A plant factory production operation inspection device

Abstract

This invention relates to an inspection device for plant factory production operations. The inspection device includes at least a mobile platform, a detection unit, and a remote control module. The mobile platform is configured to move within an area to be inspected. The detection unit is disposed on the mobile platform and configured to at least collect first data on the plants within the area to be inspected. The remote control module is configured to control the movement of the mobile platform. When the remote control module is able to acquire the first data, it is configured to at least predict the yield of the plants within the area to be inspected based on the acquired first data.

Description

Technical Field

[0001] This invention relates to the field of facility agriculture technology, and in particular to an inspection device for plant factory production operations. Background Technology

[0002] Diseases have always been a major factor restricting agricultural production. Vegetables, flowers, and other plants suffer from numerous diseases, causing diverse and complex symptoms. Disease occurrence not only leads to decreased yield and quality but also increases pesticide input and control costs, raising production costs and hindering the pollution-free and green production of vegetables and flowers. It also creates obstacles for export trade and poses potential environmental and health risks. In general, traditional greenhouse / plant factory disease inspection methods are either inefficient, inaccurate, or difficult to apply in practice in the short term. Therefore, establishing a rapid, accurate, non-destructive, and widely applicable plant disease inspection method is an urgent problem to be solved in my country's integrated plant disease management. Machine vision technology involves acquiring images through machine vision products, processing and analyzing the images using certain algorithms, and then determining the results and controlling the equipment. Computers can use machine vision technology to perform image processing and recognition to achieve the purpose of disease detection; therefore, machine vision technology can be widely used in vegetable disease detection. According to the growth patterns of vegetables, the metabolism of infected vegetables undergoes certain changes, leading to alterations in the pigment content, intercellular spaces, and water content within the leaf cells. This results in significant changes in the external morphology of the leaves, specifically the appearance of lesions. From an image perspective, these lesions exhibit color, shape, and texture characteristics. These characteristics are related to the type and severity of the disease affecting the vegetable, and this can be used to achieve disease detection based on machine vision.

[0003] For example, Chinese patent CN109154978A discloses a system and method for detecting plant diseases. This system involves extracting plant characteristics for identifying plant diseases, starting with color distribution to determine lesions, etc. However, in actual plant factory production, large-scale plant image scanning / capturing via detection units is costly, cumbersome, and time-consuming, especially when capturing color or full-color images for identification. The moving platform in the inspection device needs to constantly adjust its focus between multiple different locations to capture color or full-color images suitable for image recognition. Because color or full-color images have high resolution, they occupy a large amount of memory. The amount of data processed by the remote control module for subsequent image recognition is also enormous. Furthermore, the processing requires periodic or ad-hoc identification and analysis of plants at different locations at different times. This results in not only a massive amount of data to process for image recognition using color or full-color images, but also a slow processing speed for the remote control module. This limits the maximum speed of the plant monitoring device's mobile platform to the image processing speed of the remote control module, significantly impacting the area of ​​the plant factory / greenhouse that can be inspected per unit time. In other words, it reduces the efficiency of the plant monitoring device's inspections, or necessitates the simultaneous deployment of more plant monitoring devices to meet the daily inspection needs of the plant factory / greenhouse. Therefore, improvements are necessary to address the shortcomings of the existing technology.

[0004] Furthermore, on the one hand, there are differences in understanding among those skilled in the art; on the other hand, the applicant studied a large number of documents and patents when making this invention, but due to space limitations, not all details and contents were listed in detail. However, this does not mean that the present invention does not possess the features of these prior art. On the contrary, the present invention already possesses all the features of the prior art, and the applicant reserves the right to add relevant prior art to the background art. Summary of the Invention

[0005] In the actual production process of a plant factory, scanning / capturing plant images over a large area using detection units is costly, tedious, and time-consuming. In particular, capturing color or full-color images for identification using detection units requires the mobile platform in the inspection device to continuously adjust the focus between multiple different locations in order to capture color or full-color images that can be used for image recognition. Because color or full-color images have high resolution, they occupy a large amount of memory. The amount of data processed by the remote control module for subsequent image recognition of these color or full-color images is also enormous. Furthermore, the processing requires periodic or irregular identification and analysis of plants at different locations at different times. This results in not only a massive amount of data to be processed for image recognition using color or full-color images, but also a slow processing speed for the remote control module. This limits the maximum speed of the plant patrol monitoring device's mobile platform to the image processing speed of the remote control module, significantly impacting the area of ​​the plant factory / greenhouse that can be inspected per unit time. In other words, it reduces the efficiency of the plant patrol monitoring device's inspections, or necessitates the simultaneous deployment of more plant patrol monitoring devices to meet the daily inspection needs of the plant factory / greenhouse.

[0006] To address the shortcomings of existing technologies, this invention provides an inspection device for plant factory production operations. The inspection device includes at least a mobile platform, a first detection unit, and a remote control module.

