An all-terrain crop phenotyping multi-source detection device

Abstract

This invention is applicable to the field of crop phenotyping technology and provides an all-terrain, multi-source crop phenotyping detection device, comprising a forward system, a multi-source sensor quick-connect module, and a crop phenotyping detection system. This device is suitable for automated crop phenotyping throughout the entire growth cycle in the field or greenhouse, freely switching between wheel-track modes for different ground environments, improving the overall efficiency of crop phenotyping. The sensors are easily replaceable and highly scalable. Based on active optical noise reduction and multi-source synchronous detection technology, they accurately analyze crop physiological information and morphological characteristics, building a multidimensional data system for crop phenotypic research, growth dynamics tracking, and environmental factor regulation.

Description

Technical Field

[0001] The present invention belongs to the technical field of crop phenotyping detection, and in particular relates to an all-terrain crop phenotyping multi-source detection device. Background Art

[0002] Crop phenotypes are measurable characteristics resulting from the interaction of genes and the environment. These include comprehensive indicators such as plant morphology (such as plant height, crown width, and leaf area), physiological characteristics, and spectral response. They are the core basis for understanding crop growth patterns. High-quality crop phenotyping and analysis technologies provide critical data support for variety improvement, gene function verification, environmental adaptation mechanism research, and smart agricultural management. Developing efficient and accurate crop phenotyping systems in conjunction with automated equipment is becoming a key technical direction driving the development of digital agriculture and smart breeding.

[0003] Existing crop phenotyping equipment mainly acquires phenotypic information by moving forward on wheels or transporting potted plants on belts. The wheels are prone to slipping when traveling in the field, which greatly compacts the soil. The belt-type equipment is only suitable for fixed-point detection of potted plants. The limited variety of sensors used for phenotyping leads to poor analysis results, and sensor switching relies on manual labor. The coordination between mechanical automation control and data detection modules is insufficient, making it difficult to meet the combined requirements of high-throughput crop phenomics for precision and efficiency. Summary of the Invention

[0004] The purpose of the embodiments of the present invention is to provide an all-terrain crop phenotyping multi-source detection device, aiming to solve the problems raised in the above background technology.

[0005] The embodiment of the present invention is implemented as follows: an all-terrain crop phenotyping multi-source detection device includes an advancing system, a multi-source sensor quick-connect module, and a crop phenotyping detection system;

[0006] The forward system includes a frame, on which a crawler-type mobile module and a mobile wheel module are provided. When the crawler-type mobile module is in use, the mobile wheel module is in a retracted state and suspended in the air; when the mobile wheel module is in use, the mobile wheel module is in an expanded state and lifts the crawler-type mobile module.

[0007] The multi-source sensor quick-connect module is used for automatic and quick replacement and docking of multi-source sensors;

[0008] The crop phenotype detection system is used to collect crop phenotype graphics and collect different information through the multi-source sensors in the multi-source sensor quick-connect module.

[0009] According to a further technical solution, the frame is of a trapezoidal structure, a square opening is opened at the front end of the frame, and a detachable end cover is connected thereto.

[0010] A further technical solution is that the crawler-type mobile module is symmetrically installed on both sides of the frame, specifically including two triangular crawlers symmetrically installed on both sides of the frame, and small tension wheels and large tension wheels are symmetrically distributed inside the triangular crawler, wherein three small tension wheels are provided, and two large tension wheels are provided. The small tension wheel is provided above the inner side of the triangular crawler, and the large tension wheel is provided below the upper side of the inner side of the triangular crawler. The triangular crawler forms an isosceles pentagonal structure through the small tension wheel and the large tension wheel; each of the small tension wheels is respectively installed on a second transmission shaft, and the second transmission shaft is connected to the output end of the brushless motor through a synchronous belt.

