Dry ice cleaning method and device

Abstract

The invention discloses a dry ice cleaning method and device, and the method comprises the steps: carrying out the stain recognition of a building surface image of a to-be-cleaned building, and carrying out the dry ice cleaning based on the stain information after the existence of the stain is recognized, because the dry ice cleaning does not need to use a water source, and does not need to use a mechanical tool to make contact with the building surface. Therefore, stains on the surface of a building can be quickly removed, secondary pollution cannot be caused, the surface of the building cannot be damaged, and the cleaning effect and the integrity of the surface of the building can be considered at the same time. In addition, mixed gas in a dry ice spraying area is recycled based on a closed circulation grading recycling mode, resource consumption is reduced, and excessive emission of carbon dioxide is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of building cleaning, and in particular, to a dry ice cleaning method and device. Background Art

[0002] Building maintenance is crucial for extending the service life of buildings, maintaining their aesthetics, and ensuring their functionality. Existing building cleaning methods usually rely on high-pressure water jets, chemical cleaning agents, and mechanical tools. These methods have problems such as time-consuming, water-consuming, environmental pollution, and easy damage to the building surface. For some special buildings, such as historical buildings, due to their unique cultural and artistic values, it is necessary to avoid damage to their surfaces, and existing building cleaning methods are difficult to balance both cleaning effect and the integrity of the building surface. Summary of the Invention

[0003] Embodiments of the present invention provide a dry ice cleaning method and device to solve the technical problem that existing building cleaning methods are difficult to balance both cleaning effect and the integrity of the building surface.

[0004] To solve the above technical problem, a first aspect of an embodiment of the present invention provides a dry ice cleaning method, including:

[0005] Obtaining an image of the building surface of the building to be cleaned, and performing stain recognition on the building surface image;

[0006] When stains are recognized in the building surface image, obtaining stain information of the stains;

[0007] Determining current dry ice spraying parameters according to the stain information;

[0008] Based on the dry ice spraying parameters, spraying dry ice on the stains.

[0009] As a preferred solution, the obtaining an image of the building surface of the building to be cleaned, and performing stain recognition on the building surface image specifically includes:

[0010] Obtaining a multispectral image of the building surface of the building to be cleaned, and using the multispectral image as the building surface image;

[0011] Using a pre-trained stain recognition model to perform stain recognition on the multispectral image to obtain a stain recognition result; wherein, the stain recognition model is obtained by training a multi-channel convolutional neural network with multispectral training images with stain marking information;

[0012] When it is determined that there are any type of stains based on the stain recognition result, it is determined that there are stains in the building surface image.

[0013] As a preferred solution, using the pre-trained stain recognition model to perform stain recognition on the multi-spectral image to obtain a stain recognition result, specifically including:

[0014] Extract features from the multi-spectral image to obtain multi-modal stain features;

[0015] Use the CBAM module to enhance the features of the multi-modal stain features to obtain enhanced stain features;

[0016] Use a classifier to perform stain recognition on the enhanced stain features to generate a stain probability heat map;

[0017] According to the stain probability heat map, determine the stain bounding box and the target stain type;

[0018] Adopt the FAST corner detection algorithm to extract a preset number of candidate points within the stain bounding box, and based on the gradient magnitudes of each candidate point, obtain a number of stain feature points;

[0019] Based on the gray gradient method, iteratively optimize the positions of the several stain feature points until the change amount of the positions of the stain feature points is less than the preset pixel value, and determine the target stain center position according to the centroid of each stain feature point after iterative optimization;

[0020] According to the target stain type and the target stain center position, obtain the stain recognition result.

[0021] As a preferred solution, obtaining the stain information of the stain, specifically including:

[0022] Use a lidar to scan the building surface of the building to be cleaned to construct a three-dimensional building surface model;

[0023] According to the point cloud data in the three-dimensional building surface model and the positions of each stain feature point after iterative optimization, determine the stain area in the three-dimensional building surface model;

[0024] According to the point cloud data and the stain area, calculate the stain area of the stain area and the building surface area of the building to be cleaned, and calculate the maximum distance difference between the stain area and the building surface;

[0025] According to the stain area and the building surface area, calculate the area ratio of the stain area, and based on the preset stain stubbornness levels corresponding to different area ratios and distance differences, determine the target stain stubbornness level corresponding to the area ratio of the stain area and the maximum distance difference;

[0026] Determine the stain information of the stain based on the target stain type, the target stain center position, and the target stain stubbornness level.

[0027] As a preferred solution, determining the current dry ice jetting parameters according to the stain information specifically includes:

[0028] Determine the current dry ice jetting pressure, dry ice particle density, and dry ice jetting time according to the dry ice jetting parameters corresponding to different preset stain types and stain stubbornness levels;

[0029] Determine the current dry ice jetting angle according to the target stain center position;

[0030] Determine the current dry ice jetting parameters according to the dry ice jetting pressure, the dry ice particle density, the dry ice jetting time, and the dry ice jetting angle.

[0031] As a preferred solution, based on the dry ice jetting parameters, performing dry ice jetting on the stain specifically includes:

[0032] Adjust the current dry ice jetting nozzle according to the dry ice jetting angle so that the dry ice jetting nozzle is aligned with the target stain center position;

[0033] Determine the stain density at different positions according to the stain probability heat map;

[0034] According to the preset cleaning area boundary information and motion constraint parameters, starting from the target stain center position, perform path planning in the order of decreasing stain density to obtain a dry ice jetting trajectory;

[0035] Perform dry ice jetting using the dry ice jetting nozzle according to the dry ice jetting pressure, the dry ice particle density, and the dry ice jetting time, and move the dry ice jetting nozzle according to the dry ice jetting trajectory.

[0036] As a preferred solution, after performing dry ice jetting on the stain based on the dry ice jetting parameters, the method further includes:

[0037] Recover the mixed gas in the dry ice jetting area; wherein the mixed gas includes dry ice particles, dirt, and carbon dioxide;

[0038] Perform a primary separation treatment on the mixed gas through a cyclone separator to separate and obtain the dirt and the dry ice particles from the mixed gas, and collect the dirt and the dry ice particles in a preset dirt storage tank and a dry ice storage tank respectively;

[0039] The remaining mixed gas after the primary separation treatment is subjected to a secondary separation treatment through an electrostatic precipitator to separate and obtain the carbon dioxide from the remaining mixed gas, and the carbon dioxide is liquefied and collected in a preset carbon dioxide storage tank, or the carbon dioxide is adsorbed and stored through a molecular sieve.

