Intelligent medicine box accurate medicine discharging control method and system based on multi-mode perception

Abstract

The invention discloses an intelligent medicine box accurate medicine discharging control system and method based on multi-mode perception, and belongs to the technical field of intelligent medical equipment. According to the system, tablet data are collected in real time through a multi-mode sensing module (comprising a high-precision weight sensor, machine vision and an infrared counter), and high-precision tablet dispensing is achieved through a three-stage tablet dispensing mechanism. The control method is fused with an adaptive Kalman filtering algorithm for data correction, has the functions of blockage detection and emergency recovery, and can dynamically adjust operation parameters. Experimental verification shows that the total accuracy rate of 1000 times of medicine discharging tests on five kinds of tablets by the system reaches 99.2% (the accuracy rate of special-shaped tablets is 98.7%), and the blockage occurrence rate is 1t; the emergency recovery success rate is 100%, and the standby power consumption is 1t; the method is suitable for complex medication scenes of old patients, and has the advantages of high precision, self-adaption, high reliability and the like.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent medical equipment, and in particular relates to a method and system for accurately dispensing medicine from an intelligent medicine box based on multimodal sensing. Background Art

[0002] Traditional medicine boxes rely on manual sorting and reminders and are unable to adapt to complex medication scenarios. In particular, the sorting errors for odd-shaped tablets are large, which can easily lead to missed or accidental doses. Existing automatic medicine boxes, while incorporating sensor technology, often use a single mode of perception (such as weight or infrared counting), making it difficult to cope with the diversity of tablet shapes, sizes, and materials, resulting in insufficient sorting accuracy. Furthermore, traditional medicine boxes lack a real-time calibration mechanism, leading to significant error accumulation after long-term use. Furthermore, blockage handling relies on manual intervention, resulting in low reliability. Therefore, a high-precision, adaptive smart medicine box system is urgently needed to meet the complex medication needs of elderly patients. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a precise dispensing control method and system for intelligent pill boxes based on multimodal sensing. By integrating multimodal data from a high-precision weight sensor, a micro-machine vision module, and an infrared counter, and employing an adaptive Kalman filter algorithm to correct dispensing parameters in real time, the system addresses the issue of error in sorting irregular-shaped tablets. The system includes a three-stage dispensing mechanism driven by a stepper motor, achieving ±0.1mm positioning accuracy. Combined with a jam detection and emergency recovery mechanism, the system achieves a dispensing accuracy rate of 99.2%, significantly improving the reliability of the automated pill box. Technical Solution

[0004] A smart medicine box precision dispensing control system includes a main processor module, a multimodal sensing module, a three-level medicine dispensing mechanism, a communication module, and a power module. The main processor module uses an ARM Cortex-M7 chip with an integrated real-time operating system to coordinate the work of each module and perform data fusion processing. The multimodal sensing module includes: (1) a high-precision weight sensor: using an HBM U10M weighing sensor with an accuracy of ±0.01g, installed at the bottom of the medicine bin to monitor the weight changes of the medicine in real time; (2) a micro machine vision module: equipped with a Sony IMX290 image sensor, an LED ring light source, and a resolution of 1920×1080, used to identify the shape, color, and position of the tablets; (3) an infrared counter: using a Sharp GP2Y0A21YK0F infrared pair tube to detect the number of tablets passing through, with a response time of <1ms. The three-stage pill separation mechanism is driven by a stepper motor and includes three levels: coarse separation, fine separation, and fine adjustment: (1) Coarse separation: a two-phase hybrid stepper motor (42BYGH401) is used to achieve preliminary separation of tablets through a spiral slide; (2) Fine separation: a high-precision screw module (THK SRG15) is used in conjunction with a vacuum adsorption gripper (SMC ZQ2 series) for single-piece grasping; (3) Fine separation: an integrated voice coil motor (Hiwin VCM-0603) is used to achieve ±0.1mm position compensation.

[0005] Control methods Multimodal data fusion: The main processor collects weight, visual, and infrared data in real time and uses an adaptive Kalman filter algorithm for data fusion. The algorithm flow is as follows: (1) State equation: X_k = A X_{k_-1} + B U_{k_-1} + W_{k_-1}; (2) Observation equation: Z_k = H X_k + V_k; (3) Where the state vector X_k contains the tablet's position, velocity, and mass, and the observation vector Z_k contains the weight, visual coordinates, and infrared counts.

