Edible fungus cultivation covering soil quality detection device based on visual identification

Abstract

The invention relates to the technical field of edible mushroom cultivation covering soil quality detection equipment, in particular to an edible mushroom cultivation covering soil quality detection device based on visual identification. According to the technical scheme, the edible mushroom cultivation covering soil quality detection device based on visual recognition comprises a detection device body, movable universal wheels, a terminal controller, a heat dissipation module, a pushing handle, a storage bin and a cleaning module, the movable universal wheels are arranged at the four corners of the bottom face of the detection device body, and the terminal controller is arranged on one side of the detection device body; a heat dissipation module is arranged on one side of the terminal controller; according to the invention, the degree-of-freedom mechanical arm and the binocular structured light scanning unit are introduced, and through spiral track scanning and phase deflection technologies, full-area high-precision three-dimensional shape acquisition is realized, and meanwhile, the multispectral imaging module synchronously obtains spectral characteristics of covering soil color, humidity and impurities through a pulse type composite light source and automatic white balance calibration, so that the detection precision of the covering soil is improved. And the detection efficiency and the impurity identification accuracy are effectively improved.

Description

Technical Field

[0001] The invention relates to the technical field of edible fungus cultivation covering soil quality detection equipment, in particular to an edible fungus cultivation covering soil quality detection device based on visual recognition. Background Art

[0002] The edible fungus cultivation cover soil quality detection device is an agricultural intelligent equipment that integrates multimodal sensing technology, intelligent algorithms and automatic control technology. It is specifically used for non-contact, high-precision detection and evaluation of the physical properties, chemical properties and impurity contamination of the cover soil layer during the edible fungus cultivation process.

[0003] Existing edible fungus cultivation soil quality detection equipment relies on manual sampling and detection. The single-bed detection cycle is as long as 20 minutes and can only cover a limited area. It is impossible to achieve fully automated coverage. In addition, the fixed equipment has a single viewing angle and has detection blind spots, making it difficult to obtain the complete three-dimensional morphology of the soil layer.

[0004] Existing equipment for detecting the quality of the covering soil for edible fungus cultivation relies on manual sampling and detection. The detection cycle for a single bed is as long as 20 minutes, and it can only cover a limited area, making it impossible to achieve fully automated coverage. In addition, the fixed equipment has a single viewing angle and has detection blind spots, making it difficult to obtain the complete three-dimensional morphology of the covering soil layer. This solution introduces a free-degree-of-freedom robotic arm and a binocular structured light scanning unit. Through spiral trajectory scanning and phase deflection technology, it can achieve high-precision three-dimensional morphology acquisition of the entire area. At the same time, the multispectral imaging module uses a pulsed composite light source and automatic white balance calibration to synchronously obtain the spectral characteristics of the covering soil color, humidity, and impurities, effectively improving detection efficiency and impurity identification accuracy. Summary of the Invention

[0005] In order to overcome the problems raised by the above background technology, this solution proposes a device for detecting the quality of edible fungus cultivation soil based on visual recognition.

[0006] The technical solution of the present invention is: a device for detecting the quality of soil covering for cultivating edible fungi based on visual recognition, comprising a detection device body, movable universal wheels, a terminal controller, a heat dissipation module, a pushing handle, a storage bin, a rotating adjustment base, an intelligent robotic arm, a visual recognition detector and a cleaning module. The four corners of the bottom surface of the detection device body are provided with movable universal wheels, one side of the detection device body is provided with a terminal controller, one side of the terminal controller is provided with a heat dissipation module, a pushing handle is provided at the top of one side of the detection device body, a storage bin is provided inside the detection device body, the top surface of the detection device body is provided with a rotating adjustment base, the top surface of the rotating adjustment base is provided with an intelligent robotic arm, one end of the intelligent robotic arm is provided with a visual recognition detector, and cleaning modules are provided at both upper and lower ends of the visual recognition detector.

[0007] Preferably, the universal wheels are moved to facilitate the multi-directional movement of the detection device body, the terminal controller is used to control the various components inside the detection device body, the heat dissipation operation is performed on the inside of the detection device body through the heat dissipation module, the pushing handle is used to facilitate the operator to push the detection device body, the storage bin is used to store and place the tools and materials required for the soil cover quality detection, the rotation and adjustment base is used to facilitate the intelligent robotic arm to rotate and adjust the position, the intelligent robotic arm is used to dynamically adjust the soil cover quality detection position of the visual recognition detector, the visual recognition detector is used to scan and analyze the soil cover quality data, and the visual recognition detector is cleaned regularly through the cleaning module.

