Barreled water automatic selling system of unmanned water changing cabinet

Abstract

The invention relates to the technical field of intelligent selling, and discloses a barreled water automatic selling system of an unmanned water changing cabinet, the system comprises an identity authentication module, an empty barrel recovery module, a full barrel water detection module, a full barrel confirmation module, a payment module and a water taking module, the identity authentication certificate of the unmanned water changing cabinet is generated according to an operation instruction; the recycling module is used for opening a recycling bin door of the unmanned water changing cabinet based on the identity authentication credential, detecting the weight of an empty barrel thrown by a user and generating a recycling credential; activating a depth camera and a weighing element of a full bin in the unmanned water changing cabinet, and synchronously obtaining a depth image and real-time weight data; verifying the state of the full water, and outputting a barrel body integrity identifier of the full water; when the barrel integrity identifier passes verification, a dynamic payment voucher generator is triggered, an encrypted payment instruction is generated in combination with the identity authentication voucher and the barrel integrity identifier, and the encrypted payment instruction is sent to a payment gateway; according to the invention, the accuracy of automatic vending of barreled water can be improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent vending technology, and in particular to an automated bottled water vending system with an unmanned water tank. Background Technology

[0002] In the field of automated bottled water vending machines with unmanned water tanks, existing technologies have significant limitations in the user authentication process. The authentication process is cumbersome and lacks accuracy, often resulting in long waiting times for users due to low efficiency in biometric matching or delays in generating dynamic vouchers, severely impacting the user experience. Simultaneously, the empty bottle recycling mechanism is inadequate, lacking precise filtering and noise reduction for the weight detection of empty bottles, making it susceptible to errors caused by external interference and difficult to reliably generate valid recycling vouchers, thus hindering the smooth operation of the subsequent full-bottle vending process.

[0003] Existing technologies also have shortcomings in the verification of full water status and transaction settlement. The detection of full water relies heavily on single-dimensional data and lacks a comprehensive analysis of the three-dimensional shape of the bucket, making it difficult to accurately determine the integrity of the bucket. This may result in defective products entering the user's end. The encryption mechanism of the payment process is imperfect, transaction information is easily tampered with, and the response speed of payment instruction generation and transmission is slow. This not only poses security risks but also reduces the overall system's operating efficiency, failing to meet the efficient and secure operational requirements of unmanned vending scenarios. Summary of the Invention

[0004] This invention provides an automated bottled water vending system with an unmanned water exchange tank to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides an automated bottled water vending system with an unmanned water exchange cabinet, characterized in that the system includes an information extraction module, a product verification module, a verification failure module, a verification success module, a product settlement module, and a settlement success module, wherein:

[0006] The system includes an identity authentication module, an empty bucket recycling module, a full bucket detection module, a full bucket confirmation module, a payment module, and a water dispensing module, among which:

[0007] The identity authentication module is used to generate identity authentication credentials for the unmanned water changing tank based on the user's operation instructions;

[0008] The empty bucket recycling module is used to open the recycling compartment door of the unmanned water exchange cabinet based on the identity authentication credentials, detect the weight of the empty bucket put in by the user, and generate a recycling certificate for the empty bucket.

[0009] The full water detection module is used to activate the depth camera and weighing unit in the full water compartment of the unmanned water exchange cabinet according to the empty water recycling certificate, and simultaneously acquire the depth image and real-time weight data of the full water.

[0010] The full bucket confirmation module is used to verify the state of the full bucket based on the preset bucket shape rules, the depth image and the real-time weight data, and output the bucket integrity identifier of the full bucket.

[0011] The payment module is used to trigger a dynamic payment credential generator when the bucket integrity identifier is verified, and to generate an encrypted payment instruction by combining the identity authentication credential and the bucket integrity identifier, and send it to the payment gateway.

[0012] The water intake module is used to receive the payment success signal returned by the payment gateway and control the electromagnetic lock of the water intake port to open.

[0013] In a preferred embodiment, when the identity authentication module generates an identity authentication credential for the unmanned water changing tank based on the user's operation instructions, it is specifically used for:

[0014] Receive biometric data from users in unmanned water changing tanks;

[0015] The biometric data is matched and verified against a pre-stored template;

[0016] When a match is successful, an encrypted dynamic token is generated for the user as an identity authentication credential.

[0017] The identity authentication credential is synchronized to the access control controller of the unmanned water exchange tank.

[0018] In a preferred embodiment, when the empty bucket recycling module performs the following actions: opening the recycling compartment door of the unmanned water exchange tank based on the identity authentication credential, detecting the weight of the empty bucket deposited by the user, and generating a recycling credential for the empty bucket, it is specifically used for:

[0019] The pressure sensing element is activated after the recovery compartment door is opened;

[0020] Acquire the weight analog signal continuously output by the pressure sensing element;

[0021] The simulated weight signal is filtered and denoised to obtain a stable sample value of the empty bucket;

[0022] When the stable sampled value falls within the preset empty bucket weight range, a recycling certificate for the empty bucket is generated.

[0023] In a preferred embodiment, after generating a recycling certificate for the empty bucket, the full water detection module is specifically used for:

[0024] Scan the barcode label on the empty bucket;

[0025] Extract the manufacturer code from the barcode identifier;

[0026] By associating the manufacturer code with the recycling certificate, the traceability record of the empty bucket is obtained.

[0027] In a preferred embodiment, when the full water detection module activates the depth camera and weighing unit of the full water compartment in the unmanned water exchange cabinet according to the empty water recycling certificate, and simultaneously acquires the depth image and real-time weight data of the full water, it is specifically used for:

[0028] The infrared ranging sensor is triggered to calibrate the focal length of the depth camera based on the recycling certificate;

[0029] The zero-point drift compensation program of the weighing element is started simultaneously;

[0030] After compensation is completed, the depth image and real-time weight data of the full bucket of water are obtained.

[0031] In a preferred embodiment, when the full-bucket confirmation module verifies the full-bucket status based on preset bucket shape rules, the depth image, and the real-time weight data, and outputs a full-bucket integrity identifier, it is specifically used for:

[0032] A three-dimensional point cloud model of the full bucket of water is reconstructed based on the depth image;

[0033] Extract the curvature distribution features of the three-dimensional point cloud model of the barrel;

[0034] The state of the full bucket of water is determined based on the curvature distribution characteristics and the real-time weight data.

[0035] When the state is complete, a tank integrity identifier for the full tank of water is generated.

[0036] In a preferred embodiment, when the full-bucket confirmation module performs the reconstruction of the 3D point cloud model of the full-bucket body based on the depth image, it is specifically used for:

[0037] The depth image is subjected to multi-scale Gaussian filtering to eliminate noise, thereby obtaining the image of the full bucket of water.

