Soft stool recognition and cleaning method, system and intelligent cat litter box
By using a smart litter box to identify and clean soft feces, and employing image acquisition and rotational motion technology, the problem of soft feces adhesion is solved, achieving efficient cleaning, reducing maintenance costs, and improving the user experience.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANGHAI LIANCHONG INTELLIGENT TECH CO LTD
- Filing Date
- 2026-04-16
- Publication Date
- 2026-06-09
Smart Images
Figure CN122162713A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of pet product technology, and in particular to a soft stool recognition and cleaning method, system and intelligent cat litter box. Background Technology
[0002] With the increasing popularity of pet ownership, smart litter boxes, as automated cleaning devices, are gradually becoming common appliances in multi-cat households. Existing smart litter boxes typically use methods such as roller turning, sieving, or conveyor belts to separate cat litter from excrement and automatically collect it into a collection bin, reducing the frequency of manual cleaning. However, in actual use, cats often excrete soft stools due to diet, stress, or health problems; these stools are characterized by high water content, high viscosity, and unstable consistency. Existing cleaning methods have significant shortcomings when dealing with soft stools: Soft feces easily adhere to the inner walls of the ball chamber / roller, the feces passage, the screen, or the tumbling path, forming stubborn residue. This residue is not only difficult to remove automatically during subsequent cleaning, but it can also be stepped on by cats using the litter box later, sticking to their paws or fur, leading to contamination and the spread of bacteria. Users are forced to frequently disassemble and clean the ball chamber and passage, significantly increasing maintenance costs and severely impacting the user experience.
[0003] Furthermore, in multi-cat households, a lengthy cleaning process can lead to the litter box being occupied, affecting other cats' normal toilet use and potentially causing elimination disorder or inappropriate defecation. Therefore, cleaning time should be minimized to avoid conflicts arising from cleaning.
[0004] Therefore, there is an urgent need for a soft stool identification and cleaning method to reduce residual pollution, shorten cleaning time, and improve the user experience for both users and pets. Summary of the Invention
[0005] The purpose of this invention is to overcome the problem that existing soft stool cleaning relies on a single physical method and cannot dynamically adjust the strategy according to the state of the excrement, resulting in soft stool residue. This invention provides a soft stool identification and cleaning method that can automatically switch to a high-efficiency anti-adhesion cleaning mode when a risk of soft stool is detected, thereby reducing residual pollution, shortening cleaning time, and improving the user and pet experience.
[0006] In a first aspect, embodiments of this application provide a method for soft stool identification and cleaning, including: Collect images of excrement; The excrement image is processed for soft stool recognition to obtain the soft stool recognition result; When the soft stool identification result indicates soft stool, perform the soft stool cleaning operation; The soft stool cleaning procedure includes: Control the excrement to reciprocate and rotate along a preset trajectory, so that the cat litter forms a continuous coating layer on the outer periphery of the excrement, thus obtaining cat litter clumps; Control the cat litter clump to rotate to the preset dropping angle, so that the cat litter clump falls into the collection bin.
[0007] In some embodiments, before controlling the excrement to reciprocate along a preset trajectory, the method further includes: Identify areas of excrement exposed on the surface of the litter box; The target coverage area is determined based on the area of excrement, and the target coverage area includes the area of excrement. The litter dispenser controls the litter to spread cat litter over the target area.
[0008] In some embodiments, the soft stool identification result includes: soft stool determination result and soft stool severity level; the soft stool severity level includes at least mild soft stool, moderate soft stool and severe soft stool; Controlling the litter dispenser to spread cat litter over the target area includes: Based on the level of soft stool, determine the lower limit of the litter cover thickness corresponding to that level; Obtain the area of the target coverage area, and determine the amount of cat litter to be spread based on the area and the minimum thickness. Control the litter dispenser to distribute cat litter to the target coverage area at the specified amount.
[0009] In some embodiments, the preset trajectory is: rotating forward to a first angle and rotating backward to a second angle; Controlling the excrement to reciprocate and rotate along a preset trajectory includes: Control the excrement to reciprocate rotational motion between the first angle and the second angle; The first angle and the second angle are greater than the angle that causes the cat litter to begin to slide, and the first angle is less than the angle that causes the cat litter to slide onto the filter.
[0010] In some embodiments, the first angle is greater than 30° and less than 60°; the second angle is greater than 30°.
[0011] In some embodiments, the reciprocating rotational motion includes: At least once, rotate the excrement to a stopping position, which is a position where the excrement is exposed above the litter layer and separated from the litter, so as to capture images of the fecal clump; After a first preset period of reciprocating rotation, the cat litter forms a continuous coating layer around the excrement, resulting in a cat litter clump.
[0012] In some embodiments, acquiring images of excrement includes: Images of pet excrement were collected at multiple stages during the toileting process, including: The first phase involves capturing the first images of pets using the toilet. The second phase involved collecting a second image of the pet from the time it used the toilet until its excrement was covered. The third stage involves collecting images from the time the pet excrement is covered until the automatic cleaning process. Soft stool recognition processing is performed on excrement images to obtain soft stool recognition results, including: Construct a soft stool recognition model; Input at least one of the first image, the second image, and the third image into the soft stool recognition model; The processing steps of the soft stool recognition model include: The images at each stage are analyzed to obtain the corresponding soft stool identification sub-results for each stage; Based on the preset confidence weights of the images at each stage, the soft stool recognition sub-results are weighted and fused to output the soft stool recognition result.
