Ridge-crossing type variable-spraying-width strawberry spraying robot structure and operation control method thereof

Abstract

The invention discloses a ridge-crossing type variable-spraying-width strawberry spraying robot structure and an operation control method thereof, and aims to meet the requirement for precise operation in a complex environment of a strawberry garden, improve the operation efficiency and reduce the labor cost. The chassis adopts a hub motor driving wheel set and a universal wheel set and is combined with a height adjusting device, so that flexible running and terrain adaptive capacity between ridges of the strawberry garden are realized; the spraying system supports dynamic adjustment of the spraying width and the spraying amount through the design of a carbon fiber folding rod and a modular nozzle, and the variable spraying requirement is met. The robot utilizes a depth camera to collect strawberry plant and inter-ridge environment information, accurately recognizes plant positions, growth states and pest and disease damage conditions through a Fast R-CNN image recognition model, and realizes autonomous navigation and path planning in combination with a GNSS navigation module and an MPC model prediction control algorithm. The composite sensor group monitors the running state and the working environment of the robot in real time, and provides data support for spray volume adjustment and path optimization.

Description

Technical Field

[0001] The present invention belongs to the technical field of agricultural robots, and specifically relates to a cross-ridge intelligent spray robot structure design for strawberry garden operations, which is particularly suitable for precise operation scenarios such as spraying and irrigation in strawberry gardens. Background Art

[0002] Strawberry is a high-value-added economic crop. In the greenhouse cultivation stage, pest control and precision spraying have a significant impact on its yield and quality. At present, the spraying operation in the strawberry garden is mainly completed manually. This manual operation method is inefficient and difficult to adapt to large-scale planting needs. Moreover, working in a narrow, high-temperature and closed greenhouse environment requires great labor intensity, and long-term exposure to pesticides will threaten physical health. In addition, it is difficult to achieve precise control of manual application of pesticides, and it is very easy to have heavy spraying, missed spraying, uneven spraying, etc., which leads to waste of pesticides, increases production costs, and may also cause environmental pollution.

[0003] Existing agricultural robots have many limitations when used in strawberry gardens. The positioning accuracy of the navigation systems of some robots decreases in complex strawberry garden environments, such as changes in light and occlusion of strawberry plants. They are unable to accurately plan the operation path, which greatly reduces the operation effect. Its spray system is also unable to accurately adjust according to factors such as the growth status of strawberry plants, the degree of pests and diseases, and soil fertility, making it difficult to meet the differentiated needs of strawberries in different regions and at different growth stages. Therefore, it is very necessary to develop a robot that can adapt to the special environment of strawberry gardens and achieve efficient and accurate automatic navigation operations. Summary of the invention

[0004] Based on the current situation and needs of the above-mentioned strawberry garden operation, the present invention aims to provide a cross-ridge type strawberry intelligent operation robot structure. The robot structure of the present invention includes a cross-ridge type variable height mobile chassis (1), a composite sensor group (3), a battery, an image sensor module (5), a control box system (6), a variable spray width spray system (2), and a support platform (7). The cross-ridge type variable height mobile chassis (1) is composed of a hub motor drive wheel group (101), a universal wheel group (102), a chassis bracket (103) and a height adjustment device (104). It has the characteristics of high strength and light weight, and can flexibly travel between narrow ridges in the strawberry garden. The variable spray width spray system (2) adopts a carbon fiber folding rod design, combined with multiple folding joints and lightweight materials, which is easy to store and can achieve a wide range of coverage.

[0005] The present invention aims to design a cross-ridge type strawberry intelligent operation robot structure and a control method thereof, so that the robot has autonomous navigation capability in the complex environment of a strawberry garden, can realize precise variable spray operation according to the real-time growth status and operation requirements of the strawberry plants, improve operation efficiency, reduce labor costs, reduce pesticide waste, ensure the safety of operators, and promote the intelligent development of the strawberry planting industry.