[0011] According to a further technical solution, the mobile wheel module includes a retractable outer wheel and a servo motor, wherein the retractable outer wheel includes an outer wheel hub, an X-shaped hinge structure and an inner wheel hub disk, wherein the outer sides of the two arms of the X-shaped hinge structure are hinged to the outer wheel hub, and the inner sides of the two arms of the X-shaped hinge structure are hinged to the inner wheel hub disk and the spiral sleeve respectively;

[0012] The output end of the servo motor is connected to the first transmission shaft through a double-stage reduction gearbox, the first transmission shaft is connected to a T-shaped lead screw, and an electromagnetic clutch is further provided between the first transmission shaft and the T-shaped lead screw, and the spiral sleeve is mounted on the T-shaped lead screw;

[0013] The spiral sleeve is also connected to the second transmission shaft through a sprocket chain mechanism.

[0014] According to a further technical solution, the arc-shaped outer surface of the outer hub is evenly covered with a styrene-butadiene rubber layer.

[0015] A further technical solution is that the multi-source sensor quick-connect module includes a quick-connect module rack installed on a rack, a fixed plate is fixedly installed on the quick-connect module rack, a slide is slidably installed on the quick-connect module rack, and an electric telescopic rod for driving the slide to slide is provided on the quick-connect module rack, the fixed plate and the slide are hinged with an L-shaped rod, the free ends of the two L-shaped rods are hinged to the lifting plate in the quick-connect module rack, two quick-connect brackets are placed on the lifting plate, and the top of the quick-connect bracket is an upwardly protruding arc structure, and each of the quick-connect brackets is installed with a multi-source sensor;

[0016] A double slide rail is also provided above the quick-connect module rack, and a movable slider is slidably installed on the double slide rail. A spring locking mechanism is provided at the bottom of the movable slider, and the spring locking mechanism includes two turning handles symmetrically installed in the movable slider, each of the turning handles is connected to the inner wall of the movable slider through a spring, and an electromagnet is correspondingly provided at the turning handle, and the lower end of the turning handle is a protruding arc structure.

[0017] A further technical solution is that two groups of crop phenotyping detection systems are symmetrically arranged on the frame, and the crop phenotyping detection system includes a stainless steel bracket installed on the frame, and a gantry-type synchronous belt module with vertical freedom is installed on the stainless steel bracket. The synchronous belt module is connected to a detection darkroom through a flange rod, and the two detection darkrooms are interconnected through a square aluminum tube, and the double slide rails are installed in the square aluminum tube. An adjustable light group is installed in the detection darkroom, and an annular guide rail is also installed inside the detection darkroom, and a connecting slider is installed on the annular guide rail, and an RGB-D camera is installed on the connecting slider.

[0018] In a further technical solution, the RGB-D camera is used to scan and obtain a point cloud with color information, and the original data is lightweighted by adopting a voxel downsampling method, and then the point cloud filter is used to remove noise. Assuming that the point cloud The coordinates of the points are , then point To any other point The distance is :

[0019] ;

[0020] The average distance between each point in the point cloud and any other point :

[0021] ;

[0022] The corresponding standard deviation :

[0023] ;

[0024] Based on the multiple of the determined standard deviation , calculate whether the average distance from each point to other points in the neighborhood is within the truncation range , outliers are determined and removed to achieve denoising, and the ICP (Iterative Closest Point) algorithm is used for point cloud registration. The extreme value iterative search is performed on the maximum and minimum differences between vertical spatial coordinates and horizontal amplitudes in the plant canopy point cloud. The three-dimensional mesh of the leaf point cloud is reconstructed and its surface area integral is calculated. Finally, the three-dimensional morphological parameters of the crop, such as plant height, crown width and leaf area, are obtained.

[0025] A further technical solution is that a black annular soft cloth is installed at the bottom of the detection darkroom through a fixed ring, and a crank slider mechanism is also provided at the top of the detection darkroom for driving the fixed ring to rise and fall. A stepper motor for driving the crank slider mechanism is installed on the top of the detection darkroom, and the fixed ring is connected to the moving end of the crank slider mechanism.