[0008] Preferably, the edible fungus cultivation soil quality detection device based on visual recognition also includes the following modules:

[0009] 3D topography acquisition module: used to obtain 3D topography data of the overburden layer and provide thickness analysis and topography assessment data;

[0010] Multispectral imaging module: obtains spectral characteristic information of the soil cover layer’s color, moisture and impurities;

[0011] Self-cleaning maintenance module: cleans the optical system and maintains detection accuracy;

[0012] Dynamic calibration module: real-time monitoring and adjustment of the measurement accuracy and stability of the detection system;

[0013] Control and data processing module: performs data analysis and intelligent component control on multiple data.

[0014] Preferably, the three-dimensional shape acquisition module includes:

[0015] A1001: Free-range robotic arm unit, including a servo motor, reducer, encoder, harmonic reducer, and carbon fiber connecting rod, used for precise positioning and posture adjustment of the scanning head in space, covering the entire area of the cultivation bed;

[0016] A1002: Binocular structured light scanning unit, including projection lens, DLP projection chip, polarizer, CMOS sensor and heat sink fins, used to obtain soil surface point cloud data through sinusoidal fringe projection and phase deflection technology;

[0017] A1003: A micro air curtain protection unit, including a micro air pump, an air nozzle array, an air flow duct and a pressure sensor, is used to form a 0.3mm stable air curtain in front of the scanning head to isolate flying soil particles and water mist.

[0018] Preferably, the three-dimensional shape acquisition module comprises the following steps when operating:

[0019] S1001: The main control unit calculates the motion trajectory of the six-degree-of-freedom robotic arm using an inverse kinematics algorithm based on the size of the cultivation bed and the detection area;

[0020] S1002: The robotic arm starts from the initial position, 300 mm above the edge of the cultivation bed, and scans along a preset spiral trajectory with a track spacing of 50 mm, covering the entire soil covering area;

[0021] S1003: The binocular structured light scanning unit is activated, and the DLP projection chip projects a sinusoidal fringe pattern onto the soil surface through the projection lens. The frequency is 20 Hz, and the phase offset is π / 2.

[0022] S1004: The CMOS sensor synchronously collects the deformed fringe image, with the resolution set to 1280 × 1024 pixels and the exposure time automatically adjusted to 1 / 500 second to adapt to different lighting conditions;

[0023] S1005: The edge computing unit performs phase calculation on the collected fringe image and calculates the depth value of each pixel by using a phase deflection method;

[0024] S1006: The depth data is integrated with the robot arm's posture information to generate a three-dimensional point cloud of the soil surface, with a point spacing of 0.15 mm and a vertical resolution better than 0.08 mm.

[0025] S1007: During the scanning process, the micro air pump supplies air to the air nozzle array at a pressure of 0.3 MPa, forming a stable air curtain with an air flow speed of 15 m / s, which effectively isolates the flying soil particles;

[0026] S1008: The pressure sensor monitors the air curtain pressure in real time. When it detects that the pressure drops to 0.25 MPa, the pressurization program is automatically triggered.

[0027] Preferably, the multispectral imaging module includes:

[0028] A2001: Multi-spectral optical window unit, including sapphire glass window, electrochromic film layer, ITO conductive layer and sealing rubber ring, used to achieve switching between visible light and near-infrared bands through voltage regulation;

[0029] A2002: Chopper lighting unit, including a micro-stepping motor, a chopper disk, a halogen lamp bead, an LED array, and a heat sink, is used to generate a pulsed composite light source, suppress ambient light interference, and improve the signal-to-noise ratio;

[0030] A2003: Spectral spectrometer unit, including diffraction grating, linear array CCD, focusing lens group and filter wheel, is used to decompose reflected light into components of different wavelengths for multi-parameter synchronous detection.

[0031] Preferably, the multispectral imaging module comprises the following steps when operating:

[0032] S2001: The electrochromic film layer applies a 3V voltage through the ITO conductive layer to switch the optical window to the visible light band for 20ms;

[0033] S2002: The chopper lighting unit is started, the micro-stepping motor drives the chopper disk to rotate, and the halogen lamp beads and the LED array flash alternately at a frequency of 50 Hz, generating a pulsed composite light source;

[0034] S2003: The CMOS sensor is triggered by a pulse light source synchronization signal and collects light reflected from the soil surface with an exposure time of 1 / 1000 second;

[0035] S2004: After completing the visible light band acquisition, the electrochromic film layer switches the voltage to 5V, enters the near-infrared band, and repeats the above acquisition process;

[0036] S2005: The diffraction grating decomposes the reflected light into 256 wavelength channels, and the linear array CCD scans the spectrum after decomposition at a line frequency of 10kHz;

[0037] S2006: The spectral spectrometer performs automatic white balance calibration using a standard white plate, and the consistency of radiant brightness across different wavelengths is better than 98%.

