This represents the k-th value in the ascending sequence. This indicates the number of elements in the ascending sequence. This represents the mean of all transformation coefficients; Step 3.4: Set the transform coefficients that are less than the denoising threshold to 0, and construct the denoised frequency domain signal based on the remaining transform coefficients; The reconstruction module is used to reconstruct the denoised frequency domain signal to obtain the denoised underwater image; The color compensation module is used to perform color compensation on the denoised underwater image to obtain the compensated underwater image. The training module is used to train a deep learning model using compensated underwater images to obtain a silt recognition model; The silt detection module is used to detect silt in the water at the target dredging location using a silt identification model; The control module is used to complete the dredging operation using preset automatic operation instructions when the dredging robot approaches the silt.

7. An electronic device comprising a bus, a transceiver, a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the transceiver, the memory, and the processor are connected via the bus, characterized in that, When the computer program is executed by the processor, it implements the steps in the multi-degree-of-freedom dredging and obstacle removal operation trajectory planning and control method as described in any one of claims 1-5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps in the multi-degree-of-freedom dredging and obstacle removal operation trajectory planning and control method as described in any one of claims 1-5.

Images