[0007] The first shuttle was configured to move within the area to be detected.

[0008] A first detection unit is disposed on the first shuttle vehicle. Preferably, the first shuttle vehicle is capable of moving along a shuttle track detachably connected to the three-dimensional cultivation rack, so as to be able to perform preliminary identification of the growth status of the plants planted on each cultivation board in each cultivation layer through the first detection unit. Preferably, the first detection unit does not need to use a high-resolution camera to capture color or full-color images, but only a low-resolution camera to capture black and white images of the plants in the cultivation board. Preferably, the first detection unit establishes a data connection with the remote control module of the intelligent plant factory / system via the first communication unit of the shuttle vehicle itself. Preferably, the first shuttle vehicle is capable of performing preliminary identification of the growth status of the plants planted on each cultivation board in each cultivation layer periodically or irregularly along the shuttle track.

[0009] Particularly preferably, the first shuttle vehicle can directly perform preliminary analysis and recognition of the black and white image through the first detection unit installed on the first shuttle vehicle, without sending the black and white image to the remote control module or other modules for recognition and analysis.

[0010] Preferably, the second shuttle cart is disposed within a shuttle track detachably connected to the three-dimensional cultivation rack. Preferably, the second shuttle cart is capable of moving within the shuttle track detachably connected to the three-dimensional cultivation rack, so as to further identify the type and severity of plant diseases and / or pests. When the black-and-white image identifies an abnormal growth state of a plant growing on a certain cultivation board, the second detection unit disposed on the second shuttle cart starts and takes a picture of the plant on the cultivation board corresponding to the black-and-white image, so as to further identify the type and severity of plant diseases and / or pests. When the remote control module obtains the preliminary identification judgment result of the first detection unit as abnormal, the remote control module can send a control signal to the second shuttle cart of the plant cruise monitoring device to move to the cultivation board corresponding to the black-and-white image of the abnormal plant growth state.

[0011] Furthermore, since most plants undergo abnormal phenotypic changes during disease progression, such as leaf shrinkage and slow growth, the detection unit can avoid using color or full-color images / videos for preliminary disease and / or pest identification in plant factories / greenhouses (e.g., whether plant growth is normal, requiring further imaging, identification, and analysis only after abnormal growth is identified). While there are many mature existing image recognition technologies, using color or full-color images throughout the entire disease and / or pest identification process would require sophisticated hardware and software, inevitably leading to unnecessary production costs for plant factories / greenhouses. For example, a black and white image occupies 1600*900*2 bits of memory and is a single channel; a grayscale image occupies 1600*900*8 bits of memory; a color image occupies 1600*900*16 bits of memory; and a full-color image occupies 1600*900*24 bits of memory. Considering the limited memory capacity of the first detection unit carried by the first shuttle, its data processing capability for images is limited. However, the data processing / compression requirements for black and white images are significantly reduced compared to grayscale and color images, especially compared to color and full-color images, where the reduction is an order of magnitude. Furthermore, the first detection unit on the first shuttle does not require high-precision focusing when capturing black and white images, thus enabling faster inspection. Therefore, during the inspection of plant factories / greenhouses by plant patrol monitoring devices to photograph the planted plants, the plant's tissue morphology can be the first priority in judging the plant's growth status. That is, when the plant patrol monitoring device photographs the plants in the cultivation racks, it captures black and white images of the plants so that the remote control module can extract the approximate tissue shape / outline of the plant (e.g., stems / leaves) from the black and white images, and uses this as a basis for judging whether the plant has diseases and / or pests. When plants are suffering from diseases and / or pests, they will exhibit obvious changes such as leaf shrinkage and stem bending, and these characteristics can be extracted using black and white images without using color or full-color images. The technical problem this invention aims to solve is how to achieve preliminary identification of whether the plant's growth status is normal by capturing black and white images at a lower cost, while simultaneously increasing the monitoring speed of the plant patrol monitoring device.

[0012] Preferably, the factors affecting plant growth status are mainly divided into first priority, second priority and third priority.

[0013] Preferably, the first priority includes stem and / or leaf morphology / outline.

[0014] Preferably, the second priority includes lesion morphology, lesion location, and lesion area.

[0015] Preferably, the third priority includes the distribution of foreign matter. The distribution of foreign matter can be the location of pests / fungi, etc., on the plant.

[0016] Preferably, the first detection unit can send the coordinates / location information of the black and white image that indicates an abnormal growth state of the corresponding plant to the remote control module.

