Pet health condition optimization method, pet feeder and storage medium

Abstract

The invention discloses a pet health condition optimization method, a pet feeder and a storage medium, and relates to the technical field of pet feeding, and the method comprises the steps: dynamically collecting two-dimensional image information and three-dimensional depth image information of a target pet through a two-dimensional image collection module and a depth image collection module; according to the three-dimensional depth image information, identifying pet body and weight information, and according to the two-dimensional image information and the three-dimensional depth image information, carrying out comprehensive analysis to obtain pet motion state information; and inputting the pet motion state information and the pet body shape and weight information into a machine learning model, predicting to obtain health assessment information, thereby determining a feeding optimization strategy, and adjusting set parameters in a preset feeding plan according to adjustment parameters in the feeding optimization strategy, so that a feeding module executes a feeding task according to the adjusted set parameters. According to the method, the behavior data of the pet is effectively converted into a targeted health improvement strategy, and the intelligence and accuracy of the pet feeder in the aspect of health management are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of pet feeding, and in particular to a pet health condition optimization method, a pet feeder and a storage medium. BACKGROUND

[0002] Most of the existing pet feeders still mainly have the basic feeding function, and although some high-end models integrate a camera and a networking module, enabling users to remotely observe the daily state of the pet, the overall function is still limited to state monitoring and remote interaction.

[0003] In the process of feeding a pet using an existing pet feeder, if the pet has health problems such as obesity, decreased activity, or abnormal behavior, since the existing pet feeder product has obvious limitations in the health management of the pet, there is still a deficiency in the continuous monitoring and active intervention of the pet's health condition, and there is a lack of ability to convert the pet's behavior data into effective health management strategies, therefore, it is difficult to provide targeted improvement means for the pet, and only artificial intervention can be used to achieve the corresponding processing of the pet's health condition. SUMMARY

[0004] The main purpose of the present application is to provide an invention name, aiming to solve the technical problem of how to improve the health condition of a pet through a pet feeder.

[0005] To achieve the above-mentioned purpose, the present application provides a pet health condition optimization method, which is applied to a pet feeder, the pet feeder comprises a lower feeding module, a two-dimensional image acquisition module and a depth image acquisition module, and the method comprises: dynamically acquiring two-dimensional image information of a target pet through the two-dimensional image acquisition module, and dynamically acquiring three-dimensional depth image information of the target pet through the depth image acquisition module; According to the three-dimensional depth image information, identifying pet body shape and weight information of the target pet, and comprehensively analyzing pet activity state of the target pet according to the two-dimensional image information and the three-dimensional depth image information, to obtain pet motion state information, wherein the pet motion state information is used to represent walking posture, running posture or jumping posture of the target pet; inputting the pet motion state information and the pet body shape and weight information into a trained health assessment machine learning model to predict health assessment information of the target pet; determining a feeding optimization strategy based on the health assessment information, and adjusting the set parameters in the preset feeding plan according to the adjustment parameters in the feeding optimization strategy, so that the lower feeding module executes the feeding task for the target pet according to the adjusted set parameters.

[0006] In an embodiment, the pet feeder further comprises an RGB camera module, a first infrared camera module, an infrared radiation lamp, and an ambient light intensity (ALS) sensor, and the method further comprises: detecting a current ambient light intensity value through the ALS sensor; in a case where the ambient light intensity value reaches a preset light intensity threshold, taking the RGB camera module as the two-dimensional image acquisition module, and starting the RGB camera module to acquire two-dimensional image information of the target pet; in a case where the ambient light intensity value does not reach the preset light intensity threshold, taking the first infrared camera module as the two-dimensional image acquisition module, starting the infrared radiation lamp to irradiate the target pet, and starting the first infrared camera module to acquire two-dimensional image information of the target pet.

[0007] In an embodiment, the depth image acquisition module comprises a second infrared camera module and a VCSEL projector, and the step of dynamically acquiring three-dimensional depth image information of the target pet through the depth image acquisition module comprises: projecting an infrared speckle pattern onto the target pet through the VCSEL projector, and acquiring actual image information corresponding to the infrared speckle pattern projected onto the target pet through the second infrared camera module; comparing standard image information corresponding to a preset standard infrared speckle pattern with the actual image information based on the actual image information, to obtain infrared speckle pattern deformation information generated by the actual image information compared with the standard image information; determining three-dimensional depth image information of the target pet according to the infrared speckle pattern deformation information.

[0008] In an embodiment, the health assessment information comprises obesity degree information and limb movement disorder information, the obesity degree information represents an obesity level in terms of a body weight and body shape index, and the limb movement disorder information represents a movement disorder level and / or a movement disorder type in terms of a limb movement function, and the method further comprises: Obtaining learning sample data, the learning sample data comprising first pet motion state information and first pet body type weight information of a plurality of healthy pets, and second pet motion state information and second pet body type weight information of a plurality of unhealthy pets, wherein the healthy pet is a pet with a weight index meeting a preset standard and without limb movement disorder, and the unhealthy pet is a pet with a weight index not meeting a preset standard and / or with limb movement disorder, wherein the first pet motion state information and the first pet body type weight information are used as first learning samples, obesity degree information and limb movement disorder information of the healthy pet are used as sample labels of the first learning samples, the second pet motion state information and the second pet body type weight information are used as second learning samples, and obesity degree information and limb movement disorder information of the unhealthy pet are used as sample labels of the corresponding second learning samples. Dividing the learning sample data into training sample data and verification sample data according to a preset ratio. Training a preset health assessment machine learning model through the training sample data, and verifying an identification accuracy of the preset health assessment machine learning model through the verification sample data. If the identification accuracy reaches a preset accuracy threshold, stopping training of the health assessment machine learning model, and obtaining a trained health assessment machine learning model.

[0009] In an embodiment, the step of adjusting a set parameter in a preset feeding plan according to an adjustment parameter in the feeding optimization strategy comprises: In a case where the health assessment information is a first unhealthy condition, reducing a feeding amount in the preset feeding plan according to the adjustment parameter, so that the feeding module executes the feeding task according to the reduced feeding amount, wherein the first unhealthy condition comprises a pet body type index of the target pet being greater than a preset first pet body type threshold or a pet weight index of the target pet being greater than a preset first pet weight threshold. In a case where the health assessment information is a second unhealthy condition, increasing the feeding amount in the preset feeding plan according to the adjustment parameter, so that the feeding module executes the feeding task according to the increased feeding amount, wherein the second unhealthy condition comprises a pet body type index of the target pet being less than a preset second pet body type threshold or a pet weight index of the target pet being less than a preset second pet weight threshold, the second pet body type threshold being less than the first pet body type threshold, and the second pet weight threshold being less than the first pet weight threshold.

[0010] In an embodiment, the pet feeder further comprises at least two feeders corresponding to the lower food mold module and a weighing sensor at the bottom of the feeder, and the pet health condition optimization method further comprises: monitoring the food weight in each of the feeders through the weighing sensor; evaluating the preference degree of the target pet for the food in each of the feeders according to the change of the food weight; adjusting the lower food proportion of each of the feeders in the preset feeding plan based on the preference degree, so that the lower food mold module performs the feeding task according to the adjusted lower food proportion.

[0011] In an embodiment, the pet feeder further comprises an angle calculation module, and before the step of evaluating the preference degree of the target pet for the food in each of the feeders according to the change of the food weight, the method further comprises: monitoring the inclination angle of the feeder through the angle calculation module; in the case that the inclination angle is less than a preset angle threshold, correcting the food weight according to the inclination angle to determine the change of the food weight based on the corrected food weight.

[0012] In an embodiment, the pet feeder further comprises a laser projection module, and after the step of predicting the health evaluation information of the target pet, the method further comprises: determining a motion optimization strategy based on the health evaluation information, calling the laser projection module to project a pet interesting pattern to a preset area according to the motion optimization strategy, and monitoring the actual position of the target pet through the two-dimensional image acquisition module and / or the depth image acquisition module; in the case that the actual position coincides with the preset area, determining that the target pet chases the pet interesting pattern, and changing the projection position of the pet interesting pattern; wherein the motion optimization strategy affects the calling frequency of the laser projection module, the distance between the projection position and the target pet, or the change frequency of the projection position.

[0013] In addition, to achieve the above-mentioned purposes, the present application further provides a pet feeder, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the computer program is configured to implement the steps of the pet health condition optimization method as described above.

[0014] In addition, to achieve the above-mentioned purpose, the application further provides a storage medium, which is a computer readable storage medium, and a computer program is stored on the storage medium, and the computer program is executed by a processor to implement the steps of the pet health condition optimization method.