[0040] As a preferred solution, the method further includes:

[0041] When it is recognized that there are no stains in the building surface image, calculate the distance between the current position and the edge of the building surface of the building to be cleaned;

[0042] When the distance is greater than a first preset distance, move along the normal of the building surface of the building to be cleaned towards the edge of the building surface, and calculate the distance in real time; when it is detected that the distance remains unchanged within a preset time period, retreat a preset distance, then switch the moving angle and continue to move towards the edge of the building surface until it is detected that the distance becomes smaller within the preset time period, re-acquire the building surface image of the building to be cleaned, and perform stain recognition on the building surface image;

[0043] When the distance is less than or equal to the first preset distance and greater than or equal to a second preset distance, move towards the edge of the building surface at the approaching speed corresponding to the building surface material of the preset building to be cleaned, and update the distance at a frequency of 10 Hz during the movement, perform multi-spectral scanning on the building surface to re-acquire the building surface image of the building to be cleaned, and perform stain recognition on the building surface image;

[0044] When the distance is less than the second preset distance, move towards the edge of the building surface until the distance reaches a preset minimum distance, use a contact sensor array to detect the change in the pressure distribution on the building surface, and calculate the standard deviation of the contact pressure on the building surface based on the change in the pressure distribution; when the standard deviation of the contact pressure is greater than a preset pressure standard deviation threshold, emit a vibration signal to the building surface to identify whether there are stains on the building surface; if there are, re-acquire the building surface image of the building to be cleaned, and perform stain recognition on the building surface image; if not, perform dry ice spraying on the building surface according to the preset dry ice cleaning parameters.

[0045] As a preferred solution, the method further includes:

[0046] Real-time detect the remaining amount of dry ice in the dry ice storage tank;

[0047] When it is detected that the remaining amount of dry ice is less than a preset remaining amount threshold, an alarm message indicating insufficient remaining amount is issued.

[0048] In a second aspect of the embodiments of the present invention, a dry ice cleaning device is provided, which includes a housing, a moving module, a storage module, a recovery module, a telescopic arm module, a dry ice spraying module, and a control module;

[0049] The moving module is arranged on both sides of the housing; the storage module, the recovery module, and the telescopic arm module are all arranged on the housing;

[0050] The dry ice spraying module is arranged at the end of the telescopic arm module. The dry ice spraying module includes a high-pressure air generating module, a dry ice spraying nozzle, and a vacuum adsorption module; the input end of the dry ice spraying nozzle is connected to the gas output end of the high-pressure air generating module, and the input end of the dry ice spraying nozzle is connected to the recovery module through a dry ice delivery pipeline; the vacuum adsorption module is arranged close to the dry ice spraying nozzle, and the vacuum adsorption module is connected to the recovery module;

[0051] The storage module includes a dirt storage tank, a dry ice storage tank, and a carbon dioxide storage tank. The recovery module has a dirt separation port, a dry ice separation port, and a carbon dioxide separation port. The dirt separation port, the dry ice separation port, and the carbon dioxide separation port are respectively connected to the dirt storage tank, the dry ice storage tank, and the carbon dioxide storage tank;

[0052] The control module is arranged inside the housing, and the control end of the control module is respectively connected to the controlled ends of the moving module, the storage module, the recovery module, the telescopic arm module, and the dry ice spraying module; the control module is used to execute the dry ice cleaning method described in any one of the first aspects.

[0053] Compared with the prior art, the beneficial effect of the embodiments of the present invention is that by identifying stains on the building surface image of the building to be cleaned, and after identifying the existence of stains, dry ice cleaning is performed based on the stain information. Since dry ice cleaning does not require the use of water and does not need to contact the building surface with mechanical tools, it can quickly remove the stains on the building surface while not causing secondary pollution and not damaging the building surface, thus being able to take into account both the cleaning effect and the integrity of the building surface. Description of the Drawings

[0054] Figure 1 is a schematic flow chart of the dry ice cleaning method in the embodiments of the present invention;

[0055] Figure 2 is a schematic diagram of the dry ice spraying parameters corresponding to different stain types in the embodiments of the present invention;

[0056] Figure 3 is a schematic flow chart of the recovery of the mixed gas during the dry ice cleaning process in the embodiments of the present invention;

[0057] Figure 4 is the flowchart of the dry ice cleaning method in the embodiment of the present invention;

[0058] Figure 5 is the structural schematic diagram of the dry ice cleaning device in the embodiment of the present invention;

[0059] Figure 6 is the architecture diagram of the dry ice cleaning system in the embodiment of the present invention;

[0060] Wherein, 1. Outer shell; 2. Moving module; 3. Dry ice delivery pipeline; 4. Telescopic arm module; 5. Dry ice spraying module. Specific implementation manners

[0061] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present invention.

[0062] Please refer to Figure 1 , the first aspect of the embodiment of the present invention provides a dry ice cleaning method, including the following steps S1 to S4:

[0063] Step S1, obtain an image of the building surface to be cleaned, and perform stain recognition on the building surface image;

[0064] Step S2, when stains are recognized in the building surface image, obtain the stain information of the stains;

[0065] Step S3, determine the current dry ice spraying parameters according to the stain information;

[0066] Step S4, perform dry ice spraying on the stains based on the dry ice spraying parameters.

[0067] It should be noted that there are various factors that may damage the exterior walls of buildings in the traditional high-pressure water jet cleaning method, chemical cleaning agent cleaning method, and mechanical tool cleaning method. Specifically, when high-pressure water jets are used to wash the exterior walls of buildings, a large impact force will be generated. This impact force may damage the surface structure of the exterior walls of buildings, especially for some historical buildings with relatively fragile materials and long ages. For example, high-pressure water jets may cause cracks, peeling, etc. on the stone surface of the exterior walls. Long-term use may also cause the joints of the walls to become loose, affecting the overall stability of the walls. Chemical cleaning agents usually contain various chemical components, and these components may react chemically with the materials of the exterior walls of buildings. For example, acidic cleaning agents may corrode the calcium carbonate components in the stone, causing the stone surface to lose its luster and become rough; alkaline cleaning agents may damage the coatings on the exterior walls of buildings, resulting in coating peeling and color change. Moreover, if the chemical cleaning agents are not thoroughly rinsed after cleaning, the residual chemical substances will continue to erode the exterior walls, accelerating the aging and damage of the exterior walls. When using mechanical tools to clean the exterior walls of buildings, the tools are in direct contact with the exterior wall surface, and friction is likely to occur. During the cleaning process, the mechanical tools may scratch the exterior wall surface. For exterior walls of buildings with decorative carvings and coatings, this kind of damage is more obvious. For example, tools such as wire brushes may damage the details of the carvings, damaging the artistic value of the building and also causing the exterior wall coatings to peel off, reducing the protective performance of the exterior walls. Dry ice cleaning does not require the use of water sources, avoiding secondary pollution and wall residues of traditional water washing methods. It also does not need to use mechanical tools to contact the building surface. At the same time, it is non-toxic and non-conductive, will not cause environmental pollution, and ensures the integrity of the building surface.