[0038] The disparity map of the full bucket of water is calculated based on the image of the bucket, wherein the formula for calculating the disparity map is as follows:

[0039]

[0040] In the formula, D(x,y) is the disparity map, (x,y) is the pixel coordinate, d is the disparity search range, C(*) is the sum of gray-level absolute differences, and P(*) is the disparity smoothing constraint term.

[0041] The disparity map is converted into an initial depth map using the principle of triangulation.

[0042] By fusing the initial depth map with the outline features of the full bucket of water, an optimized depth map is obtained;

[0043] A triangular mesh is constructed by fusing the optimized depth map with the RGB texture information of the full bucket of water;

[0044] A Poisson reconstruction is performed on the triangular mesh to obtain a three-dimensional point cloud model of the full bucket of water.

[0045] In a preferred embodiment, the barrel shape is regular, including:

[0046] Check the integrity of the sealing ring at the barrel opening;

[0047] The flatness of the barrel bottom is verified by fitting the barrel bottom point cloud to the three-dimensional point cloud model of the barrel body and the corresponding plane.

[0048] The barrel wall concavity is determined based on the number of extreme points of edge curvature in the three-dimensional point cloud model of the barrel.

[0049] By comparing the distance between the actual outline and the standard template, the deformation of the bucket shoulder when the bucket is full of water is obtained;

[0050] When the integrity of the sealing ring, the flatness of the bottom of the bucket, the concavity of the bucket wall, and the deformation of the bucket shoulder are all within the integrity threshold, the full bucket of water is determined to be intact.

[0051] In a preferred embodiment, when the payment module triggers a dynamic payment credential generator upon successful verification of the bucket integrity identifier, and generates an encrypted payment instruction by combining the identity authentication credential and the bucket integrity identifier, it is specifically used for:

[0052] By combining the identity authentication credential with the bucket integrity identifier, the user's original transaction code is obtained;

[0053] The original transaction code is asymmetrically encrypted, and a timestamp and geolocation information are added to the asymmetrically encrypted original transaction code to generate the user's encrypted payment instruction.

[0054] In a preferred embodiment, when the water intake module receives a payment success signal returned by the payment gateway and controls the opening of the water intake electromagnetic lock, it is specifically used for:

[0055] Monitor the displacement sensor signal of the full bucket of water;

[0056] When the displacement sensor signal is detected as a displacement, the cabinet door locking procedure is initiated.

[0057] Update the inventory status database of the unmanned water exchange tank.

[0058] Compared with the prior art, the present invention has the following beneficial effects:

[0059] 1. The unmanned water vending machine system for bottled water of this invention improves the accuracy and efficiency of user identity verification by combining biometric matching with encrypted dynamic tokens in its identity authentication module. This ensures the security and reliability of the identity authentication credentials and simultaneously synchronizes with the access control controller for rapid response, laying an efficient foundation for subsequent operation processes. The empty bottle recycling module accurately obtains stable sampling values ​​of empty bottles through filtering and noise reduction processing of pressure sensing elements, ensuring the stability of recycling certificate generation. Furthermore, it associates manufacturer codes to achieve traceability records, enhancing the standardization and traceability of the empty bottle recycling process.

[0060] 2. This invention excels in the full-bucket water detection and transaction settlement stages. The full-bucket water detection module, combining a depth camera and weighing unit, accurately outputs a bucket integrity identifier through 3D point cloud model reconstruction and multi-dimensional morphological rule verification, ensuring the reliable quality of the full-bucket water. The payment module uses asymmetric encryption combined with timestamps and geographic location information to generate encrypted payment instructions, enhancing transaction security. The water dispensing module updates the inventory status promptly after completing the transaction, optimizing the overall system's operational efficiency and achieving a highly efficient, secure, and stable unmanned vending process. Attached Figure Description

[0061] Figure 1 This is a system architecture diagram of an automated bottled water vending system with an unmanned water changing tank provided in an embodiment of the present invention;

[0062] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments belong to some, but not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0064] The terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms “said” and “the” as used in the embodiments of this invention and the appended claims are also intended to include the plural forms, and “multiple” generally includes at least two unless the context clearly indicates otherwise.

[0065] Depending on the context, the word "if" or "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0066] Furthermore, the timing of the steps in the following method embodiments is merely an example and not a strict limitation.

[0067] In reality, the server-side equipment deployed in an automated bottled water vending system may consist of one or more devices. This automated bottled water vending system can be implemented as a business instance, a virtual machine, or hardware devices. For example, it can be implemented as a business instance deployed on one or more devices in a cloud node. Simply put, it can be understood as software deployed on a cloud node to provide automated bottled water vending services to various users. Alternatively, it can be implemented as a virtual machine deployed on one or more devices in a cloud node, with application software installed to manage each user. Or, it can also be implemented as a server composed of numerous identical or different types of hardware devices, with one or more devices configured to provide automated bottled water vending services to various users.

[0068] In terms of implementation, the automated bottled water vending system with unmanned water lockers and the user terminal are mutually compatible. That is, if the automated bottled water vending system with unmanned water lockers is an application installed on a cloud service platform, then the user terminal is a client that establishes a communication connection with the application; or if the automated bottled water vending system with unmanned water lockers is implemented as a website, then the user terminal is implemented as a webpage; or if the automated bottled water vending system with unmanned water lockers is implemented as a cloud service platform, then the user terminal is implemented as a mini-program in an instant messaging application.

[0069] like Figure 1 The figure shown is a system architecture diagram of an automated bottled water vending system with an unmanned water changing tank provided in an embodiment of the present invention.

[0070] The unmanned water vending machine system 100 described in this invention can be installed on a cloud server. In terms of implementation, it can be one or more service devices, or it can be installed as an application on the cloud (e.g., a mobile service operator's server, server cluster, etc.), or it can be developed into a website. Depending on the functions implemented, the unmanned water vending machine system 100 may include an identity authentication module 101, an empty bottle recycling module 102, a full bottle detection module 103, a full bottle confirmation module 104, a payment module 105, and a water dispensing module 106. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, stored in the memory of the electronic device.

[0071] In this embodiment of the invention, in the automated bottled water vending system with unmanned water changer cabinets, each of the above-mentioned modules can be implemented independently and can call other modules. Here, "calling" can be understood as one module connecting to multiple modules of another type and providing corresponding services to those connected modules. In the automated bottled water vending system with unmanned water changer cabinets provided by this embodiment of the invention, the applicable scope of the system architecture can be adjusted by adding modules and directly calling them without modifying the program code, achieving cluster-based horizontal expansion to quickly and flexibly expand the automated bottled water vending system. In practical applications, the above-mentioned modules can be set in the same device or different devices, or they can be set in a virtual device, such as a service instance in a cloud server.

[0072] The following describes the components and specific workflow of an automated bottled water vending system with unmanned water tank replacement, using specific embodiments as examples:

[0073] The identity authentication module 101 is used to generate an identity authentication credential for the unmanned water changing tank based on the user's operation instructions.

