Warehousing rice quality monitoring robot based on near infrared spectrum

Abstract

The invention discloses a robot for monitoring the quality of stored rice. The technical blank of an existing method for monitoring the quality of the stored rice in automation and nondestructive testing is filled. The robot comprises a timing wake-up system, a navigation positioning and driving system, a spectrum acquisition and quality detection system and an automatic recharging system. The timed awakening system triggers and awakens the robot according to different task requirements; the navigation positioning and driving system realizes path planning and autonomous navigation of the robot in a storage environment; the spectrum acquisition and quality detection system realizes nondestructive monitoring of the quality of the stored rice through a near infrared spectrum technology; the automatic recharging system is used for monitoring the electric quantity state of the robot and achieving automatic recharging when the electric quantity is low. The storage rice quality monitoring efficiency is remarkably improved, the labor cost is reduced, and powerful technical support is provided for modernized granary management.

Description

Technical Field

[0001] The invention belongs to the field of storage rice quality monitoring and relates to a storage rice quality monitoring robot based on near infrared spectroscopy. Background Art

[0002] Monitoring the quality of stored rice is an important part of ensuring subsequent processing and food safety. However, traditional manual sampling and detection methods have problems such as slow detection speed and poor sample representativeness, and can no longer meet the needs of modern agriculture for efficient and accurate monitoring. In addition, changes in temperature and humidity in the storage environment have a significant impact on rice quality. If there is a lack of real-time monitoring and response mechanisms, it may lead to quality degradation and even serious economic losses. Existing monitoring equipment is mostly single-point measurement or semi-automatic devices, which do not have the ability to fully cover and dynamically adapt to complex storage environments, and are difficult to meet the needs of intelligent warehouse management. For this reason, there is an urgent need for an intelligent device that can realize non-destructive real-time monitoring of rice quality, and at the same time has the ability of autonomous navigation and environmental parameter monitoring, thereby significantly improving monitoring efficiency, data accuracy and system automation level.

[0003] Near-infrared spectroscopy is a rapid, non-destructive, and environmentally friendly optical detection method. The sum and harmonic absorption of the vibrational groups of most organic compounds can be detected in the near-infrared spectrum, providing information about the chemical bonds or functional groups within the molecule, and thus revealing the structure and composition of the substance. Therefore, it can be used for rapid, non-destructive testing of rice quality, such as moisture and fatty acids. Combining near-infrared spectroscopy with robotics will enable rapid, real-time, non-destructive monitoring of stored rice quality. Summary of the Invention

[0004] In order to solve the problems existing in the background technology and actual detection, the purpose of the present invention is to provide a storage rice quality monitoring robot based on near-infrared spectroscopy, which can autonomously navigate and monitor rice quality in real time in a storage environment. At the same time, it has functions such as timed wake-up, multi-task scheduling, environmental data collection and data analysis, ensuring the efficient execution of monitoring tasks and the accuracy of collected data.

[0005] The technical solution adopted in the present invention is as follows:

[0006] The robot includes a timed wake-up system, a navigation, positioning, and driving system, a spectrum acquisition and quality detection system, and an automatic recharge system. The timed wake-up system triggers the robot to wake up according to different task requirements; the navigation, positioning, and driving system enables the robot to plan paths, navigate autonomously, and drive in a storage environment; the spectrum acquisition and quality detection system uses near-infrared spectroscopy technology to achieve non-destructive monitoring of the quality of stored rice, and performs quality analysis and generates reports; the automatic recharge system monitors the robot's battery status and enables autonomous recharge when the battery is low.

[0007] The robot hardware structure of the present invention is composed of a car body, an STM32 single-chip microcomputer, a raspberry pie, a laser radar, an inertial measurement unit (IMU), a data storage module, a real-time clock (RTC) module, a low-power power management unit (PMU), a temperature and humidity sensor, a spectrum acquisition device, an intelligent lifting device, an on-board charging device, a charging pile, and a touch screen display. The car body includes an upper load plate, a lower load plate, two motors, and a track wheel. The upper load plate carries a laser radar, a raspberry pie, an STM32 single-chip microcomputer, a touch screen display, a spectrum acquisition device, and an on-board charging device. The lower load plate is used to carry a spectrometer, a light source, and a power supply. The two motors are located between the upper and lower load plates. The track wheel is located on both sides of the car, and crawler-type movement gives the robot good mobility performance and can adapt to a variety of complex terrains. The STM32 single-chip microcomputer is equipped with a real-time clock (RTC) module, a low-power power management unit (PMU), and a temperature and humidity sensor, and is located at the tail of the upper load plate. The raspberry pie is equipped with an inertial measurement unit (IMU) and a data storage module, and is located directly above the STM32 single-chip microcomputer. The lidar is located in the center of the upper support plate. The upper support plate occupies a high position, preventing obstruction by devices in front, ensuring ranging accuracy and data integrity. The spectral acquisition device includes a near-infrared spectrometer, two quartz optical fibers, a light source, and a diffuse reflectance P-TIP probe. The light source is a tungsten-halogen lamp. A quartz optical fiber connects the light source outlet to one port of the diffuse reflectance P-TIP probe via a fiber optic connector. The other port of the diffuse reflectance P-TIP probe is connected to the near-infrared spectrometer via another quartz optical fiber. The spectrometer is connected to the Raspberry Pi via a USB port. The intelligent lifting device integrates a motorized lead screw, flat shovel, guide rails, and a laser ranging module. It is deployed directly in front of the robot to enable adaptive spectral acquisition. The motorized lead screw is mounted perpendicular to the ground, with the diffuse reflectance P-TIP probe on the left and the laser ranging module on the right. The flat shovel is fixed at the bottom. The onboard charging station is located at the rear of the robot, with a charging station located at the intersection of the central aisle of the granary. The touchscreen display is located at the rear of the robot, above the Raspberry Pi.