[0006] The technical solution of the present invention is:

[0007] A cross-ridge type strawberry variable spray width spray robot structure comprises a cross-ridge type variable height mobile chassis (1), and a variable spray width spray system (2), a composite sensor group (3), a battery group (4), an image sensor module (5), a control box system (6), and a support platform (7) installed thereon; the cross-ridge type variable height mobile chassis (1) is used to realize stable driving and position adjustment of the robot between ridges in a strawberry garden; the variable spray width spray system (2) is installed on the support platform (7) and realizes variable adjustment of the spray width through a carbon fiber folding rod; the image sensor module (5), mainly a depth camera, is used to collect image information of the environment between ridges in a strawberry garden; the composite sensor group (3) is used to monitor the robot's operating state and operating environment; the control box system (6) is used to control the robot's movement and spraying; the battery group (4) provides power support for the robot; and the support platform (7) is used to carry the above components and connect to the chassis support (103).

[0008] Furthermore, the cross-ridge type variable height mobile chassis (1) is composed of a wheel hub motor driven wheel group (101), a universal wheel group (102), a chassis bracket (103) and a height adjustment device (104); a bracket platform (7) is provided on the chassis bracket (103); the height adjustment device (104) is connected to the chassis bracket (103); one end of the chassis bracket (103) is movably connected to the left front and right front wheel hub motor driven wheel groups (101), and the other end is movably connected to the left rear and right rear universal wheel groups (102).

[0009] Furthermore, the carbon fiber rod (201) of the variable spray width spray system (2) is fixedly connected by a hinge structure composed of a multi-section folding joint 1 (207) and a folding joint 2 (208), thereby realizing three-section unfolding and folding. The whole carbon fiber rod (201) is rigidly connected to the support platform (7) through a connecting bracket (204), thereby forming a complete mechanical system with the robot main structure. The three-way pipe fitting (209) with a fixing ring is fixed on the carbon fiber rod (201) to form a modular assembly structure. The nozzle (205) adopts an atomizing nozzle, and the nozzle (205) is connected to the adjustable spacing water pump (203) and the water tank (202) through a PVC transparent hose (206). The adjustable spacing water pump (203) is a variable frequency centrifugal pump, and the flow rate is adjusted in real time. The spray flow rate and spray amount are adjusted in real time according to the travel speed of the robot.

[0010] Further, the composite sensor group (3) mainly includes ultrasonic sensors, liquid level monitoring sensors, spray flow sensors, motor encoders, and GNSS navigation modules. The ultrasonic sensors are installed at the chassis brackets (103), with two installed on each side of the left and right drive wheels, and the height can be adjusted to adapt to different ridge height ranges. Additionally, one ultrasonic sensor is vertically installed below the bracket platform (7) to detect the relative height between the platform and the ridge horizontal plane and the distance between the vehicle tires and the ridge obstacles, assisting in path planning and obstacle avoidance. The liquid level monitoring sensor is installed inside the water tank (202) to detect the liquid level of the liquid medicine and feedback it to the control system. The spray flow sensor is installed at the liquid outlet of the nozzle (205) to detect the spraying flow and feedback it to the control system. The motor encoder is embedded in the hub motor to detect the rotational speed and feedback it to the control system to achieve speed closed-loop control. The GNSS navigation module is installed on the bracket platform (7) to be responsible for obtaining the instant GPS positioning signal and obtaining the absolute coordinate information of the robot during operation.

[0011] Further, the battery pack (4) includes lithium batteries, which are fixed in the bracket structure of the cross-ridge variable-height mobile chassis (1), equipped with an intelligent battery management system to monitor the power, voltage, and current parameters in real time, achieve charge and discharge protection and optimization management, extend the battery life, and stably supply power to the variable-height mobile chassis (1), the image sensing module (5), and the control box system (6).

[0012] Further, the image sensing module (5) includes a depth camera, which can obtain image information under different spectra between the ridges in the strawberry field for inter-ridge positioning, tracking control, and autonomous navigation.

[0013] Further, the control box system (6) includes a Jetson Nano B01 central processing module, an STM32F407VGT6 single-chip microcomputer bottom-layer MCU main control board, and a TTL-RS485 signal conversion module inside.

[0014] Further, the bracket platform (7) is formed by connecting aluminum profiles and aluminum plates through metal angle pieces, and corresponding installation holes are set according to the design requirements to tightly fix the variable spray width spray system (2), the battery pack (4), the image sensing module (5), and the control box system (6) on the bracket platform (7). While providing support and load-bearing for the entire system, the bracket platform (7) minimizes the platform weight and is connected to the variable-height mobile chassis (1) through metal angle pieces, playing a role of connecting the upper and lower parts.