[0017] When the first detection unit identifies that the plant corresponding to the black-and-white image is in an abnormal growth state, the remote control module can control the second shuttle vehicle to use the second detection unit to capture grayscale / color / full-color images of the plants in the cultivation board corresponding to the abnormal black-and-white image, based on the coordinate / position information corresponding to the black-and-white image. Preferably, the second shuttle vehicle can send the grayscale / color / full-color images to the remote control module for processing and analysis. Preferably, the grayscale image can be used to identify / distinguish the morphology, location, and area of ​​lesions belonging to the second priority, and to capture color or full-color images of the lesion locations of some plants.

[0018] Preferably, color or full-color images are primarily used to identify / discover foreign objects belonging to the third priority category. Foreign objects may include the location of pests / fungi, etc., on the plant.

[0019] The first detection unit performs preliminary identification by recognizing the plant's tissue morphology or outline within the black-and-white image. For example, if the outline of the plant leaves, plant height, etc., in the black-and-white image are significantly different from the normal outline and plant height of that type of plant, and / or exhibit obvious changes such as leaf shrinkage or stem bending, the first detection unit determines the plant's growth status in the black-and-white image as abnormal; if no obvious changes in morphology / outline features are observed in the black-and-white image, the first detection unit determines the plant's growth status as normal. Preferably, the first detection unit can send the preliminary identification result to the remote control module of the plant factory / greenhouse.

[0020] When an abnormal growth status of a plant growing on a cultivation board is identified through a black-and-white image, the second detection unit on the plant patrol monitoring device is activated and takes a picture of the plant on the cultivation board corresponding to the black-and-white image.

[0021] Compared to directly acquiring color images of plants within the cultivation trays via the detection unit, this configuration offers several advantages. First, the large memory required for color images / videos leads to a massive amount of data processing for image recognition by the remote control module. This results in slow processing speed for disease and / or pest analysis based on image recognition by the detection unit, consequently limiting the maximum speed of the mobile platform on its dedicated track and making it unsuitable for the daily inspection and monitoring needs of large-scale plant factories. Second, using the first detection unit to capture black-and-white images of plants within the cultivation racks for preliminary disease and / or pest identification allows for direct identification of plant growth status without sending the images to the remote control module. This significantly improves the processing speed of preliminary identification. Furthermore, the increased processing speed of the remote control module also significantly increases the speed of the first shuttle vehicle. This, while ensuring the preliminary identification of normal plant growth status, increases the monitoring rate of plants within the plant factory and expands the monitored area of ​​plants growing on the cultivation trays per unit time.

[0022] At the same time, when the remote control module obtains the preliminary identification result of the first detection unit as abnormal, the remote control module can send a control signal to the second shuttle of the plant cruise monitoring device to move to the cultivation board corresponding to the black and white image of the abnormal plant growth state, so that the second detection unit set on the second shuttle can further take grayscale and / or color images of the plants on the cultivation board corresponding to the black and white image of the abnormality.

[0023] Particularly preferably, the second shuttle vehicle is capable of sending the grayscale image and / or color image to the remote control module of the plant factory / greenhouse, whereby the remote control module further identifies and analyzes the grayscale image and / or color image.

[0024] Preferably, the first shuttle vehicle can transmit the second location information—the location of the cultivation board corresponding to the black-and-white image or the location of the mobile platform indicating plant diseases and / or pests—to the remote control module via a wireless transmission module. Preferably, the remote control module can adjust the lighting scheme of the LED lights in the light source unit corresponding to the second location information based on the second location information.

[0025] Preferably, the second location information may include the number or location of the cultivation board where the abnormally growing plant is located. For example, the second location information may be: the first cultivation board located on the first layer of the first cultivation rack.

[0026] Preferably, the second location information may further include the two-dimensional or three-dimensional coordinates of the disease and / or pest on the plant.

[0027] For example, the first shuttle trolley travels normally along the shuttle track and takes black and white pictures through the first detection unit to preliminarily determine whether the growth status of the plants photographed by the first detection unit is normal. When the remote control module does not receive the second location information, the light source unit provides corresponding lighting to the plants in the plant factory / greenhouse in the first mode (i.e., the normal light supply scheme of the light source unit for the corresponding growth stage of the corresponding plant species). When the remote control module receives the second location information, the remote control module adjusts the lighting scheme of the LED lights in the light source unit corresponding to the second location information. For example, the brightness of the LED lights in the light source unit corresponding to the second location information is increased; when the first shuttle trolley leaves the cultivation board area corresponding to the second location information, the LED lights in the light source unit corresponding to the second location information are restored to the first mode.

[0028] For example, when the second detection unit is photographing the morphology of lesions on the leaves, the LED lights in the light source unit corresponding to the cultivation board of the plant leaf should be reduced to medium brightness. That is, the brightness should not be too high to avoid overexposure and blurring of the lesion morphology.

[0029] Particularly preferably, the plant patrol detection device is further equipped with a disease-removing robotic arm to grasp the diseased and / or pest-infested parts or the entire diseased plant corresponding to the second location information. This method can prevent the spread of the aforementioned diseases and / or pests to other areas, thus avoiding cross-infection between plants.