[0013] In some embodiments, images of excrement are captured at multiple stages during a pet's toileting process, including: In the first phase, a first set of images is acquired from the time the pet enters the smart litter box until it finishes using the litter box, and a first image is selected from the first set of images; the first image is an image in which excrement is directly exposed. In the second stage, a second set of images was acquired from the end of the pet's toilet training until the excrement was covered, and a second image was selected from the second set of images; the second image is an image of the excrement initially mixed with cat litter and not yet covered. In the third stage, a third image set is obtained from the time the excrement is covered until the automatic cleaning, and a third image is selected from the third image set. The third image is the image of the excrement after it has been covered by cat litter.
[0014] In some embodiments, acquiring images of excrement at multiple stages during a pet's toileting process further includes: When a pet is detected, it is determined that the pet has entered the smart litter box, and the first stage of image acquisition is initiated. When the pet changes from a defecation posture to a standing posture, the toileting is considered complete, the first stage of image acquisition is terminated, and the second stage of image acquisition is started. When excrement is detected exposed on the surface of the litter layer, the litter adding device is controlled to add litter to the surface of the excrement until the excrement is covered by litter, the second stage of image acquisition is terminated, and the third stage of image acquisition is started. The litter-covered excrement is rotated to a preset shooting angle, and after a second preset time, the third stage of image acquisition is terminated.
[0015] In some embodiments, an image acquisition device is installed at a preset location in the smart litter box, and the preset location includes at least one of the following: Image acquisition point located at the entrance of the smart litter box; Two image acquisition points are located above the entrance of the smart litter box and on the inside side directly opposite the entrance; Image acquisition points located on the rotating axis of the inner wall of the smart litter box.
[0016] In some embodiments, the confidence weights are set according to at least one of the following rules: When the first image, the second image, and the third image are obtained, the confidence weight of the second image shall not be less than 50%, and the confidence weight of the third image shall not be greater than 20%. When only the first and third images are obtained, the confidence weight of the first image shall not be less than 50%.
[0017] When only the second and third images are obtained, the confidence weight of the second image shall not be less than 70%.
[0018] When only the third image is obtained, its confidence weight is 100%.
[0019] Secondly, embodiments of this application provide a soft stool identification and cleaning system for implementing the aforementioned soft stool identification and cleaning method, including: The image acquisition module is used to acquire images of excrement; The soft stool recognition module is connected to the image acquisition module and is used to perform soft stool recognition processing on excrement images to obtain soft stool recognition results. The control module, connected to the soft stool recognition module, is used to perform a soft stool cleaning operation when the soft stool recognition result indicates that the stool is soft. The litter dispenser, connected to the control module, is used to dispense cat litter onto the surface of excrement under the control of the control module. A rotary drive device, connected to a control module, is used to drive the excrement to reciprocate along a preset trajectory under the control of the control module, so that the cat litter forms a continuous coating layer on the outer periphery of the excrement, thus obtaining a cat litter ball. The rotary drive device is also used to rotate the cat litter clump to a preset dropping angle under the control of the control module, so that the cat litter clump falls into the collection bin.
[0020] Thirdly, embodiments of this application provide a smart litter box, comprising: Basin; An image acquisition device is installed at a preset position on the basin to acquire images of excrement; A litter dispenser, located on the basin, is used to dispense cat litter onto the surface of excrement. The rotating mechanism, located inside the basin, is used to carry and drive the excrement or fecal bolus to reciprocate and rotate. The litter box, located on one side of the basin, is connected to the basin via a litter channel and is used to collect cat litter pellets. The controller is electrically connected to the image acquisition device, the sand adding device, and the rotating mechanism, and is configured to perform the aforementioned soft stool identification and cleaning method.
[0021] Upon detecting soft feces, this application first applies cat litter to the surface of the excrement to prevent the soft feces from directly contacting and adhering to the inner wall of the litter box, the rotation path, and the screen during subsequent reciprocating rotation. Then, through the reciprocating rotation, the cat litter forms a continuous coating layer around the excrement, completely encapsulating the highly viscous, high-moisture soft feces into a litter clump with a certain structural strength. This coating layer effectively isolates the soft feces from direct contact with the internal structure of the device, preventing not only stubborn residue caused by fecal adhesion but also secondary pollution caused by pets stepping on or contaminating the feces.
[0022] This application determines the minimum litter coverage thickness based on the degree of soft stool and calculates the amount of litter to be distributed based on the area of the target coverage area. Less litter is used for mild soft stool, while the amount of litter is automatically increased for severe soft stool, thus ensuring effective coverage while avoiding litter waste and achieving a balance between cleaning quality and material consumption.
[0023] This application utilizes images from three different stages during a pet's toilet training process to extract features of excrement under different states, and then performs weighted fusion using preset confidence levels. This approach overcomes the susceptibility of single images to factors such as occlusion, lighting changes, and litter color interference, significantly improving the accuracy and robustness of soft stool recognition. Furthermore, in the second stage, when excrement is detected exposed on the litter layer surface, litter is added until it is completely covered, effectively preventing direct contact between excrement and the internal structure of the device during subsequent rotation to a preset shooting angle. Attached Figure Description
[0024] Figure 1 This is a flowchart illustrating the soft stool identification and cleaning method according to an embodiment of this application; Figure 2 This is a schematic diagram of the structure of the smart litter box according to an embodiment of this application; Figure 3 This is a structural diagram of the litter box in the litter box state according to an embodiment of this application; Figure 4 This is a schematic diagram of the litter box in the non-rotating state according to an embodiment of this application; Figure 5 This is a schematic diagram of the structure of a cat litter box rotated to a first angle according to an embodiment of this application; Figure 6 This is a schematic diagram of the structure of the litter box rotated to the second angle according to an embodiment of this application; Figure 7 This is a schematic diagram of the structure of a litter box rotated to the desired angle for feces placement, according to an embodiment of this application.