[0068] Specifically, in order to perform precise dry ice cleaning on the building to be cleaned in this embodiment, it is first necessary to obtain the building surface image and perform stain recognition on it. When stains are identified in the building surface image, it is necessary to obtain the stain information corresponding to the stains, such as stain type, location, size, stubbornness level, etc., so as to be able to perform precise dry ice spraying based on the location of the stains, and based on the size and stubbornness level of the stains, adopt different dry ice spraying parameters for spraying to effectively remove the stains from the building surface. Further, after determining the stain information, it is necessary to determine the dry ice spraying parameters for the current stain to be cleaned, such as dry ice spraying angle, dry ice spraying pressure, dry ice particle density, and dry ice spraying time, etc., and perform dry ice spraying based on this dry ice spraying parameter, so as to achieve the corresponding stain removal effect.

[0069] The dry ice cleaning method provided by the embodiments of the present invention identifies stains by analyzing the surface images of the building to be cleaned. After stains are identified, dry ice cleaning is performed based on the stain information. Since dry ice cleaning does not require the use of water and does not need to contact the building surface with mechanical tools, it can quickly remove stains on the building surface without causing secondary pollution or damaging the building surface, thus taking into account both the cleaning effect and the integrity of the building surface.

[0070] Compared with traditional hand-held dry ice cleaning machines, operators need to wear appropriate protective equipment, such as gloves and goggles, during operation to ensure safety, which makes the operation difficult. However, the embodiments of the present invention can realize an intelligent and automated dry ice cleaning process, replacing manual labor to complete high-risk, repetitive, and heavy dry ice cleaning tasks.

[0071] As a preferred solution, obtaining the surface image of the building to be cleaned and performing stain identification on the surface image of the building specifically includes:

[0072] Obtaining a multi-spectral image of the surface of the building to be cleaned and using the multi-spectral image as the surface image of the building;

[0073] Using a pre-trained stain identification model to perform stain identification on the multi-spectral image to obtain a stain identification result; wherein, the stain identification model is obtained by training a multi-channel convolutional neural network with multi-spectral training images having stain marking information;

[0074] When it is determined that there is any type of stain based on the stain identification result, it is determined that there are stains in the surface image of the building.

[0075] Specifically, this embodiment combines computer vision technology, sensor fusion, and intelligent algorithms to collect multi-modal stain data through sensors. For example, a multi-spectral camera is used to obtain multi-spectral images of the building surface of the building to be cleaned. It can be understood that different types of stains on the wall are identified by using light of different wavelengths. Different stains will exhibit different characteristics in the multi-spectral image. For example, oil stains will have a unique reflection spectrum in certain bands, while mold will have obvious absorption characteristics in other bands. By analyzing the multi-spectral image, various stains can be quickly distinguished. Further, a pre-trained stain recognition model is used to perform stain recognition on the multi-spectral image to obtain a stain recognition result. Among them, this embodiment can prepare a multi-spectral training image set with stain marking information in advance, that is, the stain positions and stain types in different multi-spectral training images can be marked, so that the multi-channel convolutional neural network can learn the feature information of different types of stains. Exemplarily, the stain types include oil stains (grease, industrial pollutants), biofilms (mold, algae, moss), dust (sand dust, PM2.5 deposition), paint residues (paint, graffiti), and mixed stains (i.e., mutual mixtures of the above various stains).

[0076] As a preferred solution, the using of the pre-trained stain recognition model to perform stain recognition on the multi-spectral image to obtain a stain recognition result specifically includes:

[0077] Performing feature extraction on the multi-spectral image to obtain multi-modal stain features;

[0078] Using the CBAM module to perform feature enhancement on the multi-modal stain features to obtain enhanced stain features;

[0079] Using a classifier to perform stain recognition on the enhanced stain features to generate a stain probability heat map;

[0080] According to the stain probability heat map, determining the stain bounding box and the target stain type;

[0081] Adopting the FAST corner detection algorithm to extract a preset number of candidate points within the stain bounding box, and obtaining a number of stain feature points based on the gradient magnitudes of each of the candidate points;

[0082] Based on the gray gradient method, iteratively optimizing the positions of a number of the stain feature points until the change amount of the positions of the stain feature points is less than a preset pixel value, and determining the target stain center position according to the centroids of the stain feature points after iterative optimization;

[0083] According to the target stain type and the target stain center position, obtaining the stain recognition result.

[0084] Specifically, in the process of stain recognition in this embodiment, first, feature extraction is performed on the multi-spectral image to obtain multi-modal stain features. For some small target stains, such as bird droppings, in order to achieve accurate stain recognition, this embodiment introduces an attention mechanism for feature enhancement. The CBAM module (Convolutional Block Attention Module) can enhance local features by introducing the attention mechanism in both the channel and spatial dimensions. In the channel dimension, the CBAM module can automatically learn the importance of different channel features and give higher weights to the channels containing small target stain features. In the spatial dimension, it can focus on the area where the small target stain is located and enhance the feature representation of this area. For example, when a bird dropping point is detected, the CBAM module will highlight the local area where the bird dropping point is located, making the network pay more attention to this small target, thereby improving the recognition and classification ability of small target stains. Further, the classifier calculates the probability of each pixel point belonging to each stain type based on the enhanced stain features in different bands of the image, and then generates a stain probability heat map.