[0030] Particularly preferably, the third shuttle cart is also disposed within the shuttle track detachably connected to the three-dimensional cultivation rack. Preferably, the third shuttle cart is equipped with a third detection unit. Preferably, the third shuttle cart can move within the shuttle track detachably connected to the three-dimensional cultivation rack, so as to be able to detect at least the color temperature and / or color rendering index of the LED lamps of the light source unit through the third detection unit. Preferably, the third detection unit can also be used to detect whether the LED lamps of the light source unit are damaged. Preferably, when the third detection unit on the third shuttle cart detects a fault in the LED lamps of the light source unit, the third shuttle cart can send the coordinates / location information of the faulty LED lamps of the light source unit to the remote control module of the intelligent plant factory / greenhouse through a wireless communication module disposed inside the third shuttle cart, so that a fourth shuttle cart equipped with repair components can replace the faulty components (e.g., LED lamp components) of the LED lamps of the light source unit.

[0031] According to a preferred embodiment, the remote control module is able to identify the plant images / videos collected by the first detection unit based on image recognition technology / artificial intelligence technology in order to obtain the first data.

[0032] According to a preferred embodiment, the mobile platform includes a PC control module. The first detection unit can send the collected first data to the PC control module in real time. The PC control module can acquire the first data sent by the first detection unit in real time. The PC control module can send the first data to the remote control module in real time / non-real time.

[0033] According to a preferred embodiment, the mobile platform is equipped with a forward navigation detection unit and a local environmental factor sensor. The forward navigation detection unit is configured to determine the location of the mobile platform. The local environmental factor sensor is configured to monitor the local environmental factors around the area traversed by the mobile platform in real time.

[0034] According to a preferred embodiment, the first detection unit can be mounted on the side of the mobile platform away from the ground via a multi-degree-of-freedom robotic arm. The remote control module can control the multi-degree-of-freedom robotic arm to move so that the first detection unit, located at the end of the multi-degree-of-freedom robotic arm near the area to be detected, can face the plants within the area to be detected.

[0035] According to a preferred embodiment, the mobile platform further includes a tracking module. The tracking module is configured to control the mobile platform to automatically follow a track along pre-planned navigation markers on the ground within the detection area.

[0036] According to a preferred embodiment, the PC control module disposed inside the mobile platform can be electrically connected to the first detection unit, the multi-degree-of-freedom robotic arm, the front navigation detection unit, the local environmental factor sensor, and the global environmental factor sensor, respectively.

[0037] According to a preferred embodiment, the mobile platform detects obstacles encountered during its movement via a forward navigation detection unit. The PC control module can acquire images / videos collected by the forward navigation detection unit and identify the obstacles to achieve automatic detour.

[0038] According to a preferred embodiment, the remote control module is electrically connected to the PC control module and the global environmental factor sensor respectively, so as to remotely control the mobile platform to move in the area to be detected, and receive and analyze the feature extraction image, disease occurrence location and local environmental factors transmitted back by the PC control module, as well as the global environmental factors transmitted back by the global environmental factor sensor set in the plant factory.

[0039] According to a preferred embodiment, the system further includes a display module. The display module is capable of acquiring the location of the mobile platform as collected by the front navigation detection unit, and displaying the inspection trajectory of the inspection device in real time. The display module can also display one or more of the images, videos, and environmental data collected by the first detection unit, the local environmental factor sensor, and the global environmental factor sensor. Attached Figure Description

[0040] Figure 1 This is a simplified schematic diagram of the module connection relationship of a preferred embodiment provided by the present invention;

[0041] Figure 2 This is a simplified schematic diagram of the module connection relationship of a preferred embodiment of the display module provided by the present invention.

[0042] List of reference numerals

[0043] 1: Mobile platform; 2: First detection unit; 3: Remote control module;

[0044] 4: PC control module; 5: Front navigation detection unit; 6: Local environmental factor sensor;

[0045] 7: Tracking module; 8: Global environmental factor sensor; 9: Display module. Detailed Implementation

[0046] The following is a detailed explanation with reference to the accompanying drawings.

[0047] Figure 1 An inspection device for plant factory production operations is shown. The inspection device includes at least a mobile platform 1, a first detection unit 2, and a remote control module 3. The mobile platform 1 is configured to move within the area to be inspected.

[0048] The first detection unit 2 is disposed on the mobile platform 1. The first detection unit 2 is configured to be able to collect first data on plants in the area to be detected traversed by the mobile platform 1.

[0049] The remote control module 3 is capable of controlling the movement of the mobile platform 1. When the remote control module 3 is able to acquire the first data, it is configured to at least predict the yield of plants within the detection area based on the acquired first data.