[0015] The one or more technical solutions provided by the application have at least the following technical effects: The application dynamically collects two-dimensional image information and three-dimensional depth image information of the target pet through the two-dimensional image acquisition module and the depth image acquisition module integrated in the pet feeder, then identifies pet body shape and weight information based on the three-dimensional depth image information, and comprehensively analyzes the activity state of the pet in combination with the two-dimensional image information and the three-dimensional depth image information to obtain pet motion state information for representing a walking posture, a running posture or a jumping posture. Then, the pet motion state information and the pet body shape and weight information are input into a pre-trained health assessment machine learning model to predict health assessment information, so as to determine a feeding optimization strategy according to the health assessment information, and adjust the set parameters in the preset feeding plan based on the adjustment parameters in the strategy, so that the feeding module executes the feeding task according to the adjusted parameters. Thus, through multi-modal data acquisition and machine learning analysis, active intervention and optimization of the pet health condition are realized, the behavior data of the pet is effectively converted into a targeted health improvement strategy, the limitation of relying on human intervention to intervene the pet health condition is overcome, continuous and automatic monitoring and management of the pet health condition based on the pet feeder are realized, and the intelligence and precision of the pet feeder in health management are improved. BRIEF DESCRIPTION OF DRAWINGS

[0016] The accompanying drawings, which are incorporated into and form part of the specification, illustrate embodiments consistent with the present application and, together with the specification, serve to explain the principles of the application.

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the accompanying drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, for those skilled in the art, other drawings can also be obtained based on these drawings without creative labor.

[0018] Figure 1 A flowchart is provided for the pet health condition optimization method embodiment one of the application; Figure 2 A flowchart is provided for the pet health condition optimization method embodiment two of the application; Figure 3 A flowchart is provided for the pet health condition optimization method embodiment three of the application; Figure 4The flowchart provided for the pet health condition optimization method embodiment four of the present application; Figure 5 The brief flowchart of the pet health condition optimization method provided for the embodiment four of the present application; Figure 6 The appearance diagram of the pet health condition optimization method provided for the embodiment four of the present application; Figure 7 The laser projection scene diagram of the pet health condition optimization method provided for the embodiment four of the present application; Figure 8 The system architecture diagram of the pet health condition optimization method provided for the embodiment four of the present application; Figure 9 The device structure diagram of the hardware running environment involved in the pet health condition optimization method in the embodiment of the present application.

[0019] The purpose implementation, functional features and advantages of the present application will be further described with reference to the accompanying drawings in conjunction with the embodiments. DETAILED DESCRIPTION

[0020] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application, and are not used to limit the present application.

[0021] In order to better understand the technical solutions of the present application, the following will be described in detail in conjunction with the drawings and specific embodiments of the specification.

[0022] The main solution of the embodiment of the present application is: through the two-dimensional image acquisition module, the two-dimensional image information of the target pet is dynamically acquired, and through the depth image acquisition module, the three-dimensional depth image information of the target pet is dynamically acquired; according to the three-dimensional depth image information, the pet body type weight information of the target pet is recognized, and according to the two-dimensional image information and the three-dimensional depth image information, the pet activity state of the target pet is comprehensively analyzed to obtain pet motion state information, wherein the pet motion state information is used to represent the walking posture, running posture or jumping posture of the target pet; the pet motion state information and the pet body type weight information are input into the trained health evaluation machine learning model, and the health evaluation information of the target pet is predicted; based on the health evaluation information, the feeding optimization strategy is determined, and according to the adjustment parameters in the feeding optimization strategy, the set parameters in the preset feeding plan are adjusted, so that the feeding module executes the feeding task for the target pet according to the adjusted set parameters.

[0023] Since in the process of feeding pets by using existing pet feeders, if the pets have health problems such as obesity, decreased vitality or abnormal behavior, the existing pet feeder products have obvious limitations in the health management of pets, and there are still deficiencies in the continuous monitoring and active intervention of the health status of pets, and the ability to convert the behavior data of pets into effective health management strategies is lacking, therefore, it is difficult to provide targeted improvement means for pets, and only artificial intervention can be used to realize the corresponding processing of the health status of pets.

[0024] The present application provides a solution, through the two-dimensional image acquisition module and the depth image acquisition module integrated in the pet feeder, the two-dimensional image information and the three-dimensional depth image information of the target pet are dynamically acquired, then the pet body shape and weight information is identified based on the three-dimensional depth image information, and the activity state of the pet is comprehensively analyzed combined with the two-dimensional image information and the three-dimensional depth image information, to obtain the pet motion state information for representing walking posture, running posture or jumping posture. Then the pet motion state information and the pet body shape and weight information are input into the pre-trained health assessment machine learning model to predict the health assessment information, so as to determine the feeding optimization strategy according to the health assessment information, and adjust the set parameters in the preset feeding plan based on the adjustment parameters in the strategy, so that the feeding module executes the feeding task according to the adjusted parameters. Thus, through multi-modal data acquisition and machine learning analysis, active intervention and optimization of the health status of the pet are realized, the behavior data of the pet is effectively converted into targeted health improvement strategies, the limitation of relying only on artificial intervention of the health status of the pet is overcome, the continuous, automatic monitoring and management of the health status of the pet based on the pet feeder is realized, and the intelligence and precision of the pet feeder in health management are improved.

[0025] It should be noted that the execution subject of the present embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device capable of realizing the above functions, a pet feeder, etc. The present embodiment and the following embodiments will be described below taking the pet feeder as an example.

[0026] Based on this, the present embodiment provides a pet health status optimization method, which is described with reference to Figure 1 , Figure 1 The flowchart of the first embodiment of the pet health status optimization method of the present application is shown in the figure.

[0027] In the present embodiment, the pet health status optimization method comprises steps S10-S40: Step S10, dynamically acquiring two-dimensional image information of the target pet through the two-dimensional image acquisition module, and dynamically acquiring three-dimensional depth image information of the target pet through the depth image acquisition module; It should be noted that the two-dimensional image acquisition module is a hardware component provided on the pet feeder and used for dynamically acquiring planar image data of the target pet. The functional core of the module is to obtain two-dimensional image information without depth dimension. The specific implementation can include but is not limited to an RGB (Red Green Blue, a camera that captures color images through red, green, and blue three-color channels) camera module or an infrared camera module, etc., which can provide a basic visual information source for subsequent analysis of the pet activity state. The two-dimensional image information is planar image data representing the appearance and action posture of the target pet, which is output by the two-dimensional image acquisition module. The information contains color, texture, contour, and dynamic change of the pet in the two-dimensional pixel space, and can be used in combination with three-dimensional depth image information to comprehensively analyze the pet motion state information such as walking posture, running posture, or jumping posture.

[0028] The depth image acquisition module is a hardware component provided on the pet feeder and used for dynamically acquiring spatial depth information of the target pet. The functional core of the module is to obtain distance data of each pixel point in the scene from the camera. The specific implementation can be based on the principles of structured light, time-of-flight method, or binocular vision, etc. The core function is to provide three-dimensional depth image information for non-contact identification of pet size and weight information. The three-dimensional depth image information is a data set representing the position and shape structure of the target pet in three-dimensional space, which is output by the depth image acquisition module. The information is essentially a point cloud or depth map containing depth values, which can record the spatial coordinates of each point on the pet surface, so as to calculate the volume, contour, and other pet size and weight information of the pet, and is the key three-dimensional data for precise health assessment.

[0029] Exemplarily, after a preset time interval or triggered by the motion sensor, the pet feeder synchronously starts the RGB camera equipped as the two-dimensional image acquisition module to record a color video stream of the target pet as two-dimensional image information. At the same time, the depth image acquisition module composed of a VCSEL (Vertical-Cavity Surface-Emitting Laser) projector and an IR (Infrared) camera is started. The VCSEL projector projects an invisible infrared speckle pattern to the pet, and the IR camera acquires the image of the deformed pattern due to the pet body contour, and then generates three-dimensional point cloud data describing the distance of each point of the pet body, which is used as three-dimensional depth image information.

[0030] In step S20, pet body shape and weight information of the target pet is identified according to the three-dimensional depth image information, and pet motion state information is obtained by comprehensively analyzing pet activity state of the target pet according to the two-dimensional image information and the three-dimensional depth image information, wherein the pet motion state information is used to represent walking posture, running posture or jumping posture of the target pet. It should be noted that the pet body shape and weight information refers to a series of parameters for reflecting the shape size and relative mass of the target pet, which are identified and calculated by an algorithm based on the three-dimensional depth image information dynamically collected by the depth image acquisition module. These information sources are derived from the measurement of the three-dimensional contour of the pet, and can non-contact estimate the volume, body length and other body shape indicators and weight changes of the pet, thereby providing a quantitative basis for evaluating the obesity degree or growth and development status of the pet.

[0031] The pet motion state information refers to data for describing the behavior pattern and posture characteristics of the target pet, which is obtained by comprehensively analyzing the two-dimensional image information collected by the two-dimensional image acquisition module and the three-dimensional depth image information collected by the depth image acquisition module. The information specifically represents the dynamic posture of the pet during activity, such as walking posture, running posture or jumping posture, and the purpose is to identify whether the pet has gait abnormalities, lack of vitality or potential limb movement dysfunction through these specific posture analysis.

[0032] It can be understood that, since the existing pet feeder lacks precise and non-contact quantitative ability for the physiological state of the pet, and cannot automatically obtain key body shape and dynamic behavior data, the embodiment calculates the volume and contour of the pet by using the three-dimensional depth image information, and identifies the pet body shape and weight information in a non-contact manner. Meanwhile, the texture, contour details provided by the two-dimensional image and the spatial structure data provided by the three-dimensional depth image are fused to comprehensively analyze the dynamic behavior of the pet, so as to extract the pet motion state information for representing the walking posture, running posture or jumping posture of the pet. Thus, the problem that the traditional scheme can only rely on manual observation or contact measurement and cannot continuously and automatically obtain the core health indicators of the pet is avoided, and the automatic and high-precision monitoring of the physical condition and motion ability of the pet is realized, thereby providing a reliable data basis for subsequent health evaluation.