[0074] In this embodiment of the invention, when the identity authentication module generates an identity authentication credential for the unmanned water exchange tank based on the user's operation instructions, it is specifically used for:

[0075] Receive biometric data from users in unmanned water changing tanks;

[0076] The biometric data is matched and verified against a pre-stored template;

[0077] When a match is successful, an encrypted dynamic token is generated for the user as an identity authentication credential.

[0078] The identity authentication credential is synchronized to the access control controller of the unmanned water exchange tank.

[0079] Specifically, the biometric data acquisition device of the unmanned water changing tank includes a fingerprint scanner and a facial recognition camera. When a user operates the device, the fingerprint scanner captures the image of the fingerprint surface through optical sensing, converts the texture of the fingerprint into a grayscale image, and then converts the image into a digital signal composed of 0s and 1s through a built-in analog-to-digital converter. The facial recognition camera captures the user's facial image through a high-definition lens, extracts facial contours, the position of facial features, and other features, and also converts them into digital signals. These digital signals are transmitted to the main control module via an internal USB data cable. The main control module temporarily stores the received biometric data in its internal random access memory in a specific binary file format, awaiting further processing.

[0080] Furthermore, the main control module's storage unit includes a dedicated database pre-stored biometric templates entered during user registration. Each template corresponds to a registered user, indexed by the user's registration number. During matching verification, the main control module retrieves newly received biometric data from random access memory. First, it extracts key feature points, such as fingerprint endpoints and bifurcation points, and facial features like the distance between the corners of the eyes and the length of the bridge of the nose. Then, it retrieves pre-stored templates potentially related to the user from the database and compares the feature points of the new data with those in the templates one by one, checking whether the positional deviation of each corresponding point is within the allowable range. When more than a preset number of feature points match and the overall consistency reaches 90% or higher, the main control module determines that the match is successful.

[0081] Furthermore, once the main control module confirms a successful match, it immediately triggers the encrypted dynamic token generation program. This program first obtains the current system time, using a timestamp composed of year, month, day, hour, minute, and second as a dynamic factor, for example, "20250730153022"; simultaneously, it retrieves the unique identifier generated during user registration, i.e., the user ID consisting of 8 digits, for example, "12345678". The original string is formed by combining the "timestamp + user ID" rule, such as "2025073015302212345678"; then, the original string is encrypted by replacing every third character with a character shifted 3 positions to the right of the alphabet (or 3 positions to the right of numbers). The processed string is then reversed to generate a 20-character sequence, i.e., the encrypted dynamic token, such as "87654321553051035202", and stored in the main control module's temporary buffer.

[0082] Furthermore, the main control module connects to the access controller of the unmanned water exchange tank via an RS485 serial communication interface. During connection, the main control module first sends a handshake signal containing the device address to the access controller. Upon receiving the signal, the access controller returns an acknowledgment signal, completing the connection establishment. The main control module extracts the encrypted dynamic token from the temporary buffer, converts it into binary data recognizable by the access controller, and assembles it into a data frame according to the structure of "start bit (1 byte) + token data (20 bytes) + check bit (1 byte)". The data frame is sent to the access controller via the RS485 interface. After receiving the data frame, the access controller verifies the check bit to confirm that the data has not been tampered with. It then stores the encrypted dynamic token in its own flash memory and simultaneously returns a successful storage signal to the main control module, completing the synchronization of the authentication credential.

[0083] In summary, the uniqueness of biometric data ensures the accuracy of identity authentication, reduces the possibility of misidentification, and improves the reliability of user identification. On the other hand, encrypted dynamic tokens, as authentication credentials, enhance information security and reduce the risk of credential forgery or theft. Simultaneously, synchronizing authentication credentials to the access control system enables rapid response to subsequent operational commands, providing a fundamental guarantee for the efficient operation of unmanned water changing tanks and improving the overall system's smoothness and security.

[0084] The empty bucket recycling module 102 is used to open the recycling compartment door of the unmanned water exchange cabinet based on the identity authentication certificate, detect the weight of the empty bucket put in by the user, and generate a recycling certificate for the empty bucket.

[0085] In this embodiment of the invention, when the empty bucket recycling module performs the following actions: opening the recycling compartment door of the unmanned water exchange cabinet based on the identity authentication credential, detecting the weight of the empty bucket deposited by the user, and generating a recycling credential for the empty bucket, it is specifically used for:

[0086] The pressure sensing element is activated after the recovery compartment door is opened;

[0087] Acquire the weight analog signal continuously output by the pressure sensing element;

[0088] The simulated weight signal is filtered and denoised to obtain a stable sample value of the empty bucket;

[0089] When the stable sampled value falls within the preset empty bucket weight range, a recycling certificate for the empty bucket is generated.

[0090] After generating the empty bucket recycling certificate, the full water detection module is specifically used for:

[0091] Scan the barcode label on the empty bucket;

[0092] Extract the manufacturer code from the barcode identifier;

[0093] By associating the manufacturer code with the recycling certificate, the traceability record of the empty bucket is obtained.

[0094] Specifically, a mechanical limit switch is installed on the edge of the recycling bin door. When the user pulls the bin door handle to rotate the door to the fully open position, the door squeezes the trigger rod of the limit switch, causing the metal contacts inside the switch to close and forming a conductive circuit. This circuit sends a high-level signal to the control module. After receiving the signal, the control module immediately outputs a 12V DC power supply voltage to the pressure sensing element. The strain gauge circuit inside the pressure sensing element is powered on and starts, entering the weight detection state.

[0095] Furthermore, the strain gauge of the pressure sensing element deforms under pressure, causing a change in its resistance value. The resistance change is converted into a 0-5V weight analog signal through a Wheatstone bridge circuit. This signal is continuously transmitted to the analog signal input terminal of the control module through shielded wires. The AD converter built into the control module samples the analog signal every 100 milliseconds, converts the voltage value into the corresponding digital quantity, and continuously stores it in the module's random access memory to form a real-time updated weight data sequence.

[0096] Furthermore, the control module processes the digital weight analog signal stored in the random access memory using a moving average filtering method. Each time, it removes the two maximum and two minimum values ​​from the latest 30 stored data, adds the values ​​of the remaining 26 data, and divides by 26 to obtain the result of a single smoothing process. This operation is repeated until the difference between the results of three consecutive smoothing processes is less than 0.1 kg. At this point, the last processing result is determined as the stable sample value of the empty bucket and stored in the module's dedicated register.

[0097] Furthermore, the read-only memory of the control module pre-stores a preset range of empty bucket weights, which exists in the form of a minimum and a maximum value, for example, 1.2 kg to 1.5 kg. The control module retrieves a stable sample value from a dedicated register and compares it with the minimum and maximum values. When the stable sample value is greater than the minimum value and less than the maximum value, the control module immediately starts the voucher generation program. The program automatically records the current time, the stable sample value, and the bin door number, combining them into a string containing 18 characters. This string is the empty bucket recycling voucher and is synchronously stored in the module's flash memory database.