[0050] Preferably, the mobile platform 1 can be a tracked robot. Preferably, the mobile platform 1 can also be a wheeled mobile device. Preferably, the mobile platform 1 can rotate and move 360 ​​degrees through the wheels at its bottom, so as to smoothly and automatically inspect, collect data at fixed points, turn automatically, return automatically, and recharge automatically within the area to be inspected. Preferably, the mobile platform 1 detects obstacles encountered during its movement through the front navigation detection unit 5, and the PC control module 4 can acquire the images / videos collected by the front navigation detection unit 5 and identify the obstacles to achieve automatic detour. Preferably, the power supply module inside the mobile platform 1 can provide power to the various components of the inspection device.

[0051] Preferably, the area to be tested can be an area where plants are grown inside a plant factory / greenhouse. Preferably, the area to be tested can also be other outdoor areas where plants are grown.

[0052] Preferably, the remote control module 3 can be a control module for a plant factory / greenhouse.

[0053] Preferably, the first detection unit 2 can be used to collect first data on plants within the area to be detected.

[0054] Preferably, the first data includes at least one or more of the following: the number / diameter / length / fullness of fruits / buds of the plant within the detection area. Preferably, the first data may also include the number / diameter / length of stems / leaves / roots of the plant within the detection area. Preferably, the first data may also include the number, diameter, color, gloss, and fullness of fruits / buds / stems / leaves / roots of the plant within the detection area.

[0055] Preferably, the first detection unit 2 can send the collected first data to the remote control module 3 in real time.

[0056] Preferably, the remote control module 3 can acquire the first data sent by the first detection unit 2 in real time. Preferably, the remote control module 3 can predict the yield of plants in the detection area based on the acquired first data. Preferably, the remote control module 3 can identify the plant images / videos collected by the first detection unit 2 based on image recognition technology / artificial intelligence technology to obtain the first data.

[0057] Preferably, the inspection device is capable of periodically inspecting the plants planted within the inspection area. For example, the inspection device periodically (e.g., daily) inspects the inspection area A. The remote control module 3 uses image recognition / artificial intelligence technology to identify the plant images / videos collected by the first detection unit 2, thereby identifying and statistically determining the number and diameter of fruits / buds of the plants (e.g., tomatoes, cucumbers, etc.) planted in the inspection area A. Based on historical data of the same period's fruit / bud production and the final harvest yield, the remote control module 3 can accurately predict / estimate the current batch's harvest yield. For instance, if the current batch's fruit / bud production increases by 30% compared to the previous batch with similar or identical growing conditions, the remote control module 3 predicts or estimates that the current batch's harvest yield of the same type of plants increases by 30% compared to the previous batch. Because the planting environment within the entire testing area (e.g., a greenhouse or plant factory) is precisely controlled and monitored by the plant factory's control module, the growth environment and historical data of each batch of plants can be traced and precisely adjusted. Therefore, the yield of plants grown in the greenhouse or plant factory under similar or identical growing conditions can be predicted relatively accurately. For example, the nutrient solution ratio and dosage, light formula, temperature and humidity, and carbon dioxide supply for each batch of plants are optimized through multiple rounds of selection, resulting in minimal or identical changes in the growth environment for each batch. Even if the growth plan / environmental factors of a certain type of plant are changed to optimize its yield, the remote control module 3 can record the adjusted / changed environmental factors. Then, the remote control module 3 can analyze and compare the historical yields of multiple batches corresponding to the adjusted / changed environmental factors, thereby summarizing the relationship between each environmental factor and the yield of the plant. This allows for further correction of the initial plant yield prediction made by the remote control module 3. For example, if the current batch of lettuce's growth program only optimizes one environmental factor (e.g., the carbon dioxide concentration in the current batch increases by 5% compared to the previous batch), and the actual yield of the current batch of lettuce increases by 10%, then if other environmental factors are optimized in the above lettuce growth program, such as lowering the temperature around the plant roots and / or stems and leaves during the seedling stage by two degrees compared to the optimal lettuce program, and the actual yield of the current batch of lettuce still only increases by 10%, then the remote control module 3 will not include the environmental factor of the temperature around the plant roots and / or stems and leaves during the seedling stage in the lettuce yield prediction model in the next lettuce yield prediction.With this configuration, the PC control module 4 or the remote control module 3 can analyze and compare the historical yields of multiple batches corresponding to the adjusted / changed environmental factors through multiple comparisons and analyses of historical yields, thereby summarizing the relationship between the changes in the yield of each environmental factor and the plant. This allows for further correction of the plant yield initially predicted by the PC control module 4 or the remote control module 3.

[0058] According to a preferred embodiment, the remote control module 3 can identify the plant images / videos collected by the first detection unit 2 based on image recognition technology / artificial intelligence technology to obtain the first data. Since those skilled in the art can easily find relevant image recognition technology / artificial intelligence technology for plant image / video identification in the prior art, the process of how the remote control module 3 performs plant image / video identification will not be described in detail here.