[0006] Blockage detection and emergency recovery: (1) Blockage detection: Install an Omron E3Z-LS63 photoelectric sensor in the medicine dispensing channel. When no tablet movement is detected for 500ms, a blockage alarm is triggered; (2) Emergency recovery: Start the reverse stepper motor drive and turn on the vacuum adsorption device at the same time to suck the blocked tablets back to the temporary storage bin to avoid drug waste.

[0007] Adaptive parameter correction: Dynamically adjusts the stepper motor pulse frequency and vacuum pressure based on historical dispensing data. For example, for irregularly shaped tablets with smooth surfaces, the vacuum pressure is automatically increased to 80kPa while the stepper motor speed is reduced by 20%.

[0008] Beneficial effects Compared with the prior art, the present invention has the following significant advantages: Multimodal perception fusion: By leveraging the complementary capabilities of weight, vision, and infrared data, it effectively solves the problem of sorting irregular-shaped tablets, increasing the accuracy of tablet delivery to 99.2%. Adaptive Kalman filter: real-time correction of dispensing parameters, eliminating error accumulation during long-term use, and ensuring accurate dispensing every time; Three-stage pill dispensing mechanism: Combining a spiral chute, vacuum adsorption, and voice coil motor, it achieves ±0.1mm positioning accuracy and is suitable for all types of tablets with a diameter of 2-20mm. Intelligent blockage handling: Photoelectric sensors are combined with reverse drive to achieve automatic blockage detection and drug recovery, reducing manual intervention. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 is a schematic diagram of the system modules of the present invention; Figure 2 is a schematic diagram of the three-level medicine distribution mechanism; Figure 3 is a flow chart of the multimodal data fusion algorithm; Figure 4 is a schematic diagram of the congestion detection and emergency recovery mechanism; Figure 5 is a schematic diagram of the adaptive Kalman filter algorithm. DETAILED DESCRIPTION

[0010] System hardware design The smart pill box utilizes a modular design, comprising a medicine compartment module, a dispensing module, a sensing module, and a control module. The medicine compartment module contains eight independent medicine storage compartments, each with a capacity of 50ml and equipped with a motorized flap (driven by an SG90 servo). The dispensing module integrates a three-stage dispensing mechanism and is connected to a stepper motor via a coupling. It is installed below the medicine compartment. Sensing modules are located at the bottom of the medicine compartment, along the dispensing channel, and at the medicine outlet, responsible for weight monitoring, pill identification, and counting, respectively. The control module utilizes an STM32F767IGT6 development board, communicating with each module via the CAN bus. An integrated WiFi module (ESP8266) enables remote monitoring.

[0011] Workflow Medication entry: Users scan the QR code on the medicine box through the mobile app to enter the medication management interface, take a photo of the medication, and enter information such as the name and dosage. The system automatically identifies the tablet features using a micro-machine vision module and establishes a medication database.

[0012] Drug dispensing control: The main processor drives the corresponding medicine compartment cover to open according to the preset medication plan; The coarse separation stepper motor starts and pushes the drugs to the spiral chute to achieve preliminary separation; The precision layering screw module drives the vacuum gripper to move to the target position and grab a single tablet; The micro-layered voice coil motor adjusts its position based on visual feedback to ensure that the tablets fall accurately into the outlet; The infrared counter records the number of medicines discharged, and the weight sensor verifies the weight of the medicines in real time. If the deviation exceeds ±0.5%, re-sorting is triggered.

[0013] Exception handling: When the photoelectric sensor detects a blockage, the main processor immediately stops dispensing and starts reverse drive and vacuum adsorption recovery; The system pushes alarm information through the APP and displays the fault code on the medicine box LCD screen (such as E001 for channel blockage); The recovered tablets are temporarily stored in a separate warehouse, and users can re-sort or empty them by manually pressing buttons.

[0014] Algorithm Implementation The adaptive Kalman filter algorithm was implemented on the STM32F767 chip, using fixed-point arithmetic to optimize processing speed. The state transition matrix A and observation matrix H were dynamically adjusted based on the physical characteristics of the tablets. For example, for round tablets, A = \begin{bmatrix}1&T&0 \\ 0&1&0 \\ 0&0&1\end{bmatrix} , where T is the sampling period (50ms). The noise covariance matrices Q and R were continuously optimized through online learning, with initial values ​​set to Q = diag([0.1, 0.01, 0.001]) and R = diag([0.02, 0.05, 0.1]) .

[0015] Experimental verification In a test simulating medication use for elderly patients, the system performed 1,000 dispensing tests on five types of tablets, including aspirin (round, 8mm diameter) and gliclazide (irregular, 12mm diameter). The results showed an overall accuracy rate of 99.2%, including 98.7% for irregular-shaped tablets. The blockage rate was less than 0.3%, and the emergency retrieval success rate was 100%. The system consumes less than 0.5W in standby mode and less than 3W in operation. The built-in lithium battery provides 72 hours of continuous use.