[0098] In summary, opening the recycling bin door based on identity authentication credentials ensures that only verified users can deposit empty bins, effectively preventing unauthorized personnel from accidentally depositing non-standard containers and guaranteeing the standardization and security of the recycling process.

[0099] In summary, when detecting the weight of empty containers, by activating the pressure sensing element and filtering and reducing noise in the weight analog signal, a stable sampling value of the empty container can be accurately obtained. When this value falls within a preset range, a recycling voucher is generated. This not only ensures the accuracy of weight detection and reduces misjudgments caused by signal interference, but also enables standardized recording of empty container recycling through the generation of recycling vouchers. This provides reliable data support for the subsequent linkage of full-bottle water sales, improving the coherence and efficiency of the entire system process.

[0100] Specifically, a fixed laser scanner is installed inside the recycling bin, with its lens facing the empty bin placement area. When an empty bin is placed in the recycling bin and triggers the position sensor, the position sensor sends a signal to the control module. The control module then activates the laser scanner. The laser beam emitted by the scanner illuminates the barcode marking on the surface of the empty bin. The black and white stripes of the barcode produce different reflections of the laser light. The reflected light is received by the scanner's photoelectric converter and converted into high and low level signals. These signals form a pulse sequence according to the width and spacing of the barcode and are transmitted to the decoding unit of the control module via a data line.

[0101] Furthermore, the decoding unit of the control module converts the received pulse sequence into a corresponding digital character sequence, which is the complete information contained in the barcode identifier. The first 8 characters are the manufacturer code, representing the manufacturer information of the empty bucket. The decoding unit extracts 8 consecutive characters starting from the beginning of the digital character sequence according to the preset character truncation rules. After removing any possible separators at the beginning, the manufacturer code composed of numbers and letters is obtained and stored in the temporary data area of ​​the control module.

[0102] Furthermore, the control module retrieves the manufacturer code from the temporary data area and extracts the recycling voucher corresponding to the current recycling operation from the flash memory database. The associated program of the control module concatenates the character sequence of the manufacturer code and the recycling voucher, separated by the special symbol "|", to form a new string, such as "ABC12345|2025073015302212345678". This string is the traceability record of the empty bucket. The program writes the traceability record into the traceability table of the database and stores it in association with information such as the stable sampling value and recycling time of the empty bucket, thus completing the entire process of generating the traceability record.

[0103] The full water detection module 103 is used to activate the depth camera and weighing unit of the full water compartment in the unmanned water exchange cabinet according to the empty water recycling certificate, and simultaneously acquire the depth image and real-time weight data of the full water.

[0104] In this embodiment of the invention, when the full water detection module activates the depth camera and weighing unit of the full water compartment in the unmanned water exchange cabinet according to the empty water recycling certificate, and simultaneously acquires the depth image and real-time weight data of the full water, it is specifically used for:

[0105] The infrared ranging sensor is triggered to calibrate the focal length of the depth camera based on the recycling certificate;

[0106] The zero-point drift compensation program of the weighing element is started simultaneously;

[0107] After compensation is completed, the depth image and real-time weight data of the full bucket of water are obtained.

[0108] Specifically, the control module retrieves the recycling voucher from the flash memory, identifies the operation identifier within it, and sends a synchronous start signal to the infrared ranging sensor and the depth camera. The infrared ranging sensor is installed 10 centimeters directly below the depth camera, with both aligned with the center of the tray containing the full bucket of water. The infrared ranging sensor emits an infrared beam with a wavelength of 850 nanometers. After the beam is reflected by the surface of the tray, it is received by the sensor. The sensor calculates the actual distance to the tray based on the round-trip time of the beam and transmits this distance data to the focus adjustment unit of the depth camera via the I2C bus. The stepper motor in the adjustment unit drives the camera lens group to move along the optical axis. Every 0.1 millimeters of movement, the image sharpness detection circuit judges the sharpness of the tray edge contour until the pixel grayscale change at the contour edge reaches the preset maximum difference. At this point, the focus calibration of the depth camera is completed, and the current lens position parameters are stored.

[0109] Furthermore, at the same moment the control module sends the synchronization start signal, it sends a zero-point drift compensation command to the weighing unit via the SPI interface. At this time, the weighing unit is in an unloaded state (there is no object on the pallet). Its internal Wheatstone bridge outputs an initial voltage signal. The compensation program compares this initial voltage with the stored standard zero-point voltage (the unloaded voltage calibrated at the factory), calculates the voltage difference, and changes the bridge balance by adjusting the variable resistor inside the unit to make the current output voltage consistent with the standard zero-point voltage. During the compensation process, the voltage deviation is detected every 0.5 seconds until three consecutive deviations are less than 0.01 volts. At this time, the compensation program returns a compensation completion signal to the control module.

[0110] Furthermore, after receiving the compensation completion signal from the weighing element, the control module immediately sends an image acquisition command to the depth camera via the GPIO interface. The camera, which has completed focus calibration, takes a picture of the full bucket of water placed on the tray. The lens captures spatial information such as the water level and the outline of the bucket, converting it into a grayscale image containing depth coordinates, i.e., the depth image of the full bucket of water, which is stored in the camera's built-in SD card. At the same time, the control module sends a weight acquisition command to the weighing element. The strain gauge of the weighing element deforms due to the pressure of the full bucket of water, outputting an analog voltage corresponding to the weight. This voltage is converted into a digital quantity, i.e., real-time weight data, by the internal AD converter and transmitted to the weight data buffer of the control module via the UART serial port, thus completing the acquisition of the depth image and real-time weight data of the full bucket of water.

[0111] In summary, using empty bucket recycling certificates as activation signals enables precise coordination between empty bucket recycling and full bucket water detection, avoiding ineffective equipment operation, improving the utilization efficiency of system resources, ensuring that full bucket water detection is only activated when necessary, and reducing energy consumption and equipment wear and tear.

[0112] In summary, in terms of data acquisition accuracy, simultaneously acquiring depth images and real-time weight data provides multi-dimensional and highly reliable foundational information for subsequent verification of the full water tank status. Depth images reflect the details of the tank's shape, while real-time weight data indicates whether the water volume meets standards. The combination of these two elements lays a data foundation for accurately determining the full water tank status, ensuring the smooth progress of subsequent processes.

[0113] The full bucket confirmation module 104 is used to verify the state of the full bucket of water based on the preset bucket shape rules, the depth image and the real-time weight data, and output the bucket integrity identifier of the full bucket of water.

[0114] In this embodiment of the invention, when the full-bucket confirmation module verifies the state of the full bucket based on preset bucket shape rules, the depth image, and the real-time weight data, and outputs a bucket integrity identifier indicating that the bucket is full, it is specifically used for:

[0115] A three-dimensional point cloud model of the full bucket of water is reconstructed based on the depth image;

[0116] Extract the curvature distribution features of the three-dimensional point cloud model of the barrel;

[0117] The state of the full bucket of water is determined based on the curvature distribution characteristics and the real-time weight data.