[0033] Exemplarily, the processing system of the pet feeder first processes the three-dimensional depth image information of the continuous frames, calculates the contour dimensions such as the volume and back length of the pet through point cloud segmentation and three-dimensional reconstruction algorithm, estimates the weight of the current target pet according to the pre-established relationship model of pet breed, volume and weight, and thus identifies the pet body shape and weight information. At the same time, the system performs spatio-temporal alignment and fusion of the two-dimensional image information captured by the RGB camera and the three-dimensional depth image information, identifies the gait cycle, joint bending angle and motion trajectory of the pet by analyzing the continuous position changes of the joints of the pet in the three-dimensional space, and thus comprehensively analyzes whether the pet is currently in a normal walking posture, or in an unstable gait running posture or a jumping posture with insufficient jumping height, and takes these quantitative analysis results as the pet motion state information.

[0034] In step S30, the pet motion state information and the pet body shape and weight information are input into the trained health assessment machine learning model to predict the health assessment information of the target pet. It should be noted that the trained health assessment machine learning model is a machine learning model that has completed the training process and can perform evaluation. The model takes the pet body shape and weight information and the pet motion state information as input data, and outputs the evaluation result of the health status of the target pet through the complex mapping relationship learned by the model, i.e., the health assessment information. The health assessment information is the prediction result output by the trained health assessment machine learning model, and the information is a comprehensive and structured judgment conclusion of the current health status of the target pet. This information will be used as a direct basis for determining whether the feeding plan needs to be adjusted and how to adjust.

[0035] It can be understood that pure body shape and behavior data cannot be directly converted into a comprehensive judgment of the health status of the pet, and relying on fixed rules or artificial experience for evaluation is neither accurate nor difficult to cover complex situations. Therefore, the embodiment inputs the multi-dimensional and quantified pet motion state information and pet body shape and weight information obtained above into a machine learning model trained using a large amount of pet health data in advance, so as to comprehensively analyze and judge by using the complex nonlinear relationship learned by the model, thereby avoiding the one-sidedness, hysteresis and inaccuracy in health judgment based on simple threshold or artificial rules, and realizing intelligent and accurate identification of the health status of the pet.

[0036] Exemplarily, the system combines the formatted pet motion state information and pet body size and weight information into a feature vector, and inputs the feature vector into a trained health assessment machine learning model, such as a support vector machine classification model. The model is trained in the background based on a large amount of data of healthy and sick pets, and can output a comprehensive health assessment information, such as a classification including “obesity level: moderate” and “exercise function: suspected joint discomfort” and a probability result, after inputting the feature vector.

[0037] In step S40, a feeding optimization strategy is determined based on the health assessment information, and a set parameter in a preset feeding plan is adjusted according to an adjustment parameter in the feeding optimization strategy, so that the feeding module executes a feeding task for the target pet according to the adjusted set parameter.

[0038] It should be noted that the feeding optimization strategy is a specific action plan generated based on the health assessment information, which defines what changes should be made to the current feeding method in order to improve or maintain the health of the target pet, and the core content includes the adjustment operation to be performed. The adjustment parameter is a specific numerical instruction included in the feeding optimization strategy, which directly specifies the direction and amplitude of the modification of each set parameter in the preset feeding plan, for example, the feeding amount can be increased by a specific value, or the feeding amount can be reduced by a certain percentage.

[0039] The preset feeding plan is a basic feeding plan preset for the pet without considering the health assessment information, which specifies, but is not limited to, a standard process followed in a regular feeding task, food types, and time arrangement. The set parameter is a series of configurable variables that constitute the preset feeding plan, which specifically controls the execution details of the feeding task, such as the amount of food fed each time, the number of feeding times per day, the ratio of different types of food, etc. The parameter is the direct object of the adjustment parameter in the feeding optimization strategy.

[0040] It can be understood that, since even if the health assessment result is obtained, it is still impossible to achieve closed-loop management and active intervention of the pet's health, by generating a feeding optimization strategy containing specific adjustment parameters based on the health assessment information, and then using the adjustment parameters to dynamically adjust the set parameters in the preset and static feeding plan, and finally driving the feeding module to execute the feeding task according to the new parameters, the passive situation of health monitoring and feeding control being disconnected and still requiring the pet owner to manually adjust the feeding plan is avoided, and an automatic closed loop from “health problem identification” to “precise feeding intervention” is realized, which can actively and timely help the pet to improve or maintain the health condition by adjusting the diet.

[0041] Exemplarily, after the system receives the health assessment information "obesity level: moderate", according to the preset strategy mapping table, it is determined that the target optimization strategy is the feeding optimization strategy, and the adjustment parameter in the feeding optimization strategy is calculated as "reduce the single food intake by 20%" based on the health assessment information. Then, the preset feeding plan is adjusted according to the adjustment parameter, that is, the set parameter "single food intake is 50 grams" in the preset feeding plan is modified to "single food intake is 40 grams". Finally, the feeding module executes the feeding task according to the adjusted set parameter, so as to realize the purpose of controlling the pet's heat intake to optimize its health status.

[0042] The embodiment provides a pet health status optimization method. Two-dimensional image information and three-dimensional depth image information of a target pet are dynamically collected through a two-dimensional image collection module and a depth image collection module integrated in a pet feeder. Then, pet body shape and weight information is recognized based on the three-dimensional depth image information, and pet activity status is comprehensively analyzed based on the two-dimensional image information and the three-dimensional depth image information, so as to obtain pet motion state information for representing a walking posture, a running posture or a jumping posture. The pet motion state information and the pet body shape and weight information are input into a pre-trained health assessment machine learning model to predict health assessment information, so as to determine a feeding optimization strategy according to the health assessment information, and adjust set parameters in a preset feeding plan based on adjustment parameters in the strategy, so that a feeding module executes a feeding task according to the adjusted parameters. Thus, through multi-modal data collection and machine learning analysis, active intervention and optimization of the pet's health status are realized, pet behavior data is effectively converted into a targeted health improvement strategy, the limitation of relying on human intervention to intervene in the pet's health status is overcome, continuous and automatic monitoring and management of the pet's health status based on the pet feeder is realized, and the intelligence and accuracy of the pet feeder in health management are improved.

[0043] In a feasible implementation, the pet feeder further comprises an RGB camera module, a first infrared camera module, an infrared radiation lamp and an ambient light intensity ALS sensor. The pet health status optimization method can further comprise steps S01-S03: Step S01, detecting a current ambient light intensity value through the ALS sensor; It should be noted that the ALS (Ambient Light Sensor, ambient light sensor) sensor refers to an ambient light sensor, which is an electronic element for detecting the intensity of ambient light. Its specific function is to measure the current ambient light intensity value to provide a basis for the system to decide which image collection module to enable.

[0044] Step S02, in the case that the ambient light intensity value reaches the preset light intensity threshold, the RGB camera module is used as the two-dimensional image acquisition module, the RGB camera module is started to acquire two-dimensional image information of the target pet; It should be noted that the RGB camera module is a camera assembly that captures color images through red, green and blue three primary color light channels. The module is enabled when the ambient light is sufficient, used as a two-dimensional image acquisition module, responsible for acquiring two-dimensional image information of the target pet containing color and texture details.

[0045] Step S03, in the case that the ambient light intensity value does not reach the preset light intensity threshold, the first infrared camera module is used as the two-dimensional image acquisition module, the infrared radiation lamp is turned on to irradiate the target pet, and the first infrared camera module is started to acquire two-dimensional image information of the target pet.

[0046] It should be noted that the first infrared camera module is a camera assembly sensitive to infrared light. The module is enabled when the ambient light is insufficient, replacing the RGB camera module as a two-dimensional image acquisition module, responsible for acquiring black and white infrared two-dimensional image information of the target pet under the illumination of the infrared radiation lamp. The infrared radiation lamp can work with the first infrared camera module to provide illumination for the first infrared camera module in a dimly lit environment to ensure that it can clearly acquire two-dimensional image information of the target pet.

[0047] It can be understood that due to the different light conditions of the pet feeder in day and night, a single image acquisition scheme cannot guarantee the continuous high-quality acquisition of two-dimensional image information, especially in low light environments, it may not be able to effectively monitor the pet activity state. Therefore, in this embodiment, the ALS sensor detects the ambient light intensity and makes a judgment accordingly: in sufficient light, the RGB camera module is enabled to acquire color two-dimensional image information, and in insufficient light, the infrared radiation lamp is turned on and the first infrared camera module is enabled to acquire infrared two-dimensional image information. Thus, the problem of interruption or quality degradation in two-dimensional image information acquisition caused by changes in ambient light is effectively avoided, and stable and high-quality two-dimensional image information can be acquired under different light conditions in day and night, providing reliable data guarantee for subsequent pet activity state analysis.