[0085] Further, apply a threshold segmentation algorithm (such as the Otsu threshold method) to the generated stain probability heat map to convert the heat map into a binary image, thereby distinguishing the area where stains may exist and the background area. In the binary image, use a contour detection algorithm (such as the findContours function in OpenCV) to find the contours of the stain area. Then, calculate the minimum rectangular bounding box that can enclose the stain area based on these contours. The coordinates of this bounding box are the initial values of the stain bounding box. Based on the calculated initial values of the bounding box, determine the dynamic region of interest (ROI) in the image, that is, the stain bounding box in this embodiment. Subsequent processing operations will mainly be carried out within this stain bounding box, which can reduce the amount of calculation and improve the processing efficiency.

[0086] Further, use the FAST corner detection algorithm to extract a preset number of candidate points within the stain bounding box, and obtain a number of stain feature points based on the gradient magnitudes of each candidate point. For example, extract 50 to 100 candidate points and retain the 20% high-confidence points with the largest gradient magnitudes as stain feature points.

[0087] Further, based on the gray gradient method, iteratively optimize the positions of a number of stain feature points until the change amount of the positions of the stain feature points is less than a preset pixel value. For example, until the change amount of the positions of the stain feature points < 0.01 pixel, so that the centroid iteratively approaches, thereby determining the central position of the stain within the stain bounding box.

[0088] As a preferred solution, obtaining the stain information of the stain specifically includes:

[0089] Use a lidar to scan the building surface of the building to be cleaned and construct a three-dimensional building surface model;

[0090] Determine the stain area in the three-dimensional building surface model according to the point cloud data in the three-dimensional building surface model and the positions of the respective stain feature points after iterative optimization;

[0091] Calculate the stain area of the stain area and the building surface area of the building to be cleaned according to the point cloud data and the stain area, and calculate the maximum distance difference between the stain area and the building surface;

[0092] Calculate the area ratio of the stain area according to the stain area and the building surface area, and determine the target stain stubbornness level corresponding to the area ratio of the stain area and the maximum distance difference based on the preset stain stubbornness levels corresponding to different area ratios and distance differences;

[0093] Determine the stain information of the stain according to the target stain type, the target stain center position and the target stain stubbornness level.

[0094] Specifically, use a 3D lidar to comprehensively scan the building wall. The lidar emits laser beams towards the wall and measures the time it takes for the reflected light to return, thereby calculating the distance of each point from the lidar, and thus obtaining the three-dimensional coordinate information of a large number of points on the wall surface to form point cloud data. These point cloud data record the geometric shape and spatial position of the wall, so that a three-dimensional building surface model can be constructed. Since the positions of the respective stain feature points, which are equivalent to the stain area, have been confirmed in the previous steps, considering that the coordinate system of the three-dimensional building surface model is different from that of the two-dimensional image, the position of the stain area in this two-dimensional image can be converted into the point cloud position in the three-dimensional building surface model, and the point cloud position can be directly obtained based on the point cloud data in the three-dimensional building surface model, thereby determining the stain area in the three-dimensional building surface model.

[0095] Furthermore, since the position of each point cloud can be directly obtained based on the point cloud data, the stain area of the stain area and the building surface area of the building to be cleaned can be directly calculated. By calculating the maximum distance difference between the stain area and the building surface, the depth value of the stain area can be determined, thereby reflecting the stubbornness of the stain to a certain extent.

[0096] Further, since the proportion of different stain areas and the distance difference between the stain area and the building surface indicate different levels of stain stubbornness, this embodiment can preset the stain stubbornness levels corresponding to different area proportions and distance differences, so as to directly determine the target stain stubbornness level corresponding to the area proportion and distance difference of the current stain area, and use the target stain type, the target stain center position, and the target stain stubbornness level together as the stain information of the stain.

[0097] In addition, this embodiment can further consider other quantitative indicators. For example, color contrast, that is, calculate the ΔE value between the stain area and the building surface in the LAB color space. The ΔE value reflects the degree of difference between the two colors. The larger the ΔE value, the higher the color contrast. Texture complexity, that is, calculate parameters such as energy and entropy based on the gray-level co-occurrence matrix. Energy reflects the uniformity of the gray-scale distribution of the image, and entropy represents the randomness of the image. These parameters can quantify the texture complexity of the stain area.

[0098] As a preferred solution, determining the current dry ice spraying parameters according to the stain information specifically includes:

[0099] Determine the current dry ice spraying pressure, dry ice particle density, and dry ice spraying time according to the preset dry ice spraying parameters corresponding to different stain types and stain stubbornness levels;

[0100] Determine the current dry ice spraying angle according to the target stain center position;

[0101] Determine the current dry ice spraying parameters according to the dry ice spraying pressure, the dry ice particle density, the dry ice spraying time, and the dry ice spraying angle.

[0102] Specifically, due to different stain types and stain stubbornness levels, different stains have different characteristics, which determines that the dry ice spraying parameters are also different, so as to achieve the cleaning effect. Therefore, this embodiment presets the dry ice spraying parameters corresponding to different stain types and stain stubbornness levels. Taking the stain type as an example, the dry ice spraying pressure and dry ice spraying density corresponding to different stain types can be set, as Figure 2 shown.

[0103] Furthermore, the central position of the identified stain in the image has been determined, usually represented by pixel coordinates. For example, the center of the image is taken as the origin, or a specific point is used as a reference. Then, the position in this two-dimensional image is converted into a position in three-dimensional space, and combined with the position and orientation of the current dry cleaning device, the angle that the dry ice injection nozzle needs to be adjusted is calculated. Specifically, first, the pixel coordinates in the image need to be converted into three-dimensional coordinates in the camera coordinate system. This involves the internal parameters of the camera (such as focal length, principal point coordinates) and external parameters (the position and orientation of the camera relative to the base of the dry cleaning device). These parameters can be obtained through camera calibration. Then, the three-dimensional points in the camera coordinate system need to be converted to the base coordinate system of the dry cleaning device. This requires knowing the position and orientation of the camera relative to the base of the dry cleaning device, that is, the external parameter matrix. Next, the points in the base coordinate system of the dry cleaning device need to be converted to the coordinate system of the dry ice injection nozzle. The position and orientation of the dry ice injection nozzle relative to the base of the dry cleaning device are known, or can be calculated through the kinematic model of the base of the dry cleaning device. Once the position of the center point of the target stain in the coordinate system of the dry ice injection nozzle is determined, the angle that the dry ice injection nozzle needs to be adjusted can be calculated through inverse kinematics, so that the injection direction is aligned with the center position of the target stain.