[0025] Explanation of reference numerals in the attached figures: 1. Basin; 2. Roller; 3. Image acquisition device; 4. Sand adding device; 5. Stool collection chamber; 6. Stool discharge channel. Detailed Implementation
[0026] The following specific embodiments illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. Although the description of this application will be presented in conjunction with some embodiments, this does not mean that the features of this application are limited to this embodiment. On the contrary, the purpose of describing the application in conjunction with embodiments is to cover other options or modifications that may be derived based on the claims of this application. To provide a thorough understanding of this application, many specific details will be included in the following description. This application may also be implemented without using these details. Furthermore, to avoid confusion or obscuring the focus of this application, some specific details will be omitted in the description. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other.
[0027] It should be noted that in this specification, similar reference numerals and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.
[0028] In the description of this application, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this application 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 application. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.
[0029] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" 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 between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.
[0030] In the description of this application, it should be understood that "electrical connection" can be understood as physical contact and electrical conduction between components; it can also be understood as the form of connection between different components in a circuit structure through physical lines that can transmit electrical signals, such as copper foil or wires on a printed circuit board (PCB). "Coupled through..." can be understood as electrical conduction through indirect coupling. Indirect coupling can be understood as contactless coupling. Those skilled in the art will understand that coupling refers to the phenomenon where there is a close cooperation and mutual influence between the inputs and outputs of two or more circuit elements or electrical networks, and energy is transferred from one side to the other through interaction. To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be further described in detail below with reference to the accompanying drawings.
[0031] It's important to note that in actual use, pets often excrete soft stools with high water content, high viscosity, and unstable consistency due to diet, stress, or health issues. Existing cleaning methods have significant drawbacks when dealing with soft stools: soft stools easily adhere to the inner walls of the ball chamber or roller, the stool passage, the screen, and the tumbling path, forming stubborn residues. These residues are not only difficult to remove automatically during subsequent cleaning but can also be stepped on by pets later and adhere to their paws or fur, leading to the spread of bacteria. Users are forced to frequently disassemble and clean the ball chamber and passage, significantly increasing maintenance costs and severely impacting the user experience.
[0032] Therefore, this application provides a soft stool identification and cleaning method, which can be applied to smart cat litter boxes to automatically detect soft stool and perform targeted anti-adhesion cleaning operations.
[0033] like Figure 1-3 As shown in the embodiment of this application, a soft stool identification and cleaning method is provided, which includes: acquiring an image of excrement; performing soft stool identification processing on the excrement image to obtain a soft stool identification result; when the soft stool identification result indicates soft stool, performing a soft stool cleaning operation; the soft stool cleaning operation includes: controlling the excrement to reciprocate and rotate along a preset trajectory, so that the cat litter forms a continuous coating layer on the outer periphery of the excrement to obtain a cat litter clump; controlling the cat litter clump to rotate to a preset dropping angle, so that the cat litter clump falls into the collection bin.
[0034] In this embodiment, upon detecting soft feces, a reciprocating rotational motion causes the cat litter to form a continuous coating layer around the excrement, completely encapsulating the highly viscous, high-moisture soft feces into a litter clump with a certain structural strength. This coating layer effectively isolates the soft feces from direct contact with the internal structure of the device, not only avoiding stubborn residue caused by feces adhesion but also preventing secondary pollution caused by subsequent pet trampling or contamination.
[0035] In this embodiment of the application, when the soft stool identification result indicates soft stool, a soft stool cleaning operation is performed. This operation specifically includes the following sub-steps: Step 1: First, identify the areas of excrement exposed on the surface of the litter box.
[0036] Then, the target coverage area is determined based on the area of excrement. The target coverage area typically includes the area of excrement.
[0037] Finally, control the litter dispenser to spread cat litter over the target area.
[0038] In one implementation, the outline of the excrement is segmented from the current image using an image segmentation algorithm to obtain the boundary of the excrement area; the boundary of the excrement area is extended outward by 5-10mm to obtain the target coverage area; then, cat litter is evenly spread on the target coverage area using a spiral litter feeding mechanism or a vibrating litter spreader.
[0039] In this embodiment of the application, the soft stool identification result includes: soft stool determination result and soft stool severity level; the soft stool severity level includes at least mild soft stool, moderate soft stool and severe soft stool.
[0040] Furthermore, when covering the excrement with cat litter, it is necessary to determine the amount of litter to use. The specific method for determining this is as follows: Based on the level of soft stool, determine the lower limit of the litter cover thickness corresponding to that level.
[0041] Obtain the area of the target coverage region.
[0042] Calculate the amount of cat litter to be spread. The amount of litter to be spread is equal to the product of the area of the target coverage area and the lower limit of the thickness.
[0043] Optionally, if the density of the cat litter is known, the amount of litter spread can be further converted into the mass of cat litter. The litter dispenser is controlled to dispense cat litter to the target coverage area at a calculated amount.
[0044] By using the above method, cat litter is placed on the surface of the excrement before it reciprocates along a preset trajectory, preventing soft feces from directly contacting and adhering to the inner wall of the litter box, the rotation path, and the screen during subsequent reciprocating rotation. At the same time, less cat litter is used for mildly soft feces, and the amount of cat litter is automatically increased for severely soft feces, thus ensuring effective coverage while avoiding litter waste.
[0045] In one implementation, the minimum thickness for mildly soft stool is 3 mm, for moderately soft stool is 5 mm, and for severely soft stool is 8 mm.
[0046] In this step, the cat litter and excrement are initially mixed.