[0048] Exemplarily, the microcontroller unit on the pet feeder continuously reads the ambient light intensity value detected by the ALS sensor integrated on the pet feeder. When the value is higher than the preset 50 lux threshold, the microcontroller unit determines that the ambient light is sufficient, and then drives the RGB camera module to start to collect the two-dimensional image of the target pet containing color information. Conversely, when the ambient light intensity value is lower than 50 lux, the microcontroller unit turns off the RGB camera module, and at the same time triggers the 940 nm infrared LED lamp (i.e. infrared radiation lamp) located in front of the device to light up to provide invisible illumination, and synchronously starts the IR camera (i.e. first infrared camera module) sensitive to infrared light to collect the black-and-white two-dimensional image of the target pet under infrared illumination, thereby ensuring that effective two-dimensional image information can be obtained under different ambient light.

[0049] In the embodiment, by using the ALS sensor to detect the ambient light intensity and intelligently switching the RGB camera module or the combination of the infrared radiation lamp and the first infrared camera module, the problem that a single image acquisition module cannot continuously obtain clear two-dimensional image information when the ambient light is too dark or too strong is avoided, and the two-dimensional image information of the target pet can be stably and clearly collected under all-weather complex lighting conditions.

[0050] In a possible implementation, the depth image acquisition module includes a second infrared camera module and a VCSEL projector. The step S10 of dynamically collecting the three-dimensional depth image information of the target pet by the depth image acquisition module can include steps S11-S13. In step S11, the VCSEL projector projects an infrared speckle pattern onto the target pet, and the second infrared camera module collects actual image information corresponding to the infrared speckle pattern projected onto the target pet. It should be noted that the second infrared camera module is an IR camera in the depth image acquisition module, which cooperates with the VCSEL projector to capture the infrared light pattern generated by the VCSEL projector and modulated by the target pet's body surface. The VCSEL projector is a vertical cavity surface emitting laser, which is used as an active light source of the depth image acquisition module and is responsible for projecting an infrared speckle pattern with specific and random distribution characteristics onto the area where the target pet is located. The infrared speckle pattern is a known pattern composed of a large number of small and randomly distributed infrared laser spots formed in space by the VCSEL projector after passing through an optical diffraction element, and the degree of deformation of the pattern can be used to calculate depth information.

[0051] In addition, the second infrared camera module can be consistent with the first infrared camera module, that is, reuse the same infrared camera module, or inconsistent with the first infrared camera module, that is, the pet feeder is provided with two independent infrared camera modules, one of which is used to shoot two-dimensional image information, corresponding to the first infrared camera module, and the other is used to collect the actual image information of the infrared speckle pattern on the pet, corresponding to the second infrared camera module.

[0052] The actual image information refers to the image data of the infrared speckle pattern actually captured by the second infrared camera module from a specific perspective, which has been distorted due to the concave-convex fluctuation of the target pet's body surface. This information records the true state of the original infrared speckle pattern projected by the VCSEL projector after interacting with the three-dimensional object surface.

[0053] In step S12, based on the actual image information, the standard image information corresponding to the preset standard infrared speckle pattern is compared to obtain the infrared speckle pattern deformation information generated by the actual image information compared with the standard image information; It should be noted that the standard image information refers to the original reference pattern image captured and stored by the second infrared camera module during the pre-calibration process when the VCSEL projector projects the same infrared speckle pattern onto a flat reference plane at a known distance. The infrared speckle pattern deformation information is obtained by comparing and analyzing each speckle point in the actual image information with the standard image information. The data set about the offset direction and offset amount of each speckle point relative to its original reference position is directly related to the depth and is the direct basis for calculating the three-dimensional depth image information.

[0054] In step S13, the three-dimensional depth image information of the target pet is determined according to the infrared speckle pattern deformation information.

[0055] It can be understood that in order to stably and accurately dynamically collect the three-dimensional depth image information of the target pet and overcome the shortcomings of ordinary two-dimensional vision technology in depth perception and adaptability to complex environments, in this embodiment, the VCSEL projector actively projects a known infrared speckle pattern, and the second infrared camera module collects the actual image information modulated by the pet's body surface. Then, by comparing the actual image information with the preset standard image information to extract the infrared speckle pattern deformation information, the three-dimensional depth image information is finally calculated accordingly. This can avoid the problems of passive vision technology being greatly affected by environmental texture and light, and inaccurate depth measurement. It realizes active, reliable and high-precision acquisition of pet three-dimensional contour data, that is, three-dimensional depth image information, under various environmental lighting conditions.

[0056] Exemplarily, when the depth information of the target pet needs to be collected, the system first starts the VCSEL projector in the depth image collection module, which projects a preset infrared speckle pattern with a wavelength of 940 nm and containing thousands of randomly distributed infrared spots onto the target pet through the diffraction optical element in front of it. At the same time, the second infrared camera module that is strictly calibrated in position with the VCSEL projector is started synchronously to capture the deformed infrared speckle pattern modulated by the pet's body contour from a specific viewing angle, which is taken as the actual image information. The system microcontroller unit then calls the undeformed original pattern collected on the calibration plane pre-existing in the memory as the standard image information, and obtains the complete infrared speckle pattern deformation information by calculating the pixel offset of each spot in the actual image information relative to the corresponding reference position in the standard image information. Finally, based on the principle of triangulation, the depth values of each point on the pet's body surface are accurately calculated according to these offsets, thereby generating the required three-dimensional depth image information.

[0057] In the present embodiment, by actively projecting a known infrared speckle pattern by the VCSEL projector, collecting the actual image information modulated by the object surface by the second infrared camera module, and then comparing and analyzing the actual image information with the preset standard image information to extract the pattern deformation information and further calculate the three-dimensional depth image, the problem of passive vision technology being greatly affected by environmental texture and light and being difficult to perform accurate depth measurement in weak light or non-texture areas is avoided, and the three-dimensional contour information of the target conflict is actively, reliably and highly accurately obtained under various environmental conditions.

[0058] In a feasible embodiment, the step of adjusting the set parameters in the preset feeding plan according to the adjustment parameters in the feeding optimization strategy in step S40 can include steps S41-S42: Step S41, in the case that the health assessment information is a first non-healthy condition, reducing the lower food amount in the preset feeding plan according to the adjustment parameters, so that the lower food module executes the feeding task according to the reduced lower food amount, wherein the first non-healthy condition includes that the pet body shape index of the target pet is greater than a preset first pet body shape threshold, or the pet body weight index of the target pet is greater than a preset first pet body weight threshold. It should be noted that the pet body shape index refers to one or more quantitative data calculated after the target pet is scanned non-contactly and three-dimensionally by the depth image acquisition module, which is used to quantitatively represent the external shape characteristics of the pet, and the specific source can be that after the three-dimensional depth image information of the target pet is captured by the depth image acquisition module, a three-dimensional contour model of the pet is reconstructed by an algorithm, and then one or more quantitative data such as the volume, the ratio of the body length to the height, or the cross-sectional area of the torso of the pet are calculated based on the model, which are the pet body shape index. The index, as a key health indicator, is directly used to compare with a preset pet body shape threshold to determine whether the pet is in a non-healthy condition such as obesity or emaciation, and the health assessment information.

[0059] The pet weight index refers to an estimated value for quantitatively representing the weight of the pet, which is derived by the system microcontroller unit based on the volume of the three-dimensional model of the pet and combined with a preset animal density parameter after the target pet is scanned non-contactly and three-dimensionally by the depth image acquisition module, and the specific source can be that the system creates a three-dimensional model of the pet and calculates its volume using the depth information obtained by the depth image acquisition module, and then multiplies the volume value with an approximate density value preset according to factors such as the pet species and age, thereby indirectly calculating the target pet weight index. The index, as a key health indicator, is directly used to compare with a preset pet weight threshold to determine whether the pet is in a non-healthy condition such as obesity or emaciation, and the health assessment information.

[0060] In step S42, in the case that the health assessment information is a second non-healthy condition, the lower food amount in the preset feeding plan is increased according to the adjustment parameter, so that the lower food module performs the feeding task according to the increased lower food amount, wherein the second non-healthy condition includes that the pet body shape index of the target pet is less than a preset second pet body shape threshold, or the pet weight index of the target pet is less than a preset second pet weight threshold, the second pet body shape threshold is less than the first pet body shape threshold, and the second pet weight threshold is less than the first pet weight threshold.

[0061] In this embodiment, by explicitly distinguishing the unhealthy conditions into the first unhealthy condition corresponding to obesity and the second unhealthy condition corresponding to emaciation, and setting opposite adjustment actions for the two opposite conditions, i.e., reducing the lower food amount in the preset feeding plan for the first unhealthy condition and increasing the lower food amount for the second unhealthy condition, the abstract adjustment parameter is concretized into operation instructions with explicit mathematical symbols that match specific health problems, thereby avoiding intervention failure and even health risks that may be caused by the unclear binding of health status judgment and feeding amount adjustment direction, such as mistakenly reducing the food amount for an emaciated pet, and preventing the system from making logically correct decisions when facing opposite health problems, so as to realize precise and directional feeding intervention for common and opposite pet health problems. By mapping a specific unhealthy condition to a unique and correct lower food amount adjustment direction, the pertinence, reliability and safety of the feeding optimization strategy are significantly improved.

