CN-122024030-A - Multi-degree-of-freedom dredging and obstacle removing operation track planning and control method

CN122024030ACN 122024030 ACN122024030 ACN 122024030ACN-122024030-A

Abstract

The invention relates to a track planning and control method for multi-degree-of-freedom dredging and obstacle removing operation, which belongs to the technical field of dredging and obstacle removing operation and comprises the steps of dividing an underwater image acquired by 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, denoising to obtain a denoised underwater image, performing color compensation on the denoised underwater image to obtain a compensated underwater image, training a deep learning model to obtain a silt identification model, and detecting silt in water at a target dredging position by using the silt identification model to finish dredging operation. According to the invention, the deep learning model is trained by using the denoised underwater image, and the silt identification model aiming at the silt characteristics is constructed, so that compared with the traditional image processing methods based on threshold segmentation, edge detection and the like, the deep characteristics of the silt in multiple dimensions such as texture, color and shape can be adaptively extracted, and the high-precision detection and segmentation of the silt region are realized.

Inventors

ZHANG SHENG
WANG JING
GUO PENG
ZHANG HANCE
WANG XIAOFENG
MA QILONG
SUN CHENGXIANG
LI YAKUN

Assignees

鸿基骏业环保科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260413

Claims (10)

1. The multi-degree-of-freedom dredging obstacle-removing operation track planning and controlling method is characterized by comprising the following steps: Step 1, acquiring an underwater image of a dredging position by using multi-beam image sonar; dividing the underwater image into a plurality of sub-images, and performing two-dimensional Fourier transform on each sub-image to obtain a frequency domain signal after Fourier transform; step 3, constructing a corresponding denoising threshold value based on each sub-image, and removing noise in the frequency domain signal to obtain a denoised frequency domain signal; step 4, reconstructing the denoised frequency domain signal to obtain a denoised underwater image; Step 5, performing color compensation on the denoised underwater image to obtain a compensated underwater image; training a deep learning model by using the compensated underwater image to obtain a silt identification model; Step 7, detecting sludge in water at a target dredging position by using a sludge identification model; and 8, finishing dredging operation by using a preset automatic operation instruction when the dredging robot approaches to the sludge.
2. The method for planning and controlling the track of the multi-degree-of-freedom dredging operation according to claim 1, wherein in the step 2, the formula is adopted: performing two-dimensional Fourier transform on each sub-image to obtain a frequency domain signal after Fourier transform, wherein, Representing sub-images in The pixel value at which it is located, The length of the sub-image is represented, The width of the sub-image is indicated, Representing the frequency domain signal after the fourier transform, Representing the frequency coordinates.
3. The method for planning and controlling the track of the multi-degree-of-freedom dredging operation according to claim 2, wherein the step 3 is to construct a corresponding denoising threshold value based on each sub-image, remove noise in the frequency domain signal, and obtain a denoised frequency domain signal, and comprises the following steps: Step 3.1, obtaining a Fourier spectrogram according to the frequency domain signal after Fourier transformation; And 3.2, performing short-time Fourier transform on the Fourier spectrogram to obtain a transformation coefficient, wherein the short-time Fourier transform process is as follows: ; Wherein, the The transform coefficients are represented by a set of coefficients, The frequency coordinates are represented as such, Representing Fourier spectrum in Amplitude at; Step 3.3, determining a denoising threshold value of each sub-image by using the transformation coefficient; And 3.4, setting the transformation coefficient corresponding to the denoising threshold value smaller than the denoising threshold value to be 0, and constructing a denoised frequency domain signal according to the residual transformation coefficient.
4. The method for planning and controlling a track of a multi-degree-of-freedom dredging operation according to claim 3, wherein in step 3.3, the method comprises: the transformation coefficients are ordered according to ascending order to form ascending order sequences, and denoising thresholds are calculated according to the ascending order sequences, wherein the calculation formula of the denoising thresholds is as follows: Wherein, the Representing the threshold value of the de-noising, Representing the kth value in the ascending sequence, Representing the kth value in the ascending sequence, Representing the ranking function, Representing the kth value in the ascending sequence, Representing the number of elements in the ascending sequence, Representing the average of all transform coefficients.