[0104] As a preferred solution, the dry ice injection for the stain based on the dry ice injection parameters specifically includes:

[0105] Adjust the current dry ice injection nozzle according to the dry ice injection angle so that the dry ice injection nozzle is aligned with the center position of the target stain;

[0106] Determine the stain density at different positions according to the stain probability heat map;

[0107] According to the preset cleaning area boundary information and motion constraint parameters, starting from the center position of the target stain, perform path planning in the order of decreasing stain density to obtain the dry ice injection trajectory;

[0108] Perform dry ice injection using the dry ice injection nozzle according to the dry ice injection pressure, the dry ice particle density, and the dry ice injection time, and move the dry ice injection nozzle along the dry ice injection trajectory.

[0109] Specifically, in this embodiment, a coverage path (such as a spiral shape or a grid method) is generated based on the stain distribution, and the cleaning priority areas are divided based on the stain density. The areas with severe stains are preferentially processed. It can be understood that the boundary information of the cleaning area is the vertex coordinates of the polygon area, which defines the current dry ice spraying area. For the determination of the stain density, it can be determined according to the proportion of the stain area within each preset area. The motion constraint parameters include the maximum speed, acceleration, and turning radius. Thus, starting from the current target stain center position, path planning is carried out in the order of decreasing stain density to obtain the dry ice spraying trajectory.

[0110] As a preferred solution, after dry ice is sprayed on the stains based on the dry ice spraying parameters, the method further includes:

[0111] Recycling the mixed gas within the dry ice spraying area; wherein, the mixed gas includes dry ice particles, dirt, and carbon dioxide;

[0112] Performing a primary separation process on the mixed gas through a cyclone separator to separate and obtain the dirt and the dry ice particles from the mixed gas, and respectively collecting the dirt and the dry ice particles in a preset dirt storage tank and a dry ice storage tank;

[0113] Performing a secondary separation process on the remaining mixed gas after the primary separation process through an electrostatic precipitator to separate and obtain the carbon dioxide from the remaining mixed gas, and liquefying the carbon dioxide and collecting it in a preset carbon dioxide storage tank, or adsorbing and storing the carbon dioxide through a molecular sieve.

[0114] Specifically, as Figure 3 shown, this embodiment proposes a hierarchical recycling method based on a closed cycle to recycle the mixed gas within the dry ice spraying area, reduce resource consumption, and avoid excessive carbon dioxide emissions. First, it is difficult to distinguish between dry ice particles and dirt after mixing. The cyclone separator is used for gas-solid separation. Based on the principle of density difference (the density of dry ice is 1.5 g / cm 3 , and dirt is usually lighter), large particle dirt and dry ice debris respectively fall into the dirt storage tank and the dry ice storage tank, and micron-sized particles are filtered. The remaining mixed gas enters the electrostatic precipitator, which can capture micron-sized particles with a size range of 0.1 micron to 50 microns, thereby separating carbon dioxide. The separated carbon dioxide is recycled through a compression / adsorption system. The recycled carbon dioxide can be remade into dry ice (requiring external equipment) or directly used for secondary spraying (gaseous assisted cleaning).

[0115] In addition, this embodiment can also perform activated carbon filtration on the recycled carbon dioxide through a filtration and purification device to remove odors and volatile organic compounds (VOCs).

[0116] As a preferred solution, the method further includes:

[0117] When it is recognized that there is no stain in the building surface image, calculate the distance between the current position and the edge of the building surface of the building to be cleaned;

[0118] When the distance is greater than a first preset distance, move along the normal of the building surface of the building to be cleaned towards the edge of the building surface, and calculate the distance in real time; when it is detected that the distance remains unchanged within a preset time period, retreat a preset distance, then switch the moving angle and continue to move towards the edge of the building surface until it is detected that the distance becomes smaller within the preset time period, re-acquire the building surface image of the building to be cleaned, and perform stain recognition on the building surface image;

[0119] When the distance is less than or equal to the first preset distance and greater than or equal to a second preset distance, move towards the edge of the building surface at the approaching speed corresponding to the building surface material of the building to be cleaned as preset, and update the distance at a frequency of 10 Hz during the movement, perform multi-spectral scanning on the building surface to re-acquire the building surface image of the building to be cleaned, and perform stain recognition on the building surface image;

[0120] When the distance is less than the second preset distance, move towards the edge of the building surface until the distance reaches a preset minimum distance, use the contact sensor array to detect the change in the pressure distribution on the building surface, and calculate the standard deviation of the contact pressure on the building surface based on the change in the pressure distribution; when the standard deviation of the contact pressure is greater than a preset pressure standard deviation threshold, emit a vibration signal to the building surface to identify whether there is a stain on the building surface; if there is, re-acquire the building surface image of the building to be cleaned, and perform stain recognition on the building surface image; if not, perform dry ice spraying on the building surface according to the preset dry ice cleaning parameters.

[0121] Specifically, when no stains are detected in the building surface image, this embodiment adopts a distance-based control strategy for movement. When the distance between the current position and the edge of the building surface to be cleaned is greater than the first preset distance (e.g., 50 cm), it indicates that the current distance is relatively far, and the core goal should be to quickly approach the edge and avoid path deviation. At this time, the control priority of navigation accuracy is higher than that of cleaning efficiency. Therefore, the map is corrected through the SLAM algorithm, the straight-line distance between the current position and the edge of the building surface to be cleaned is set as the navigation path, and the edge approximation mode is enabled (reducing the obstacle avoidance sensitivity to 30%, and preferentially moving along the normal direction of the wall). The lidar is used to detect the large-scale wall contour, and the infrared sensor compensates for the edge angle deviation. If the detected distance remains unchanged for 5 consecutive seconds, the anti-jamming mechanism is triggered: after retreating 1 m, switch to an oblique approach at 45°, until the detected distance becomes smaller within 5 seconds, re-acquire the building surface image, and perform stain recognition on the building surface image.

[0122] When the distance is less than or equal to the first preset distance (e.g., 50 cm) and greater than or equal to the second preset distance (e.g., 10 cm), the core goal at this time should be to precisely adjust the posture and prepare for contact detection. The control priority of sensor calibration is higher than that of the control priority of the moving speed. Switch to the millimeter-wave radar + visual servo system to update the distance data at a frequency of 10 Hz. The approaching speed is dynamically adjusted according to the building surface material. It can be understood that for some relatively fragile surface materials, such as gypsum materials, a relatively low approaching speed is preset to prevent wall damage caused by too high a speed; if it is a solid surface material, such as concrete material, a relatively high approaching speed can be set to improve the cleaning efficiency. An extendable and retractable camera (3 cm - 5 cm away from the wall) is extended to perform multispectral scanning. If a suspected stain (such as an area with abnormal reflection) is found, it is marked as a point to be confirmed.