[0008] The robot system of the present invention adopts the collaborative architecture of raspberry pie and STM32 single-chip microcomputer, wherein the STM32 single-chip microcomputer is equipped with real-time clock (RTC) module, low-power power management unit and temperature and humidity sensor, is responsible for the low-latency tasks such as real-time time management, sensor data acquisition, motor drive of hardware level, and realizes energy-saving control by power management unit; raspberry pie runs ROS system as main control unit, completes complex task planning, algorithm processing and decision logic based on the distributed computing ability of ROS, and all ROS codes are stored in raspberry pie; both realize data interaction through serial communication, raspberry pie sends control instruction to STM32 single-chip microcomputer according to task scheduling result, STM32 single-chip microcomputer performs hardware operation and feeds back execution state, forms complete closed-loop control system, both ensures the real-time response capability of underlying hardware, and gives full play to the advantage of raspberry pie in complex computing and task scheduling.

[0009] The timed wake-up system includes a real-time clock (RTC) module, a low-power power management unit, a scheduled task module, and a multi-task priority scheduling module. The real-time clock (RTC) module and the low-power power management unit are installed in the STM32, while the scheduled task module and the multi-task priority scheduling module are integrated into the Raspberry Pi. The scheduled task module of the timed wake-up system has a periodic task scheduling function. Users can customize different task cycles (such as daily, weekly, or monthly) and execution times through the configuration interface. Based on the parameters set by the user, the scheduled task module for periodic task scheduling automatically generates a task scheduling plan, triggers the monitoring task at the specified time point, and ensures that the task is executed efficiently according to the cycle. In addition, the scheduled task module of the timed wake-up system can also automatically calculate and adjust the execution time of the next task based on the set periodic task characteristics to ensure the efficient execution of periodic tasks.

[0010] The scheduled wakeup system's multitasking priority scheduling module prioritizes periodic and burst tasks by setting priority tags. User-defined periodic tasks are assigned a low priority, while burst tasks are assigned a high priority. Burst tasks are triggered by changes in environmental parameters monitored in real time by the built-in temperature and humidity sensors. When a sudden change in environmental parameters exceeds a set threshold, the scheduled wakeup system automatically raises the task priority and wakes the Raspberry Pi via a hardware interrupt, quickly switching the robot from low-power mode to normal operation and ensuring timely response to burst tasks. If a periodic task conflicts with a burst task, the scheduled wakeup system prioritizes the burst task and records any unfinished periodic tasks. The system's built-in task backtracking mechanism records unfinished periodic tasks and automatically executes the periodic task after the burst task is processed, ensuring the integrity and continuity of the periodic tasks.

[0011] When not in a task, the robot operates in low-power mode. The Raspberry Pi and most of its non-essential peripheral modules are shut down or put into hibernation. Only essential core components, such as the real-time clock (RTC) module, temperature and humidity sensors, and a low-power power management unit (PMU), remain operational. This ensures minimal monitoring functionality and energy consumption in low-power mode, and allows the robot to quickly resume operation upon receiving a wake-up signal. When the robot receives a wake-up signal, the scheduled wake-up system triggers a hardware interrupt to wake up the main processor, restoring the robot from low-power mode to normal operation. The specific wake-up process is as follows: After the Real-Time Clock (RTC) module is triggered, it outputs a low-level pulse via a hardware pin, directly waking the Raspberry Pi. When the Raspberry Pi receives this signal, its internal PMU is activated, gradually restoring power to the core and peripheral modules. The system activates each functional module sequentially according to the preset wake-up sequence, ensuring a smooth startup of the scheduled wake-up system and rapid entry into task execution.

[0012] The navigation, positioning, and driving system includes a map-building and path-planning module, a target point management module, and a loop control module. The map-building and path-planning module primarily uses lidar and an inertial measurement unit (IMU) as its primary sensor data sources. The lidar uses a 360° rotational scanning method, boasts a 30-meter range, and 0.1° angular resolution, capturing information about the warehouse environment. The inertial measurement unit (IMU) utilizes a 9-axis sensor and supports a 200Hz sampling rate, monitoring the robot's posture and acceleration for precise navigation and stable positioning. The lidar is deployed on top of the robot and connected to a Raspberry Pi via an STM32 microcontroller. The inertial measurement unit (IMU) is integrated into the STM32 microcontroller development board and transmits posture data to the Raspberry Pi via a serial port. Laser ranging allows for rapid construction of a three-dimensional environmental map, real-time obstacle sensing for dynamic obstacle avoidance, and high-precision positioning based on map matching, supporting the robot's autonomous navigation. The Raspberry Pi uses the laser SLAM algorithm to build a high-precision environmental model for precise spatial structure perception. Combined with the A* algorithm or the dynamic windowing algorithm (DWA), it implements path planning and dynamic obstacle avoidance, ensuring stable robot operation in complex scenarios. Furthermore, the target point management module and loop control module are integrated into the Raspberry Pi, supporting the system's target point prioritization and loop task execution, significantly improving task scheduling efficiency and navigation flexibility. The system optimizes global path planning and local obstacle avoidance capabilities in warehouse environments through multi-sensor data fusion, forming a comprehensive intelligent solution from environmental modeling to task execution.

[0013] The navigation, positioning, and driving system's map-building and path-planning modules use an adaptive extended Kalman filter (EKF) algorithm to fuse the robot's multi-sensor data. This algorithm can adjust filter parameters in real time based on changes in sensor noise, making sensor data fusion more accurate and improving the robustness of navigation and positioning.

[0014] The navigation, positioning, and driving system's map-building and path-planning modules utilize the Gmapping algorithm, using LiDAR to collect spatial data about the warehouse environment and employing particle filtering to generate a high-precision map of the warehouse environment. Based on this, the system employs the A* algorithm for global path planning, calculating the optimal path to the target point to ensure the robot's efficient arrival. Furthermore, the system employs the dynamic windowing algorithm (DWA) for local path adjustments, detecting obstacles in real time and dynamically optimizing the path, enabling the robot to flexibly avoid dynamic and static obstacles.