[0038] S2007: The edge computing unit performs pixel-level registration of the visible light image and the near-infrared spectrum to generate a multispectral data cube with spatial resolution aligned with the three-dimensional point cloud.

[0039] Preferably, the self-cleaning maintenance module includes:

[0040] A3001: Eddy air knife cleaning unit, including a high-pressure blower, Laval nozzle, air duct straightener, and dust collection box, is used to generate spiral airflow to remove debris from the window surface;

[0041] A3002: Precision wiping unit, including a stepper motor, eccentric wheel mechanism, medical-grade silicone scraper and force sensor, used for physical wiping of stubborn stains, with pressure controlled within the range of 0.5-2N;

[0042] A3003: Ultrasonic atomization cleaning unit, including ultrasonic generator, cleaning liquid storage tank, micro pump and atomization nozzle, is used for in-situ deep cleaning using special disinfectant for edible fungi.

[0043] Preferably, the dynamic calibration module includes:

[0044] A4001: Composite calibration plate unit, including ceramic standard balls, high-precision checkerboard, temperature sensor and fixed bracket, used to provide a reference for spatial coordinate system and optical parameters;

[0045] A4002: Automatic calibration unit, including laser interferometer, six-axis force sensor and temperature and humidity compensation module, used to automatically perform parameter calibration and error compensation before detection;

[0046] A4003: Online monitoring unit, including vibration sensor, inclinometer and data logger, is used to monitor the working status of equipment in real time and trigger early warning.

[0047] Preferably, the control and data processing module includes:

[0048] A5001: Main control unit, including ARM Cortex-A53 processor, FPGA coprocessor, DDR4 memory and eMMC storage, used to execute motion control algorithms and detection process management;

[0049] A5002: Edge computing unit, including the NVIDIA Jetson AGX Xavier module, cooling module, and SSD hard drive, for real-time processing of point cloud data and spectral images and running deep learning models;

[0050] A5003: Communication unit, including 5G module, LoRaWAN module, industrial Ethernet interface and antenna, used to achieve device networking and remote data transmission.

[0051] Preferably, the control and data processing module comprises the following steps when operating:

[0052] S3001: After the main control unit is started, the free robot arm joint status, optical window sealing, and air curtain pressure sensor reading are checked in sequence;

[0053] S3002: The dynamic calibration module automatically performs composite calibration plate detection to generate an initial calibration matrix of the spatial coordinate system and optical parameters;

[0054] S3003: The main control unit sends a position command to the robot servo driver according to the detection task priority, and the trajectory interpolation period is 1ms;

[0055] S3004: During the movement of the robotic arm, the sensor monitors the end load in real time. When abnormal resistance (>5N) is detected, the emergency stop procedure is triggered;

[0056] S3005: The edge computing unit receives the 3D point cloud and multispectral data and runs the improved YOLOv7 algorithm for impurity detection, with the minimum identifiable particle diameter of 0.3 mm.

[0057] S3006: Predicting soil moisture using a random forest regression model, with input features including near-infrared reflectance and its first-order derivative;

[0058] S3007: The control module encapsulates the processing results into JSON format, including the thickness distribution map, humidity cloud map, and impurity location coordinates;

[0059] The S3008:5G module uploads data to the cloud platform at a rate of 100Mbps and sends irrigation instructions to the field controller via LoRaWAN.

[0060] Beneficial effects of the present invention:

[0061] 1. Compared with traditional edible fungus cultivation soil quality detection equipment, which relies on manual sampling and detection, has a single bed detection cycle of up to 20 minutes, and can only cover a limited area, it cannot achieve fully automated coverage. In addition, fixed equipment has a single viewing angle, has detection blind spots, and it is difficult to obtain the complete three-dimensional morphology of the soil layer. This solution introduces a free-degree-of-freedom robotic arm and a binocular structured light scanning unit. Through spiral trajectory scanning and phase deflection technology, it can achieve high-precision three-dimensional morphology acquisition of the entire area. At the same time, the multispectral imaging module uses a pulsed composite light source and automatic white balance calibration to simultaneously obtain the spectral characteristics of the soil color, humidity, and impurities, effectively improving detection efficiency and impurity identification accuracy.