5. The method for planning and controlling the track of the multi-degree-of-freedom dredging and obstacle removing operation according to claim 1, wherein the step 5 is characterized in that the step of performing color compensation on the denoised underwater image to obtain the compensated underwater image comprises the following steps: Step 5.1, mapping the denoised underwater image into a pseudo-color image, and calculating the average value of each color channel; Step 5.2, sorting the average value of each color channel to screen out the maximum value, the minimum value and the median value; step 5.3, calculating a weight coefficient by using the maximum value, the minimum value and the median value; and 5.4, respectively compensating the pixel values of the target color channels by using the weight coefficients to obtain the compensated underwater image.
6. The method for planning and controlling the track of the multi-degree-of-freedom dredging and obstacle removing operation according to claim 5, wherein in step 5.4, the color channel corresponding to the median value and the color channel corresponding to the minimum value are compensated by using the weight coefficient to obtain the compensated underwater image, wherein the compensation process is as follows: Wherein, the A first weight coefficient is represented and a second weight coefficient is represented, A second weight coefficient is represented and is used to represent, The color channel corresponding to the median value is represented, Representing the compensation value of the color channel corresponding to the minimum value, Representing the compensation value of the color channel corresponding to the maximum value, The value of the minimum value is indicated, The median value is indicated.
7. The method for planning and controlling the track of the multi-degree-of-freedom dredging and obstacle-removing operation according to claim 1, wherein the step 6 is to train a deep learning model by using the compensated underwater image to obtain a silt identification model, and comprises the following steps: inputting the compensated underwater image into YOLOv network for training to obtain a silt identification model, wherein in the training process of YOLOv network, the formula is adopted: ) ) updating model parameters of YOLOv network, wherein, The current time step is indicated and the current time step, The previous time step is indicated and is indicated, Is a time step A first moment estimate of the time period, Is a time step A second moment estimate of the time period, In the form of a gradient, As a first super-parameter, the first super-parameter, As a second super-parameter, the second super-parameter, In order for the rate of learning to be high, Is a constant value, and is used for the treatment of the skin, Is a time step A first moment estimate of the time period, Is a time step A second moment estimate of the time period, In order for the parameters to be regularized, For the offset correction value of the first moment estimate at time step t, Is a time step The parameters of the model at the time of the calculation, Is a time step Model parameters at that time.
8. Multi freedom desilting clearance arm operation orbit planning and control system, its characterized in that includes: The data acquisition module is used for acquiring an underwater image of the dredging position by using the multi-beam image sonar; the frequency domain transformation module is used for dividing the underwater image into a plurality of sub-images and carrying out two-dimensional Fourier transformation on each sub-image to obtain a frequency domain signal after Fourier transformation; the denoising module is used for constructing a corresponding denoising threshold value based on each sub-image and removing noise in the frequency domain signal to obtain a denoised frequency domain signal; the reconstruction module is used for reconstructing the denoised frequency domain signal to obtain a denoised underwater image; The color compensation module is used for performing color compensation on the denoised underwater image to obtain a compensated underwater image; the training module is used for training the deep learning model by using the compensated underwater image to obtain a silt identification model; the silt detection module is used for detecting silt in water at a target dredging position by using a silt identification model; And the control module is used for completing dredging operation by using a preset automatic operation instruction when the dredging robot approaches to the sludge.
9. An electronic device comprising a bus, a transceiver, a memory, a processor and a computer program stored on the memory and operable on the processor, the transceiver, the memory and the processor being connected by the bus, characterized in that the computer program when executed by the processor implements the steps of the multiple degree of freedom dredging barrier cleaning operation trajectory planning and control method according to any one of claims 1-7.
10. A computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements the steps of the multiple degree of freedom dredging obstacle clearance job trajectory planning and control method according to any one of claims 1-7.