[0015] The target point management module of the navigation, positioning, and driving system is primarily responsible for recording and dynamically updating target point locations. It supports adding, removing, and prioritizing targets, enabling efficient robot switching between different monitoring targets. The loop control module, incorporating a path planning algorithm, enables the robot to navigate autonomously between multiple targets, ensuring the continuity and efficient execution of multi-target monitoring tasks.

[0016] The spectral acquisition and quality inspection system consists of a spectral acquisition device, an intelligent lifting device, a temperature and humidity compensation model, a rice quality quantitative analysis model, and a data storage module. The spectral acquisition device primarily comprises a near-infrared spectrometer, two quartz optical fibers, a light source, and a diffuse reflectance P-TIP probe. The light source is a halogen tungsten lamp, whose output light is connected to a 600μm core diameter quartz optical fiber. This fiber utilizes the low loss characteristics of quartz fiber in the 360-2000nm wavelength range to efficiently transmit the optical signal generated by the halogen tungsten lamp. The other end of the quartz optical fiber is connected to a diffuse reflectance P-TIP probe. The light signal from the light source is transmitted to the probe via the optical fiber and then focused to illuminate the sample. The diffusely reflected light from the sample is collected by the probe and then transmitted back via another optical fiber to a high-precision near-infrared spectrometer (900-1700nm wavelength range), which splits the light and converts it into an electrical signal to obtain spectral data. The spectrometer is connected to a Raspberry Pi via a USB port, uploading spectral data in real time for subsequent data processing and analysis.

[0017] The intelligent lifting device integrates an electric lead screw, a flat shovel, guide rails, and a laser ranging module, deployed directly in front of the robot to enable adaptive spectral acquisition. The laser ranging module monitors the vertical distance between the diffuse reflectance probe and the rice grains on the ground in real time. When the detection value deviates from the preset threshold, the main control unit (Raspberry Pi) sends a command to the STM32 microcontroller, which drives the electric lead screw motor to adjust the probe height. If the distance is too large, the lead screw rotates forward to push the probe down, while if the distance is too small, the lead screw rotates in the reverse direction to raise the probe until the initial set distance is restored. The synchronously operating flat shovel mechanism maintains a horizontal position via the guide rails, flattening the rice grain surface in front of the robot as it moves, forming a continuous, flat detection reference surface. This effectively eliminates interference from terrain fluctuations on spectral data acquisition and ensures that the multispectral sensor can stably acquire uniform sample data in dynamic environments.

[0018] The temperature and humidity compensation model, the rice quality quantitative analysis model, and the data storage module are all installed on the Raspberry Pi. The temperature and humidity compensation model is an environmental compensation model constructed based on a multivariate linear regression algorithm to correct for the effects of ambient temperature and humidity changes on the collected rice spectra. The rice quality quantitative analysis model analyzes the corrected rice spectra to output quality information such as rice moisture and fatty acids. The data storage module is used to store all original and corrected spectral data and rice quality predictions, and to send this data to a cloud server for data backup and remote access. The Raspberry Pi, serving as the main control unit, reads data from the temperature and humidity sensor connected to the STM32 microcontroller in real time through a communication interface and simultaneously acquires raw spectral data collected by the spectrometer. The temperature and humidity compensation model is then invoked, taking real-time environmental parameters (temperature and humidity) and the raw spectral data as input, and outputs corrected spectral data to eliminate the effects of ambient temperature and humidity. The corrected spectral data is then input into the rice quality quantitative analysis model to output quality indicators such as rice moisture and fatty acids. All raw and corrected spectral data and rice quality predictions are stored in a data storage module and synchronized to a cloud server for data backup and remote access. The system is equipped with a touchscreen display that visualizes rice quality data and environmental parameters in real time, supports historical data query and abnormality alarms, and forms a closed loop from data acquisition, compensation processing, model analysis, to cloud storage.

[0019] The automatic recharging system consists of an on-board charging unit, an infrared receiver module, a mission recovery module, and a fixed charging station in the granary. The on-board charging unit is integrated into the rear of the robot. The built-in infrared receiver module communicates with the STM32 microcontroller via an interface, monitoring the battery voltage and remaining charge in real time. When the battery level falls below a preset threshold, a navigation task is triggered. Combining the SLAM algorithm with the positioning signal from the charging station's infrared transmitter module, the robot plans an optimal path back to the charging station at the intersection of the granary's central aisle. The infrared transmitter module in the charging station transmits infrared signals, and the infrared receiver module in the on-board charging unit, combined with a triangulation algorithm, achieves short-range positioning and fine-tunes the posture, ensuring precise alignment of the charging contacts. During charging, the STM32 microcontroller continuously monitors the battery status. Once charging is complete, it automatically wakes up the mission recovery module, loads unfinished mission data (such as inspection routes and spectral acquisition points) from the local computer, and synchronously restores the robot's operating state through the state machine, achieving a seamless transition between charging and mission, ensuring the continuity and data integrity of the granary monitoring mission.

[0020] Through the coordinated work of the above systems, the present invention can efficiently and accurately complete the rice quality monitoring task and meet the real-time and reliability requirements of storage environment monitoring.