[0047] Step 2: Control the excrement to rotate back and forth along a preset trajectory, so that the cat litter forms a continuous coating layer on the outer periphery of the excrement, thus obtaining a cat litter ball. like Figure 4-6 As shown in the embodiment of this application, the preset trajectory is: rotating forward to a first angle and rotating backward to a second angle; controlling the excrement to reciprocate between the first angle and the second angle; the first angle and the second angle are greater than the angle that causes the cat litter to start sliding, and the first angle is less than the angle that causes the cat litter to slide onto the filter.
[0048] In one embodiment, the first angle is set to be greater than 30° and less than 60°. By limiting the first angle to greater than 30°, it is ensured that the roller tilt is sufficient to drive the cat litter to slide effectively relative to the excrement surface, thereby achieving full coverage of the exposed excrement. By limiting the first angle to less than 60°, sufficient forward residual stroke is reserved after the forward litter wrapping stage is completed, so that the roller can be controlled to continue rotating forward to the filter separation position to perform the sieving action of cat litter and fecal clumps, thereby allowing the formed cat litter clumps to fall into the fecal collection bin under the action of gravity.
[0049] The second angle is set to be greater than 30° so that there is sufficient angular displacement during the reverse swing process, driving the cat litter to cover the uncovered areas of the excrement surface in the reverse direction, avoiding uneven litter coverage due to insufficient swing amplitude.
[0050] In one embodiment, the first angle is preferably set to 45°, and the second angle is preferably set to 60°. Choosing the first angle to 45° ensures sufficient relative sliding efficiency between the excrement and the litter to achieve uniform litter coating, while avoiding excessively long single rotation strokes of the roller, thus reducing the time required for a single cleaning cycle and balancing litter coating quality and cleaning efficiency. Setting the second angle to 60° allows for a larger reverse swing amplitude to fully compensate for any blind spots that might be left during forward litter coating, further improving the integrity of the final litter clump. It also avoids excessively long rotation strokes, thereby reducing cleaning time and balancing litter coating quality and cleaning efficiency.
[0051] In one embodiment, the speed of the reciprocating rotation is configured to maintain the shape integrity of the excrement during movement and to allow it to slide relative to the covering litter without scattering. For example, the angular velocity can be set to 5° / s to 15° / s.
[0052] In this embodiment, multiple pauses are incorporated during the reciprocating rotational motion. Specifically, the excrement is rotated to a pause position at least once. This pause position is where the excrement is exposed above the litter layer and separated from the litter, allowing the image acquisition device to capture real-time images of the fecal clump for evaluating the litter coating effect. After a first preset duration of reciprocating rotational motion, the litter gradually forms a continuous and dense coating layer around the excrement, resulting in a litter clump with a certain structural strength.
[0053] In one embodiment, the first preset duration can be set to 20 to 60 seconds. Furthermore, the first preset duration can be determined according to the degree of stool softness: for mild soft stool, the first preset duration is set to 20 seconds; for moderate soft stool, the first preset duration is set to 40 seconds; and for severe soft stool, the first preset duration is set to 60 seconds.
[0054] In this step, the cat litter and excrement are completely mixed to form a litter ball.
[0055] Step 3: The waste falls into the collection tank. like Figure 7 As shown, once the cat litter clump forms, it is controlled to rotate to a preset dropping angle, allowing the cat litter clump to fall into the collection chamber by its own weight, thus completing the cleaning process.
[0056] In one embodiment, a filter angle is provided between the dropping angle and the first angle of the reciprocating rotation to separate the litter clump from the litter. Specifically: after the litter clump is formed, the mixture of the litter clump and the litter is rotated to the first angle, and then rotated to the filter angle, causing the litter clump to separate from the litter at the filter; then it continues to rotate to the dropping angle, at which point the litter clump falls into the collection chamber through the dropping channel on the filter, thus completing the entire cleaning process.
[0057] In one embodiment, the filter angle is set to 60° to 90°; the waste discharge angle is set to 90° to 150°.
[0058] Setting the filter angle to 60° to 90° allows the excrement to have sufficient sliding and wrapping path and time during the sand-coating stage before it rotates to the filter angle, i.e., when it is within the first angle range, to ensure that the sand coating is fully completed. At the same time, the upper limit of this angle range ensures that the clean cat litter has sufficient sieving stroke when passing through the filter area, thereby achieving effective separation of litter clumps and preventing clean cat litter from being mistakenly brought into the waste collection bin.
[0059] Set the dropping angle to 90° to 150°. By setting a larger dropping angle, provide sufficient gravity-driven displacement for the feces, allowing it to enter the dropping channel intact and fall into the collection chamber. This prevents the feces from getting stuck in the filter area due to insufficient rotation, causing blockage or secondary mixing with clean cat litter.
[0060] In one implementation, if the cat litter still does not adequately cover the excrement after multiple reciprocating rotations, an abnormal state is recorded, and the fecal mass is controlled to rotate to a preset dropping angle so that it falls into the collection chamber.
[0061] In this embodiment of the application, the collection of excrement images includes collecting excrement images at multiple stages during the pet's toileting process.
[0062] It's important to note that a pet's body posture during littering can easily obstruct the camera's view, making it difficult to capture complete and clear images of excrement. This is especially true when the pet is using the litter box or before burying it; body obstruction often leads to failed image capture during crucial moments of excretion or the initial mixing of excrement with the litter. While capturing images after the pet has finished burying and left the litter box avoids obstruction, the original shape of the litter clump has already been disrupted, increasing the difficulty of subsequent image analysis.