[0015] An operation control method for a cross-ridge strawberry variable spray pattern spraying robot structure of the present invention realizes the autonomous operation of the robot between the ridges in the strawberry field by using a navigation scheme combining a neural network visual recognition model and an MPC model predictive control, and performs variable spray control by combining the robot motion model and image recognition technology; the specific steps are as follows:

[0016] Step 1: Image recognition and analysis: The image sensing module (5) uses a depth camera combined with an optical filter to collect image data of strawberry plants and the environment between the ridges. Through the depth camera installed on the support platform (7), multi-spectral image information of the strawberry field between the ridges is obtained. The image data is processed in real time by the Jetson Nano B01 central processing module, and image recognition is realized by combining the Fast R-CNN network to identify the position, size, growth state and pest and disease conditions of the strawberry plants;

[0017] The image recognition model based on the Fast R-CNN network is constructed as:

[0018] Dataset construction: Use a depth camera to collect the characteristics of strawberry plants and the environment between the ridges, obtain RGB images and use the annotation tool LabelImg to annotate the images of the strawberry plant areas, and respectively annotate the positions of the strawberry plants and the categories of the strawberries (healthy plants and pest and disease plants); enhance the annotated data, use geometric transformation, brightness adjustment and noise addition of the images to obtain the enhanced dataset for training to improve the robustness of the model, and divide the constructed dataset into a training set, a validation set and a test set according to the ratio of 7:2:1;

[0019] Select the backbone network based on ResNet as the feature extraction part of Fast R-CNN. ResNet has strong feature extraction ability and can effectively capture the multi-scale features of strawberry plants. Use the constructed dataset to train the Fast R-CNN model, set appropriate hyperparameters, and use the pre-trained model for fine-tuning to accelerate convergence;

[0020] Divide the width of the camera image into 6 regions, and sequentially set the 1st, 2nd, 3rd, 4th, 5th, and 6th regions from left to right; the middle 1 region (the 3rd region) is narrower and controls 2 nozzles (205), and the other 5 regions (the 1st, 2nd, 4th, 5th, and 6th regions) are wider, each controlling 4 nozzles (205). According to the strawberry plants detected by Fast R-CNN, frame and return their center points, and according to the formula:

[0021]

[0022] Calculate the total number of plants N in the kth interval region,k ; N krepresents the total number of the central points of the strawberry plant coordinates in the k-th region; x i represents the abscissa of the image;

[0023] where the δ(x) function is defined as: the center coordinates (x c , y c ) recognized by the FastR-CNN network fall within the k-th region, then the value is 1, otherwise 0; x min,k and x max,k represent the coordinate lower limit and upper limit of the k-th region, and the calculation formula is:

[0024]

[0025] where W is the field of view width of the camera, then the width of the middle region is expressed as:

[0026]

[0027] The widths of the remaining 5 regions are respectively:

[0028]

[0029] According to the division, the physical area A region,k of each region is:

[0030]

[0031] where H is the field of view height of the camera;

[0032] Calculate the plant density of each divided region as:

[0033]

[0034] And normalize to get the spray importance weight W k :

[0035]

[0036] where S region,k is the average severity of the pest and disease plants in this region, that is, the proportion of the pest and disease plants recognized by the network in the total number of recognized plants; D max represents the maximum plant density in the six divided regions; α is the weight factor of the pest and disease severity, and this weight coefficient is an important indicator for the subsequent preset spray model decision-making;

[0037] Convert the trained model into the ONNX format supported by the Jetson Nano platform, and use TensorRT to optimize the model to improve the inference speed, achieve real-time inference on the Jetson Nano, combine the image data collected by the depth camera, identify the positions of strawberry plants, and mark healthy plants and plants with pests and diseases, providing real-time decision-making support for the intelligent management of strawberry gardens;

[0038] Step 2: Spray volume calculation: Based on the image recognition results, the central processing module combines the robot's traveling speed, the growth status of strawberry plants, the distribution of pests and diseases, and a preset spray model to calculate the spray volume of each nozzle (205); in the composite sensor group (3), the ultrasonic sensor detects the width of the ridge and the plant spacing, the liquid level monitoring sensor monitors the liquid medicine height in the water tank (202), and the spray flow sensor monitors the spray volume of the nozzle (205), providing data support; for areas with strong growth and severe pests and diseases, the central processing module instructs to increase the spray volume; for sparse or healthy areas, reduce the spray volume;

[0039] The preset spray model is:

[0040]

[0041] Among them, Q k represents the flow rate that each nozzle needs to provide in the kth area; β is the proportional amplification adjustment gain; W k is the calculated normalized spray importance weight; Q max represents the maximum spray flow rate of each nozzle; n k represents the total number of nozzles covering the kth area;

[0042] Step 3: Spray volume control: The central processing module passes the PWM control signal to the motor drive module of the water pump (203) through the STM32F407VGT6 single-chip microcomputer bottom-layer MCU main control board to adjust the output flow rate and pressure. The liquid medicine is transported to the nozzle (205) through the PVC transparent hose (206) to achieve variable spray operation; the spray rod structure composed of the carbon fiber rod (201), the folding joint 1 (207), and the folding joint 2 (208) unfolds or folds as needed, and the nozzles (205) are evenly distributed through the tee pipe fitting (209) to achieve flexible adjustment of the spray width. At the same time, a new type of atomizing nozzle is used to support dynamic adjustment of the spray angle and spray volume to ensure spraying uniformity;

[0043] Step 4: Chassis dynamic lifting control: By driving the wheel set (101) and the caster wheel set (102) through the hub motors, the chassis support (103) ensures the flexible movement and stable support of the robot between the ridges in the strawberry field; the central processing module uses the MPC model predictive control algorithm to perform differential control on the driving wheel set to achieve precise navigation of the robot under different terrain conditions. The ultrasonic sensors installed under the support platform (7) detect the relative height between the robot and the ridge, and adjust the height adjustment device (104) in real time to ensure that the distance between the spray boom and the plants is always appropriate; the telescopic motor adjusts the chassis height through the guide rail and the support rod to ensure that the spraying device always maintains the optimal working height; the central processing module controls the telescopic motor to complete the height adjustment by calculating the error between the current height and the target height to adapt to the terrain changes of different ridges.

[0044] Step 5: Implement closed-loop control: According to the motion state of the robot and the feedback information of the spraying system, the spraying parameters are adjusted in real time to ensure the uniformity and accuracy of spraying; when the robot turns or encounters special terrain, the spraying mode is automatically adjusted to avoid waste and uneven distribution of the liquid medicine; the control box system (6) converts the PWM signal into a driving signal through the TTL-RS485 signal conversion module to precisely control the operating states of the water pump motor and the hub motors, ensuring the stable operation of the robot in a complex environment.

[0045] The present invention has the following beneficial effects:

[0046] The present invention provides a cross-ridge intelligent strawberry operation robot, which has a variable-height mobile chassis (1) and a variable-spraying-width spraying system (2), and can adapt to the complex terrain in the strawberry field and the height changes of different ridges, realizing functions such as autonomous navigation and precise variable spraying. Through the combination of the image sensing module (5) and the composite sensor group (3), the robot can real-time sense the growth status and pest and disease conditions of strawberry plants, and use deep learning algorithms to optimize and control the spraying amount, significantly improving the resource utilization rate and reducing the waste of liquid medicine. At the same time, the robot adopts a lightweight design and an efficient driving system, has good obstacle-crossing ability and operation stability, can greatly reduce the manual labor intensity, and improve the management efficiency and economic benefits of strawberry planting.

[0047] Compared with traditional manual operations and some existing robot operation methods, the present invention has significant advantages.

[0048] In terms of precise operation, through the visual navigation scheme based on the deep learning image segmentation model, the stability of the path planning of the robot between the ridges in the strawberry field is greatly improved, and it can accurately avoid obstacles and strawberry plants to avoid collision damage. Combining advanced image recognition technology for variable spraying control, the spraying amount can be precisely adjusted according to the actual situation of strawberry plants, improving the utilization rate of pesticides, reducing waste, reducing environmental pollution, and ensuring the yield and quality of strawberries.

[0049] In terms of efficient operation, the robot adopts a high-power drive wheel set and a flexible omnidirectional wheel set (102), with good obstacle-crossing ability and steering performance, and can quickly shuttle between the ridges in the strawberry field for operation. The adjustable design of the telescopic spray system enables the spray coverage range to be adjusted according to actual needs, improving the operation efficiency.

[0050] In terms of intelligent control, the central processing module combines various sensor data and advanced control algorithms to achieve intelligent control of the robot's movement and spraying. The intelligent battery management system ensures stable power supply during the operation of the robot and extends the battery life.