[0021] The beneficial effects of the present invention are:

[0022] The present invention realizes unmanned real-time monitoring of rice quality through the process design of timed wake-up, autonomous navigation and driving, spectrum acquisition and quality detection, and automatic recharging, which significantly improves the monitoring efficiency and automation level. The timed wake-up system of the present invention supports flexible scheduling of periodic and sudden tasks, and ensures timely response to tasks by setting priorities; the built-in task backtracking mechanism can automatically supplement unfinished periodic tasks after sudden tasks to ensure continuous and complete tasks. The user interface can flexibly set the task cycle and execution time to adapt to different storage scenarios. The robot of the present invention can monitor the environment in real time with the help of temperature and humidity sensors, and use the built-in multivariate linear regression temperature and humidity compensation model of the spectrum acquisition and quality detection system to dynamically eliminate the impact of the environment on the spectral data, ensuring accurate and consistent data acquisition; the rice quality quantitative analysis model in the system can calculate rice quality indicators in real time and generate high-precision quality reports; and the system is equipped with an intelligent lifting device, which can automatically adjust the height of the probe detection port according to the height of the rice surface to ensure stable spectrum acquisition and high monitoring accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 This is the system module block diagram of this robot;

[0024] Figure 2 This is an isometric view of this robot;

[0025] Figure 3 This is a schematic diagram of the chassis structure of this robot;

[0026] Figure 4 This is a schematic diagram of the robot's spectrum collection and intelligent lifting device structure;

[0027] Figure 5 This is the navigation path planning flow chart of this robot;

[0028] Figure 6 This is the automatic recharging flow chart of this robot;

[0029] In the figure: 1. Charging pile; 2. On-board charging device; 3. Raspberry Pi; 4. STM32 microcontroller; 5. Touch screen; 6. LiDAR; 7. Track wheel; 8. Diffuse reflectance P-TIP probe; 9. Intelligent lifting device; 10. Laser ranging module; 11. Halogen tungsten light source; 12. Near-infrared spectrometer; 13. Motor I; 14. Power supply; 15. Flat shovel; 16. Guide rail; 17. Electric screw; 18. Optical fiber I; 19. Optical fiber II; 20. Upper load plate; 21. Lower load plate; 22. Motor II DETAILED DESCRIPTION

[0030] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0031] An embodiment of the present invention provides a robot for monitoring the quality of stored rice based on near-infrared spectroscopy. The robot achieves real-time, non-destructive monitoring of the quality of stored rice through the coordinated operation of a timed wake-up system, a navigation, positioning, and driving system, a spectral acquisition and quality detection system, and an automatic recharge system. Specifically, the timed wake-up system, acting as a trigger for task initiation, precisely activates the robot at a preset time, rapidly switching it from standby mode to working mode and simultaneously sending initialization instructions to the navigation and positioning system. The navigation and positioning system then activates, relying on a laser radar 6 and a SLAM algorithm to generate a high-precision environmental model, and combines the A* algorithm or a dynamic window method to perform path planning and dynamic obstacle avoidance, accurately reaching the target spectral acquisition point. The spectral acquisition and quality detection system automatically activates after the robot reaches the specified coordinates. The high-precision near-infrared spectrometer 12 it carries scans the target sample, obtains spectral data, analyzes the data in real time, generates a quality report, and uploads it to the user end via a wireless communication module. The automatic recharge system completes the anchoring of the charging pile coordinates during the initialization phase, marking it as a global priority target point and permanently storing it in the environmental map. When the robot completes all spectral collection tasks or detects that the battery power is lower than the preset threshold, the system automatically activates the recharge mode. The navigation and positioning system then intervenes to perform path planning and dynamic obstacle avoidance. It reaches charging station 1 for charging and automatically switches to low-power mode after being fully charged, waiting for the next scheduled wake-up command, forming a seamless task cycle.

[0032] The robot hardware structure of the present invention consists of a vehicle body, an STM32 single-chip microcomputer 4, a Raspberry Pi 3, a laser radar 6, an inertial measurement unit (IMU), a data storage module, a real-time clock (RTC) module, a low-power power management unit, a temperature and humidity sensor, a spectrum acquisition device, an intelligent lifting device 9, an on-board charging device 2, a charging station 1, and a touch screen 5. The vehicle body includes an upper load plate 20, a lower load plate 21, a motor I13, a motor II 22, and track wheels 7. The upper load plate 20 carries the laser radar 6, Raspberry Pi 3, STM32 single-chip microcomputer 4, a touch screen 5, a spectrum acquisition device, and an on-board charging device 2; the lower load plate 21 is used to carry a near-infrared spectrometer 12, a halogen tungsten light source 11, and a power supply 14; the motors I13 and II 22 are located between the upper and lower load plates; and the track wheels 7 are located on both sides of the vehicle. The crawler-type movement gives the robot excellent mobility and enables it to adapt to various complex terrains. The STM32 microcontroller 4, equipped with a real-time clock (RTC) module, a low-power power management unit, and a temperature and humidity sensor, is located at the rear of the upper support plate. The Raspberry Pi 3, equipped with an inertial measurement unit (IMU) and a data storage module, is located directly above the STM32 microcontroller 4. The lidar 6 is located in the center of the upper support plate 20. Using the upper support plate 20 to occupy a high position prevents obstruction by devices in front, ensuring ranging accuracy and data integrity. The spectral acquisition device includes a near-infrared spectrometer 12, optical fiber I 18, optical fiber II 19, a halogen light source 11, and a diffuse reflectance P-TIP probe 8. The halogen light source 11 is a tungsten halogen lamp. The halogen light source 11's output port is connected to one end of optical fiber I 18, the other end of optical fiber I 18 is connected to one end of diffuse reflectance P-TIP probe 8, the other end of diffuse reflectance P-TIP probe 8 is connected to one end of optical fiber I 19, and the other end of optical fiber I 19 is connected to the near-infrared spectrometer 12. The near-infrared spectrometer 12 is connected to the Raspberry Pi 3 via a USB port. The intelligent lifting device 9 integrates a motorized screw 17, a flat shovel 15, a guide rail 16, and a laser ranging module 10. It is deployed directly in front of the robot to enable adaptive spectral acquisition. The motorized screw 17 is mounted perpendicular to the ground, equipped with a diffuse reflectance P-TIP probe 8 on the left and a laser ranging module 10 on the right. The flat shovel 15 is fixed at its bottom. The onboard charging device 2 is located at the rear of the robot, with the charging station 1 located at the intersection of the central aisle of the granary. The touchscreen display 5 is located at the rear of the robot, above the Raspberry Pi 3.