[0063] Therefore, based on the behavioral characteristics of pet toileting, this application divides the data collection process into the following three stages to cover the complete time sequence from the occurrence of toileting to the formation of fecal pellets: Phase 1: Toileting Phase. This phase involves the pet defecating while the camera captures images of the excrement in real time. Images from this phase capture the excrement in its raw state, before it is fully mixed with the litter. The morphological features, texture, and adhesion characteristics are relatively realistic, making it an important basis for identifying soft stools.
[0064] Among them, morphological characteristics include the clarity of the excrement boundary, whether there is collapse or diffusion, and whether it is pasty; texture characteristics include the reflective properties caused by high moisture content, surface continuity, and the coverage of cat litter granules; adhesion signs include the size of the contact area between the excrement and the bottom cat litter, and whether there is a stringy or trailing pattern.
[0065] Phase Two: From toilet training to before excrement is covered. This phase is the window of opportunity between when the pet finishes toilet training and before the excrement is covered. The camera captures images of the excrement in real time during this period. At this time, the excrement has been initially mixed with the cat litter, but has not yet been buried or destroyed. The excrement in the image is relatively intact, with minimal interference from the cat litter, making it easy to perform image recognition and analysis. This is a crucial phase for obtaining clear images.
[0066] Phase 3: From when the excrement is covered to before automatic cleaning. In this phase, the excrement has been initially covered, and the pet has left the litter box, so there is no issue of the pet's body obstructing the view, thus ensuring that a clear image can be captured.
[0067] By dividing the pet's toileting process into three stages and collecting images of excrement at each stage, even if no valid images are collected at a certain stage due to the pet obstructing the view or exhibiting abnormal behavior, the system can still output analysis results based on images from other stages, thereby significantly improving the robustness and applicability of soft stool analysis.
[0068] In this embodiment, for each stage, instead of capturing a single image, a continuous video stream or image sequence is captured: In the first stage, a first image set is acquired from the time the pet enters the smart litter box until it finishes using the litter box, and a first image is selected from the first image set; the first image is an image of excrement directly exposed; In the second stage, a second image set is acquired from the time the pet finishes using the litter box until the excrement is covered, and a second image is selected from the second image set; the second image is an image of excrement initially mixed with litter and not covered; In the third stage, a third image set is acquired from the time the excrement is covered until automatic cleaning, and a third image is selected from the third image set; the third image is an image of excrement covered by litter.
[0069] By acquiring continuous image sets at each stage and selecting images from them, the problem of missing effective images caused by poor timing of single-frame image acquisition, momentary occlusion by pets, or image blurring is avoided. This improves the success rate of acquiring effective images at each stage, provides a richer and more reliable image data foundation for subsequent soft stool recognition, and enhances the robustness of the method.
[0070] The first image selected is an image of the excrement directly exposed; this excrement image has not yet been mixed with cat litter, and its morphological features, texture features and adhesion signs are relatively realistic, avoiding the interference of cat litter on the excrement features, providing the most accurate analytical basis for soft stool identification, and improving the sensitivity and accuracy of soft stool identification.
[0071] The second image selected is an image of excrement initially mixed with cat litter and not covered; the second image reflects the true form of excrement after initial mixing with cat litter. At this time, the excrement has not been buried or destroyed, the image is clear and has less interference, making it easy to extract stable visual features.
[0072] The third image selected shows the excrement covered by cat litter. Although the excrement has been initially covered by cat litter, this image can still be used to assess the degree of soft stool: for example, for mild soft stool, the surface of the litter is relatively dry and the clumps are intact; for severe soft stool, even after initial covering, the surface of the litter may still show obvious signs of moisture penetration or adhesion. Therefore, by analyzing the morphological characteristics of the litter covering layer in the third image, the degree of soft stool can be identified, and the third image has important supplementary value.
[0073] By complementing the images from the three stages, the accuracy of the analysis is ensured, and effective images can still be acquired for health monitoring even in complex scenarios.
[0074] In this embodiment of the application, acquiring images of excrement at multiple stages during a pet's toilet training process also includes the following steps: When a pet is detected, it is determined that the pet has entered the smart litter box, and the first stage of image acquisition is initiated. This stage begins when the pet enters and continues to acquire images until the pet finishes using the litter box.
[0075] During the pet's toilet training, image recognition algorithms analyze the pet's posture features in real time, including back curvature, tail posture, and hind leg position. When the pet transitions from a posture of arched back, raised tail, and half-crouched hind legs to a standing position, the toilet training is considered complete. At this point, the first stage of image acquisition terminates, and the second stage begins. This second stage starts at the end of toilet training and continuously acquires images until the excrement is covered by litter.
[0076] In the second stage, the state of excrement on the surface of the litter layer is continuously monitored. When excrement is detected exposed on the surface of the litter layer, i.e., not yet covered by litter, the litter adding device is controlled to add litter to the surface of the excrement until the excrement is completely covered by litter. Once coverage is complete, the controller terminates the second stage of image acquisition and starts the third stage of image acquisition.
[0077] In the third stage, the excrement covered with cat litter is rotated to a preset shooting angle to obtain a clear image. After a second preset period of stable acquisition, the third stage of image acquisition is terminated.
[0078] In one embodiment, the shooting angle is the angle directly facing the image acquisition device; the second preset duration ranges from 5s to 10s.
[0079] Through the above steps, the system can automatically switch the image acquisition stage based on events such as the pet's entry, changes in toilet posture, and the state of excrement exposure, ensuring that the images at each stage can reflect the key state characteristics of the excrement, and providing high-quality input data for subsequent soft stool recognition.