[0051] In terms of environmental protection and energy conservation, precise variable spraying reduces the amount of pesticides used and decreases environmental pollution. The lightweight structural design and efficient power system reduce the energy consumption of the robot, meeting the requirements of green agricultural development. Brief Description of the Drawings

[0052] Figure 1 Axonometric view of the structure of the cross-ridge strawberry variable spray width spraying robot;

[0053] Figure 2 Top view of the variable spray width spray system of the cross-ridge strawberry variable spray width spraying robot based on a carbon fiber folding rod;

[0054] Figure 3 Side view of the cross-ridge variable height mobile chassis structure of the cross-ridge strawberry variable spray width spraying robot;

[0055] Figure 4 Schematic diagram of the unfolded state of the carbon fiber folding rod structure of the cross-ridge strawberry variable spray width spraying robot;

[0056] Figure 5 Figure of extreme case 1 of the cross-ridge strawberry variable spray width spraying robot during operation and collision;

[0057] Figure 6 Simplified mechanical structure model diagram of the cross-ridge strawberry variable spray width spraying robot under extreme case 1 of collision during operation;

[0058] Figure 7 Figure of extreme case 2 of the cross-ridge strawberry variable spray width spraying robot during operation and collision;

[0059] Figure 8 Simplified mechanical structure model diagram of the cross-ridge strawberry variable spray width spraying robot under extreme case 2 of collision during operation. Detailed Implementation Manner

[0060] The following specifically describes a cross-ridge strawberry variable spray width spraying robot of the present invention with reference to the schematic diagrams.

[0061] A cross-ridge strawberry variable spray pattern spraying robot structure proposed by the present invention, as Figure 1 shown, includes a cross-ridge variable height mobile chassis (1), a variable spray pattern spraying system (2), a composite sensor group (3), a battery pack (4), an image sensing module (5), a control box system (6), and a support platform (7).

[0062] The variable spray pattern spraying system (2), the composite sensor group (3), the battery pack (4), and the image sensing module (5) are electrically connected to the control box system (6) to process the collected image data and provide relevant computing power for controlling the drive wheel group motor and the water pump motor; the support platform (7) is provided with the variable spray pattern spraying system (2), and its overall structure is as Figure 2 shown. The variable spray pattern spraying system (2) is composed of multi-section carbon fiber spray pipes and folding joints. The multi-section carbon fiber spray pipes are folded and placed flat on both sides of the support platform (7), and the folding balance double rods are unfolded during the spraying stage; the nozzle (205) is connected to the water pump (203) and the water tank (202) through a PVC transparent hose. The water pump (203) adjusts the flow rate of the water pump (203) and the spraying amount of the nozzle (205) in real time according to the traveling speed of the robot. The connecting bracket (204) and the support platform (7) are connected by metal corner pieces.

[0063] The cross-ridge variable height mobile chassis (1) is composed of a hub motor drive wheel group (101), a universal wheel group (102), a chassis bracket (103), and a height adjustment device (104), as Figure 3As shown in the figure. A support platform (7) is provided on the chassis support (103). One end of the support platform is movably connected to the left front and right front hub motor drive wheel sets (101), and the other end is movably connected to the left rear and right rear universal wheel sets (102). The drive wheels are driven by high-power and high-torque motors to ensure that the robot has sufficient driving force in the complex terrain of the strawberry orchard; the universal wheels are selected as high-precision and wear-resistant universal wheels. The chassis support (103) is made of high-strength aluminum alloy material, which reduces the overall weight while ensuring the structural stability. This cross-ridge variable-height mobile chassis (1) is used for the steering, attitude and position adjustment of the robot, and provides a stable mounting platform for the operation system. Through the image sensing module (5) for environmental perception, the central processing module uses the MPC model predictive control algorithm to perform differential control on the drive wheel sets, and can accurately adjust the rotation speed and steering of the drive wheels according to different working environments and path planning, so as to realize the stable driving of the robot between the ridges in the strawberry orchard. The height adjustment module uses the chassis height adjustment device (104) which includes a plurality of telescopic motors, guide rails, support rods and transmission mechanisms installed on the chassis support (103) to realize the chassis height adjustment of the robot in different working environments. Each telescopic motor is connected to the support rod through a screw drive or a gear drive mechanism, and the other end of the support rod is connected to the four corner support points of the chassis. The telescopic motor is driven by a control signal to realize the telescopic movement of the support rod, thereby adjusting the overall height of the chassis. During the lifting and lowering process of the chassis, the support rod needs to bear the total weight W of the robot and the additional load that may be applied by the external environment. The thrust F p can be calculated by the following formula:

[0064]

[0065] where W is the total weight of the robot; n is the number of support rods; f r is the friction of the ball screw and the sliding resistance of the slide rail.