[0118] When the state is complete, a tank integrity identifier for the full tank of water is generated.

[0119] When the full bucket confirmation module performs the reconstruction of the 3D point cloud model of the full bucket based on the depth image, it is specifically used for:

[0120] The depth image is subjected to multi-scale Gaussian filtering to eliminate noise, thereby obtaining the image of the full bucket of water.

[0121] The disparity map of the full bucket of water is calculated based on the image of the bucket, wherein the formula for calculating the disparity map is as follows:

[0122]

[0123] In the formula, D(x,y) is the disparity map, (x,y) is the pixel coordinate, d is the disparity search range, C(*) is the sum of gray-level absolute differences, and P(*) is the disparity smoothing constraint term.

[0124] The disparity map is converted into an initial depth map using the principle of triangulation.

[0125] By fusing the initial depth map with the outline features of the full bucket of water, an optimized depth map is obtained;

[0126] A triangular mesh is constructed by fusing the optimized depth map with the RGB texture information of the full bucket of water;

[0127] A Poisson reconstruction is performed on the triangular mesh to obtain a three-dimensional point cloud model of the full water bucket.

[0128] Specifically, each pixel in the depth image contains distance information from the surface of the full water bucket to the depth camera. The control module reads the pixel data of the depth image and establishes a three-dimensional coordinate system with the optical center of the depth camera as the origin, where the horizontal direction is the X-axis, the vertical direction is the Y-axis, and the lens orientation is the Z-axis. Based on the image resolution and the camera focal length, the two-dimensional coordinates (X pixel value, Y pixel value) of each pixel are converted into the actual lengths on the X and Y axes. The distance information of the pixel is then used as the Z-axis coordinate. Each pixel corresponds to a three-dimensional spatial coordinate point. The control module summarizes the three-dimensional coordinate points corresponding to all pixels to form a set of a large number of discrete points, namely the three-dimensional point cloud model of the full water bucket, which is stored in the graphics processing unit of the control module.

[0129] Furthermore, the control module retrieves the 3D point cloud model of the barrel from the graphics processing unit. For each point in the model, it selects the 10 nearest neighboring points and fits a local surface using the 3D coordinates of these neighboring points. It then calculates the curvature value of the surface at that point. The curvature value of each point is represented by a specific numerical value, reflecting the curvature of the barrel surface at that location. For example, the curvature value of points at the edge of the barrel opening is larger, while the curvature value of points on the side of the barrel body is smaller. The control module counts the curvature values ​​of all points, records the number of times different curvature values ​​occur and the corresponding location areas, forming a curvature distribution feature that describes the curvature distribution of the barrel surface, and stores it in the feature data buffer.

[0130] Furthermore, the control module's storage unit pre-stores the standard curvature distribution characteristics and standard weight range of a complete full bucket of water. The control module compares the curvature distribution characteristics in the feature data buffer with the standard curvature distribution characteristics to check whether the difference in curvature values ​​in each region is within the preset allowable range. At the same time, it retrieves the real-time weight data in the weight data buffer and compares it with the standard weight range to determine whether the real-time weight is within the range. When the difference in curvature distribution characteristics is within the allowable range and the real-time weight is within the standard weight range, the control module determines that the full bucket of water is complete; when either condition is not met, the state is determined to be incomplete.

[0131] Furthermore, when the control module determines that the full water tank is complete, it immediately starts the identifier generation program. The program automatically extracts the current system time (accurate to the second), the unique identification code of the 3D point cloud model of the tank (automatically assigned when the model is generated), and the "complete" status identifier, and combines them into a 24-bit character sequence in the format of "time + identification code + status".

[0132] For example, the character sequence “20250730165030_BT87654321_complete” is the integrity identifier of a full water tank. The program writes this identifier into the traceability database of the control module and stores it in association with the corresponding recycling voucher and traceability record.

[0133] Specifically, the control module loads the depth image into the image processing unit and performs three filtering processes on the image using Gaussian convolution kernels of three different sizes: 3×3, 5×5, and 7×7. During each filtering process, the center of the convolution kernel is aligned with each pixel in the image, and a new pixel value is calculated by weighted averaging. Small-sized convolution kernels eliminate small noise in the image, medium-sized convolution kernels smooth edge areas, and large-sized convolution kernels process large areas of noise. After the three filtering processes are completed, the results are superimposed to obtain a full bucket image with random noise removed and the edge information of the bucket retained. This image is then stored in the temporary buffer area of ​​the image processing unit.

[0134] Furthermore, the control module retrieves the bucket image from the temporary buffer and inputs it into the disparity calculation unit. The disparity calculation unit uses a block-based matching method to select an 8×8 pixel reference block in the left view and search for the most similar matching block from left to right in the same row of the right view. The similarity is determined by calculating the difference in grayscale values ​​of corresponding pixels in the two blocks. After finding the best matching block, its horizontal displacement relative to the reference block is calculated. This displacement is the disparity value of the reference block. The disparity calculation unit repeats the above process for each pixel block in the bucket image, and finally generates a two-dimensional matrix composed of disparity values, i.e., the disparity map of a full bucket of water, which is stored in the disparity data buffer.

[0135] Furthermore, the control module transmits the disparity map from the disparity data buffer to the depth calculation unit. The depth calculation unit converts the disparity values ​​into depth values ​​based on the triangulation principle. In a binocular camera system, there is a fixed distance between the optical centers of the two cameras, namely the baseline distance. When light enters the two cameras from the same point on the object, it will form image points at different positions on the imaging plane. The disparity between these two image points is inversely proportional to the distance from the object to the camera. The depth calculation unit converts each disparity value in the disparity map using the known baseline distance and the camera focal length to obtain the corresponding depth value. The two-dimensional image composed of these depth values ​​is the initial depth map, which is stored in the depth data buffer.

[0136] Furthermore, the control module reads the initial depth map from the depth data buffer and extracts the contour features of the barrel image using an edge detection algorithm to obtain the set of edge pixels of the barrel. The control module then fuses the initial depth map with the barrel contour features. For edge pixels in the contour features, the depth value of the corresponding position in the initial depth map is corrected using the precise location information obtained from edge detection. For non-edge pixels, the depth value of the initial depth map is retained. A weighted average method is used during the fusion process, with the weight of edge pixels being 0.8 and the weight of non-edge pixels being 0.2. Finally, an optimized depth map with more accurate depth values ​​and clearer barrel boundaries is obtained and stored in the optimized data buffer.