[0059] According to a preferred embodiment, the mobile platform 1 is internally equipped with a PC control module 4. The first detection unit 2 can send the collected first data to the PC control module 4 in real time. The PC control module 4 can acquire the first data sent by the first detection unit 2 in real time. The PC control module 4 can send the first data to the remote control module 3 in real time / non-real time.

[0060] According to a preferred embodiment, the mobile platform 1 is provided with a forward navigation detection unit 5 and a local environmental factor sensor 6. The forward navigation detection unit 5 is configured to determine the position of the mobile platform 1. The local environmental factor sensor 6 is configured to monitor the local environmental factors around the area traversed by the mobile platform 1 in real time.

[0061] Preferably, the front navigation detection unit 5 is disposed at the front end of the mobile platform 1 to determine the location of the mobile platform 1.

[0062] Preferably, the local environmental factor sensor 6 is mounted on the mobile platform 1 for real-time monitoring of local environmental factors. Preferably, the local environmental factor sensor 6 includes at least a carbon dioxide sensor, a temperature sensor, and a humidity sensor. Preferably, the local environmental factor sensor 6 may further include a photosynthetically active radiation sensor to monitor the photosynthetically active radiation that plants can receive.

[0063] Preferably, the PC control module 4 installed inside the mobile platform 1 is electrically connected to the multi-degree-of-freedom robotic arm, the first detection unit 2, the forward navigation detection unit 5, the local environmental factor sensor 6, and the global environmental factor sensor 8 installed in the greenhouse / plant factory.

[0064] According to a preferred embodiment, the first detection unit 2 can be mounted on the side of the mobile platform 1 away from the ground via a multi-degree-of-freedom robotic arm. The remote control module 3 can control the multi-degree-of-freedom robotic arm to move so that the first detection unit 2, located at the end of the multi-degree-of-freedom robotic arm near the area to be detected, can face the plants within the area to be detected.

[0065] Preferably, the PC control module 4 is used to control the movements of the multi-degree-of-freedom robotic arm. Preferably, the bottom of the multi-degree-of-freedom robotic arm is mounted on the side of the mobile platform 1 away from the ground. Preferably, the degrees of freedom of the multi-degree-of-freedom robotic arm can be flexibly selected according to the needs of the actual scenario. For example, the multi-degree-of-freedom robotic arm can be a three-degree-of-freedom or six-degree-of-freedom robotic arm.

[0066] Preferably, the first detection unit 2, which has a built-in wireless image transmission module, is located at the end of the multi-degree-of-freedom robotic arm near the area to be detected.

[0067] Preferably, the remote control module 3 can control the multi-degree-of-freedom robotic arm to move via the PC control module 4, so that the first detection unit 2 set at the end of the multi-degree-of-freedom robotic arm can face the plant to be detected and collect images / videos of the plant in the area to be detected.

[0068] According to a preferred embodiment, the mobile platform 1 is further provided with a tracking module 7. The tracking module 7 is configured to control the mobile platform 1 to automatically follow a track along navigation markers pre-planned on the ground within the detection area.

[0069] Preferably, the tracking module 7 installed inside the mobile platform 1 can control the mobile platform 1 to automatically track along pre-planned navigation markers on the ground within the detection area via the PC control module 4. Preferably, the navigation markers can be manually set according to actual needs.

[0070] According to a preferred embodiment, the PC control module 4 disposed inside the mobile platform 1 can be electrically connected to the first detection unit 2, the multi-degree-of-freedom robotic arm, the front navigation detection unit 5, the local environmental factor sensor 6, and the global environmental factor sensor 8, respectively.

[0071] Preferably, the global environmental factor sensor 8 installed in the greenhouse / plant factory includes a carbon dioxide sensor, a photosynthetically active radiation sensor, a temperature sensor, and a humidity sensor.

[0072] According to a preferred embodiment, the mobile platform 1 detects obstacles encountered during its journey by the front navigation detection unit 5, and the PC control module 4 can acquire the images / videos collected by the front navigation detection unit 5 and identify the obstacles to achieve automatic detour.

[0073] According to a preferred embodiment, the remote control module 3 can be electrically connected to the PC control module 4 and the global environmental factor sensor 8 respectively, so as to remotely control the mobile platform 1 to move in the area to be detected, and receive and analyze the feature extraction image, disease occurrence location and local environmental factors returned by the PC control module 4, as well as the global environmental factors returned by the global environmental factor sensor 8 set in the plant factory.

[0074] Preferably, the local and global environmental factors include carbon dioxide concentration, photosynthetically active radiation, temperature, and humidity.

[0075] Preferably, the carbon dioxide concentration, photosynthetically active radiation, temperature and humidity environmental factors at the location of the mobile platform 1 are monitored in real time by a local environmental factor sensor 6 installed on the mobile platform 1, and the data is transmitted to the PC control module 4 in real time.