[0066] According to the thrust required by the motor, a closed-loop control is performed in the control system so that the motor can support the total weight of the overall support platform (7).

[0067] Height change calculation:

[0068] By calculating the current height H c and the safety height H t , calculate the error ΔH between the current height and the target height

[0069] ΔH = H c - H t (10)

[0070] According to the pitch p of the telescopic motor, calculate the number of turns required for the motor to rotate:

[0071]

[0072] According to the number of pulses P output per revolution of the encoder rev , the total number of pulses to be output is obtained as follows:

[0073] P total = N·P rev (12)

[0074] Feed this signal back to the control module to output the corresponding control pulses for the telescopic motor, so that the motor can complete the closed-loop control of the relative height. Within the stroke range of the telescopic motor, ensure that the robot can be applied to work on ridgelines of different heights.

[0075] The variable spray width spray system (2) based on the carbon fiber folding rod is composed of a carbon fiber rod (201), a water tank (202), a water pump (203), a connecting bracket (204), a nozzle (205), a PVC transparent hose (206), a folding joint 1 (207), a folding joint 2 (208), and a tee fitting (209). One end of the folding joint 1 (207) is a cylindrical opening, which is coaxially connected to the carbon fiber rod (201); the other end is a joint connection structure with a hinge base and a connecting shaft, and is supported and connected to another folding joint 2 (208) with a hinge head through a hinge structure. The water tank (202) is made of corrosion-resistant and high-strength plastic material and is fixed on the bracket platform (7) to ensure stable and reliable operation during the movement of the robot. The water pump (203) is a high-precision and variable-frequency centrifugal pump, which can accurately adjust the flow rate and pressure in real time according to the traveling speed of the robot, the pest and disease situation of strawberry plants in the operation area, and the water demand. The telescopic spray rod is made of carbon fiber material, combined with a foldable structure design, which reduces the weight while ensuring the strength, and is convenient for storage and extension. The nozzle (205) is a new type of atomizing nozzle, which can adjust the spray angle and spray volume according to the operation requirements, and is fixed on the carbon fiber rod (201) through a threaded structure by a tee fitting (209) with a fixed ring, and is evenly fixed at the lower end of the carbon fiber rod (201). The PVC transparent hose (206) is coaxially fixed through the tee fitting (209) and is used to connect the water pump (203), the water tank (202), and the nozzle (205) to ensure smooth water flow.

[0076] The composite sensor group (3) mainly includes an ultrasonic sensor, a liquid level monitoring sensor, a spray flow sensor, a motor encoder, and a GNSS navigation module. The ultrasonic sensors are installed at the chassis bracket (103), with two installed on each side of the left and right drive wheels, and their height can also be adjusted to adapt to different ridge channel height ranges. At the same time, one is also installed vertically below the bracket platform (7) perpendicular to the ground to detect the relative height of the current platform from the ridge channel horizontal plane. It is used to detect the distance between the vehicle tires and the obstacles on both sides of the ridge channel, so as to assist the robot in path planning and obstacle avoidance control. The liquid level monitoring sensor is installed inside the water tank (202) to detect the current liquid level height of the liquid medicine and feedback the liquid level height signal back to the control system. The spray flow sensor is installed inside the liquid outlet of the nozzle (205) to judge the current spray flow rate and feedback the flow signal back to the control system. The motor encoder is embedded in the hub motor, responsible for detecting the rotation speed and signal and feedbacking the speed signal back to the control system for the control calculation of the motor speed closed-loop. The GNSS navigation module is installed on the bracket platform (7) and is responsible for obtaining real-time GPS positioning signals to obtain the absolute position information of the robot during operation in real time. This module can accurately measure the longitude and latitude information of the robot in the global coordinate system and convert it into a plane coordinate suitable for local navigation through a built-in algorithm.