[0062] 2. Compared with traditional edible fungus cultivation soil quality detection devices, humid environments can easily lead to contamination of optical components, affecting detection accuracy and equipment stability. This solution integrates eddy current air knife, precision wiping and ultrasonic atomization cleaning through a self-cleaning maintenance module to achieve three-level cleaning. At the same time, the micro air curtain protection unit forms a 0.3mm-level stable air curtain in front of the scanning head, isolating flying soil particles and water mist, effectively reducing equipment failure rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 Shown is a first three-dimensional structural schematic diagram of the edible fungus cultivation soil quality detection device based on visual recognition of the present invention;

[0064] Figure 2 Shown is a second three-dimensional structural schematic diagram of the edible fungus cultivation soil quality detection device based on visual recognition of the present invention;

[0065] Figure 3 Shown is a schematic side view of the stereoscopic structure of the edible fungus cultivation soil quality detection device based on visual recognition of the present invention;

[0066] Figure 4 Shown is a schematic diagram of the workflow of the three-dimensional morphology acquisition module of the edible fungus cultivation soil quality detection device based on visual recognition of the present invention;

[0067] Figure 5 Shown is a schematic diagram of the workflow of the multispectral imaging module of the edible fungus cultivation soil quality detection device based on visual recognition of the present invention;

[0068] Explanation of the accompanying symbols: 1. Detection device body; 2. Mobile universal wheel; 3. Terminal controller; 4. Heat dissipation module; 5. Push handle; 6. Storage compartment; 7. Rotating adjustment base; 8. Intelligent robotic arm; 9. Visual recognition detector; 10. Cleaning module. DETAILED DESCRIPTION

[0069] The present invention will be further described below with reference to the accompanying drawings and examples.

[0070] See also Figure 1-3 The present invention provides an embodiment: a device for detecting the quality of soil covering for cultivating edible fungi based on visual recognition, comprising a detection device body 1, movable universal wheels 2, a terminal controller 3, a heat dissipation module 4, a pushing handle 5, a storage bin 6, a rotating adjustment base 7, an intelligent robotic arm 8, a visual recognition detector 9 and a cleaning module 10. The four corners of the bottom surface of the detection device body 1 are all provided with movable universal wheels 2, one side of the detection device body 1 is provided with a terminal controller 3, one side of the terminal controller 3 is provided with a heat dissipation module 4, a pushing handle 5 is provided at the top of one side of the detection device body 1, a storage bin 6 is opened inside the detection device body 1, a rotating adjustment base 7 is provided on the top surface of the detection device body 1, an intelligent robotic arm 8 is provided on the top surface of the rotating adjustment base 7, a visual recognition detector 9 is provided at one end of the intelligent robotic arm 8, and cleaning modules 10 are provided at both the upper and lower ends of the visual recognition detector 9.

[0071] See also Figure 4-5 In this embodiment, the device for detecting the quality of edible fungus cultivation soil based on visual recognition also includes the following modules:

[0072] 3D topography acquisition module: used to obtain 3D topography data of the overburden layer and provide thickness analysis and topography assessment data;

[0073] Multispectral imaging module: obtains spectral characteristic information of the soil cover layer’s color, moisture and impurities;

[0074] Self-cleaning maintenance module: cleans the optical system and maintains detection accuracy;

[0075] Dynamic calibration module: real-time monitoring and adjustment of the measurement accuracy and stability of the detection system;

[0076] Control and data processing module: performs data analysis and intelligent component control on multiple data.

[0077] Preferably, the three-dimensional shape acquisition module includes:

[0078] A1001: Free-range robotic arm unit, including a servo motor, reducer, encoder, harmonic reducer, and carbon fiber connecting rod, used for precise positioning and posture adjustment of the scanning head in space, covering the entire area of the cultivation bed;

[0079] A1002: Binocular structured light scanning unit, including projection lens, DLP projection chip, polarizer, CMOS sensor and heat sink fins, used to obtain soil surface point cloud data through sinusoidal fringe projection and phase deflection technology;

[0080] A1003: A micro air curtain protection unit, including a micro air pump, an air nozzle array, an air flow duct and a pressure sensor, is used to form a 0.3mm stable air curtain in front of the scanning head to isolate flying soil particles and water mist.

[0081] Preferably, the three-dimensional shape acquisition module comprises the following steps when operating:

[0082] S1001: The main control unit calculates the motion trajectory of the six-degree-of-freedom robotic arm using an inverse kinematics algorithm based on the size of the cultivation bed and the detection area;

[0083] S1002: The robotic arm starts from the initial position, 300 mm above the edge of the cultivation bed, and scans along a preset spiral trajectory with a track spacing of 50 mm, covering the entire soil covering area;

[0084] S1003: The binocular structured light scanning unit is activated, and the DLP projection chip projects a sinusoidal fringe pattern onto the soil surface through the projection lens. The frequency is 20 Hz, and the phase offset is π / 2.