[0062] Based on the first embodiment of the present application, the same or similar contents as the above embodiment one can be referred to the above introduction, and will not be described in detail hereinafter. On this basis, please refer to Figure 2 , the health assessment information includes obesity degree information and limb movement disorder information, the obesity degree information represents the obesity level of the body type index, and the limb movement disorder information represents the movement disorder level and / or movement disorder type of the limb movement function, and the pet health condition optimization method further includes steps S001-S004: Step S001, obtaining learning sample data, the learning sample data includes first pet movement state information and first pet body type weight information of a plurality of healthy pets, and second pet movement state information and second pet body type weight information of a plurality of unhealthy pets, wherein the healthy pet is a pet with a body weight index meeting a preset standard and without limb movement disorder, and the unhealthy pet is a pet with a body weight index not meeting a preset standard and / or with limb movement disorder, wherein the first pet movement state information and the first pet body type weight information are used as first learning samples, the obesity degree information and the limb movement disorder information of the healthy pet are used as sample labels of the first learning samples, the second pet movement state information and the second pet body type weight information are used as second learning samples, and the obesity degree information and the limb movement disorder information of the unhealthy pet are used as sample labels corresponding to the second learning samples; It should be noted that the learning sample data refers to a set of labeled data used to train and verify the health assessment machine learning model, which explicitly includes historical data of two types of pets: one is related data of a plurality of healthy pets, and the other is related data of a plurality of unhealthy pets. These data together constitute the knowledge base for the model to distinguish between healthy and unhealthy conditions.

[0063] The first pet movement state information is a part of movement data of healthy pets in the learning sample data, which records the typical walking posture, running posture or jumping posture and other characteristics of healthy pets with normal weight indicators and no limb movement disorders during activities, and is used as a feature input of a positive sample in model training; and the first pet body shape and weight information is the body shape and weight data of the same batch of healthy pets matched with the first pet movement state information in the learning sample data, which reflects the standard body shape and weight parameter range that healthy pets should have, and can be used together with the first pet movement state information to form a complete first learning sample.

[0064] The second pet movement state information is a part of movement data of non-healthy pets in the learning sample data, which records the abnormal walking posture, running posture or jumping posture and other characteristics of pets with health problems such as obesity or limb movement disorders, and is used as a feature input of a negative sample in model training; and the second pet body shape and weight information is the body shape and weight data of the same batch of non-healthy pets matched with the second pet movement state information in the learning sample data, which directly reflects the body shape and weight parameters of pets in non-healthy states such as obesity or emaciation, and can be used together with the second pet movement state information to form a second learning sample, and the sample label is the corresponding specific obesity degree information and limb movement disorder information.

[0065] The obesity degree information is a quantitative output of the health assessment machine learning model for the weight and body shape condition of the target pet, which is based on the pet body shape and weight information, and generates a classification or score about the pet's obesity condition through model analysis, such as "normal", "overweight" or "obesity" levels, which is used to quantitatively represent the degree of deviation of the pet from the health standard in terms of weight indicators.

[0066] The limb movement disorder information is a structured output of the health assessment machine learning model for the movement function condition of the target pet, which is based on the pet movement state information, identifies abnormal patterns in the walking, running or jumping posture of the pet through the model, and then judges whether there is a movement disorder and outputs specific movement disorder levels (such as mild, moderate) and / or movement disorder types (such as hip dysplasia, lameness), which is used to accurately describe the health problems of the pet in terms of limb movement function.

[0067] It can be understood that, since only one unverified machine learning model framework cannot achieve reliable evaluation of pet health status, the embodiment systematically collects two types of data that have been explicitly classified, that is, the body shape and weight and movement state information of healthy pets are labeled as "healthy" samples, and the corresponding information of unhealthy pets is bound with the specific obesity level, movement disorder type and other diagnosis results to form a labeled negative sample, thereby constructing a high-quality learning sample dataset, thereby avoiding the problems of inaccurate model evaluation and poor generalization caused by insufficient training data, unclear labeling or single sample, and providing a reliable data foundation for achieving high accuracy and reliable prediction of the health evaluation machine learning model.

[0068] Exemplarily, the historical data of two types of pets are systematically collected; the first type is healthy pets confirmed by comprehensive evaluation as having moderate weight, ideal body condition score and no orthopedic disease record, and the standard walking and running posture data collected by a 3D camera and the corresponding standard body shape and weight data thereof are taken as first learning samples, and all these samples are uniformly labeled with sample labels of "obesity level: normal, and limb movement disorder: none"; the second type is pets diagnosed with specific health problems, for example, the related data of pets diagnosed with obesity, and the abnormal gait data and body shape data of pets diagnosed with arthritis-induced lameness are taken as second learning samples, and the corresponding specific sample labels such as "obesity level: level 3" or "movement disorder type: left hind limb lameness" are accurately labeled, thereby constructing a structured dataset for model training.

[0069] In step S002, the learning sample data is divided into training sample data and verification sample data according to a preset proportion; In step S003, the preset health evaluation machine learning model is trained by using the training sample data, and the identification accuracy of the preset health evaluation machine learning model is verified by using the verification sample data. In step S004, if the identification accuracy of the verification reaches a preset accuracy threshold, the training of the health evaluation machine learning model is stopped, and a trained health evaluation machine learning model is obtained.

[0070] It can be understood that, since there is a risk of overfitting in health assessment using only an unverified machine learning model, which can lead to a model performing well on training data but having low accuracy in actual application, by dividing the learning sample data into a training set and a validation set according to a preset ratio, using the training set to train the parameters of the model, and simultaneously using the validation set to independently evaluate the recognition accuracy, and stopping training once a preset threshold is reached, the problem of poor generalization ability and unstable actual prediction performance caused by over-learning of the noise and specific rules of the training sample can be effectively avoided, and finally an optimized and verified health assessment machine learning model with high accuracy and strong generalization ability is obtained, providing a reliable guarantee for reliable prediction of the health status of the pet.

[0071] Exemplarily, the system divides the acquired structured learning sample data into training sample data and validation sample data according to a preset ratio such as 7:3 after randomizing the data, and then uses the training sample data to perform multi-round iterative training on a preset machine learning model to learn the complex mapping relationship between the pet motion state information, pet body shape and weight information, and health label. After each round of training is completed, the performance of the current model is tested using the reserved validation sample data, and the recognition accuracy for health, obesity and various motion disorders is calculated. When the comprehensive recognition accuracy of the model on the validation set reaches and stabilizes at a preset accuracy threshold of 95% for three consecutive evaluations, the system automatically stops the training process, and the model parameters at this moment are solidified as the final trained health assessment machine learning model for subsequent actual prediction tasks.

[0072] In this embodiment, by acquiring the motion state and body shape and weight information of healthy and unhealthy pets as labeled learning sample data, and dividing the data into a training set and a validation set, the training set is used to train the preset model while the validation set is used to continuously evaluate the recognition accuracy until the training is stopped when the preset threshold is reached, avoiding the problem of inaccurate health assessment results and poor generalization ability caused by insufficient and non-standard training and verification of the model, and realizing an optimized and verified health assessment machine learning model with high accuracy and strong generalization ability, which provides a core guarantee for reliable prediction of the health status of the pet.

[0073] Based on the first and / or second embodiments of the present application, in the third embodiment of the present application, the same and / or similar contents as the above-mentioned first and second embodiments can be referred to the above introduction, and will not be described in detail hereinafter. On this basis, please refer to Figure 3 , the pet feeder further comprises at least two feeders corresponding to the lower food mold module, and a weighing sensor at the bottom of the feeder, and the pet health status optimization method further comprises steps S401-S403: Step S401, monitoring the food weight in each feeder through the weighing sensor; It should be noted that the weighing sensor is a mechanical measuring element installed at the bottom of at least two feeders corresponding to the lower food module in the pet feeder, which is used to convert the pressure on the feeder and its contents into an electrical signal. By real-time sensing the total weight change of the feeder (including the food contained therein), the food weight is accurately monitored, thereby providing basic data for subsequent calculation of food consumption.

[0074] Step S402, evaluating the preference degree of the target pet for the food in each feeder according to the change of the food weight; It should be noted that the preference degree refers to the evaluation result obtained by comparing and analyzing the change of the food weight in each feeder monitored by the weighing sensor, such as the reduction rate or total amount of food weight per unit time, which is used to quantitatively represent the love level of the target pet for different food types in different feeders. For example, in the same time, the feeder with faster food weight reduction is determined to have higher preference degree.

[0075] Step S403, adjusting the food-down ratio of each feeder in the preset feeding plan based on the preference degree, so that the lower food module executes the feeding task according to the adjusted food-down ratio.

[0076] It should be noted that the food-down ratio refers to the proportion or ratio of the respective food-down amount of at least two feeders in the total food-down amount in a single or specific period, which is set as a key parameter in the preset feeding plan and can be adjusted based on the preference degree. For example, the proportion of the feeder corresponding to the high preference degree food is increased, and the proportion of the feeder corresponding to the low preference degree food is reduced, so as to guide or meet the eating tendency of the pet.

[0077] In addition, it should be noted that during the adjustment of the food-down ratio of each feeder in the preset feeding plan based on the preference degree, the system also sets an adjustable range, i.e. upper limit and lower limit, for the food-down ratio of each feeder. For example, even if the preference degree of a certain food is very high, the adjusted food-down ratio of the corresponding feeder should not exceed the preset upper limit, such as 70% of the total feeding amount, to prevent the pet from causing nutritional imbalance due to severe food preference; similarly, even if the preference degree of a certain food is very low, the adjusted food-down ratio of the corresponding feeder should not be lower than the preset lower limit, such as 10% of the total feeding amount, to ensure that the pet can intake the necessary basic nutrients, thereby meeting the eating preference of the pet while ensuring the rationality and health safety of the diet structure.