[0123] When the distance is less than the second preset distance (e.g., 10 cm), it indicates that the current distance is relatively close, and the core goal is to maintain stable contact. Depth scanning is started, and the control priority of mechanical stability is higher than that of energy consumption. Move towards the edge of the building surface until the distance reaches the preset minimum distance. The contact sensor array (such as piezoelectric film, conductive rubber) is used to detect the change in the pressure distribution on the building surface. When the standard deviation of the contact pressure is greater than the preset pressure standard deviation threshold (e.g., 15%), it indicates that the surface is uneven, and the high-frequency vibration mode (50 Hz) is started to assist in detecting stains. Even if no stains are recognized, a cleaning spray is still performed once according to the preset dry ice cleaning parameters. It can be understood that the dry ice cleaning parameters can set fixed and general dry ice spraying pressure, dry ice particle density, and dry ice spraying time, and a fixed dry ice spraying path can be set, such as a horizontal path, a vertical path, a spiral path.

[0124] Therefore, the overall dry ice cleaning process can be asFigure 4 as shown

[0125] As a preferred solution, the method further includes:

[0126] real-time detection of the remaining amount of dry ice in the dry ice storage tank;

[0127] When it is detected that the remaining amount of dry ice is less than a preset remaining amount threshold, an alarm message indicating insufficient remaining amount is issued.

[0128] Specifically, in this embodiment, a sensor is arranged in the dry ice storage tank to real-time detect the remaining amount of dry ice in the dry ice storage tank. When it is detected that the remaining amount of dry ice is less than the preset remaining amount threshold, an audible and visual alarm signal indicating insufficient remaining amount can be emitted through a configured audible and visual alarm, or an alarm message indicating insufficient remaining amount can be sent to relevant terminal devices through communication technology, so that relevant personnel can replace the dry ice storage tank in time.

[0129] Please refer to Figure 5 , a second aspect of the embodiments of the present invention provides a dry ice cleaning device, including a housing 1, a moving module 2, a storage module (not shown in the figure), a recovery module (not shown in the figure), a telescopic arm module 4, a dry ice spraying module 5, and a control module (not shown in the figure);

[0130] The moving module 2 is arranged on both sides of the housing 1; the storage module, the recovery module, and the telescopic arm module 4 are all arranged on the housing 1;

[0131] The dry ice spraying module 5 is arranged at the end of the telescopic arm module 4. The dry ice spraying module 5 includes a high-pressure air generating module, a dry ice spraying nozzle, and a vacuum adsorption module; the input end of the dry ice spraying nozzle is connected to the gas output end of the high-pressure air generating module, and the input end of the dry ice spraying nozzle is connected to the recovery module through a dry ice conveying pipeline 3; the vacuum adsorption module is arranged close to the dry ice spraying nozzle, and the vacuum adsorption module is connected to the recovery module;

[0132] The storage module includes a dirt storage tank, a dry ice storage tank, and a carbon dioxide storage tank. The recovery module has a dirt separation port, a dry ice separation port, and a carbon dioxide separation port. The dirt separation port, the dry ice separation port, and the carbon dioxide separation port are respectively connected to the dirt storage tank, the dry ice storage tank, and the carbon dioxide storage tank;

[0133] The control module is arranged in the housing 1, and the control end of the control module is respectively connected to the controlled ends of the moving module 2, the storage module, the recovery module, the telescopic arm module 4, and the dry ice spraying module 5; the control module is used to execute the dry ice cleaning method described in any one of the embodiments in the first aspect.

[0134] Specifically, the dry ice storage tank is further composed of a vacuum insulation storage tank and several modular storage tanks. The modular storage tanks play a role in precisely controlling the dry ice injection volume. They are connected to the dry ice separation port of the recovery module to receive the recovered dry ice particles, and their delivery ports are connected to the interior of the vacuum insulation storage tank. Granular dry ice is stored in the vacuum insulation storage tank, which is connected to the dry ice injection nozzle through the dry ice delivery pipeline 3. The inner wall of the delivery pipeline needs to be treated for anti-static, such as coating with a conductive / anti-static material coating to reduce the resistivity, conduct static charges, and prevent dry ice caking.

[0135] Furthermore, the high-pressure air generation module generates high-pressure air flow (pressure range 5 bar - 15 bar) through an air compressor to push the dry ice particles to impact the wall at high speed. The dry ice injection nozzle is designed to be angle-adjustable, so as to flexibly control the dry ice injection trajectory and coverage. The dry ice injection nozzle connects the high-pressure air generation module and the dry ice delivery pipeline 3, and sprays out the mixture of high-pressure air and dry ice particles. When the dry ice particles impact the wall, they instantly sublimate, generating a micro-explosion effect to peel off dirt. Combining with the low-temperature embrittlement effect, pollutants such as oil stains and coatings are separated from the wall. Different injection nozzles are intelligently adjusted according to different cleaning requirements for dry ice injection cleaning to achieve the best cleaning effect. For example, a round nozzle is used when small-area fine treatment of stains is required, and a flat nozzle is used when large-area rapid cleaning is required.

[0136] Furthermore, the vacuum adsorption module is arranged close to the dry ice injection nozzle, so that during dry ice cleaning, the remaining mixed gas after cleaning can be adsorbed and recycled to the recovery module at the same time. It uses a negative pressure suction port and a vacuum pump to recover the un-sublimated dry ice particles, dirt, and carbon dioxide gas. The recovery module further includes a cyclone separator that uses centrifugal force to separate large-particle dirt and dry ice debris, and an electrostatic precipitator that captures micron-sized particles (such as dust). In addition, a mechanical sorting mechanism is also set in the recovery module, including an air lock ash discharge valve, an eddy current separator, and a vibrating screen. Among them, the air lock ash discharge valve alternately opens the dirt separation port and the dry ice separation port through a rotary vane valve (5 rpm - 15 rpm), so that the leakage rate of dirt in the dry ice storage tank is less than 0.1%; the eddy current separator repels non-magnetic metal pollutants with an alternating magnetic field (frequency of 50 Hz), so that the metal removal rate is greater than 99%; the vibrating screen passes through a double-layer screen (aperture 1 mm / 3 mm) to ensure that the dry ice particles return to the dry ice storage tank and the dirt returns to the dirt storage tank, and the screening efficiency is greater than 95%. In addition, this embodiment is also equipped with a sensor network for anti-mixing, including an infrared spectrometer, a capacitive level gauge, and a laser particle size analyzer. Among them, the infrared spectrometer is set at the outlet of the cyclone separator to detect the dry ice content in real time, the capacitive level gauge is set inside each modular storage tank to monitor the dry ice storage volume, and the laser particle size analyzer is set at the inlet of the electrostatic precipitator to classify micron-sized particles.