[0080] In one implementation, the recognition of a pet entering or leaving the smart litter box can be achieved using one or more of the following sensors. Specifically: The system uses a camera to capture images of the pet in real time and uses an image recognition algorithm to determine whether the pet has entered or left the litter box.
[0081] A weight sensor is installed at the base or support leg of the litter box to monitor changes in overall weight in real time. The detected weight increases when the pet enters the litter box and decreases when the pet leaves. By setting thresholds for weight changes and the duration of these changes, it is possible to effectively distinguish between pet activity and human intervention or environmental disturbances.
[0082] Employing a millimeter-wave radar sensor, this device is installed above or on the inner side wall of the litter box entrance. The sensor accurately detects the pet's direction, speed, and trajectory, unaffected by environmental factors such as light, temperature, and humidity. During different stages of the pet's movement, such as approaching, entering, turning around, or leaving, the radar sensor continuously outputs motion status signals and effectively filters out false triggers caused by non-biological targets.
[0083] An infrared sensor is placed at the entrance of the litter box to determine whether a pet is entering or leaving by the on / off state of the infrared beam or by the reflected signal. For example, when a pet blocks the infrared beam, it is determined to have entered; when the beam is unblocked, it is determined to have left.
[0084] In this embodiment of the application, an image acquisition device is set at a preset position in the smart litter box, and the preset position includes at least one of the following: The image acquisition point is located at the entrance of the smart litter box. The camera is positioned at a downward angle to capture images of the area inside the litter box. Using only one camera reduces equipment costs while meeting basic image acquisition requirements.
[0085] Two image acquisition points are located above the entrance to the smart litter box and on the inner side directly opposite the entrance. The perspectives of the two cameras complement each other, effectively reducing the risk of single-camera occlusion due to changes in pet posture and position, and improving the success rate of acquiring effective images in the first and second stages. Images acquired simultaneously by the dual cameras can be used for pet behavior recognition. By analyzing behavioral characteristics such as the time when the pet enters and leaves, and changes in body posture, the system can accurately determine the start and end times of the littering stage, providing a more reliable triggering basis for stage division and image capture.
[0086] Image acquisition points are located on the rotating axis inside the smart litter box. The camera is positioned at these acquisition points along the rotating axis, adjusting its shooting position according to the pet's orientation while using the litter box to ensure comprehensive and complete image capture of the excretion area.
[0087] In this embodiment of the application, soft stool recognition processing is performed on excrement images to obtain soft stool recognition results, including: constructing a soft stool recognition model; inputting at least one of the first image, second image, and third image into the soft stool recognition model; the processing of the soft stool recognition model includes: analyzing the images at each stage to obtain the soft stool recognition sub-results corresponding to each stage; and weighting and fusing each soft stool recognition sub-result according to the preset confidence weights of the images at each stage to output the soft stool recognition result.
[0088] Before performing soft stool identification, a soft stool identification model needs to be built in advance. This model is trained on a large number of soft stool samples, and its specific construction method can adopt machine learning or deep learning modeling methods known in the field, which will not be elaborated here. It should be noted that images of excrement collected at different stages vary in reliability and analytical difficulty. Specifically: The first stage: The excrement has just been expelled and has not yet been covered by cat litter. Its shape, texture and adhesion are relatively real, which is the direct basis for identifying soft stool.
[0089] Stage Two: The excrement has been initially mixed with the cat litter, but has not yet been buried or destroyed. At this stage, the degree to which the cat litter is moistened and penetrated by the excrement can be observed, helping to determine if the stool is soft. Simultaneously, the image at this stage is relatively clear, with less interference from the cat litter, making image recognition and analysis easier.
[0090] Phase 3: Although images can be collected, the images are more complex and the analysis is relatively difficult because the excrement has been buried or partially destroyed by cat litter.
[0091] Therefore, in this embodiment, the soft stool identification sub-results are weighted and fused according to the confidence weights of the images at each stage, thereby improving the overall accuracy and robustness of soft stool identification. This confidence weight is used to weight the determination result of "whether it is soft stool". The confidence weight is set according to at least one of the following rules: When the first image, the second image, and the third image are obtained, the confidence weight of the second image shall not be less than 50%, and the confidence weight of the third image shall not be greater than 20%. When only the first and third images are obtained, the confidence weight of the first image shall not be less than 50%; When only the second and third images are obtained, the confidence weight of the second image shall not be less than 70%; When only the third image is obtained, its confidence weight is 100%.
[0092] In one implementation, when the first image, the second image, and the third image are acquired, the confidence weight of the first image is 30%, the second image is 60%, and the third image is 10%; when only the first image and the third image are acquired, the confidence weight of the first image is 60% and the third image is 40%; when only the second image and the third image are acquired, the confidence weight of the second image is 80% and the third image is 20%.
[0093] The weight values mentioned above are for illustrative purposes only and can be adjusted according to specific scenarios and sample statistical results in actual applications.
[0094] Finally, the final soft stool recognition result is output.
[0095] In one implementation, if the first image is identified as soft stool (denoted as 1, confidence level 30%), the second image is soft stool (denoted as 1, confidence level 60%), and the third image is not soft stool (denoted as 0, confidence level 10%), then the weighted fusion score is 0.3×1 + 0.6×1 + 0.1×0 = 0.9. This score is compared with a preset threshold. If the score exceeds the threshold, the final output is soft stool; otherwise, the output is not soft stool.
[0096] The threshold here represents the minimum overall score required to classify a patient as having soft stool. Adjusting this threshold balances the sensitivity and specificity of the analysis method: a lower threshold (e.g., 0.3) results in more samples being classified as having soft stool, reducing false negatives and improving sensitivity; a higher threshold (e.g., 0.7) imposes stricter criteria for classifying a patient as having soft stool, reducing false positives and improving specificity. In practical applications, the threshold can be flexibly set according to specific needs to achieve optimal soft stool identification results.