[0033] This robot system adopts the collaborative architecture of Raspberry Pi 3 and STM32 MCU 4. The STM32 MCU 4 is equipped with a real-time clock (RTC) module, a low-power power management unit and a temperature and humidity sensor. It is responsible for low-latency tasks such as real-time time management, sensor data acquisition, and motor drive at the hardware level, and realizes energy-saving control through the power management unit. The Raspberry Pi 3 runs the ROS system as the main control unit, and completes complex task planning, algorithm processing and decision logic based on the distributed computing capability of ROS. All ROS code is stored in the Raspberry Pi 3. The two realize data exchange through serial communication. The Raspberry Pi 3 sends control instructions to the STM32 MCU 4 based on the task scheduling results. The STM32 MCU 4 executes hardware operations and feedbacks the execution status, forming a complete closed-loop control system. This not only ensures the real-time responsiveness of the underlying hardware, but also gives full play to the advantages of the Raspberry Pi in complex computing and task scheduling.

[0034] The timed wake-up system includes a real-time clock (RTC) module, a low-power power management unit (PMU), a timed task module, and a multi-task priority scheduling module. The real-time clock (RTC) module and the PMU are installed on the STM32 MCU 4, while the timed task module and the multi-task priority scheduling module are integrated on the Raspberry Pi 3. The timed wake-up system includes a timed task module for periodic task scheduling. Users can customize different task cycles (such as daily, weekly, or monthly) and execution times through the configuration interface. Based on the parameters set by the user, the timed task module for periodic task scheduling automatically generates a task scheduling plan, triggers the monitoring task at the specified time point, and ensures that the task is executed efficiently according to the cycle. In addition, the timed task module for periodic task scheduling in the timed wake-up system can also automatically calculate and adjust the execution time of the next task based on the set periodic task characteristics, ensuring the efficient execution of periodic tasks.

[0035] The scheduled wakeup system's multitasking priority scheduling module prioritizes periodic and burst tasks by setting priority tags. User-defined periodic tasks are assigned a low priority, while burst tasks are assigned a high priority. Burst tasks are triggered by changes in environmental parameters monitored in real time by the built-in temperature and humidity sensors. When a sudden change in environmental parameters exceeds a set threshold, the scheduled wakeup system automatically raises the task priority and wakes the Raspberry Pi 3 via a hardware interrupt, enabling the robot to quickly switch from low-power mode to normal operation and ensure timely response to burst tasks. If a periodic task conflicts with a burst task, the scheduled wakeup system prioritizes the burst task and records any unfinished periodic tasks. The system's built-in task backtracking mechanism records unfinished periodic tasks and automatically replaces them with the periodic task after the burst task is processed, ensuring the integrity and continuity of the periodic tasks.

[0036] When not in a task, the robot operates in low-power mode. The Raspberry Pi 3 and most of its non-essential peripheral modules are shut down or put into hibernation. Only essential core components, such as the real-time clock (RTC) module, temperature and humidity sensors, and a low-power power management unit (PMU), remain operational. This ensures minimal monitoring functionality and energy consumption in low-power mode, and allows the robot to quickly resume operation upon receiving a wake-up signal. When the robot receives a wake-up signal, the scheduled wake-up system triggers a hardware interrupt to wake up the main processor, restoring the robot from low-power mode to normal operation. The specific wake-up process is as follows: After the Real-Time Clock (RTC) module is triggered, it outputs a low-level pulse via a hardware pin, directly waking the Raspberry Pi 3. Upon receiving this signal, the PMU within the Raspberry Pi 3 is activated, gradually restoring power to the core and peripheral modules. The system activates each functional module sequentially according to a pre-set wake-up sequence, ensuring a smooth startup of the scheduled wake-up system and rapid entry into task execution.

[0037] The spectrum acquisition and quality detection system includes a spectrum acquisition device, an intelligent lifting device 9, a rice quality quantitative analysis model, and a data storage module to realize spectrum data acquisition, quality prediction through a quantitative prediction model, and storage of related data.

[0038] The spectral acquisition device uses a 600μm core diameter quartz optical fiber as the core transmission medium. The light outlet of the halogen tungsten light source 11 (output power is 9W) is connected to one end of the optical fiber I 18, and the low-loss characteristics of the quartz optical fiber in the 360-2000nm band are used to efficiently transmit the light source light signal; the other end of the optical fiber I 18 is connected to the diffuse reflection P-TIP probe 8, so that the light source light is focused by the probe to illuminate the sample. The generated diffuse reflection light is transmitted back through the optical fiber II 19 to the high-precision near-infrared spectrometer 12 (900-1700nm band), which splits the light and converts the optical signal into an electrical signal to obtain spectral data; the spectrometer 12 is connected to the Raspberry Pi 3 via a USB interface to upload the spectral data in real time for subsequent data processing and analysis.

[0039] The intelligent lifting device 9 consists of an electric screw 17, a flat shovel 15, a guide rail 16, and a laser ranging module 10. It is deployed directly in front of the robot to realize the adaptive spectral acquisition function. The diffuse reflection P-TIP probe 8 and the laser ranging module 10 are installed on the intelligent lifting device 9. The laser ranging module 10 monitors the vertical distance between the detection port of the diffuse reflection P-TIP probe 8 and the rice on the ground in real time. When the detection value deviates from the preset threshold, the main control unit (Raspberry Pi 3) sends a command to the STM32 microcontroller 4, which drives the electric screw motor to adjust the probe height. If the distance is too large, the screw rod rotates forward to push the probe down. If the distance is too small, the screw rod rotates in the reverse direction to drive the probe up until the initial set distance is restored. The synchronous flat shovel mechanism is maintained in a horizontal position by the guide rail. During the robot's movement, it flattens the surface of the rice in front, forming a continuous and flat detection reference surface. This effectively eliminates the interference of terrain fluctuations on spectral data acquisition and ensures that the multispectral sensor can stably obtain uniform sample data in a dynamic environment.