[0137] Furthermore, a filtering and purification device is fixedly installed inside the carbon dioxide storage tank to perform activated carbon filtration on the recovered carbon dioxide gas to remove odors and volatile organic compounds (VOCs).

[0138] Furthermore, the control module can be powered by a lithium battery pack, and a solar panel (lightweight flexible photovoltaic film) is fixedly covered on the surface of the housing 1 as an auxiliary power source for the control module.

[0139] Furthermore, the mobile module 2 can further include crawlers, motors, drive gears, and combined gears. The motors operate to drive the crawlers and drive gears to rotate at high speeds, providing strong and stable friction, enabling the dry ice cleaning device to move vertically, horizontally, or obliquely along the ground. The telescopic arm module 4 is used to clean areas at different heights.

[0140] The dry ice cleaning device provided by the embodiment of the present invention identifies stains by analyzing the building surface images of the building to be cleaned, and after detecting the presence of stains, performs dry ice cleaning based on the stain information. Since dry ice cleaning does not require the use of water and does not need to contact the building surface with mechanical tools, it can quickly remove stains on the building surface without causing secondary pollution or damaging the building surface, thus taking into account both the cleaning effect and the integrity of the building surface.

[0141] In addition, based on the closed-loop recycling method, the mixed gas in the dry ice spraying area is recycled to reduce resource consumption and avoid excessive carbon dioxide emissions.

[0142] In an alternative embodiment, this embodiment can also provide a dry ice cleaning system, as Figure 6 shown. Preferably, the dry ice cleaning system can include a dry ice cleaning device and a server. The sensors of the dry ice cleaning device include cameras, ultrasonic sensors, infrared sensors, wall type sensors, environmental sensors, obstacle sensors, collision sensors, etc., and are connected to the main control board.

[0143] Preferably, the configuration server of the dry ice cleaning system selects appropriate server hardware, such as CPU, memory, disk, and network interface, etc., and installs an appropriate operating system, such as Linux. Then, according to the requirements of the dry ice cleaning system, appropriate server software, such as ROS (operating system for cleaning devices), OpenCV (computer vision library), programming languages (such as Python, C++), etc., are installed.

[0144] Preferably, the control program of the dry ice cleaning device is written in a programming language. According to the dry ice cleaning requirements and the data of the sensors, appropriate algorithms are written to implement the dry ice cleaning function, including path planning algorithms, obstacle avoidance algorithms, dynamic adjustment, dry ice residue monitoring, remote control, etc.

[0145] Preferably, the dry ice cleaning device and the server are connected via Bluetooth or wired connection to link the dry ice cleaning device to the server, ensuring that the sensors and actuators of the dry ice cleaning device can be controlled by the server and the sensor data can be transmitted to the server.

[0146] Preferably, the realization of the basic functions of the dry ice cleaning device includes dry ice quantitative spraying, path movement, emergency stop, obstacle avoidance, automatic filling of the storage tank, and remote control. Among them, for dry ice quantitative spraying, the parameters such as the angle and pressure of the dry ice spraying nozzle, the conveying speed and spraying time of dry ice particles can be adjusted through the program, and the working mode of the dry ice cleaning device is adjusted according to the wall material and dirt degree feedback by the sensor data to adapt to different cleaning objects and environmental requirements; for path movement, the movement control of the dry ice cleaning device is realized through the program, including basic actions such as forward, backward, turning, left and right movement. At the same time, the map drawing and navigation functions of the dry ice cleaning device can be realized, and a cleaning plan or a specified cleaning area can be formulated; for emergency stop, all operations of the dry ice cleaning device can be immediately stopped in case of an emergency through the program; for obstacle avoidance, the obstacle avoidance function of the dry ice cleaning device is realized through the program. When the dry ice cleaning device detects an obstacle, it will automatically stop or bypass the obstacle; for automatic filling of the storage tank, considering the continuity of dry ice supply, dry ice is replenished without interrupting the cleaning according to the feedback of the dry ice sensor and the remaining amount sensor, and the dry ice recovery, rapid replacement or automatic filling of the modular storage tank are realized through the program; for remote control, the real-time monitoring and control functions of the dry ice cleaning device are realized by remotely accessing the server.

[0147] The above are the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art of this technology, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements are also regarded as the protection scope of the present invention.

Claims

1. A dry ice cleaning method, characterized in that: include: Acquire a building surface image of a building to be cleaned, and perform stain recognition on the building surface image; When stains are identified in the building surface image, obtaining stain information of the stains; Determining current dry ice blasting parameters according to the stain information; The stain is dry ice blasted based on the dry ice blasting parameters.

2. The dry ice cleaning method according to claim 1, characterized in that: The step of obtaining a building surface image of a building to be cleaned and identifying stains on the building surface image specifically includes: Acquire a multispectral image of the building surface of the building to be cleaned, and use the multispectral image as the building surface image; Using a pre-trained stain recognition model to perform stain recognition on the multispectral image to obtain a stain recognition result; wherein the stain recognition model is obtained by training a multi-channel convolutional neural network using a multispectral training image with stain marking information; When it is determined based on the stain recognition result that any type of stain exists, it is determined that stains exist in the building surface image.

3. The dry ice cleaning method according to claim 2, characterized in that: The using a pre-trained stain recognition model to perform stain recognition on the multispectral image to obtain a stain recognition result specifically includes: Performing feature extraction on the multispectral image to obtain multimodal stain features; Using the CBAM module to enhance the multimodal stain feature to obtain an enhanced stain feature; Using a classifier to perform stain recognition on the enhanced stain features to generate a stain probability heat map; Determining a stain bounding box and a target stain type according to the stain probability heat map; A preset number of candidate points are extracted within the stain boundary box using the FAST corner point detection algorithm, and a number of stain feature points are obtained based on the gradient amplitude of each of the candidate points; Iteratively optimizing the positions of a plurality of the stain feature points based on a grayscale gradient method until the position change of the stain feature point is less than a preset pixel value, and determining the target stain center position according to the centroid of each stain feature point after iterative optimization; The stain recognition result is obtained according to the target stain type and the target stain center position.