[0026] According to a further technical solution, the multi-source sensor includes a thermal imaging sensor and a hyperspectral camera, and the thermal imaging sensor and the hyperspectral camera are respectively installed in two quick-connect brackets;

[0027] Thermal imaging sensors are used to detect thermal images of leaves in the crop canopy and then combine them with visible light images for temperature calibration, image registration, and denoising. Subsequently, by calculating the average canopy temperature and the dynamic temperature change curve during the day, key physiological parameters of crop water are inverted, ultimately obtaining crop phenotypic parameters.

[0028] Hyperspectral cameras are used to acquire continuous narrow-band spectral images. By extracting spectral reflectance, first-order derivative spectra, and spectral vegetation indices, and with the help of machine learning algorithms, a model of the quantitative relationship between spectral features and crop physiological characteristic parameters is constructed to invert crop phenotypic information.

[0029] An embodiment of the present invention provides an all-terrain crop phenotyping multi-source detection device, which is suitable for automated detection of crop phenotyping throughout the entire growth cycle in the field or greenhouse. It can freely switch wheel modes for different ground environments, thereby improving the overall efficiency of crop phenotyping detection. The multi-source sensor quick-connect module equipped with the device automatically docks and locks with the crop phenotyping detection system, making the replacement of multi-source sensors convenient and highly scalable. The device is equipped with multi-source sensors that utilize dual-track exchange sliding and surround scanning to acquire crop phenotypic information in real time. Based on active light noise reduction and multi-source synchronous detection technology, it accurately analyzes the physiological information and morphological characteristics of crops, and constructs a multi-dimensional data system for crop phenotypic research, growth dynamics tracking, and environmental factor regulation. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 A schematic structural diagram of an all-terrain crop phenotyping multi-source detection device provided by an embodiment of the present invention;

[0031] Figure 2 A schematic structural diagram of a retractable outer wheel in an all-terrain crop phenotyping multi-source detection device provided by an embodiment of the present invention;

[0032] Figure 3 A schematic structural diagram of a small tension wheel transmission scheme in an all-terrain crop phenotyping multi-source detection device provided by an embodiment of the present invention;

[0033] Figure 4 A schematic structural diagram of a multi-source sensor quick-connect module in an all-terrain crop phenotyping multi-source detection device provided by an embodiment of the present invention;

[0034] Figure 5 A schematic structural diagram of a crop phenotyping system in an all-terrain crop phenotyping multi-source detection device provided by an embodiment of the present invention;

[0035] Figure 6 A schematic diagram of the internal structure of a detection chamber in an all-terrain crop phenotyping multi-source detection device provided by an embodiment of the present invention;

[0036] Figure 7 for Figure 4 A in the enlarged view.

[0037] In the figure: forward system 1; multi-source sensor quick-connect module 2; crop phenotyping detection system 3; triangular crawler 4; frame 5; detachable end cover 6; small tension pulley 7; large tension pulley 8; retractable outer wheel 9; servo motor 10; two-stage reduction gearbox 11; cylindrical helical gear set 12; first transmission shaft 13; electromagnetic clutch 14; T-type lead screw 15; first sprocket 16; spiral sleeve 17; X-shaped hinge structure 18; outer hub 19; inner hub disc 20; brushless motor 21; synchronous belt 22; second transmission shaft 23; second sprocket 24; quick-connect module Block frame 25; fixed plate 26; L-shaped rod 27; lifting plate 28; multi-source sensor 29; quick-connect tray 30; slide plate 31; electric telescopic rod 32; double slide rails 33; movable slider 34; spring locking mechanism 35; turning handle 36; electromagnet 37; spring 38; stainless steel bracket 39; square aluminum tube 40; detection darkroom 41; black annular soft cloth 42; crank slider mechanism 43; stepper motor 44; flange rod 45; synchronous belt module 46; connecting slider 47; RGB-D camera 48; annular guide rail 49; adjustable light group 50. DETAILED DESCRIPTION

[0038] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0039] The specific implementation of the present invention is described in detail below with reference to specific embodiments.