[0040] The spectrum acquisition system works in conjunction with the temperature and humidity sensor to record the ambient temperature and humidity while collecting rice spectral data. The system has a built-in two-layer data processing mechanism: (1) The temperature and humidity compensation module is based on a multivariate linear regression algorithm to build an environmental compensation model. It uses the real-time collected temperature and humidity data to establish a dynamic mapping relationship between environmental factors and spectral characteristic values. This module adopts a spectral preprocessing compensation strategy to correct the spectral drift caused by temperature and humidity changes (such as absorption peak shift, intensity attenuation, etc.) before the spectral data enters the quantitative analysis model to ensure the environmental adaptability of the input data. The compensation process is independent of the quantitative analysis model and only serves as a data preprocessing link. (2) The rice quality quantitative analysis model is an independent analysis model constructed using chemometric methods. Its input is standardized spectral data after temperature and humidity compensation. This model focuses on the analysis of rice quality characteristic parameters (such as moisture, fatty acid value, etc.) and achieves quality grading by comparing with the standard sample database. After the analysis results are calculated locally, a visual quality report is generated through the wireless communication module and uploaded to the user end in real time for centralized management and traceability.

[0041] The humidity compensation model, rice quality quantitative analysis model and data storage module are all installed on the Raspberry Pi. As the main control unit, the Raspberry Pi reads the temperature and humidity sensor data connected to the STM32 MCU 4 microcontroller in real time through the communication interface, and simultaneously obtains the original spectral information collected by the spectrometer. The system first calls the temperature and humidity compensation model based on multiple linear regression, takes the real-time environmental parameters (temperature, humidity) and spectral data as input, and generates compensated spectral features to eliminate environmental interference. Subsequently, the compensated spectral data is input into the rice quality quantitative analysis model, and key quality indicators such as moisture and fatty acid content are output. All raw data, compensation results and model prediction values ​​are stored in the Raspberry Pi local database and synchronized to the cloud server for data backup and remote access. The system is equipped with a touch screen that displays rice quality data and environmental parameters in real time, supports historical data query and abnormal alarm functions, and forms a closed loop of the entire process from data acquisition, compensation processing, model analysis to cloud storage.

[0042] The automatic recharging system consists of an on-board charging device 2, an infrared receiver module, a mission recovery module, and a fixed charging station 1 in the granary. The on-board charging device 8, located at the rear of the robot, has a built-in infrared receiver module and communicates with the STM32 microcontroller 4 via an interface. The infrared transmitter module is built into the charging station. Charging station 1 is located at the intersection of the central aisle of the granary to ensure that the robot can cover the maximum inspection area on a single charge, improving inspection efficiency. The mission recovery module is integrated into the Raspberry Pi 3.

[0043] The automatic recharging system works in conjunction with the navigation and positioning system. The automatic recharging system sets the location of charging pile 1 as the priority target point and stores its precise coordinates. When the automatic recharging mode is triggered, the navigation and positioning system assists in calculating the optimal path to charging pile 1. Through global and local path planning, the robot can autonomously navigate to charging pile 1.

[0044] The automatic recharging system cooperates with the infrared transmitting module equipped on the charging pile 1 and the receiving module equipped on the robot. Based on the triangulation algorithm, it realizes short-distance positioning and posture fine-tuning when the robot approaches the charging pile 1, ensuring that the charging contacts are aligned to ensure charging stability.

[0045] The automatic recharge system includes a task recovery module, which is used to record unfinished tasks when the robot interrupts the task due to low battery. After the recharge is completed, the automatic recharge system decides whether to resume the task based on the periodic task rules set by the user. When the task exceeds the set period, it will no longer be resumed to ensure the integrity and rationality of the task.

[0046] The above specific embodiments are used to illustrate the present invention rather than to limit the present invention. Any modifications and changes made to the present invention within the spirit of the present invention and the protection scope of the claims shall fall within the protection scope of the present invention.

Claims

1. A rice storage quality monitoring robot based on near-infrared spectroscopy, characterized by: The robot consists of a timed wake-up system, a navigation, positioning, and driving system, a spectrum acquisition and quality inspection system, and an automatic recharge system. The timed wake-up system triggers the robot to wake up according to different task requirements; the navigation, positioning, and driving system enables the robot to plan paths, navigate autonomously, and drive in a storage environment; the spectrum acquisition and quality inspection system uses near-infrared spectroscopy to non-destructively monitor the quality of stored rice, perform quality analysis, and generate reports; and the automatic recharge system monitors the robot's battery status and automatically recharges when the battery is low.