[0097] In one implementation, after outputting the soft stool identification result through confidence weight fusion, if it is determined to be soft stool, then the degree of soft stool classification is further performed; if it is determined to be non-soft stool, the analysis ends. The soft stool degree classification selects images for analysis according to the following priority order: the second image is used first; if the second image is not obtained, the first image is used; if only the third image is obtained, the third image is used.
[0098] Through the aforementioned phased acquisition and dynamic weight fusion mechanism, even if no effective image is acquired in a certain phase, the system can still output recognition results based on images from other phases, significantly improving the robustness and applicability of the soft recognition method.
[0099] In one embodiment, the method described in this application further includes an exception handling step: When the system detects that a pet has re-entered the smart litter box during the cleaning process, it immediately stops rotating and returns to a preset safe angle.
[0100] The system detects whether there is a jamming condition by monitoring the current fluctuations of the rotating motor or the encoder feedback signal; if jamming is detected, the system stops operation and issues an alarm signal.
[0101] If image recognition fails multiple times in a row, the system will enter conservative cleanup mode and perform soft cleanup operation by default.
[0102] Secondly, embodiments of this application provide a soft stool identification and cleaning system for implementing the aforementioned soft stool identification and cleaning method. The system includes: Image acquisition module: Used to acquire images of excrement. It may include one or more cameras, as well as corresponding image preprocessing units, such as noise reduction and enhancement.
[0103] Soft stool recognition module: Connected to the image acquisition module, it is loaded with a pre-trained soft stool recognition model, which is used to process excrement images for soft stool recognition and output the soft stool recognition result.
[0104] Control module: Connected to the soft stool recognition module, it performs a soft stool cleaning operation when the soft stool recognition result indicates soft stool. The control module can be a microcontroller or an embedded processor.
[0105] Litter dispenser: Connected to the control module, it is used to dispense cat litter onto the surface of excrement under the control of the control module. A litter dispenser typically includes a litter bin, a litter delivery motor, and a litter outlet.
[0106] Rotary drive device: connected to the control module, used to drive the excrement to reciprocate along a preset trajectory under the control of the control module, so that the cat litter forms a continuous coating layer on the outer periphery of the excrement to obtain cat litter clumps; it is also used to rotate the cat litter clumps to a preset dropping angle so that they fall into the collection bin.
[0107] The above modules can be integrated on a single circuit board and communicate via a bus or signal line.
[0108] like Figure 2 As shown, in a third aspect, embodiments of this application provide a smart litter box, comprising: Basin 1: Used to hold cat litter and collect excrement. Basin 1 has a rotating roller 2 or scoop structure inside.
[0109] Image acquisition device 3: set at a preset position in the basin 1, used to acquire images of excrement.
[0110] Sand adding device 4: It is set on the basin 1, usually located on the upper part or side of the basin 1, and its sand outlet is aligned with the inside of the basin 1.
[0111] Rotating mechanism: Located inside the basin 1, it includes a rotary motor, a drive shaft and a roller 2, used to carry and drive excrement or fecal matter to reciprocate and rotate.
[0112] Fecal collection chamber 5: Located on one side of the basin 1, it is used to collect cat litter pellets. The fecal collection chamber 5 is connected to the basin 1 via a fecal discharge channel 6.
[0113] Controller: Electrically connected to image acquisition device 3, sand adding device 4, and rotating mechanism respectively. The controller is configured to perform the above-described soft stool identification and cleaning method.
[0114] Upon detecting soft stool, the controller automatically performs litter addition, reciprocating rotation, and stool collection operations. Specifically, when the pet excretes soft stool, the image acquisition device captures an image of the soft stool, the controller determines it to be soft stool, and immediately activates the litter addition device to sprinkle an appropriate amount of cat litter onto the surface of the soft stool; then, the rotating mechanism is controlled to reciprocate at a speed of 10° / s within a range of ±45° for 30 seconds, pausing twice during this period to check the covering effect; finally, the covered cat litter clump is poured into the collection tank. The entire process can be completed within 1 minute, preventing soft stool from adhering to the inner wall.
[0115] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.
Claims
1. A method for identifying and cleaning soft stools, characterized in that, include: Collect images of excrement; The excrement image is processed for soft stool recognition to obtain soft stool recognition results; When the soft stool identification result indicates soft stool, a soft stool cleaning operation is performed; The soft stool cleaning procedure includes: Control the excrement to reciprocate and rotate along a preset trajectory, so that the cat litter forms a continuous coating layer on the outer periphery of the excrement, thus obtaining a cat litter ball; Control the cat litter ball to rotate to a preset dropping angle, so that the cat litter ball falls into the collection chamber.
2. The method as described in claim 1, characterized in that, Before controlling the excrement to reciprocate along a preset trajectory, the method further includes: Identify areas of excrement exposed on the surface of the litter box; A target coverage area is determined based on the excrement area, and the target coverage area includes the excrement area; The litter dispenser controls the litter to spread cat litter over the target area.
3. The method as described in claim 2, characterized in that, The soft stool identification result includes: soft stool determination result and soft stool severity level; the soft stool severity level includes at least mild soft stool, moderate soft stool and severe soft stool; The controlled litter distribution device spreads cat litter over the target area, including: Based on the level of soft stool, determine the lower limit of the litter cover thickness corresponding to that level; Obtain the area of the target coverage area, and determine the amount of cat litter to be spread based on the area and the lower limit of the thickness; The litter dispensing device is controlled to dispense cat litter to the target coverage area at the specified dispensing amount.