[0077] The battery pack (4) is composed of lithium batteries and is fixed in the bracket structure of the cross-ridge variable-height mobile chassis (1). It is equipped with an intelligent battery management system to monitor parameters such as the battery's power, voltage, and current in real time, realizing charge and discharge protection and optimized management of the battery, extending the battery's service life, and ensuring a stable power supply for the cross-ridge variable-height mobile chassis (1), the image sensing module (5), and the control box system (6).

[0078] The image sensing module (5) selects environmental perception combined with a depth camera and is equipped with an optical filter to obtain image information under different spectra between the ridges in the strawberry field for active perception positioning and tracking control between the ridges. The vision-based solution is easy to deploy and has rich image information. Using an autonomous navigation solution guided by vision in a structured environment is a low-cost attempt. By using CNN convolutional neural networks for deep learning to implement image segmentation technology for the roads between the ridges, the influence of environmental light changes on recognition is reduced. In addition, the distribution of strawberry plants at different growth stages on the ridges makes the collected image data complex. By modifying the structure of the network layer, the generalization ability of the network is enhanced, and it can then be applied to various places between the ridges, improving the robustness of the recognition module detection. It is used for the recognition of strawberry plants, the monitoring of growth conditions, and the detection of pests and diseases through real-time transmission and processing by the central processing module.

[0079] The control box system (6) includes a Jetson Nano B01 central processing module, an STM32F407VGT6 single-chip microcomputer bottom-layer MCU main control board, a TTL-RS485 signal conversion module, a water pump motor electronic speed control drive module, and a drive wheel hub motor drive module. All are fixed on the bracket platform (7). Its function is to quickly process and analyze the collected data, run various control algorithms, and provide computing power support for the motion control, spraying control, and pesticide application control of the robot. The water pump motor electronic speed control drive module performs PWM speed control through the PWM drive signal output by the bottom-layer MCU main control board; the drive wheel hub motor drive module transmits the control quantity signal calculated by the control algorithm through the central processing module to the bottom-layer MCU main control board through the serial port and outputs it to the signal conversion module, converts the TTL level to an RS-485 signal and then inputs it into the hub motor driver, and performs torque and speed control on the hub motor through the driver.

[0080] The bracket platform (7) is a hollow platform composed of aluminum profiles and stainless steel plates. The composite sensor group (3), the battery pack (4), the image sensing module (5), and the control box system (6) are stored on the platform. This part is responsible for combining the cross-ridge variable-height mobile chassis (1) with the upper control system and sensor group.

[0081] The variable spraying and intelligent pesticide application control steps are as follows:

[0082] Step 1: Image recognition and analysis. The image sensing module (5) uses a depth camera combined with an optical filter to collect image data of strawberry plants and the inter-row environment. Through the depth camera installed on the bracket platform (7), multi-spectral image information of the inter-rows in the strawberry field is obtained. The image data is processed in real time by the Jetson Nano B01 central processing module, and image semantic segmentation is realized by combining with the CNN convolutional neural network to identify the position, size, growth status, and pest and disease conditions of strawberry plants.

[0083] Step 2: Spraying amount calculation. Based on the image recognition results, the central processing module combines the robot's traveling speed, the growth status of strawberry plants, the distribution of pests and diseases, and a preset spraying model to calculate the spraying amount of each nozzle (205). In the composite sensor group (3): the ultrasonic sensor detects the ridge width and plant spacing, the liquid level monitoring sensor monitors the liquid medicine height in the water tank (202), and the spraying flow sensor monitors the spraying amount of the nozzle (205) to provide data support. For areas with strong growth and serious pests and diseases, the central processing module commands to increase the spraying amount; for sparse or healthy areas, the spraying amount is reduced.

[0084] Step 3: Spray volume control. The central processing module transmits the PWM control signal to the water pump motor drive module through the underlying MCU main control board of the STM32F407VGT6 single-chip microcomputer, regulating the flow rate and pressure output by the water pump (203). The liquid medicine is transported to the nozzle (205) and the spray head through the PVC transparent hose (206) to achieve variable spray operation. The spray rod structure composed of the carbon fiber rod (201), the folding joint 1 (207) and the folding joint 2 (208) unfolds or folds as needed, and the nozzles (205) are evenly distributed through the three-way pipe fitting (209) to achieve flexible adjustment of the spray width. During the operation of the robot, its carbon fiber folding rod structure unfolds to increase the spray operation area, and its operation effect is as Figure 4 shown.