[0137] Furthermore, the control module retrieves the optimized depth map from the optimized data buffer and simultaneously acquires an RGB color image of a full bucket of water. It extracts the color information of each pixel, i.e., the RGB texture information, from the RGB image. The control module associates each depth point in the optimized depth map with the corresponding pixel in the RGB image, assigning a corresponding color value to each depth point. Then, using the Delaunay triangulation algorithm, it connects three adjacent points into triangles based on the spatial relationship of the depth points, forming a continuous mesh structure covering the surface of the full bucket of water. The vertex of each triangle is a depth point, and the vertex color comes from the RGB texture information. Finally, a triangular mesh with color information is constructed and stored in the 3D modeling unit.

[0138] Furthermore, the control module transfers the triangular mesh from the 3D modeling unit to the Poisson reconstruction module. The Poisson reconstruction module first calculates the normal vector field of the triangular mesh surface. The normal vector represents the orientation of the surface at each point. Then, by solving the Poisson equation, an implicit function is reconstructed based on the normal vector field. This function has a positive value inside the object, a negative value outside, and a zero value on the surface. The Poisson reconstruction module extracts the isosurface of the implicit function using the moving cube algorithm, converting the implicit function into an explicit 3D surface model. The model is then smoothed to remove minor imperfections generated during the reconstruction process. Finally, a 3D point cloud model of a full bucket of water, composed of a large number of 3D coordinate points, is obtained and stored in the 3D model database.

[0139] Specifically, the pixel coordinates (x, y) are derived from the pixel arrangement of the barrel image. The barrel image consists of multiple pixels, with each pixel's horizontal position represented by x and its vertical position by y. Their numerical range corresponds to the resolution of the barrel image; for example, when the image width is 640 pixels, x ranges from 0 to 639, and when the height is 480 pixels, y ranges from 0 to 479. The disparity search range of d is a pre-defined fixed numerical interval, determined based on the baseline length of the binocular cameras and the shooting distance of a full barrel of water. For example, it might be set to 0 to 60, meaning the disparity value of each pixel can only be selected between 0 and 60. The data source for the sum of the absolute differences in grayscale values ​​of C(x, y, d) is the barrel images captured by the left and right cameras. The grayscale value of the pixel at coordinates (x, y) in the left image is subtracted from the grayscale value of the pixel at coordinates (xd, y) in the right image, and the absolute value is taken. This absolute value is then added together with the values ​​of multiple adjacent pixels to obtain the specific numerical value. The data source for the disparity smoothing constraint term of P(x,y,d) is the disparity values ​​of the neighboring pixels around the current pixel. It calculates the sum of the differences between the current pixel and the disparity values ​​of its four neighboring pixels (up, down, left, and right) when the current pixel takes the disparity value d. The larger the difference, the larger the value of this constraint term.

[0140] Furthermore, the function of this formula is to determine the optimal disparity value d for each pixel with coordinates (x,y) in the bucket image. By calculating the sum of C(x,y,d) and P(x,y,d) corresponding to different d values, the d value that minimizes this sum is selected as the disparity value of that pixel. C(x,y,d) is used to ensure that the matching pixels in the left and right images have a high degree of similarity in grayscale, and P(x,y,d) is used to ensure that the disparity values ​​of adjacent pixels change smoothly and avoid abrupt jumps. Finally, the disparity values ​​obtained in this way are combined to form the disparity map of the full bucket.

[0141] Furthermore, as d gradually increases within the disparity search range, C(x,y,d) will first gradually decrease, then gradually increase again after reaching a certain value. This is because as d changes, the grayscale matching degree of corresponding pixels in the left and right images first improves and then deteriorates. At the same time, P(x,y,d) will fluctuate with the difference between d and the disparity values ​​of adjacent pixels. When d is close to the disparity values ​​of surrounding pixels, P(x,y,d) is smaller, and when the difference is large, it is larger. The sum of the two will show a trend of first decreasing to a minimum value and then increasing. The d value corresponding to this minimum value is the optimal disparity value of the pixel. The optimal disparity values ​​of all pixels together constitute the disparity map.

[0142] In summary, from the perspective of verification accuracy, reconstructing a 3D point cloud model of the container using depth images, extracting curvature distribution features, and combining this with real-time weight data enables a comprehensive verification of the fullness of the water in the container from both morphological and weight perspectives. Pre-defined container morphology rules cover multiple dimensions of inspection, including the integrity of the container's sealing ring, the flatness of the bottom, wall dents, and shoulder deformation, ensuring the accuracy of the container integrity assessment and effectively preventing substandard full-bottle water from entering the distribution process.

[0143] In summary, from the perspective of system reliability, the bucket integrity identifier output by this process provides a crucial basis for the subsequent payment process. Only a full bucket that has passed verification will trigger the payment process, ensuring that the goods obtained by users meet quality standards. At the same time, it provides technical support for the standardized operation of the unmanned vending system and enhances the overall reliability and credibility of the system.

[0144] The payment module 105 is used to trigger a dynamic payment credential generator when the bucket integrity identifier passes verification, and generate an encrypted payment instruction by combining the identity authentication credential and the bucket integrity identifier, and send it to the payment gateway.

[0145] In this embodiment of the invention, the barrel shape is regular, including:

[0146] Check the integrity of the sealing ring at the barrel opening;

[0147] The flatness of the barrel bottom is verified by fitting the barrel bottom point cloud to the three-dimensional point cloud model of the barrel body and the corresponding plane.

[0148] The barrel wall concavity is determined based on the number of extreme points of edge curvature in the three-dimensional point cloud model of the barrel.

[0149] By comparing the distance between the actual outline and the standard template, the deformation of the bucket shoulder when the bucket is full of water is obtained;

[0150] When the integrity of the sealing ring, the flatness of the bottom of the bucket, the concavity of the bucket wall, and the deformation of the bucket shoulder are all within the integrity threshold, the full bucket of water is determined to be intact.

[0151] When the payment module executes the dynamic payment credential generator upon successful verification of the bucket integrity identifier, and combines the identity authentication credential with the bucket integrity identifier to generate an encrypted payment instruction, it is specifically used for:

[0152] By combining the identity authentication credential with the bucket integrity identifier, the user's original transaction code is obtained;

[0153] The original transaction code is asymmetrically encrypted, and a timestamp and geolocation information are added to the asymmetrically encrypted original transaction code to generate the user's encrypted payment instruction.

[0154] Specifically, a high-definition industrial camera is installed directly above the barrel opening, with the lens vertically aimed at the opening area. The camera captures a top-down view of the opening, and the image is transmitted to the image processing module. The module uses a color threshold to segment the black sealing ring area, and then uses an edge detection algorithm to extract the continuous contour line of the sealing ring. It checks whether there are any interruptions or gaps in the contour line. If the contour line is complete and unbroken, and the difference between the area of ​​the sealing ring area and the area of ​​the standard sealing ring is within 5%, the sealing ring is judged to be complete, and the sealing ring integrity result is qualified.