4. The method as described in claim 1, characterized in that, The preset trajectory is: rotating forward to a first angle, and rotating backward to a second angle; The control of the excrement to reciprocate and rotate along a preset trajectory includes: The excrement is controlled to reciprocate between the first angle and the second angle; The first angle and the second angle are greater than the angle that causes the cat litter to begin to slide, and the first angle is less than the angle that causes the cat litter to slide onto the filter.
5. The method as described in claim 4, characterized in that, The first angle is greater than 30° and less than 60°; the second angle is greater than 30°.
6. The method as described in claim 4, characterized in that, The reciprocating rotational motion includes: The excrement is rotated to a stopping position at least once, the stopping position being a position in which the excrement is exposed above the litter layer and separated from the litter, in order to capture images of the fecal clump; After a first preset period of reciprocating rotation, the cat litter forms a continuous coating layer around the excrement, thus obtaining the cat litter clump.
7. The method as described in claim 1, characterized in that, The collection of excrement images includes: Images of excrement collected during multiple stages of a pet's toilet training process, including: The first phase involves capturing the first images of pets using the toilet. The second phase involved collecting a second image of the pet from the time it used the toilet until its excrement was covered. The third stage involves collecting images from the time the pet excrement is covered until the automatic cleaning process. The process of performing soft stool recognition processing on the excrement image to obtain soft stool recognition results includes: Construct a soft stool recognition model; Input at least one of the first image, the second image, and the third image into the soft stool recognition model; The processing steps of the soft stool recognition model include: The images at each stage are analyzed to obtain the corresponding soft stool identification sub-results for each stage; Based on the preset confidence weights of the images at each stage, the soft stool recognition sub-results are weighted and fused to output the soft stool recognition result.
8. The method as described in claim 7, characterized in that, The acquisition of images of pet excrement at multiple stages during the toileting process includes: In the first phase, a first set of images is acquired from the time the pet enters the smart litter box until it finishes using the litter box, and a first image is selected from the first set of images; the first image is an image in which excrement is directly exposed. In the second stage, a second set of images is acquired from the end of the pet's toilet training until the excrement is covered, and a second image is selected from the second set of images; the second image is an image of the excrement initially mixed with cat litter and not yet covered. In the third stage, a third image set is obtained from the time the excrement is covered until the automatic cleaning, and a third image is selected from the third image set, wherein the third image is an image of the excrement covered by cat litter.
9. The method as described in claim 8, characterized in that, The acquisition of images of pet excrement at multiple stages during the toileting process also includes: When a pet is detected, it is determined that the pet has entered the smart litter box, and the first stage of image acquisition is initiated. When the pet changes from a defecation posture to a standing posture, the toileting is considered complete, the first stage of image acquisition is terminated, and the second stage of image acquisition is started. When excrement is detected exposed on the surface of the litter layer, the litter adding device is controlled to add litter to the surface of the excrement until the excrement is covered by litter, the second stage of image acquisition is terminated, and the third stage of image acquisition is started. The litter-covered excrement is rotated to a preset shooting angle, and after a second preset time, the third stage of image acquisition is terminated.
10. The method as described in claim 9, characterized in that, An image acquisition device is installed at a preset location in the smart litter box, wherein the preset location includes at least one of the following: Image acquisition point located at the entrance of the smart litter box; Two image acquisition points are located above the entrance of the smart litter box and above the inner side directly opposite the entrance; Image acquisition points located on the rotation axis of the inner wall of the smart litter box.
11. The method as described in claim 7, characterized in that, The confidence weights are set according to at least one of the following rules: When the first image, the second image, and the third image are obtained, the confidence weight of the second image shall not be less than 50%, and the confidence weight of the third image shall not be greater than 20%. When only the first and third images are obtained, the confidence weight of the first image shall not be less than 50%; When only the second and third images are obtained, the confidence weight of the second image shall not be less than 70%; When only the third image is obtained, its confidence weight is 100%.
12. A soft stool identification and cleaning system, used to implement the soft stool identification and cleaning method as described in any one of claims 1 to 11, characterized in that, include: The image acquisition module is used to acquire images of excrement; A soft stool recognition module, connected to the image acquisition module, is used to perform soft stool recognition processing on the excrement image to obtain a soft stool recognition result; The control module, connected to the soft stool recognition module, is used to perform a soft stool cleaning operation when the soft stool recognition result indicates that the stool is soft. A litter-adding device, connected to the control module, is used to add cat litter to the surface of excrement under the control of the control module. A rotary drive device, connected to the control module, is used to drive the excrement to reciprocate and rotate along a preset trajectory under the control of the control module, so that the cat litter forms a continuous coating layer on the outer periphery of the excrement, thus obtaining a cat litter ball. The rotary drive device is also used to rotate the cat litter ball to a preset dropping angle under the control of the control module, so that the cat litter ball falls into the collection chamber.
13. A smart cat litter box, characterized in that, include: Basin; An image acquisition device is installed at a preset position on the basin to acquire images of excrement; A litter dispenser, installed on the basin, is used to dispense cat litter onto the surface of the excrement; A rotating mechanism, located inside the basin, is used to carry and drive excrement or fecal matter to reciprocate and rotate. A feces collection chamber is located on one side of the basin and is connected to the basin through a feces discharge channel; it is used to collect cat litter clumps. The controller is electrically connected to the image acquisition device, the sand adding device, and the rotating mechanism, respectively, and the controller is configured to perform the soft stool identification and cleaning method according to any one of claims 1 to 11.