[0085] S1004: The CMOS sensor synchronously collects the deformed fringe image, with the resolution set to 1280 × 1024 pixels and the exposure time automatically adjusted to 1 / 500 second to adapt to different lighting conditions;

[0086] S1005: The edge computing unit performs phase calculation on the collected fringe image and calculates the depth value of each pixel by using a phase deflection method;

[0087] S1006: The depth data is integrated with the robot arm's posture information to generate a three-dimensional point cloud of the soil surface, with a point spacing of 0.15 mm and a vertical resolution better than 0.08 mm.

[0088] S1007: During the scanning process, the micro air pump supplies air to the air nozzle array at a pressure of 0.3 MPa, forming a stable air curtain with an air flow speed of 15 m / s, which effectively isolates the flying soil particles;

[0089] S1008: The pressure sensor monitors the air curtain pressure in real time. When it detects that the pressure drops to 0.25 MPa, the pressurization program is automatically triggered.

[0090] Preferably, the multispectral imaging module includes:

[0091] A2001: Multi-spectral optical window unit, including sapphire glass window, electrochromic film layer, ITO conductive layer and sealing rubber ring, used to achieve switching between visible light and near-infrared bands through voltage regulation;

[0092] A2002: Chopper lighting unit, including a micro-stepping motor, a chopper disk, a halogen lamp bead, an LED array, and a heat sink, is used to generate a pulsed composite light source, suppress ambient light interference, and improve the signal-to-noise ratio;

[0093] A2003: Spectral spectrometer unit, including diffraction grating, linear array CCD, focusing lens group and filter wheel, is used to decompose reflected light into components of different wavelengths for multi-parameter synchronous detection.

[0094] Preferably, the multispectral imaging module comprises the following steps when operating:

[0095] S2001: The electrochromic film layer applies a 3V voltage through the ITO conductive layer to switch the optical window to the visible light band for 20ms;

[0096] S2002: The chopper lighting unit is started, the micro-stepping motor drives the chopper disk to rotate, and the halogen lamp beads and the LED array flash alternately at a frequency of 50 Hz, generating a pulsed composite light source;

[0097] S2003: The CMOS sensor is triggered by a pulse light source synchronization signal and collects light reflected from the soil surface with an exposure time of 1 / 1000 second;

[0098] S2004: After completing the visible light band acquisition, the electrochromic film layer switches the voltage to 5V, enters the near-infrared band, and repeats the above acquisition process;

[0099] S2005: The diffraction grating decomposes the reflected light into 256 wavelength channels, and the linear array CCD scans the spectrum after decomposition at a line frequency of 10kHz;

[0100] S2006: The spectral spectrometer performs automatic white balance calibration using a standard white plate, and the consistency of radiant brightness across different wavelengths is better than 98%.

[0101] S2007: The edge computing unit performs pixel-level registration of the visible light image and the near-infrared spectrum to generate a multispectral data cube with spatial resolution aligned with the three-dimensional point cloud.

[0102] Preferably, the self-cleaning maintenance module includes:

[0103] A3001: Eddy air knife cleaning unit, including a high-pressure blower, Laval nozzle, air duct straightener, and dust collection box, is used to generate spiral airflow to remove debris from the window surface;

[0104] A3002: Precision wiping unit, including a stepper motor, eccentric wheel mechanism, medical-grade silicone scraper and force sensor, used for physical wiping of stubborn stains, with pressure controlled within the range of 0.5-2N;

[0105] A3003: Ultrasonic atomization cleaning unit, including ultrasonic generator, cleaning liquid storage tank, micro pump and atomization nozzle, is used for in-situ deep cleaning using special disinfectant for edible fungi.

[0106] Preferably, the dynamic calibration module includes:

[0107] A4001: Composite calibration plate unit, including ceramic standard balls, high-precision checkerboard, temperature sensor and fixed bracket, used to provide a reference for spatial coordinate system and optical parameters;

[0108] A4002: Automatic calibration unit, including laser interferometer, six-axis force sensor and temperature and humidity compensation module, used to automatically perform parameter calibration and error compensation before detection;

[0109] A4003: Online monitoring unit, including vibration sensor, inclinometer and data logger, is used to monitor the working status of equipment in real time and trigger early warning.

[0110] Preferably, the control and data processing module includes:

[0111] A5001: Main control unit, including ARM Cortex-A53 processor, FPGA coprocessor, DDR4 memory and eMMC storage, used to execute motion control algorithms and detection process management;

[0112] A5002: Edge computing unit, including the NVIDIA Jetson AGX Xavier module, cooling module, and SSD hard drive, for real-time processing of point cloud data and spectral images and running deep learning models;

[0113] A5003: Communication unit, including 5G module, LoRaWAN module, industrial Ethernet interface and antenna, used to achieve device networking and remote data transmission.