[0078] It can be understood that, since the current scheme mainly solves the global regulation problem of feeding amount, but does not consider that the pet may have a preference for different food types, if only the total amount is adjusted and the proportion of food types is ignored, it may lead to that the pet cannot intake balanced nutrition due to picky eating, and cannot achieve real effective health optimization. Therefore, a hardware architecture with at least two feeders and a weighing sensor is introduced, the preference degree of the target pet for the food in each feeder is evaluated by monitoring the change of the food weight in each feeder through the weighing sensor, and the feeding proportion of each feeder in the preset feeding plan is adjusted accordingly, so that the feeding module executes the feeding task according to the optimized proportion, thereby avoiding the problem of nutritional imbalance that the pet may still have excessive or insufficient intake of a specific nutrient even if the total feeding amount meets the requirements due to picky eating, and the problem of reduced willingness to eat due to ignoring individual food preferences, achieving fine management of food intake structure on the basis of controlling total intake, guiding balanced diet by dynamically adapting to the pet's taste preference, and thus more effectively optimizing the pet's health condition while respecting individual habits.

[0079] Illustratively, the pet feeder is provided with two feeders for storing dry food and wet food, and the food weight of the two is continuously monitored through the weighing sensor at the bottom thereof. The system calculates that the dry food is reduced by 20 grams and the wet food is reduced by 50 grams according to the monitored food weight change in a preset time period, and accordingly evaluates that the target pet has a significantly higher preference for wet food than for dry food. Therefore, the system adjusts the feeding proportion in the preset feeding plan based on the preference, i.e., adjusts the feeding proportion of dry food and wet food from the initial 1:1 to 3:7, and finally the feeding module executes the feeding task according to the adjusted feeding proportion, so as to achieve the optimization goal of guiding eating and nutritional balance.

[0080] Illustratively, the pet feeder is provided with three feeders for storing ordinary dry food, high-protein formula food and health care nutrition food, and the food weight change of the three is continuously monitored through the weighing sensor at the bottom of each feeder. The system calculates that the ordinary dry food is reduced by 30 grams, the high-protein formula food is reduced by 60 grams, and the health care nutrition food is reduced by only 10 grams according to the monitoring data within 24 hours, and accordingly evaluates that the target pet has the highest preference for high-protein formula food, the second highest preference for ordinary dry food, and the lowest preference for health care nutrition food. Based on this preference, the system adjusts the feeding proportion in the preset feeding plan while maintaining the total heat, i.e., adjusts the feeding proportion of the three from the initial 1:1:1 to 2:6:2, and sets an upper limit (such as 70%) and a lower limit (such as 10%) for each type of food, and finally the feeding module executes the feeding task according to the adjusted feeding proportion and the range constraint, so as to achieve the optimization goal of nutritional balance and health guidance.

[0081] In this embodiment, the change of the food weight in each feeder is monitored by the weighing sensor arranged at the bottom of at least two feeders, and then the preference degree of the target pet for different food is evaluated, and the proportion of the food in each feeder in the preset feeding plan is adjusted based on the preference degree, thereby avoiding the problems of single or unbalanced nutrition intake of the pet in the feeding process due to the inherent picky eating habit of the pet, strong subjectivity, response lag and complicated operation of manual observation and manual adjustment of the food ratio, realizing the automatic and data-based fine management of the pet diet structure, intelligently guiding the pet to intake more balanced nutrition by dynamically adapting to the taste preference of the pet, thereby effectively improving the feeding willingness and efficiently optimizing the diet health level of the pet.

[0082] In a feasible implementation, the pet feeder further comprises an angle calculation module, and before step S402, steps S100-S200 can be further included. Step S100: monitoring the inclination angle of the feeder by the angle calculation module; It should be noted that the inclination angle refers to the angle between the bottom plate plane of the pet feeder and the horizontal reference plane measured by the angle calculation module (including a 6-axis sensor) in the pet feeder, such as a 3-axis accelerometer and a 3-axis gyroscope in the 6-axis sensor. The angle is a dynamic parameter for quantifying whether the feeder is in a horizontal stable state, and the core function is to determine whether the food weight data measured by the weighing sensor needs to be corrected due to the inclination of the equipment.

[0083] Step S200: in the case where the inclination angle is less than a preset angle threshold, correcting the food weight according to the inclination angle to determine the change of the food weight based on the corrected food weight.

[0084] It can be understood that, since the scheme of adjusting the food consumption ratio based on the preference degree highly depends on the accuracy of the food weight data monitored by the weighing sensor, however, when the pet feeder is tilted due to the pet collision or the ground inequality, the reading of the weighing sensor will produce a significant error, and if the inaccurate data is directly used to evaluate the preference degree, it will lead to the wrong decision of the subsequent food consumption ratio adjustment, and the expected health optimization effect cannot be achieved. Therefore, the angle calculation module is introduced to continuously monitor the tilt angle of the feeder, and a preset angle threshold is set as the judgment basis. Only when the tilt angle is less than the threshold, the food weight is corrected in real time according to the tilt angle, and the subsequent preference evaluation and ratio adjustment process is executed based on the corrected data, thereby avoiding the distortion of the weighing data due to the unstable physical posture of the feeder, and further causing a series of chain decision errors such as misjudgment of pet food preference and misadjustment of food consumption ratio, fundamentally preventing the risk of executing health intervention strategies based on incorrect data, significantly improving the robustness and reliability of the entire system data acquisition link, and ensuring the accuracy and effectiveness of the preference degree evaluation and food consumption ratio adjustment, so that the pet health condition optimization strategy can still be stably and accurately executed in complex actual use environment.

[0085] Exemplarily, the 6-axis sensor built-in the pet feeder continuously monitors the tilt angle of the device chassis. When the tilt angle is 3 degrees, which is less than the preset 5-degree angle threshold, the system first obtains the original reading of each weighing sensor, and then establishes a space coordinate system conversion model based on the tilt angle, calculates the direction cosine of the gravitational acceleration on the sensitive axis of each sensor in the tilted state, solves the error value of the gravity component caused by the device tilt, and deducts the error value from the original weight reading in real time to finally obtain the corrected accurate food weight data, thereby ensuring the accuracy of the subsequent preference degree evaluation.

[0086] In addition, in the case where the tilt angle is greater than or equal to the preset angle threshold, the system will trigger a safety protection mechanism. The specific technical scheme is that, when the tilt angle monitored by the angle calculation module reaches or exceeds the preset safety threshold (i.e. the preset angle threshold, for example, 30 degrees), the system determines that the pet feeder is in a dangerous state of possible tipping or has been overturned, immediately generates a power-off instruction and cuts off the power supply of the whole machine, and forces all modules (including the food consumption module, the laser projection module, etc.) to stop running, thereby avoiding the damage of internal components, the spilling of food, or the physical injury such as pinching or electric shock to the curious pet caused by the tipping of the device, and ensuring the safety of the pet and the device.

[0087] In this embodiment, the angle calculation module continuously monitors the inclination angle of the feeder, and when the inclination angle is less than the preset angle threshold, a mechanical model is established according to the inclination angle to correct the food weight measured by the load sensor in real time, which avoids the problem that when the pet feeder is slightly tilted due to uneven placement or collision with the pet, the measurement data of the load sensor is inaccurate due to the interference of the gravity component, which further leads to the distortion of the subsequent evaluation results of the pet food preference degree and the complete deviation of the subsequent food proportion adjustment based on the incorrect evaluation from the expected target, and ensures the accuracy and reliability of the weight data at the source of the entire health management data chain, thereby significantly improving the execution accuracy and overall optimization effect of the pet food preference analysis and subsequent feeding optimization strategy.

[0088] Based on the first embodiment and / or the second embodiment and / or the third embodiment of the present application, in the fourth embodiment of the present application, the same and / or similar contents as the above-mentioned embodiments one, embodiments two and embodiments three can be referred to the above introduction, and the subsequent will not be described in detail. On this basis, please refer to Figure 4 , the pet feeder further comprises a laser projection module, and after step S30, the pet health condition optimization method can further comprise steps S50-S60: Step S50, determining a motion optimization strategy based on the health assessment information, calling the laser projection module to project a pet interesting pattern to a preset area according to the motion optimization strategy, and monitoring the actual position of the target pet through the two-dimensional image acquisition module and / or the depth image acquisition module; It should be noted that the laser projection module refers to an optical device in the pet feeder for projecting dynamic patterns to attract the attention of pets, which can generate pet interesting patterns of preset shapes such as mice and butterflies. By changing the pattern position and moving frequency, the laser projection module can stimulate the pursuit behavior of pets and promote their movement. Among them, the pet interesting pattern refers to a dynamic image generated by the laser projection module and projected to the preset area, which is designed to stimulate the pursuit instinct of pets, such as simulating small animals or moving light points. The projection position of the pattern changes after each successful pursuit of the pet to maintain the interactivity with the target pet and prolong the movement time of the target pet.