[0040] like Figure 1 As shown, an all-terrain crop phenotyping multi-source detection device provided by an embodiment of the present invention includes a forward system 1, a multi-source sensor quick-connect module 2 and a crop phenotyping detection system 3;

[0041] The forward system 1 includes a frame 5, on which a crawler-type mobile module and a mobile wheel module are provided. When the crawler-type mobile module is in use and moves, the mobile wheel module is in a retracted state and suspended in the air; when the mobile wheel module is in use and moves, the mobile wheel module is in an expanded state and lifts the crawler-type mobile module.

[0042] The multi-source sensor quick connection module 2 is used for automatic quick replacement and docking of the multi-source sensor 29;

[0043] The crop phenotype detection system 3 is used to collect crop phenotype graphics and collect different information through the multi-source sensor 29 in the multi-source sensor quick connection module 2.

[0044] In this embodiment of the present invention, the propulsion system 1 features a freely switchable dual-mode wheel and track mode. Track mode facilitates stable propulsion in field environments, such as those with rugged terrain and sticky soil, while wheel mode is suitable for rapid travel on flat surfaces. During testing, the crop to be tested is covered by the crop phenotyping system 3, and the multi-source sensor 29 then collects crop phenotypic information.

[0045] like Figure 1 As shown in FIG. 1 , as a preferred embodiment of the present invention, the frame 5 has an overall trapezoidal structure and is made of high-strength manganese steel. An electrical box housing the power supply, control system hardware, and other equipment is located above the frame 5. The front end of the frame 5 has a square opening and is bolted to a removable end cap 6, facilitating installation and maintenance of transmission components within the frame 5.

[0046] like Figure 1-Figure 3 As shown, as a preferred embodiment of the present invention, the crawler mobile module is symmetrically installed on both sides of the frame 5, specifically including two triangular crawlers 4 symmetrically installed on both sides of the frame 5, and small tension wheels 7 and large tension wheels 8 are symmetrically distributed inside the triangular crawler 4, wherein three small tension wheels 7 are provided, and two large tension wheels 8 are provided. The small tension wheel 7 is provided above the inner side of the triangular crawler 4, and the large tension wheel 8 is provided below the upper side of the inner side of the triangular crawler 4. The triangular crawler 4 forms an isosceles pentagonal structure through the small tension wheel 7 and the large tension wheel 8; each of the small tension wheels 7 is respectively installed on a second transmission shaft 23, and the second transmission shaft 23 is connected to the output end of the brushless motor 21 through a synchronous belt 22.

[0047] In an embodiment of the present invention, the small tensioning wheel 7 located at the same height is the driving wheel, and its transmission method is: the brushless motor 21 transmits the torque to the second transmission shaft 23 through the synchronous belt 22, and the second transmission shaft 23 drives the small tensioning wheel 7 to rotate, thereby driving the triangular crawler 4 forward. The use of independent drive makes the turning radius of the forward system 1 small and the flexibility high.

[0048] like Figure 1-Figure 3 As shown in FIG. 1 , as a preferred embodiment of the present invention, the mobile wheel module includes a retractable outer wheel 9 and a servo motor 10. The retractable outer wheel 9 includes an outer hub 19, an X-shaped hinge structure 18, and an inner hub disc 20. The outer sides of the two arms of the X-shaped hinge structure 18 are hinged to the outer hub 19, and the inner sides of the two arms of the X-shaped hinge structure 18 are hinged to the inner hub disc 20 and the spiral sleeve 17 respectively.

[0049] The output end of the servo motor 10 is connected to the first transmission shaft 13 through a double-stage reduction gearbox 11. The first transmission shaft 13 is connected to a T-shaped lead screw 15. An electromagnetic clutch 14 is also provided between the first transmission shaft 13 and the T-shaped lead screw 15. The spiral sleeve 17 is mounted on the T-shaped lead screw 15.