[0114] Preferably, the control and data processing module comprises the following steps when operating:

[0115] S3001: After the main control unit is started, the free robot arm joint status, optical window sealing, and air curtain pressure sensor reading are checked in sequence;

[0116] S3002: The dynamic calibration module automatically performs composite calibration plate detection to generate an initial calibration matrix of the spatial coordinate system and optical parameters;

[0117] S3003: The main control unit sends a position command to the robot servo driver according to the detection task priority, and the trajectory interpolation period is 1ms;

[0118] S3004: During the movement of the robotic arm, the sensor monitors the end load in real time. When abnormal resistance (>5N) is detected, the emergency stop procedure is triggered;

[0119] S3005: The edge computing unit receives the 3D point cloud and multispectral data and runs the improved YOLOv7 algorithm for impurity detection, with the minimum identifiable particle diameter of 0.3 mm.

[0120] S3006: Predicting soil moisture using a random forest regression model, with input features including near-infrared reflectance and its first-order derivative;

[0121] S3007: The control module encapsulates the processing results into JSON format, including the thickness distribution map, humidity cloud map, and impurity location coordinates;

[0122] The S3008:5G module uploads data to the cloud platform at a rate of 100Mbps and sends irrigation instructions to the field controller via LoRaWAN.

[0123] Example 1

[0124] Real-time scenario: An intelligent edible fungus production base has a cultivation workshop area of 2,000 square meters and is equipped with 12 Agaricus bisporus cultivation beds. The dimensions of each bed are 100m × 1.2m × 0.6m. The soil thickness is required to be 30±2mm, and the humidity must be maintained at 65%±3%RH.

[0125] Implementation method:

[0126] S4001: The main control unit generates a scanning trajectory based on the three-dimensional model of the cultivation bed, and the robotic arm is deployed to the initial position;

[0127] S4002: The dynamic calibration module performs a self-test, and the composite calibration plate unit A4001 verifies that the spatial coordinate system accuracy is 0.05 mm;

[0128] S4003: The binocular structured light scanning unit A1002 projects sinusoidal stripes, and the CMOS sensor synchronously captures images;

[0129] S4004: The micro air curtain protection unit A1003 forms an air curtain at a pressure of 0.3 MPa to isolate the flying soil particles;

[0130] S4005: Generate point cloud data of the covering soil surface, and output the thickness distribution map in a cycle of 5 minutes per bed;

[0131] S4006: The electrochromic film layer A2001 switches to the visible light band, and the chopping lighting unit A2002 generates a 50Hz pulse light source;

[0132] S4007: After completing the dual-band acquisition, the spectral spectrometry unit A2003 generates 256 wavelength channel data, and the humidity prediction model outputs the RH value;

[0133] S4008: The impurity identification algorithm detects a plastic particle with a diameter of 0.3 mm, and the coordinates are recorded as (X = 45.2 m, Y = 0.8 m);

[0134] S4009: The vortex air knife cleaning unit A3001 removes window mist with a 15m / s airflow, and a pressure sensor monitors the air curtain stability in real time;

[0135] S4010: Every 2 hours of testing, the automatic calibration unit A4002 compensates for the thermal deformation error of the robotic arm through the laser interferometer;

[0136] S4011: The control module encapsulates the thickness distribution map, humidity cloud map, and impurity location into JSON format and uploads it to the cloud platform through the 5G module.

[0137] Comparative data:

[0138] index Existing technology This program Improvement Single bed detection time 45 minutes 5 minutes -89％ Thickness measurement error ±1.5mm ±0.08mm -94.7％ Humidity detection error ±5%RH ±1.5%RH -70％ Equipment failure rate 12 times / year 2 times / year -83.3％ Maintenance cycle 7 days 200 days -96.5％ Impurity identification accuracy 85％ 98.7％ +16.1％

[0139] Conclusion: Through the integration of modular innovation and multimodal sensing technology, the intelligent level of edible fungus cultivation soil detection has been significantly improved, achieving a leap from "sampling detection" to "full-field monitoring", providing key technical support for factory-based edible fungus production.

[0140] The embodiments of the present invention are described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Various changes can be made within the scope of knowledge of those skilled in the art without departing from the spirit of the present invention.