[0089] The actual position of the target pet refers to the space point position of a specific part of the target pet's body (such as the head or the front paw) in the preset coordinate system obtained by real-time monitoring of the target pet through the two-dimensional image acquisition module and / or the depth image acquisition module. The position is a dynamic data with three-dimensional coordinate information calculated in real time by a visual analysis algorithm based on the image or point cloud data captured by the two-dimensional image acquisition module and / or the depth image acquisition module. The core use of the position is to compare with the projection position corresponding to the preset area to determine whether the target pet successfully pursues the pet interesting pattern.

[0090] Step S60, in the case of the actual position coinciding with the preset area, determining that the target pet chases to the pet interesting pattern, changing the projection position of the pet interesting pattern; Wherein, the motion optimization strategy affects the calling frequency of the laser projection module, the distance between the projection position and the target pet, or the changing frequency of the projection position.

[0091] It should be noted that the projection position of the pet interesting pattern refers to the optical center or the optical edge of the pet interesting pattern projected by the laser projection module, which is a spatial point defined in the same preset coordinate system corresponding to the preset area. This position is a position coordinate pre-set or generated in real time by the system to instruct the laser projection module to project the pattern to a certain position. The key is that this position is a control target that can be dynamically changed to guide the target pet to move and achieve the function of optimizing the health status based on the motion.

[0092] For example, when the target optimization strategy is determined to be the motion optimization strategy, the system calls the laser projection module to project a light spot as the pet interesting pattern to the ground in front of the pet feeder (i.e., the preset area), and simultaneously starts the two-dimensional image acquisition module and / or the depth image acquisition module to continuously monitor and calculate the actual position of the target pet (e.g., the pet's head) at a frequency of several frames per second. The system compares the actual position with the current projection position of the pet interesting pattern in real time, and when it is determined that the actual position coincides with the projection position within a preset tolerance range, it is determined that the target pet chases to the pet interesting pattern, and immediately changes the projection position to another random coordinate point one meter away by the laser projection module. This cycle constitutes a continuous interactive chasing task.

[0093] In this embodiment, the pet interesting pattern is first projected to the preset area by the laser projection module, and the actual position of the target pet is continuously monitored by the two-dimensional image acquisition module and / or the depth image acquisition module. A specific determination condition is set that when the actual position coincides with the preset area, it is determined that the chasing is successful, which triggers the control instruction to change the projection position of the pet interesting pattern. Thus, the problem of incoherent interactive process and ineffective maintenance of pet interest or interactive determination failure due to lack of specific triggering determination mechanism and position control logic is avoided, and the technical defect that the system can only perform simple and fixed projection and cannot form effective dynamic guidance is prevented. The automated, closed-loop, and precise execution of the motion optimization strategy is achieved, and through the establishment of a real-time control loop of "projection-monitoring-determination-change", it is ensured that the pet can be continuously attracted to participate in the chasing motion, significantly improving the optimization effect and reliability of promoting pet motion through interactive entertainment to optimize the health status of the pet.

[0094] For example, in order to facilitate the understanding of the implementation process of the pet health status optimization method obtained after the above-mentioned embodiment one, please refer to Figure 5 , Figure 5 A brief flowchart of a pet health status optimization method is provided, specifically: In the data acquisition stage, the original image information is acquired from the two-dimensional image acquisition module and the depth image acquisition module, so as to perform denoising and calibration processing on the original image information in the image denoising stage. Then in the feature extraction stage, key features such as pet body weight information and pet motion state information are acquired from the processed original image information, so as to use the trained health assessment machine learning model to evaluate the health status of the target pet in the health assessment stage, and obtain health assessment information. If the health assessment information represents a healthy condition, the normal feeding function of the pet feeder is started, and the target pet is fed according to the standard food amount; if the health assessment information represents an unhealthy condition, such as obesity, the feeding optimization strategy and / or the exercise optimization strategy are executed. After the feeding is completed, the pet feeder can also generate a health report of the target pet regularly, and send a notification to the corresponding mobile application to inform the user of the pet-related information.

[0095] Among them, the pet feeder applied in the embodiment can refer to Figure 6 , Figure 6 A schematic diagram of the appearance of a pet health status optimization method is provided, specifically: The pet feeder includes a left food storage cabinet 1, a right food storage cabinet 2, a laser projection module 3, an ALS sensor 4, a TOF sensor 5, a microphone 6, a loudspeaker 7, a 940 nm VCSEL projector 8, a 940 nm infrared LED lamp 9, an IR camera 10, an indicator light 11, an RGB camera 12, a left feeding bowl 13, a right feeding bowl 14, and a base 15.

[0096] It should be noted that the TOF sensor 5 is used to monitor the actual distance between the target pet and the pet feeder; in the case that the actual distance is less than the preset distance threshold, the laser projection module 3 is turned off; after the laser projection module 3 is turned off, and in the case that the actual distance is greater than the preset distance threshold, the laser projection module 3 is controlled to resume projecting the pet interesting pattern to the preset area.

[0097] The IR camera 10 cooperates with the 940 nm VCSEL projector 8 to realize the function of a 3D structured light camera.

[0098] In addition, the microphone 6 and the speaker 7 construct a two-way audio interaction system, so as to realize intelligent voice control and behavior interference, and realize real-time intercom through the terminal application program corresponding to the pet feeder. The indicator light 11 can be a red-green-blue three-color light, which is used to indicate the working state of the device, and displays green light when the device is normal, displays red light when the device is faulty, and displays blue light when the device is performing firmware upgrade.

[0099] Further, the scene diagram of the pattern of interest to the pet projected by the laser projection module 3 can refer to Figure 7 , Figure 7 The pattern of interest to the pet projected by the laser projection module 3 is a mouse.

[0100] The system architecture of the system constructed based on the pet health status optimization method of the embodiment can refer to Figure 8 , Figure 8 A system architecture diagram of a pet health status optimization method is provided, in particular: The pet feeder can be divided into six parts of a sensor module, a video module, a control module, a storage module, a human-computer interaction module, and a communication module according to functions, which are controlled by an MCU (Microcontroller Unit) and coordinated by the MCU to work, and simultaneously serve as a communication bridge between different modules. The sensor module includes four components of a TOF sensor, an ALS sensor, a 6-axis sensor, and a weighing sensor; the video module includes two components of an RGB camera and an IR camera; the control module includes two motors and a motor control circuit, a left motor controls the switch of a left food outlet, a right motor controls the switch of a right food outlet, and the weighing sensor can realize accurate food feeding; the storage module includes an SD (Secure Digital) card for local storage of video streams and photos captured by the video module; the human-computer interaction module includes four components of a microphone, a speaker, a laser projection module, and an indicator light; and the communication module includes a WIFI / BT (Blue Tooth) module, which stores videos or pictures in the cloud through the WIFI / BT module, and the terminal application program is interconnected with the cloud to realize remote monitoring, build a pet community, and simultaneously realize remote upgrade of the firmware of the pet feeder.

[0101] It should be noted that the above examples are only used for understanding the present application and do not constitute a limitation on the pet health status optimization method of the present application, and more forms of simple transformation based on the technical concept are within the protection scope of the present application.

[0102] The application provides a pet feeder, comprising: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the pet health condition optimization method in the above embodiment one.

[0103] Reference will now be made to the following description Figure 9 which shows a structural schematic diagram of a pet feeder suitable for implementing the embodiments of the application. Figure 9 The pet feeder shown is merely an example and should not impose any limitation on the functions and use range of the embodiments of the application.

[0104] As Figure 9 shown, the pet feeder can include a processing device 1001 (for example, a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to programs stored in a read-only memory 1002 or programs loaded from a storage device 1003 into a random access memory 1004. In the random access memory 1004, various programs and data required for the operation of the pet feeder are also stored. The processing device 1001, the read-only memory 1002, and the random access memory 1004 are connected to each other through a bus 1005. An input / output interface 1006 is also connected to the bus. Generally, the following systems can be connected to the input / output interface 1006: input devices 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; the storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the pet feeder to communicate with other devices wirelessly or by wire to exchange data. Although the pet feeder with various systems is shown in the figure, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be alternatively implemented or provided.

[0105] In particular, according to the embodiments disclosed in the application, the processes described above with reference to the flowcharts can be implemented as a computer software program. For example, the embodiments disclosed in the application include a computer program product comprising a computer program carried on a computer readable medium, which contains program codes for executing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network through the communication device, or installed from the storage device 1003, or installed from the read-only memory 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the methods of the embodiments disclosed in the application are performed.

[0106] The pet feeder provided by the present application adopts the pet health condition optimization method in the above embodiment, and can solve the technical problem of how to improve the health condition of the pet through the pet feeder. Compared with the prior art, the pet feeder provided by the present application has the same beneficial effects as the pet health condition optimization method provided by the above embodiment, and other technical features in the pet feeder are the same as the features disclosed in the last embodiment method, which will not be described here.

[0107] It should be understood that parts of the present application can be realized by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0108] The above is merely specific implementation of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0109] The present application provides a computer readable storage medium having computer readable program instructions (i.e. computer programs) stored thereon, the computer readable program instructions being used to execute the pet health condition optimization method in the above embodiment.