[0050] The spiral sleeve 17 is also connected to the second transmission shaft 23 through a sprocket chain mechanism. Specifically, a second sprocket 24 is installed on the second transmission shaft 23, and a first sprocket 16 is installed on the spiral sleeve 17. The second sprocket 24 is connected to the first sprocket 16 through a chain, so that the torque of the second transmission shaft 23 is transmitted to the spiral sleeve 17 through chain drive.

[0051] In an embodiment of the present invention, the transmission scheme of the T-screw 15 is as follows: the power of the servo motor 10 is torque-increased by the cylindrical bevel gear set 12 in the two-stage reduction gearbox 11 and then transmitted to the first transmission shaft 13. When the friction plate in the electromagnetic clutch 14 is closed, the first transmission shaft 13 drives the T-screw 15 to rotate, pushing the spiral sleeve 17 to move, and drives the outer hub 19 to move radially along the spiral sleeve 17 through the X-shaped hinge structure 18 to control the extension and retraction of the wheel diameter. When switching to the wheel drive mode, the friction plate in the electromagnetic clutch 14 is disconnected, and the spiral sleeve 17 is driven to rotate through the first sprocket 16, driving the retractable outer wheel 9 to rotate. At this time, the T-screw 15 remains stationary with the spiral sleeve 17 due to self-locking, and the self-locking mechanism ensures the stability of the wheel diameter after adjustment. The track and wheel drive modes share a power system, reducing energy consumption and control complexity.

[0052] As a preferred embodiment of the present invention, the arc-shaped outer surface of the outer hub 19 is evenly covered with a wear-resistant styrene-butadiene rubber layer;

[0053] like Figure 1 、 Figure 4 and Figure 7 As shown, as a preferred embodiment of the present invention, the multi-source sensor quick-connect module 2 includes a quick-connect module rack 25 installed on the rack 5, a fixed plate 26 is fixedly installed on the quick-connect module rack 25, a slide 31 is also slidably installed on the quick-connect module rack 25, and an electric telescopic rod 32 is provided on the quick-connect module rack 25 for driving the slide 31 to slide, and an L-shaped rod 27 is hinged on each of the fixed plate 26 and the slide 31, and the free ends of the two L-shaped rods 27 are hinged on the lifting plate 28 in the quick-connect module rack 25, and two quick-connect brackets 30 are placed on the lifting plate 28, and the top of the quick-connect bracket 30 is an upwardly protruding arc structure, and each of the quick-connect brackets 30 is installed with a multi-source sensor 29;

[0054] A double slide rail 33 is also provided above the quick-connect module rack 25, and a movable slider 34 is slidably installed on the double slide rail 33. A spring locking mechanism 35 is provided at the bottom of the movable slider 34. The spring locking mechanism 35 includes two rotating handles 36 symmetrically installed in the movable slider 34. Each rotating handle 36 is connected to the inner wall of the movable slider 34 through a spring 38, and an electromagnet 37 is correspondingly provided at each rotating handle 36. The lower end of the rotating handle 36 is a protruding arc structure.

[0055] In an embodiment of the present invention, the electric telescopic rod 32 pushes the slide plate 31 to slide outward, thereby driving the lifting plate 28 to move upward smoothly, and the quick-connect tray 30 docks with the spring lock mechanism 35 and interlocks. The arc structure at the lower end of the handle 36 allows the top arc structure of the quick-connect tray 30 to be inserted. At this time, the spring is in a compressed state, realizing the interlocking function. When the electromagnet 37 is energized, the handle 36 is attracted to release the quick-connect tray 30, unlocking the lock. The movable slider 34 slides on the parallel double slide rails 33, thereby driving the multi-source sensor 29 in the quick-connect tray 30 to slide above the crop for detection through the spring lock mechanism 35. Then, the sensors 29 slide alternately to complete the dual-area detection of the plant, realizing real-time synchronous high-throughput detection of crop information by the multi-source sensors 29.