Claims

1. A device for detecting the quality of soil covering for edible fungus cultivation based on visual recognition; the device is characterized by: The invention comprises a detection device body (1), a movable universal wheel (2), a terminal controller (3), a heat dissipation module (4), a pushing handle (5), a storage bin (6), a rotation adjustment base (7), an intelligent mechanical arm (8), a visual recognition detector (9) and a cleaning module (10). The movable universal wheels (2) are arranged at the four corners of the bottom surface of the detection device body (1), a terminal controller (3) is arranged on one side of the detection device body (1), a heat dissipation module (4) is arranged on one side of the terminal controller (3), a pushing handle (5) is arranged on the top of one side of the detection device body (1), a storage bin (6) is opened inside the detection device body (1), a rotation adjustment base (7) is arranged on the top surface of the detection device body (1), an intelligent mechanical arm (8) is arranged on the top surface of the rotation adjustment base (7), a visual recognition detector (9) is arranged at one end of the intelligent mechanical arm (8), and cleaning modules (10) are arranged at the upper and lower ends of the visual recognition detector (9).

2. The device for detecting the quality of edible fungus cultivation soil based on visual recognition according to claim 1, characterized in that: The edible fungus cultivation soil quality detection device based on visual recognition also includes the following modules: 3D topography acquisition module: used to obtain 3D topography data of the overburden layer and provide thickness analysis and topography assessment data; Multispectral imaging module: obtains spectral characteristic information of the soil cover layer’s color, moisture and impurities; Self-cleaning maintenance module: cleans the optical system and maintains detection accuracy; Dynamic calibration module: real-time monitoring and adjustment of the measurement accuracy and stability of the detection system; Control and data processing module: performs data analysis and intelligent component control on multiple data.

3. The device for detecting the quality of edible fungus cultivation soil based on visual recognition according to claim 2, characterized in that: The 3D shape acquisition module includes: A1001: Free-range robotic arm unit, including a servo motor, reducer, encoder, harmonic reducer, and carbon fiber connecting rod, used for precise positioning and posture adjustment of the scanning head in space, covering the entire area of the cultivation bed; A1002: Binocular structured light scanning unit, including projection lens, DLP projection chip, polarizer, CMOS sensor and heat sink fins, used to obtain soil surface point cloud data through sinusoidal fringe projection and phase deflection technology; A1003: A micro air curtain protection unit, including a micro air pump, an air nozzle array, an air flow duct and a pressure sensor, is used to form a 0.3mm stable air curtain in front of the scanning head to isolate flying soil particles and water mist.

4. The device for detecting the quality of edible fungus cultivation soil based on visual recognition according to claim 3, characterized in that: The 3D shape acquisition module includes the following steps when working: S1001: The main control unit calculates the motion trajectory of the six-degree-of-freedom robotic arm using an inverse kinematics algorithm based on the size of the cultivation bed and the detection area; S1002: The robotic arm starts from the initial position, 300 mm above the edge of the cultivation bed, and scans along a preset spiral trajectory with a track spacing of 50 mm, covering the entire soil covering area; S1003: The binocular structured light scanning unit is activated, and the DLP projection chip projects a sinusoidal fringe pattern onto the soil surface through the projection lens. The frequency is 20 Hz, and the phase offset is π / 2. S1004: The CMOS sensor synchronously collects the deformed fringe image, with the resolution set to 1280 × 1024 pixels and the exposure time automatically adjusted to 1 / 500 second to adapt to different lighting conditions; S1005: The edge computing unit performs phase calculation on the collected fringe image and calculates the depth value of each pixel by using a phase deflection method; S1006: The depth data is integrated with the robot arm's posture information to generate a three-dimensional point cloud of the soil surface, with a point spacing of 0.15 mm and a vertical resolution better than 0.08 mm. S1007: During the scanning process, the micro air pump supplies air to the air nozzle array at a pressure of 0.3 MPa, forming a stable air curtain with an air flow speed of 15 m / s, which effectively isolates the flying soil particles; S1008: The pressure sensor monitors the air curtain pressure in real time. When it detects that the pressure drops to 0.25 MPa, the pressurization program is automatically triggered.

5. The device for detecting the quality of edible fungus cultivation soil based on visual recognition according to claim 2, characterized in that: The multispectral imaging module includes: A2001: Multi-spectral optical window unit, including sapphire glass window, electrochromic film layer, ITO conductive layer and sealing rubber ring, used to achieve switching between visible light and near-infrared bands through voltage regulation; A2002: Chopper lighting unit, including a micro-stepping motor, a chopper disk, a halogen lamp bead, an LED array, and a heat sink, is used to generate a pulsed composite light source, suppress ambient light interference, and improve the signal-to-noise ratio; A2003: Spectral spectrometer unit, including diffraction grating, linear array CCD, focusing lens group and filter wheel, is used to decompose reflected light into components of different wavelengths for multi-parameter synchronous detection.