[0110] The computer readable storage medium provided by the present application may, for example, be a U disk, but is not limited to an electric, magnetic, optical, electromagnetic, infrared, or semiconductor system or device, or any combination thereof. More specific examples of the computer readable storage medium can include, but are not limited to: an electric connection with one or more conductive wires, a portable computer disk, a hard disk, a random access memory (RAM: Random Access Memory), a read-only memory (ROM: Read Only Memory), an erasable programmable read-only memory (EPROM: Erasable Programmable Read Only Memory or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM: CD-Read Only Memory), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present embodiment, the computer readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system or device. The program code contained on the computer readable storage medium can be transmitted by any appropriate medium, including but not limited to: electric wires, optical cables, RF (Radio Frequency: radio frequency), etc., or any suitable combination thereof.

[0111] The computer readable storage medium can be included in the pet feeder or can exist separately from the pet feeder.

[0112] The computer readable storage medium carries one or more programs, which, when executed by the pet feeder, cause the pet feeder to: dynamically collect two-dimensional image information of a target pet through a two-dimensional image collection module, and dynamically collect three-dimensional depth image information of the target pet through a depth image collection module; identify pet body shape and weight information of the target pet according to the three-dimensional depth image information, and comprehensively analyze a pet activity state of the target pet according to the two-dimensional image information and the three-dimensional depth image information to obtain pet motion state information, wherein the pet motion state information is used to represent a walking posture, a running posture, or a jumping posture of the target pet; input the pet motion state information and the pet body shape and weight information into a trained health assessment machine learning model to predict health assessment information of the target pet; determine a feeding optimization strategy based on the health assessment information, and adjust a set parameter in a preset feeding plan according to an adjustment parameter in the feeding optimization strategy, so that the lower feeding module executes a feeding task for the target pet according to the adjusted set parameter.

[0113] Computer program code for carrying out operations of the present application can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider). The application is not limited to a specific programming language or computer system.

[0114] The computer readable program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other device to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide processes for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0115] The modules involved in the embodiments of the present application can be implemented in the form of software or hardware. In some cases, the name of the module does not constitute a limitation on the module itself.

[0116] The computer readable storage medium provided by the present application is a computer readable storage medium, which stores computer readable program instructions (i.e. computer program) for executing the pet health condition optimization method, and can solve the technical problem of how to improve the health condition of the pet through the pet feeder. Compared with the prior art, the computer readable storage medium provided by the present application has the same beneficial effects as the pet health condition optimization method provided by the above-mentioned embodiments, which will not be repeated here.

[0117] The above only describes some embodiments of the present application, and does not limit the patent scope of the present application. Any equivalent structural transformation made by using the content of the present application specification and drawings, or direct / indirect application in other related technical fields is included in the patent protection scope of the present application.

Claims

1. A method for optimizing pet health status, characterized in that, An application to a pet feeder, the pet feeder comprising a feeding module, a two-dimensional image acquisition module, and a depth image acquisition module, the method comprising: The two-dimensional image acquisition module dynamically acquires two-dimensional image information of the target pet, and the depth image acquisition module dynamically acquires three-dimensional depth image information of the target pet. Based on the three-dimensional depth image information, the pet's body size and weight information are identified, and based on the two-dimensional image information and the three-dimensional depth image information, the pet's activity status is comprehensively analyzed to obtain pet movement status information, wherein the pet movement status information is used to characterize the target pet's walking posture, running posture, or jumping posture. The pet's movement status information and the pet's body size and weight information are input into a trained health assessment machine learning model to predict the health assessment information of the target pet. Based on the health assessment information, a feeding optimization strategy is determined, and the setting parameters in the preset feeding plan are adjusted according to the adjustment parameters in the feeding optimization strategy, so that the feeding module can perform the feeding task for the target pet according to the adjusted setting parameters.

2. The pet health status optimization method as described in claim 1, characterized in that, The pet feeder also includes an RGB camera module, a first infrared camera module, an infrared radiation lamp, and an ambient light intensity ALS sensor; the method further includes: The current ambient light intensity value is detected by the ALS sensor; When the ambient light intensity value reaches a preset light intensity threshold, the RGB camera module is used as the two-dimensional image acquisition module, and the RGB camera module is turned on to acquire two-dimensional image information of the target pet; When the ambient light intensity value does not reach the preset light intensity threshold, the first infrared camera module is used as the two-dimensional image acquisition module, the infrared radiation lamp is turned on to illuminate the target pet, and the first infrared camera module is turned on to acquire the two-dimensional image information of the target pet.

3. The method for optimizing pet health status as described in claim 2, characterized in that, The depth image acquisition module includes a second infrared camera module and a VCSEL projector. The step of dynamically acquiring three-dimensional depth image information of the target pet through the depth image acquisition module includes: The infrared speckle pattern is projected onto the target pet through the VCSEL projector, and the actual image information corresponding to the infrared speckle pattern projected onto the target pet is acquired by the second infrared camera module. Based on the actual image information, it is compared with the standard image information corresponding to the preset standard infrared speckle pattern to obtain the infrared speckle pattern deformation information generated by the actual image information compared with the standard image information; Based on the infrared speckle pattern deformation information, the three-dimensional depth image information of the target pet is determined.

4. The method for optimizing pet health status as described in claim 1, characterized in that, The health assessment information includes obesity level information and limb movement disorder information. The obesity level information represents the obesity grade in terms of weight and body shape indicators, and the limb movement disorder information represents the level and / or type of movement disorder in terms of limb motor function. The method further includes: Acquire learning sample data, which includes first pet movement status information and first pet body size and weight information of various healthy pets, and second pet movement status information and second pet body size and weight information of various unhealthy pets. The healthy pets are those whose weight indicators meet preset standards and do not have limb movement disorders. The unhealthy pets are those whose weight indicators do not meet preset standards and / or have limb movement disorders. The first pet movement status information and first pet body size and weight information serve as the first learning sample; the obesity degree information and limb movement disorder information of the healthy pets serve as the sample labels for the first learning sample; the second pet movement status information and second pet body size and weight information serve as the second learning sample; and the obesity degree information and limb movement disorder information of the unhealthy pets serve as the sample labels for the corresponding second learning sample. The learning sample data is divided into training sample data and validation sample data according to a preset ratio; The preset health assessment machine learning model is trained using the training sample data, and the recognition accuracy of the preset health assessment machine learning model is verified using the verification sample data. If the verified recognition accuracy reaches a preset accuracy threshold, then training the health assessment machine learning model is stopped, and a trained health assessment machine learning model is obtained.

5. The method for optimizing pet health status as described in claim 1, characterized in that, The step of adjusting the settings in the preset feeding plan according to the adjustment parameters in the feeding optimization strategy includes: When the health assessment information indicates a first unhealthy condition, the amount of food in the preset feeding plan is reduced according to the adjustment parameters, so that the feeding module performs the feeding task according to the reduced amount of food. The first unhealthy condition includes the target pet's pet size index being greater than a preset first pet size threshold, or the target pet's pet weight index being greater than a preset first pet weight threshold. If the health assessment information indicates a second unhealthy condition, the amount of food in the preset feeding plan is increased according to the adjustment parameters, so that the feeding module performs the feeding task according to the increased amount of food. The second unhealthy condition includes the target pet's pet size index being less than a preset second pet size threshold, or the target pet's pet weight index being less than a preset second pet weight threshold, the second pet size threshold being less than the first pet size threshold, and the second pet weight threshold being less than the second pet weight threshold.

6. The method for optimizing pet health status as described in claim 1, characterized in that, The pet feeder also includes at least two feeders corresponding to the feeding module, and a weighing sensor at the bottom of the feeder. The pet health status optimization method further includes: The weight of the food in each of the feeders is monitored by the weighing sensor. Assess the target pet's preference for the food in each of the feeders based on the changes in the weight of the food; Based on the preference level, the feeding ratio of each feeder in the preset feeding plan is adjusted so that the feeding module performs the feeding task according to the adjusted feeding ratio.

7. The method for optimizing pet health status as described in claim 6, characterized in that, The pet feeder also includes an angle calculation module. Before the step of assessing the target pet's preference for food in each feeder based on changes in food weight, the method further includes: The angle calculation module monitors the tilt angle of the feeder. If the tilt angle is less than a preset angle threshold, the weight of the food is corrected according to the tilt angle, so as to determine the change in the weight of the food based on the corrected weight.

8. The method for optimizing pet health status as described in claim 1, characterized in that, The pet feeder also includes a laser projection module, and after the step of predicting the health assessment information of the target pet, the method further includes: Based on the health assessment information, a movement optimization strategy is determined. According to the movement optimization strategy, the laser projection module is invoked to project a pattern of interest to the pet onto a preset area, and the actual position of the target pet is monitored through the two-dimensional image acquisition module and / or depth image acquisition module. If the actual position coincides with the preset area, it is determined that the target pet chases the pattern of interest to the pet, and the projection position of the pattern of interest to the pet is changed; The motion optimization strategy affects the calling frequency of the laser projection module, the distance between the projection position and the target pet, or the frequency of change of the projection position.

9. A pet feeder, characterized in that, The pet feeder includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the pet health optimization method as described in any one of claims 1 to 8.

10. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the pet health status optimization method as described in any one of claims 1 to 8.

Images