Description

Multi-degree-of-freedom dredging and obstacle removing operation track planning and control method Technical Field The invention relates to the technical field of dredging and obstacle-removing operation, in particular to a track planning and control method for multi-degree-of-freedom dredging and obstacle-removing operation. Background River course, lake, reservoir and city drainage pipe network are in long-term operation in-process, are liable to be because of silt siltation, floater pile up, collapse thing and various solid foreign matter jam, lead to the water section to shrink, the flood discharge ability decline, even appear pipeline overflow, backflow and water black and odorous scheduling problem. In order to restore the normal functions of hydraulic buildings and drainage systems, dredging and obstacle clearing operations need to be carried out regularly or irregularly. The traditional dredging and obstacle removing mode mainly relies on manual well descending and water descending operations or large mechanical equipment such as an excavating ship, a grab ship and the like, and has the outstanding problems of severe operation environment, high labor intensity, low operation precision, low efficiency and high personnel safety risk, and is difficult to meet the current requirements of refined and normalized water environment treatment and facility operation and maintenance. With the development of robot technology and automatic control technology, various dredging and barrier removing devices suitable for underwater, underwater or closed pipeline environments, such as pipeline overhauling robots, underwater operation robots, dredging systems with crawler-type or wheel-type mobile platforms carrying mechanical arms and the like, appear. Such equipment is typically equipped with multiple degree of freedom robotic arms or multi-joint working mechanisms for breaking, gripping, dredging and transporting sludge or obstacles through end effectors such as grasping tools, cutting tools, suction and digging tools, high pressure water guns, and the like. Compared with the traditional mode, the robot and the multi-degree-of-freedom operation mechanism can perform operation in narrow, complex and dangerous environments, and have the advantages of high automation degree, good personnel safety, long sustainable operation time and the like. However, in the prior art, the track planning and control method of the multi-degree-of-freedom dredging and obstacle removing operation equipment is still rough, and often, a general planning and control strategy in the field of industrial mechanical arms or general mobile robots is adopted, so that the complexity of underwater or underwater dredging working conditions is difficult to fully adapt. On the one hand, the existing system mostly adopts working path planning based on manual experience or simple geometric rules, such as sequential scanning along a pipeline axis or a regular grid, linear reciprocating scraping and the like, and lacks comprehensive modeling and intelligent planning on environmental information such as sedimentation form, barrier distribution, space constraint, water flow disturbance and the like, so that the working path redundancy, insufficient coverage rate or repeated cleaning of local areas are caused, and the overall working efficiency is influenced. Disclosure of Invention In order to solve the above problems, an embodiment of the present invention is to provide a method for planning and controlling a track of a multi-degree-of-freedom dredging and obstacle removing operation. A multi-degree-of-freedom dredging obstacle-removing operation track planning and controlling method comprises the following steps: Step 1, acquiring an underwater image of a dredging position by using multi-beam image sonar; dividing the underwater image into a plurality of sub-images, and performing two-dimensional Fourier transform on each sub-image to obtain a frequency domain signal after Fourier transform; step 3, constructing a corresponding denoising threshold value based on each sub-image, and removing noise in the frequency domain signal to obtain a denoised frequency domain signal; step 4, reconstructing the denoised frequency domain signal to obtain a denoised underwater image; Step 5, performing color compensation on the denoised underwater image to obtain a compensated underwater image; training a deep learning model by using the compensated underwater image to obtain a silt identification model; Step 7, detecting sludge in water at a target dredging position by using a sludge identification model; and 8, finishing dredging operation by using a preset automatic operation instruction when the dredging robot approaches to the sludge. Preferably, in the step 2, the formula is adopted: performing two-dimensional Fourier transform on each sub-image to obtain a frequency domain signal after Fourier transform, wherein, Representing sub-images inThe pixel value at which