[0076] According to a preferred embodiment, such as Figure 2 As shown, it also includes a display module 9. The display module 9 can acquire the position of the mobile platform 1 collected by the front navigation detection unit 5, so as to display the inspection trajectory of the inspection device in real time. The display module 9 can also display one or more of the images, videos and environmental data collected by the first detection unit 2, the local environmental factor sensor 6 and the global environmental factor sensor 8.

[0077] Preferably, the display module 9 can be a personal computer, tablet, or smartphone. With this configuration, users can use a computer, tablet, or smartphone to understand the growth status of plants in the area to be detected, environmental factors, and the plant yield predicted by the remote control module 3.

[0078] A method for conducting inspections using an inspection device in plant factory production operations includes the following steps:

[0079] a. The plant disease inspection robot enters the area to be inspected and uses a multi-degree-of-freedom robotic arm to orient the first inspection unit 2 toward the plant to be inspected, thus beginning the inspection.

[0080] b. Acquisition of plant images and local environmental factors

[0081] The mobile platform 1 is controlled by the remote control module 3 or the tracking module 7 to sequentially pass through all the plants to be detected in the area to be detected; during the movement of the mobile platform 1, the first detection unit 2 installed on the multi-degree-of-freedom robotic arm collects images of the plants to be detected in each area and transmits them to the PC control module 4 in real time.

[0082] c. Disease detection

[0083] The PC control module 4 performs feature extraction image processing on the plant image to be detected acquired by the first detection unit 2 to obtain the feature extraction image;

[0084] The specific steps of feature extraction image processing are as follows:

[0085] First, the leaves of the plant image to be detected are separated from the background. Then, the lesion image on the leaves is denoised using a median filtering algorithm to complete the image preprocessing.

[0086] The preprocessed image is then converted to the HIS color space using image processing techniques to obtain the H, S, and I component images. The H component... Figure 2 The H value is set to 0 in the concentrated area and 1 in the rest area to obtain the background part. Then, by multiplying the background part with the preprocessed image, the green leaf part can be removed and the leaf lesion image can be extracted. The leaf lesion image is the feature extraction image.

[0087] The feature-extracted image is compared with plant disease data images in the database to determine whether the feature-extracted image is a suspected disease image;

[0088] If the feature-extracted image is not a suspected disease image, then determine whether the first detection unit 2 has traversed all the plant areas to be detected. If it has traversed all the plant areas to be detected, then end the inspection; otherwise, return to step b to continue the inspection.

[0089] If the feature extraction image is a suspected disease image, the suspected disease area is marked in the feature extraction image. At the same time, the moving platform 1 pauses its movement, and the first detection unit 2 is moved to the plant area to be detected corresponding to the marked suspected disease area by the multi-degree-of-freedom robotic arm. The plant magnified image of the area is collected again, and the collected plant magnified image is sent back to the PC control module 4 for a second feature extraction image processing and comparison.

[0090] If the second comparison result determines that the feature-extracted image is not a suspected disease image, then it is determined whether the first detection unit 2 has traversed all the plant areas to be detected. If it has traversed all the plant areas to be detected, then the inspection ends; otherwise, return to step b to continue the inspection. If the second comparison result determines that the feature-extracted image is a suspected disease image, then proceed to the next step to determine the location of the disease.

[0091] d. Determine the location of the disease.

[0092] The front navigation detection unit 5 identifies the navigation markers pre-planned on the ground and records the location of the mobile platform 1, i.e. the location of the disease occurrence. The PC control module 4 receives the local environmental factors transmitted back by the local environmental factor sensor 6 at this time, and transmits them back to the remote control module 3 along with the location of the disease occurrence and the feature extraction image. Then, the mobile platform 1 starts to move and continues to inspect until the first detection unit 2 has traversed all the plant areas to be inspected.

[0093] e. Disease identification

[0094] The remote control module 3, combined with the global environmental factors transmitted by the global environmental factor sensor 8 installed in the greenhouse / plant factory and the local environmental factors transmitted by the PC control module 4, further analyzes the feature extraction images to identify the type of disease occurring at the location of the disease outbreak. This guides the prompt implementation of countermeasures to prevent large-scale disease outbreaks in the greenhouse / plant factory. The global and local environmental factors are compared with the temperature, humidity, and carbon dioxide concentration ranges of common plant diseases in the database to identify the specific disease occurring at the location. For example, gray mold is a low-temperature, high-humidity disease; the pathogen grows at temperatures between 20℃ and 30℃, and the disease is most prevalent when temperatures are between 20℃ and 25℃ and humidity remains above 90%. Diseased leaves exhibit significant lesion characteristics.