[0056] like Figure 1 、 Figure 5 and Figure 6 As shown, as a preferred embodiment of the present invention, two groups of crop phenotyping detection systems 3 are symmetrically arranged on the frame 5. The crop phenotyping detection system 3 includes a stainless steel bracket 39 installed on the frame 5. A gantry-type synchronous belt module 46 with vertical freedom is installed on the stainless steel bracket 39. The synchronous belt module 46 is connected to a detection darkroom 41 through a flange rod 45. The detection darkroom 41 can be lowered to the ground to adapt to crop phenotyping detection at different growth intervals. The two detection darkrooms 41 are interconnected through a square aluminum tube 40, and the double slide rails 33 are installed in the square aluminum tube 40. An adjustable lamp group 50 is installed in the detection darkroom 41 to provide a stable light environment for phenotyping detection; an annular guide rail 49 is also installed inside the detection darkroom 41, and a connecting slider 47 is installed on the annular guide rail 49. The connecting slider 47 is installed on the connecting slider 47. An RGB-D camera 48 is installed.

[0057] In the embodiment of the present invention, a slide groove is provided at the bottom of the square aluminum tube 40 for the sliding of the movable slider 34 on the double slide rail 33. By adjusting the position of the connecting slider 47 on the annular guide rail 49, the shooting angle of the RGB-D camera 48 can be freely adjusted. The RGB-D camera 48 is used to scan and obtain a point cloud with color information. The raw data is lightweight by adopting the voxel downsampling method, which is convenient for later analysis and calculation. Then, the point cloud filtering is used to remove noise. Since the detection darkroom 41 of the device has the function of reducing light noise, the point cloud noise points that need to be filtered are mostly discrete points with edges. Assuming that the point cloud is The coordinates of the points are , then point To any other point The distance is :

[0058] ;

[0059] The average distance between each point in the point cloud and any other point :

[0060] ;

[0061] The corresponding standard deviation :

[0062] ;

[0063] Based on the multiple of the determined standard deviation , calculate whether the average distance from each point to other points in the neighborhood is within the truncation range Denoising is achieved by identifying and removing outliers, effectively removing outliers from the point cloud. The ICP (Iterative Closest Point) algorithm is used for point cloud registration. The extreme value iterative search is performed on the maximum and minimum differences between vertical spatial coordinates and horizontal amplitudes in the plant canopy point cloud. The three-dimensional mesh of the leaf point cloud is reconstructed and its surface area integral is calculated, ultimately obtaining three-dimensional morphological parameters such as crop height, crown width, and leaf area.

[0064] As a preferred embodiment of the present invention, the detection darkroom 41 is made of aluminum alloy, and its cylindrical shape makes the internal light reflection more uniform. At the same time, the inner layer is coated with a barium sulfate coating with high diffuse reflection to improve the quality of crop phenotypic imaging.

[0065] like Figure 5As shown, as a preferred embodiment of the present invention, a black annular soft cloth 42 is installed at the bottom of the detection darkroom 41 through a fixed ring, and a crank slider mechanism 43 is also provided on the top of the detection darkroom 41 for driving the fixed ring to rise and fall. A stepper motor 44 for driving the crank slider mechanism 43 is installed on the top of the detection darkroom 41, and the fixed ring is connected to the moving end of the crank slider mechanism 43.

[0066] In the embodiment of the present invention, when in use, the stepper motor 44 drives the crank slider mechanism 43 to move, thereby driving the fixed ring to rise and fall. The black ring-shaped soft cloth 42 can prevent the gap between the detection darkroom 41 and the uneven ground from leaking light, and the active light noise reduction improves the quality of phenotypic detection.

[0067] As a preferred embodiment of the present invention, the multi-source sensor 29 includes a thermal imaging sensor and a hyperspectral camera, and the thermal imaging sensor and the hyperspectral camera are respectively installed in two quick-connect brackets 30;

[0068] Thermal imaging sensors are used to detect the thermal images of each leaf in the crop canopy, and are combined with visible light images for temperature calibration, image registration, and denoising. Subsequently, by calculating the average canopy temperature and the dynamic change curve of daytime temperature, key physiological parameters such as crop water stress index and stomatal conductance are inverted, and ultimately phenotypic parameters such as crop water status, photosynthetic efficiency, and canopy temperature distribution uniformity are obtained.