6. The device for detecting the quality of edible fungus cultivation soil based on visual recognition according to claim 5, characterized in that: When the multispectral imaging module is working, The following steps are involved: S2001: The electrochromic film layer applies a 3V voltage through the ITO conductive layer to switch the optical window to the visible light band for 20ms; S2002: The chopper lighting unit is started, the micro-stepping motor drives the chopper disk to rotate, and the halogen lamp beads and the LED array flash alternately at a frequency of 50 Hz, generating a pulsed composite light source; S2003: The CMOS sensor is triggered by a pulse light source synchronization signal and collects light reflected from the soil surface with an exposure time of 1 / 1000 second; S2004: After completing the visible light band acquisition, the electrochromic film layer switches the voltage to 5V, enters the near-infrared band, and repeats the above acquisition process; S2005: The diffraction grating decomposes the reflected light into 256 wavelength channels, and the linear array CCD scans the spectrum after decomposition at a line frequency of 10kHz; S2006: The spectral spectrometer performs automatic white balance calibration using a standard white plate, and the consistency of radiant brightness across different wavelengths is better than 98%. S2007: The edge computing unit performs pixel-level registration of the visible light image and the near-infrared spectrum to generate a multispectral data cube with spatial resolution aligned with the three-dimensional point cloud.

7. The device for detecting the quality of edible fungus cultivation soil based on visual recognition according to claim 2, characterized in that: The self-cleaning maintenance module includes: A3001: Eddy air knife cleaning unit, including a high-pressure blower, Laval nozzle, air duct straightener, and dust collection box, is used to generate spiral airflow to remove debris from the window surface; A3002: Precision wiping unit, including a stepper motor, eccentric wheel mechanism, medical-grade silicone scraper and force sensor, used for physical wiping of stubborn stains, with pressure controlled within the range of 0.5-2N; A3003: Ultrasonic atomization cleaning unit, including ultrasonic generator, cleaning liquid storage tank, micro pump and atomization nozzle, is used for in-situ deep cleaning using special disinfectant for edible fungi.

8. The device for detecting the quality of edible fungus cultivation soil based on visual recognition according to claim 2, characterized in that: The dynamic calibration module includes: A4001: Composite calibration plate unit, including ceramic standard balls, high-precision checkerboard, temperature sensor and fixed bracket, used to provide a reference for spatial coordinate system and optical parameters; A4002: Automatic calibration unit, including laser interferometer, six-axis force sensor and temperature and humidity compensation module, used to automatically perform parameter calibration and error compensation before detection; A4003: Online monitoring unit, including vibration sensor, inclinometer and data logger, is used to monitor the working status of equipment in real time and trigger early warning.

9. The device for detecting the quality of edible fungus cultivation soil based on visual recognition according to claim 2, characterized in that: The dynamic calibration module includes: A4001: Composite calibration plate unit, including ceramic standard balls, high-precision checkerboard, temperature sensor and fixed bracket, used to provide a reference for spatial coordinate system and optical parameters; A4002: Automatic calibration unit, including laser interferometer, six-axis force sensor and temperature and humidity compensation module, used to automatically perform parameter calibration and error compensation before detection; A4003: Online monitoring unit, including vibration sensor, inclinometer and data logger, is used to monitor the working status of equipment in real time and trigger early warning.

10. The device for detecting the quality of edible fungus cultivation soil based on visual recognition according to claim 9, characterized in that: The control and data processing module includes the following steps when working: S3001: After the main control unit is started, the free robot arm joint status, optical window sealing, and air curtain pressure sensor reading are checked in sequence; S3002: The dynamic calibration module automatically performs composite calibration plate detection to generate an initial calibration matrix of the spatial coordinate system and optical parameters; S3003: The main control unit sends a position command to the robot servo driver according to the detection task priority, and the trajectory interpolation period is 1ms; S3004: During the movement of the robotic arm, the sensor monitors the end load in real time. When abnormal resistance (>5N) is detected, the emergency stop procedure is triggered; S3005: The edge computing unit receives the 3D point cloud and multispectral data and runs the improved YOLOv7 algorithm for impurity detection, with the minimum identifiable particle diameter of 0.3 mm. S3006: Predicting soil moisture using a random forest regression model, with input features including near-infrared reflectance and its first-order derivative; S3007: The control module encapsulates the processing results into JSON format, including the thickness distribution map, humidity cloud map, and impurity location coordinates; The S3008:5G module uploads data to the cloud platform at a rate of 100Mbps and sends irrigation instructions to the field controller via LoRaWAN.

Images