[0155] Furthermore, points with Z coordinates within the lowest 10% range are selected from the 3D point cloud model of the barrel. These points constitute the point cloud set of the barrel bottom. The control module calls a plane fitting program to process these points. The program calculates an optimal fitting plane using the least squares method, minimizing the sum of the squares of the distances from all points to this plane. Then, the vertical distance from each point on the barrel bottom to the fitting plane is calculated, and the maximum value of all distances is counted. If the maximum value is less than 0.5 mm, the flatness of the barrel bottom is determined to meet the requirements, and the barrel bottom flatness result is considered qualified.

[0156] Furthermore, the control module extracts the edge point set of the barrel wall from the 3D point cloud model of the barrel body. These points are distributed along the circumference of the barrel body. The module calculates the curvature value of each edge point. The curvature value reflects the degree of curvature of the barrel wall at that point. When the curvature value suddenly increases and exceeds the preset curvature threshold, it is recorded as a curvature extreme point. The number of all curvature extreme points is counted. If the number is less than 3, it is determined that there is no obvious indentation in the barrel wall, and the barrel wall indentation result is qualified.

[0157] Furthermore, the control module's storage unit pre-stores a 3D contour template of the standard barrel shoulder. The template contains the 3D coordinates of each point in the barrel shoulder area. The module extracts the actual contour points of the barrel shoulder area from the 3D point cloud model of the barrel body, pairs the actual contour points with the corresponding points in the standard template, calculates the spatial straight-line distance between each pair of corresponding points, and counts the maximum value among all distances. If the maximum value is less than 2 millimeters, the barrel shoulder deformation is determined to be within the allowable range, and the generated barrel shoulder deformation result is qualified.

[0158] Furthermore, the control module pre-stores complete thresholds for the integrity of the sealing ring, the flatness of the bottom of the bucket, the dent in the bucket wall, and the deformation of the bucket shoulder. These thresholds are respectively: the sealing ring is qualified, the maximum distance is 0.5 mm, the number of extreme points is less than 3, and the maximum deformation distance is 2 mm. The control module sequentially retrieves the above four test results and compares them one by one with the corresponding complete thresholds. When all four results meet their respective complete threshold requirements, the control module outputs a judgment result, determining that the full bucket of water is complete.

[0159] Specifically, the control module retrieves the authentication credential and the bucket integrity identifier from the storage unit. The authentication credential is a 20-character sequence, and the bucket integrity identifier is a 24-character sequence. The control module concatenates the first 12 characters of the bucket integrity identifier with the last 10 characters of the authentication credential to form a new 32-character sequence, which is the user's original transaction code. The control module stores the original transaction code in a temporary buffer, waiting for further processing.

[0160] Furthermore, the control module obtains a pre-allocated asymmetric encryption public key from the key management system. This public key is used to encrypt the original transaction code. The control module reads the original transaction code from the temporary buffer, divides it into sub-sequences of 8 bits each, and encrypts each sub-sequence sequentially using the asymmetric encryption public key. During encryption, the public key converts each character in the sub-sequence into another specific character, forming an encrypted sub-sequence. All encrypted sub-sequences are combined to form the encrypted original transaction code. The control module obtains the current system time as a timestamp and obtains the device's latitude and longitude coordinates from the GPS positioning module as geographical location information. It converts the timestamp and geographical location information into character format and appends them to the encrypted original transaction code, forming a complete character sequence containing the encrypted transaction code, timestamp, and geographical location information. This sequence is the user's encrypted payment instruction. The control module sends the encrypted payment instruction to the payment processing center through a secure communication channel.

[0161] In summary, from a transaction security perspective, by combining identity authentication credentials with the bucket integrity identifier to form the original transaction code and performing asymmetric encryption, while also adding timestamps and geolocation information, the security of payment instructions is significantly improved, the risk of transaction information being tampered with or stolen is reduced, and the security and reliability of the user's payment process is ensured.

[0162] In summary, from the perspective of process connectivity, using the verification of the bucket integrity mark as the payment trigger condition ensures that only qualified full buckets of water will enter the payment process. This achieves precise linkage between product quality verification and transaction settlement, avoiding transactions of unqualified products. At the same time, encrypted payment instructions are sent directly to the payment gateway, improving the efficiency and standardization of the payment process and providing key support for the smooth operation of the unmanned vending system.

[0163] The water intake module 106 is used to receive the payment success signal returned by the payment gateway and control the electromagnetic lock of the water intake port to open.

[0164] In this embodiment of the invention, when the water intake module receives the payment success signal returned by the payment gateway and controls the opening of the water intake electromagnetic lock, it is specifically used for:

[0165] Monitor the displacement sensor signal of the full bucket of water;

[0166] When the displacement sensor signal is detected as "movement out", the cabinet door locking procedure is initiated.

[0167] Update the inventory status database of the unmanned water exchange tank.

[0168] Specifically, an infrared displacement sensor is installed on the edge of the tray where the full water bucket is placed. The transmitter and receiver are located on opposite sides of the tray. The transmitter continuously emits an infrared beam to the receiver. When the full water bucket is placed on the tray, the bucket blocks the beam, and the receiver outputs a low-level signal because it cannot receive the beam. When the full water bucket is moved, the beam is turned on, and the receiver outputs a high-level signal. The control module reads the level signal output by the sensor in real time through the signal line and records the signal status every 50 milliseconds to form a continuous displacement sensor signal monitoring record.

[0169] Furthermore, when the control module detects a high-level signal output by the displacement sensor during monitoring, it determines that the signal is a move-out signal and immediately sends a locking command to the cabinet door control unit. The relay in the cabinet door control unit closes, driving the DC motor to rotate in the forward direction. The motor drives the latch to extend from the retracted state through gear transmission and insert into the latch groove on the edge of the cabinet door. When the latch is fully in place, the micro switch installed at the end of the latch is triggered, returning a locking completion signal to the control module, and the cabinet door locking procedure ends.

[0170] Furthermore, the main control unit of the unmanned water changing tank is connected to a local database server. The database contains an inventory status table that records the current inventory quantity of full water barrels, the unique identifier of each barrel, and its storage location. After receiving the signal that the cabinet door has been locked, the control module retrieves the record of the current full water barrel in the inventory status table, decrements the inventory quantity value by 1, updates the status of the barrel to "removed", and records the removal time. The updated inventory status table is backed up to the remote server through a data synchronization protocol to ensure that the inventory information in the local and remote databases is consistent, thus completing the update of the unmanned water changing tank's inventory status database.

[0171] In summary, from the perspective of process closure, using the successful payment signal as the trigger condition for opening the electromagnetic lock of the water dispenser achieves precise linkage between the "payment-water dispensing" process. This ensures that users can only obtain a full bucket of water after completing payment, guarantees the standardization and integrity of the transaction process, avoids situations where water is dispensed without payment or cannot be dispensed after payment, and improves the rigor of the unmanned vending process.