2. The robot according to claim 1, characterized in that The robot's hardware structure consists of a vehicle body, an STM32 microcontroller, a Raspberry Pi, a laser radar, an inertial measurement unit (IMU), a data storage module, a real-time clock (RTC) module, a low-power power management unit, a temperature and humidity sensor, a spectrum acquisition device, an intelligent lifting device, an on-board charging device, a charging pile, and a touch screen display. The vehicle body includes an upper load plate, a lower load plate, two motors, and track wheels. The upper load plate carries the laser radar, Raspberry Pi, STM32 microcontroller, a touch screen display, a spectrum acquisition device, and an on-board charging device. The lower carrier plate supports the spectrometer, halogen light source, and power supply; two motors are located between the upper and lower carrier plates. Track wheels are located on either side of the robot, enabling it to adapt to a variety of complex terrains. An STM32 microcontroller, equipped with a real-time clock (RTC) module, a low-power power management unit, and temperature and humidity sensors, is located at the rear of the upper carrier plate. A Raspberry Pi, equipped with an inertial measurement unit (IMU) and a data storage module, is located directly above the STM32 microcontroller. The lidar is located in the center of the upper carrier plate, using the upper carrier plate to occupy a high position to prevent obstruction by devices in front, ensuring ranging accuracy and data integrity. The spectral acquisition device includes a near-infrared spectrometer, two quartz optical fibers, a light source, and a diffuse reflectance P-TIP probe. The light source is a halogen tungsten lamp. A quartz optical fiber connects the light source output to one port of the diffuse reflectance P-TIP probe. The other port of the diffuse reflectance P-TIP probe is connected to the near-infrared spectrometer via another quartz optical fiber. The spectrometer is connected to the Raspberry Pi via a USB port. The intelligent lifting mechanism integrates a motorized lead screw, flat shovel, guide rails, and a laser ranging module. It is deployed directly in front of the robot to enable adaptive spectral acquisition. The lead screw is mounted perpendicular to the ground, equipped with a diffuse reflectance P-TIP probe on the left and a laser ranging module on the right. The flat shovel is fixed at the bottom. The onboard charging system is located at the rear of the robot, with a charging station located at the intersection of the central aisle of the granary. The touchscreen display is located at the rear of the robot, above the Raspberry Pi.

3. The robot according to claim 1, characterized in that The robot system adopts a collaborative architecture of Raspberry Pi and STM32 single-chip microcomputer, in which the STM32 single-chip microcomputer is equipped with a real-time clock (RTC) module, a low-power power management unit and a temperature and humidity sensor, and is responsible for low-latency tasks such as real-time time management, sensor data acquisition, and motor drive at the hardware level, and realizes energy-saving control through the power management unit; Raspberry Pi runs the ROS system as the main control unit, and completes complex task planning, algorithm processing and decision logic based on the distributed computing capability of ROS, and all ROS codes are stored in the Raspberry Pi; the two realize data exchange through serial communication. The Raspberry Pi sends control instructions to the STM32 single-chip microcomputer according to the task scheduling results. The STM32 single-chip microcomputer executes hardware operations and feedbacks the execution status, forming a complete closed-loop control system, which not only ensures the real-time responsiveness of the underlying hardware, but also gives full play to the advantages of Raspberry Pi in complex computing and task scheduling.

4. The robot according to claim 1, characterized in that The robot's timed wake-up system includes a real-time clock (RTC) module, a low-power power management unit, a timed task module, and a multi-task priority scheduling module. The real-time clock (RTC) module and the low-power power management unit are mounted in STM32, and the timed task module and the multi-task priority scheduling module are integrated into the Raspberry Pi. The timed task module of the timed wake-up system has a periodic task scheduling function, and users can customize different task cycles (such as daily, weekly or monthly) and execution times through the configuration interface. According to the parameters set by the user, the timed task module of the periodic task scheduling will automatically generate a task scheduling plan, trigger the monitoring task at a specified time point, and ensure that the task is executed efficiently according to the cycle. In addition, the timed task module of the timed wake-up system can also automatically calculate and adjust the execution time of the next task according to the set periodic task characteristics, ensuring the efficient execution of periodic tasks. The multi-task priority scheduling module of the timed wake-up system can realize the priority division of periodic tasks and burst tasks by setting priority tags. User-defined periodic tasks are set to low priority, while burst tasks are set to high priority. Burst tasks are triggered by changes in environmental parameters monitored in real time by the built-in temperature and humidity sensors. When a sudden change in an environmental parameter exceeds a set threshold, the scheduled wakeup system automatically raises the task priority and wakes the Raspberry Pi via a hardware interrupt, enabling the robot to quickly switch from low-power mode to normal operation, ensuring timely response to the sudden task. When periodic tasks conflict with sudden tasks, the scheduled wakeup system prioritizes the sudden task and records any unfinished periodic tasks. The system's built-in task backtracking mechanism records unfinished periodic tasks and automatically replaces them with the periodic task after the sudden task is processed, ensuring the integrity and continuity of the periodic tasks. When the robot is not tasked, it enters low-power mode. The Raspberry Pi and most of its non-essential peripheral modules are shut down or put into hibernation, leaving only essential core components such as the real-time clock (RTC), temperature and humidity sensors, and a low-power power management unit. This ensures minimal monitoring functionality and energy consumption in low-power mode, allowing the robot to quickly resume operation upon receiving a wakeup signal. Upon receiving the wakeup signal, the scheduled wakeup system triggers a hardware interrupt to wake up the main processor, restoring the robot from low-power mode to normal operation. The specific wake-up process is as follows: After the real-time clock (RTC) module is triggered, it outputs a low-level pulse through the hardware pin to directly wake up the Raspberry Pi. When the Raspberry Pi receives this signal, its internal power management unit will be activated to gradually restore power supply to the core and peripheral modules. The system activates each functional module in sequence according to the preset wake-up sequence to ensure the smooth startup of the timed wake-up system and the rapid entry into the task execution state.