[0069] Hyperspectral cameras are used to acquire continuous narrow-band spectral images. Each pixel contains a rich spectral information curve. By extracting features such as spectral reflectance, first-order derivative spectrum, and spectral vegetation index, and using machine learning algorithms such as support vector machines (SVM) and random forests (RF), a model of the quantitative relationship between spectral features and crop physiological parameters is constructed to invert key crop phenotypic information such as chlorophyll content, salinity and alkali stress, and biomass.

[0070] As a preferred embodiment of the present invention, it also includes a control system, which uses an STM32F407 microcontroller as the core processing unit and implements coordinated control of multi-threaded actuators through an integrated design. The main control unit establishes communication connections with each motor, multi-source sensor, trigger relay, etc. via the CAN (Controller Area Network) bus protocol, and simultaneously receives detection data from the multi-source sensors in real time. The system uses WiFi / 4G dual-mode wireless transmission technology to achieve two-way communication with remote terminals. The host computer uses a deep learning algorithm to integrate and analyze multi-source heterogeneous data and dynamically presents the processing results on a visual operation interface. At the same time, the design supports cross-platform interaction between mobile and PC terminals. Operators can adjust equipment parameters in real time through the cloud management platform and obtain multi-dimensional assessment reports on crop growth status, which is conducive to building a complete smart agricultural Internet of Things management and control system.

[0071] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. An all-terrain crop phenotyping multi-source detection device, characterized in that: Includes forward system, multi-source sensor quick-connect module and crop phenotyping detection system; The forward system includes a frame, on which a crawler-type mobile module and a mobile wheel module are provided. When the crawler-type mobile module is in use, the mobile wheel module is in a retracted state and suspended in the air; when the mobile wheel module is in use, the mobile wheel module is in an expanded state and lifts the crawler-type mobile module. The multi-source sensor quick-connect module is used for automatic and quick replacement and docking of multi-source sensors; The crop phenotyping detection system is used to collect crop phenotypic images and collect different information through the multi-source sensors in the multi-source sensor quick-connect module; The mobile wheel module includes a retractable outer wheel and a servo motor. The retractable outer wheel includes an outer wheel hub, an X-shaped hinge structure and an inner wheel hub disk. The outer sides of the two arms of the X-shaped hinge structure are hinged to the outer wheel hub, and the inner sides of the two arms of the X-shaped hinge structure are hinged to the inner wheel hub disk and the spiral sleeve respectively. The output end of the servo motor is connected to the first transmission shaft through a double-stage reduction gearbox, the first transmission shaft is connected to a T-shaped lead screw, and an electromagnetic clutch is further provided between the first transmission shaft and the T-shaped lead screw, and the spiral sleeve is mounted on the T-shaped lead screw; The spiral sleeve is also connected to the second transmission shaft via a sprocket chain mechanism; The multi-source sensor quick-connect module includes a quick-connect module rack installed on a rack, a fixed plate is fixedly installed on the quick-connect module rack, a slide is slidably installed on the quick-connect module rack, and an electric telescopic rod for driving the slide to slide is provided on the quick-connect module rack, an L-shaped rod is hinged on the fixed plate and the slide, and the free ends of the two L-shaped rods are hinged on the lifting plate in the quick-connect module rack, two quick-connect brackets are placed on the lifting plate, and the top of the quick-connect bracket is an upwardly protruding arc structure, and a multi-source sensor is installed in each of the quick-connect brackets; A double slide rail is further provided above the quick-connect module rack, a movable slider is slidably mounted on the double slide rail, a spring locking mechanism is provided at the bottom of the movable slider, and the spring locking mechanism includes two rotating handles symmetrically mounted in the movable slider, each of the rotating handles is connected to the inner wall of the movable slider through a spring, and an electromagnet is correspondingly provided at each rotating handle, and the lower end of the rotating handle is a protruding arc structure; Two crop phenotyping detection systems are symmetrically arranged on the frame. The crop phenotyping detection systems include a stainless steel bracket mounted on the frame, a gantry-type synchronous belt module with vertical freedom mounted on the stainless steel bracket, and a detection darkroom connected to the synchronous belt module via a flange rod. The two detection darkrooms are interconnected via a square aluminum tube, and the double slide rails are mounted in the square aluminum tube. An adjustable light group is installed in the detection darkroom, and an annular guide rail is also installed inside the detection darkroom. A connecting slider is mounted on the annular guide rail, and an RGB-D camera is mounted on the connecting slider.