4. The dry ice cleaning method according to claim 3, characterized in that: The obtaining of the stain information of the stain specifically includes: Scanning the building surface of the building to be cleaned by using a laser radar to construct a three-dimensional building surface model; Determine the stain area in the three-dimensional building surface model according to the point cloud data in the three-dimensional building surface model and the positions of each stain feature point after iterative optimization; Calculating the stain area of ​​the stain area and the building surface area of ​​the building to be cleaned according to the point cloud data and the stain area, and calculating the maximum distance difference between the stain area and the building surface; Calculating the area ratio of the stained area according to the stain area and the building surface area, and determining the target stain stubbornness level corresponding to the area ratio of the stained area and the maximum distance difference based on the preset different area ratios and distance differences corresponding to the stain stubbornness levels; The stain information of the stain is determined according to the target stain type, the target stain center position and the target stain stubbornness level.

5. The dry ice cleaning method according to claim 4, characterized in that: Determining the current dry ice blasting parameters according to the stain information specifically includes: Determine the current dry ice blasting pressure, dry ice particle density and dry ice blasting time according to the preset dry ice blasting parameters corresponding to different stain types and stain stubbornness levels; Determining a current dry ice blasting angle according to the center position of the target stain; The current dry ice blasting parameters are determined according to the dry ice blasting pressure, the dry ice particle density, the dry ice blasting time, and the dry ice blasting angle.

6. The dry ice cleaning method according to claim 5, characterized in that: The performing dry ice blasting on the stain based on the dry ice blasting parameters specifically includes: According to the dry ice blasting angle, the current dry ice blasting nozzle is adjusted so that the dry ice blasting nozzle is aligned with the center position of the target stain; Determining the stain density at different locations according to the stain probability heat map; According to the preset cleaning area boundary information and motion constraint parameters, taking the center position of the target stain as the starting point, path planning is performed in the order of the stain density from large to small to obtain the dry ice blasting trajectory; According to the dry ice blasting pressure, the dry ice particle density and the dry ice blasting time, the dry ice blasting nozzle is used to perform dry ice blasting, and the dry ice blasting nozzle is moved according to the dry ice blasting trajectory.

7. The dry ice cleaning method according to claim 1, wherein: After performing dry ice blasting on the stain based on the dry ice blasting parameters, the method further includes: Recovering the mixed gas in the dry ice blasting area; wherein the mixed gas includes dry ice particles, dirt and carbon dioxide; The mixed gas is subjected to a separation process by a cyclone separator to separate the dirt and the dry ice particles from the mixed gas, and the dirt and the dry ice particles are collected in a preset dirt storage tank and a dry ice storage tank respectively; The residual mixed gas after the primary separation treatment is subjected to a secondary separation treatment by an electrostatic precipitator to separate the carbon dioxide from the residual mixed gas, and the carbon dioxide is liquefied and collected in a preset carbon dioxide storage tank, or the carbon dioxide is adsorbed and stored by a molecular sieve.

8. The dry ice cleaning method according to claim 1, wherein: The method further comprises: When it is recognized that there is no stain in the building surface image, calculating the distance between the current position and the edge of the building surface of the building to be cleaned; When the distance is greater than a first preset distance, move toward the edge of the building surface along the normal of the building surface to be cleaned, and calculate the distance in real time; when it is detected that the distance remains unchanged within a preset time period, switch the moving angle after retreating the preset distance and continue to move toward the edge of the building surface until it is detected that the distance becomes smaller within the preset time period, reacquire the building surface image of the building to be cleaned, and perform stain recognition on the building surface image; When the distance is less than or equal to the first preset distance and greater than or equal to the second preset distance, move toward the edge of the building surface according to the preset approach speed corresponding to the building surface material of the building to be cleaned, and update the distance at a frequency of 10 Hz during the movement, perform multi-spectral scanning on the building surface to reacquire the building surface image of the building to be cleaned, and perform stain recognition on the building surface image; When the distance is less than the second preset distance, move toward the edge of the building surface until the distance reaches the preset minimum distance, use the contact sensor array to detect the pressure distribution change on the building surface, and calculate the contact pressure standard deviation of the building surface based on the pressure distribution change; when the contact pressure standard deviation is greater than the preset pressure standard deviation threshold, emit a vibration signal to the building surface to identify whether there is stain on the building surface; if so, reacquire the building surface image of the building to be cleaned, and perform stain identification on the building surface image; if not, perform dry ice blasting on the building surface according to preset dry ice cleaning parameters.

9. The dry ice cleaning method according to claim 7, characterized in that: The method further comprises: Real-time detection of the dry ice remaining in the dry ice storage tank; When it is detected that the dry ice remaining amount is less than a preset remaining amount threshold, an insufficient remaining amount alarm message is issued.

10. A dry ice cleaning device, characterized in that: It includes a housing, a mobile module, a storage module, a recovery module, a telescopic arm module, a dry ice blasting module and a control module; The mobile module is arranged on both sides of the shell; the storage module, the recovery module and the telescopic arm module are all arranged on the shell; The dry ice blasting module is arranged at the end of the telescopic arm module, and the dry ice blasting module includes a high-pressure air generating module, a dry ice blasting nozzle and a vacuum adsorption module; the input end of the dry ice blasting nozzle is connected to the gas output end of the high-pressure air generating module, and the input end of the dry ice blasting nozzle is connected to the recovery module through a dry ice delivery pipeline; the vacuum adsorption module is arranged close to the dry ice blasting nozzle, and the vacuum adsorption module is connected to the recovery module; The storage module comprises a dirt storage tank, a dry ice storage tank and a carbon dioxide storage tank, and the recovery module comprises a dirt separation port, a dry ice separation port and a carbon dioxide separation port, and the dirt separation port, the dry ice separation port and the carbon dioxide separation port are respectively connected to the dirt storage tank, the dry ice storage tank and the carbon dioxide storage tank; The control module is arranged in the shell, and the control end of the control module is respectively connected to the controlled end of the moving module, the controlled end of the storage module, the controlled end of the recovery module, the controlled end of the telescopic arm module and the controlled end of the dry ice blasting module; the control module is used to execute the dry ice cleaning method according to any one of claims 1 to 9.

Images