[0172] In summary, from the perspectives of user experience and system management, the electromagnetic lock's automatic opening response is rapid, reducing user waiting time and improving operational convenience. At the same time, this mechanism also provides a basis for the system to automatically record the transaction completion status, facilitating the smooth implementation of subsequent inventory updates and other management operations, and ensuring the efficiency and stability of the unmanned water changing tank operation.

[0173] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0174] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0175] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An automated bottled water vending system with an unmanned water tank, characterized in that, The system includes an identity authentication module, an empty bucket recycling module, a full bucket detection module, a full bucket confirmation module, a payment module, and a water dispensing module, wherein: The identity authentication module is used to generate identity authentication credentials for the unmanned water changing tank based on the user's operation instructions; The empty bucket recycling module is used to open the recycling compartment door of the unmanned water exchange cabinet based on the identity authentication credentials, detect the weight of the empty bucket put in by the user, and generate a recycling certificate for the empty bucket. The full water detection module is used to activate the depth camera and weighing unit in the full water compartment of the unmanned water exchange cabinet according to the empty water recycling certificate, and simultaneously acquire the depth image and real-time weight data of the full water. The full bucket confirmation module is used to verify the state of the full bucket based on the preset bucket shape rules, the depth image and the real-time weight data, and output the bucket integrity identifier of the full bucket. The payment module is used to trigger a dynamic payment credential generator when the bucket integrity identifier is verified, and to generate an encrypted payment instruction by combining the identity authentication credential and the bucket integrity identifier, and send it to the payment gateway. The water intake module is used to receive the payment success signal returned by the payment gateway and control the electromagnetic lock of the water intake port to open.

2. The automated bottled water vending system with unmanned water change tank as described in claim 1, characterized in that, When the identity authentication module generates an identity authentication credential for the unmanned water changing tank based on the user's operation instructions, it is specifically used for: Receive biometric data from users in unmanned water changing tanks; The biometric data is matched and verified against a pre-stored template; When a match is successful, an encrypted dynamic token is generated for the user as an identity authentication credential. The identity authentication credential is synchronized to the access control controller of the unmanned water exchange tank.

3. The automated bottled water vending system with unmanned water change tank as described in claim 1, characterized in that, When the empty bucket recycling module opens the recycling compartment door of the unmanned water exchange cabinet based on the identity authentication credential, detects the weight of the empty bucket deposited by the user, and generates a recycling voucher for the empty bucket, it is specifically used for: The pressure sensing element is activated after the recovery compartment door is opened; Acquire the weight analog signal continuously output by the pressure sensing element; The simulated weight signal is filtered and denoised to obtain a stable sample value of the empty bucket; When the stable sampled value falls within the preset empty bucket weight range, a recycling certificate for the empty bucket is generated.

4. The automated bottled water vending system with unmanned water tank as described in claim 1, characterized in that, After generating the empty bucket recycling certificate, the full water detection module is specifically used for: Scan the barcode label on the empty bucket; Extract the manufacturer code from the barcode identifier; By associating the manufacturer code with the recycling certificate, the traceability record of the empty bucket is obtained.

5. The automated bottled water vending system with unmanned water change tank as described in claim 1, characterized in that, When the full water detection module activates the depth camera and weighing unit in the full water compartment of the unmanned water exchange cabinet based on the empty water recycling certificate, and simultaneously acquires the depth image and real-time weight data of the full water, it is specifically used for: The infrared ranging sensor is triggered to calibrate the focal length of the depth camera based on the recycling certificate; The zero-point drift compensation program of the weighing element is started simultaneously; After compensation is completed, the depth image and real-time weight data of the full bucket of water are obtained.

6. The automated bottled water vending system with unmanned water tank as described in claim 1, characterized in that, When the full-bucket confirmation module verifies the full-bucket status based on preset bucket shape rules, the depth image, and the real-time weight data, and outputs a full-bucket integrity identifier, it is specifically used for: A three-dimensional point cloud model of the full bucket of water is reconstructed based on the depth image; Extract the curvature distribution features of the three-dimensional point cloud model of the barrel; The state of the full bucket of water is determined based on the curvature distribution characteristics and the real-time weight data. When the state is complete, a tank integrity identifier for the full tank of water is generated.

7. The automated bottled water vending system with unmanned water change tank as described in claim 6, characterized in that, When the full bucket confirmation module performs the reconstruction of the 3D point cloud model of the full bucket based on the depth image, it is specifically used for: The depth image is subjected to multi-scale Gaussian filtering to eliminate noise, thereby obtaining the image of the full bucket of water. The disparity map of the full bucket of water is calculated based on the image of the bucket, wherein the formula for calculating the disparity map is as follows: In the formula, D(x,y) is the disparity map, (x,y) is the pixel coordinate, d is the disparity search range, C(*) is the sum of gray-level absolute differences, and P(*) is the disparity smoothing constraint term. The disparity map is converted into an initial depth map using the principle of triangulation. By fusing the initial depth map with the outline features of the full bucket of water, an optimized depth map is obtained; A triangular mesh is constructed by fusing the optimized depth map with the RGB texture information of the full bucket of water; A Poisson reconstruction is performed on the triangular mesh to obtain a three-dimensional point cloud model of the full water bucket.

8. The automated bottled water vending system with unmanned water tank as described in claim 7, characterized in that, The barrel shape rules include: Check the integrity of the sealing ring at the barrel opening; The flatness of the barrel bottom is verified by fitting the barrel bottom point cloud to the three-dimensional point cloud model of the barrel body and the corresponding plane. The barrel wall concavity is determined based on the number of extreme points of edge curvature in the three-dimensional point cloud model of the barrel. By comparing the distance between the actual outline and the standard template, the deformation of the bucket shoulder when the bucket is full of water is obtained; When the integrity of the sealing ring, the flatness of the bottom of the bucket, the concavity of the bucket wall, and the deformation of the bucket shoulder are all within the integrity threshold, the full bucket of water is determined to be intact.

9. The automated bottled water vending system with unmanned water change tank as described in claim 1, characterized in that, When the payment module executes the dynamic payment credential generator upon successful verification of the bucket integrity identifier, and combines the identity authentication credential with the bucket integrity identifier to generate an encrypted payment instruction, it is specifically used for: By combining the identity authentication credential with the bucket integrity identifier, the user's original transaction code is obtained; The original transaction code is asymmetrically encrypted, and a timestamp and geolocation information are added to the asymmetrically encrypted original transaction code to generate the user's encrypted payment instruction.

10. The automated bottled water vending system with unmanned water tank as described in claim 1, characterized in that, When the water intake module receives a payment success signal from the payment gateway and controls the electromagnetic lock at the water intake point to open, it is specifically used for: Monitor the displacement sensor signal of the full bucket of water; When the displacement sensor signal is detected as a displacement, the cabinet door locking procedure is initiated. Update the inventory status database of the unmanned water exchange tank.

Images