[0016] Through the deep integration of multimodal perception and adaptive control, the present invention significantly improves the reliability and adaptability of the automatic medicine box. It is particularly suitable for complex medication scenarios for elderly patients and has broad application prospects.

Claims

1. A smart medicine box precise medicine dispensing control system based on multimodal sensing, characterized by: It includes a main processor module, a multimodal sensing module, a three-level medicine separating mechanism, a communication module and a power module; the main processor module adopts an ARMCortex-M7 chip to coordinate the work of each module and perform data fusion processing; the multimodal sensing module includes a high-precision weight sensor, a micro machine vision module and an infrared counter, which are respectively used to monitor the weight change of medicines in real time, identify the shape, color and position of tablets, and detect the number of tablets passing through; the three-level medicine separating mechanism is driven by a stepper motor and includes three levels: coarse separation, fine separation and fine adjustment, which are used to achieve the separation and precise positioning of tablets.

2. The intelligent medicine box precise medicine dispensing control system according to claim 1 is characterized in that: The high-precision weight sensor uses an HBM U10M weighing sensor with an accuracy of ±0.01g and is installed at the bottom of the medicine warehouse; the micro machine vision module is equipped with a Sony IMX290 image sensor, an LED ring light source, and a resolution of 1920×1080; the infrared counter uses a Sharp GP2Y0A21YK0F infrared pair tube with a response time of < 1ms.

3. The intelligent medicine box precise medicine dispensing control system according to claim 1 is characterized in that: The coarse stratification uses a two-phase hybrid stepper motor to achieve preliminary separation of tablets through a spiral slide; the fine stratification uses a high-precision screw module and a vacuum adsorption gripper to grasp single tablets; the micro stratification integrates a voice coil motor to achieve ±0.1mm position compensation.

4. The intelligent medicine box precise medicine dispensing control system according to claim 1 is characterized in that: It also includes a blockage detection and emergency recovery module. The blockage detection module installs a photoelectric sensor in the medicine dispensing channel, which triggers a blockage alarm when no tablet movement is detected for 500ms. The emergency recovery module is used to start the reverse stepper motor drive and simultaneously open the vacuum adsorption device to suck the blocked tablets back to the temporary storage bin.

5. A smart medicine box precise medicine dispensing control method based on multimodal sensing, applied to the control system according to any one of claims 1 to 4, characterized in that: The following steps are involved: (1) Multimodal data fusion: The main processor collects weight, visual and infrared data in real time and uses an adaptive Kalman filter algorithm for data fusion, where the state equation is X_k = AX_{k-1} + BU_{k-1} + W_{k-1}, the observation equation is Z_k = HX_k + V_k, the state vector X_k contains the tablet position, speed and mass, and the observation vector Z_k contains the weight, visual coordinates and infrared counts; (2) Blockage detection and emergency recovery: The photoelectric sensor is used to detect whether the dispensing channel is blocked. When no tablet movement is detected for 500ms, a blockage alarm is triggered, and the reverse stepper motor drive and vacuum adsorption device are started for emergency recovery; (3) Adaptive parameter correction: The stepper motor pulse frequency and vacuum adsorption pressure are dynamically adjusted according to historical drug dispensing data.

6. The smart medicine box precise medicine dispensing control method according to claim 5, characterized in that: In the adaptive Kalman filter algorithm, the state transfer matrix A and the observation matrix H are dynamically adjusted according to the physical characteristics of the tablets, and the noise covariance matrices Q and R are continuously optimized through online learning.

7. The smart medicine box precise medicine dispensing control method according to claim 5, characterized in that: During the drug dispensing control process, the main processor drives the corresponding medicine bin lid to open according to the preset medication plan. The coarse stratification stepper motor starts to push the drugs to the spiral chute for initial separation. The fine stratification screw module drives the vacuum adsorption gripper to grab a single tablet. The micro stratification voice coil motor adjusts its position according to visual feedback. The infrared counter records the number of drugs dispensed. The weight sensor verifies the weight of the drugs in real time. If the deviation exceeds ±0.5%, re-sorting is triggered.

8. The smart medicine box precise medicine dispensing control method according to claim 5, characterized in that: When a blockage is detected, the system pushes an alarm message through the APP and displays a fault code on the LCD screen of the medicine box. The recovered tablets are temporarily stored in an independent warehouse, and users can re-sort or empty them by manually pressing buttons.

Images