5. The robot according to claim 1, characterized in that The navigation, positioning and driving system includes a map construction and path planning module, a target point management module, and a loop control module. The map construction and path planning module uses laser radar and inertial measurement unit (IMU) as the main perception data sources, and uses the adaptive extended Kalman filter (Adaptive EKF) algorithm to realize the robot's multi-sensor data fusion. The laser radar is deployed on the top of the robot, using a 360° rotation scanning method, with a 30-meter ranging range and 0.1° angular resolution. It is responsible for capturing storage environment information and is connected to the Raspberry Pi through the STM32 microcontroller; the inertial measurement unit (IMU) is integrated into the STM32 microcontroller development board, using a 9-axis sensor, supporting a 200Hz sampling rate, for monitoring the robot's posture and acceleration, achieving precise navigation and stable positioning, and transmitting posture data to the Raspberry Pi through the serial port. Through laser ranging, a three-dimensional environment map is quickly constructed, obstacle information is perceived in real time for dynamic obstacle avoidance, and high-precision positioning is achieved based on map matching to support the robot's autonomous navigation. The Raspberry Pi uses the laser SLAM algorithm to build a high-precision environmental model for precise spatial structure perception. Combined with the A* algorithm or the dynamic windowing algorithm (DWA), it implements path planning and dynamic obstacle avoidance, ensuring stable robot operation in complex scenarios. Furthermore, the target point management module and loop control module are integrated into the Raspberry Pi, supporting the system's target point prioritization and loop task execution, significantly improving task scheduling efficiency and navigation flexibility. The system optimizes global path planning and local obstacle avoidance capabilities in warehouse environments through multi-sensor data fusion, forming a comprehensive intelligent solution from environmental modeling to task execution.

6. The robot according to claim 1, characterized in that The spectral acquisition and quality detection system includes a spectral acquisition device, an intelligent lifting device, a temperature and humidity compensation model, a rice quality quantitative analysis model, and a data storage module. The spectral acquisition device primarily comprises a near-infrared spectrometer, two quartz optical fibers, a halogen tungsten lamp light source, and a diffuse reflectance P-TIP probe. The light source's optical signal is transmitted via the quartz optical fiber to the diffuse reflectance P-TIP probe for focused illumination of the sample. The diffusely reflected light generated by the sample is collected by the probe and then transmitted back via another optical fiber to the near-infrared spectrometer, which splits the light and converts it into an electrical signal to obtain spectral data. The spectrometer is connected to a Raspberry Pi via a USB port, uploading spectral data in real time for subsequent data processing and analysis. The intelligent lifting device, integrated with a motorized screw, flat shovel, guide rails, and a laser ranging module, is deployed directly in front of the robot to enable adaptive spectral acquisition. The laser ranging module monitors the vertical distance between the diffuse reflection probe and the rice on the ground in real time. When the detection value deviates from the preset threshold, the Raspberry Pi sends a command to the STM32 microcontroller, which drives the electric screw motor to adjust the probe height. If the distance is too large, the screw will rotate forward to push the probe down. If the distance is too small, the screw will rotate in the reverse direction to drive the probe up until the initial set distance is restored. The synchronously running flat shovel mechanism maintains a horizontal posture through the guide rail, and flattens the surface of the rice in front during the movement of the robot to form a continuous and flat detection reference surface, effectively eliminating the interference of terrain undulations on spectral data collection, and ensuring that the multispectral sensor can stably obtain uniform sample data in a dynamic environment. The temperature and humidity compensation model, the rice quality quantitative analysis model, and the data storage module are all installed on the Raspberry Pi. The temperature and humidity compensation model is an environmental compensation model constructed based on the multivariate linear regression algorithm, which is used to correct the influence of changes in ambient temperature and humidity on the collected rice spectrum. The rice quality quantitative analysis model analyzes the corrected rice spectra to output quality information such as rice moisture and fatty acids. The data storage module stores all raw and corrected spectral data and rice quality predictions, and sends this data to a cloud server for data backup and remote access. A Raspberry Pi, serving as the master control unit, reads data from the temperature and humidity sensors connected to the STM32 microcontroller in real time through a communication interface and simultaneously acquires the raw spectral data collected by the spectrometer. The temperature and humidity compensation model is then invoked, taking real-time environmental parameters (temperature and humidity) and the raw spectral data as inputs, and outputs corrected spectral data to eliminate the effects of ambient temperature and humidity. The corrected spectral data is then input into the rice quality quantitative analysis model to output quality indicators such as rice moisture and fatty acids. All raw and corrected spectral data and rice quality predictions are stored in the data storage module and synchronized to a cloud server for data backup and remote access. The system is equipped with a touchscreen display that visualizes rice quality data and environmental parameters in real time, supports historical data query and abnormality alarms, and forms a closed loop from data acquisition, compensation processing, model analysis, to cloud storage.

7. The robot according to claim 1, characterized in that The automatic recharging system consists of an on-board charging device, an infrared receiving module, a task recovery module, and a fixed charging pile in the granary. The on-board charging device is integrated into the rear of the robot, and the built-in infrared receiving module communicates with the STM32 microcontroller through an interface to monitor the battery voltage and remaining power in real time. When the power level is lower than the preset threshold, the navigation task is triggered. Combining the SLAM algorithm with the positioning signal of the infrared transmitting module of the charging pile, the optimal path is planned to return to the charging pile at the intersection of the central channel of the granary. The infrared transmitting module in the charging pile transmits infrared signals, and the infrared receiving module in the on-board charging device combines the triangulation algorithm to achieve short-range positioning and posture fine-tuning, ensuring the precise docking of the charging contacts. During the charging process, the STM32 microcontroller continuously monitors the battery status. After charging is completed, it automatically wakes up the task recovery module, loads the unfinished task data (such as the inspection path and the spectrum acquisition points) from the local computer, and synchronously restores the robot's working state through the state machine, realizing the seamless connection of "charging-task" and ensuring the continuity and data integrity of the granary monitoring task.

8. The robot according to claim 1, wherein: The robot realizes real-time and non-destructive monitoring of the quality of stored rice through the coordinated work of the timing wake-up system, the navigation positioning and driving system, the spectrum acquisition and quality detection system and the automatic recharging system.

9. The robot according to claim 1, characterized in that Through the coordinated work of the above systems, the robot can efficiently and accurately complete the rice quality monitoring task and meet the real-time and reliability requirements of storage environment monitoring.

Images