[0095] It should be noted that the specific embodiments described above are exemplary. Those skilled in the art can devise various solutions inspired by the disclosure of this invention, and these solutions all fall within the scope of this invention and its protection. Those skilled in the art should understand that this specification and its accompanying drawings are illustrative and not intended to limit the scope of the claims. The scope of protection of this invention is defined by the claims and their equivalents. This specification contains multiple inventive concepts; terms such as "preferredly," "according to a preferred embodiment," or "optionally" indicate that the corresponding paragraph discloses an independent concept. The applicant reserves the right to file divisional applications based on each inventive concept.

Claims

1. An inspection device for plant factory production operations, characterized in that, include: The first shuttle trolley is set up in the area to be detected and moves along the shuttle track to periodically or irregularly identify the growth status of the plants planted on each cultivation board in each cultivation layer. The second shuttle trolley is set in the shuttle track that is detachably connected to the three-dimensional cultivation rack in the area to be tested; A remote control module for controlling the movement of the first and second shuttle trolleys; The first detection unit, set in the first shuttle car, is configured to collect the first data of the plants in the detection area through which the first shuttle car passes by by taking black and white pictures of the plants in the cultivation board and analyzing and identifying them. When the remote control module obtains the preliminary identification judgment result of the first detection unit as abnormal, the first detection unit will send the coordinate / position information corresponding to the black and white picture of the plant whose growth status is abnormal to the remote control module. When the remote control module acquires the first data, the remote control module predicts the yield of plants in the area to be detected based on the acquired first data. The remote control module sends a control signal to the second shuttle to move to the cultivation board corresponding to the black and white image of the plant with abnormal growth status. The second shuttle further identifies the type and severity of diseases and / or pests of the plants on the cultivation board corresponding to the abnormal results initially identified by the first detection unit. The second shuttle uses the second detection unit to capture grayscale / color / full-color images of the plants in the cultivation board corresponding to the abnormality. Factors affecting plant growth status are categorized into first, second, and third priorities. First priority factors include stem and / or leaf morphology / outline; second priority factors include lesion morphology, lesion location, and lesion area; and third priority factors include the distribution of foreign matter. When the inspection device photographs the plants in the cultivation rack, the plant's tissue morphology is the first priority in judging the plant's growth status. By taking black and white pictures of the plants, the remote control module can extract the shape / outline of the plant's tissues from the black and white pictures, and use this as the basis for judging whether the plants have diseases and / or pests. The second shuttle sends grayscale / color / full-color images to the remote control module, which processes and analyzes them. The grayscale images are used to identify the morphology, location, and area of ​​lesions belonging to the second priority category. Color or full-color images are then captured for the lesion locations of some plants. The color or full-color images are used to identify the distribution of foreign matter belonging to the third priority category.

2. The inspection device according to claim 1, characterized in that, The first shuttle is equipped with a PC control module (4). The first detection unit (2) can send the collected first data to the PC control module (4) in real time. The PC control module (4) can obtain the first data sent by the first detection unit (2) in real time. The PC control module (4) is capable of sending the first data to the remote control module (3) in real time or non-real time.

3. The inspection device according to claim 2, characterized in that, The first shuttle car is equipped with a front navigation detection unit (5) and a local environmental factor sensor (6). The front navigation detection unit (5) is configured to determine the position of the first shuttle car, and the local environmental factor sensor (6) is configured to monitor the local environmental factors around the area traversed by the first shuttle car in real time.

4. The inspection device according to claim 3, characterized in that, The first shuttle is also equipped with a tracking module (7), which is configured to control the first shuttle to automatically follow a navigation mark pre-planned on the ground within the detection area.

5. The inspection device according to claim 4, characterized in that, The PC control module (4) located inside the first shuttle vehicle can be electrically connected to the first detection unit (2), the multi-degree-of-freedom robotic arm, the front navigation detection unit (5), the local environmental factor sensor (6), and the global environmental factor sensor (8), respectively.

6. The inspection device according to claim 5, characterized in that, The first shuttle car detects obstacles encountered during its journey through the front navigation detection unit (5). The PC control module (4) can acquire the images / videos collected by the front navigation detection unit (5) and identify the obstacles to achieve automatic detour.

7. The inspection device according to claim 6, characterized in that, It also includes a display module (9), which can acquire the position of the first shuttle car collected by the front navigation detection unit (5) to display the inspection trajectory of the inspection device in real time. The display module (9) can also display one or more of the images, videos and environmental data collected by the first detection unit (2), the local environmental factor sensor (6) and the global environmental factor sensor (8).

8. The inspection device according to claim 7, characterized in that, It also includes a third shuttle carriage, which can also be set in the shuttle track that can be detachably connected to the three-dimensional cultivation rack. The third shuttle carriage is equipped with a third detection unit and can move in the shuttle track so that it can at least detect the color temperature and / or color rendering index of the LED lamps of the light source unit through the third detection unit.

Images