2. The all-terrain crop phenotyping multi-source detection device according to claim 1, characterized in that: The frame is in a trapezoidal structure, a square opening is opened at the front end of the frame, and a detachable end cover is connected thereto.

3. The all-terrain crop phenotyping multi-source detection device according to claim 1, characterized in that: The crawler-type mobile module is symmetrically installed on both sides of the frame, and specifically includes two triangular crawlers symmetrically installed on both sides of the frame, and small tension wheels and large tension wheels are symmetrically distributed inside the triangular crawler, wherein three small tension wheels are provided, and two large tension wheels are provided. The small tension wheel is provided above the inner side of the triangular crawler, and the large tension wheel is provided below the upper side of the inner side of the triangular crawler. The triangular crawler forms an isosceles pentagonal structure through the small tension wheel and the large tension wheel; each of the small tension wheels is respectively installed on a second transmission shaft, and the second transmission shaft is connected to the output end of the brushless motor through a synchronous belt.

4. The all-terrain crop phenotyping multi-source detection device according to claim 1, characterized in that: The arc-shaped outer surface of the outer hub is evenly covered with a styrene-butadiene rubber layer.

5. The all-terrain crop phenotyping multi-source detection device according to claim 1, characterized in that: The RGB-D camera is used to scan and obtain a point cloud with color information. The original data is lightweighted by using a voxel downsampling method, and then the point cloud filter is used to remove noise. Assuming that the point cloud The coordinates of the points are Points To any other point The distance is : ； The average distance between each point in the point cloud and any other point : ； The corresponding standard deviation : ； Based on the multiple of the determined standard deviation , calculate whether the average distance from each point to other points in the neighborhood is within the truncation range , determine and remove outliers to achieve denoising, use the ICP algorithm for point cloud registration, and iteratively search for the maximum and minimum differences between vertical spatial coordinates and horizontal amplitudes in the plant canopy point cloud. Reconstruct the three-dimensional grid of the leaf point cloud and calculate its surface area integral, and finally obtain the three-dimensional morphological parameters of the crop, such as plant height, crown width and leaf area.

6. The all-terrain crop phenotyping multi-source detection device according to claim 1, characterized in that: A black annular soft cloth is installed at the bottom of the detection darkroom through a fixed ring, and a crank slider mechanism is also provided on the top of the detection darkroom for driving the fixed ring to rise and fall. A stepper motor for driving the crank slider mechanism is installed on the top of the detection darkroom, and the fixed ring is connected to the moving end of the crank slider mechanism.

7. The all-terrain crop phenotyping multi-source detection device according to claim 1, characterized in that: The multi-source sensor includes a thermal imaging sensor and a hyperspectral camera, and the thermal imaging sensor and the hyperspectral camera are respectively installed in two quick-connect brackets; Thermal imaging sensors are used to detect thermal images of leaves in the crop canopy and then combine them with visible light images for temperature calibration, image registration, and denoising. Subsequently, by calculating the average canopy temperature and the dynamic temperature change curve during the day, key physiological parameters of crop water are inverted, ultimately obtaining crop phenotypic parameters. Hyperspectral cameras are used to acquire continuous narrow-band spectral images. By extracting spectral reflectance, first-order derivative spectra, and spectral vegetation indices, and with the help of machine learning algorithms, a model of the quantitative relationship between spectral features and crop physiological characteristic parameters is constructed to invert crop phenotypic information.

Images