[0085] Step 4: Chassis dynamic lifting control. Through the hub motor drive wheel set (101) and the universal wheel set (102), the chassis bracket (103) ensures the flexible movement and stable support of the robot between the ridges in the strawberry field. The central processing module uses the MPC model predictive control algorithm to perform differential control on the drive wheel set to achieve precise navigation of the robot under different terrain conditions. The ultrasonic sensor installed below the support platform (7) detects the relative height between the robot and the ridge, and adjusts the height adjustment device (104) in real time to ensure that the distance between the spray rod and the plants is always appropriate. The telescopic motor adjusts the chassis height through the guide rail and the support rod to ensure that the spray device always maintains the optimal operation height. The central processing module controls the telescopic motor to complete the height adjustment by calculating the error between the current height and the target height to adapt to the terrain changes of different ridges.

[0086] The specific process of the central processing module using the MPC model predictive control algorithm to perform differential control on the drive wheel set is as follows:

[0087] Construct a kinematic model of the cross-ridge strawberry variable spray width spray robot, and obtain an error prediction state model through linearization. Adopt an incremental control strategy combined with an MPC controller to optimize the objective function, solve the input control variable at the next moment in real time, and achieve path tracking control of the given input quantity. Propose the above-mentioned MPC controller, judge and call the preview control mode through the error threshold, and impose constraints on the input control quantity in combination with the special environmental conditions of the strawberry field ridge to avoid the phenomenon of "hitting the ridge"; specifically including:

[0088] Step 3.1: Obtain two input control variables and three system state variables of the cross-ridge strawberry variable spray width spray robot at the current moment. The two input control variables are the centroid movement speed of the robot and the angular velocity of the centroid rotating around the instantaneous center. The three system state variables are the x-axis coordinate and y-axis coordinate of the robot in the world coordinate system, and the angle of rotation around the z-axis coordinate of the centroid.

[0089] Before the step 3.1, it also includes:

[0090] Obtain the hub motor parameters, wheels, and robot parameters of the robot; the motor parameters mainly include the motor reduction ratio, encoder count value, and number of lines of the encoder; the wheel parameters mainly include the wheel radius and wheel speed; the robot parameters mainly include the robot's yaw angular velocity and the position of the robot in the world coordinate system, including the x-axis coordinate and y-axis coordinate.

[0091] Based on the hub motor parameters, the rotational speed of a single motor can be obtained as:

[0092]

[0093] Where M is the count value of the encoder during the sampling period, P is the number of lines of the encoder, that is, the total number of pulses triggered when the encoding disk rotates one circle, N is the motor reduction ratio, R is the radius of the wheel, and Δt is the sampling period of the encoder.

[0094] Step 3.2 also needs to obtain the planned path of the road between the ridges in the strawberry field. The present invention is based on using the GNSS navigation module to obtain the longitude, latitude, and altitude information of the robot, select the target trajectory, and calculate the position error and direction error. The position error includes: taking the longitude direction of the target trajectory as the X coordinate and the latitude direction as the Y coordinate, and the offsets of the robot in these two directions are the lateral error and longitudinal error respectively. The direction error includes: the yaw angle error of the robot with respect to the given target trajectory.

[0095] Step 3.3: Construct the kinematic model of the spraying robot according to the wheel parameters and robot parameters:

[0096]

[0097] Where the instantaneous linear velocity of the movement of the robot's center of mass is v, x is the lateral displacement; y is the longitudinal displacement; is the yaw angle; respectively represent the lateral and longitudinal velocities; ω represents the angular velocity of the robot's center of mass rotating around the instantaneous center; the state quantity can be represented by to represent; the control quantity can be represented by C = [v, ω] T to represent.

[0098] Since this model is a non-linear model, after linearization at the target state and introducing a new set of error state variables, a linearized error model is constructed:

[0099]

[0100] Where A and B are Jacobian matrices, and the state variables of the linear error model are defined as:

[0101]

[0102] where x r is the reference lateral displacement; y r is the reference longitudinal displacement; is the reference yaw angle; all are obtained from the reference values calculated by this control algorithm.

[0103] The input control quantity of the linear error model is:

[0104]

[0105] Assume that the control quantity input of the target point is 0. Where:

[0106]

[0107] The discretized error model obtained by using the forward Euler method for discretization is:

[0108]

[0109] Where: where T is the sampling time and k is the sampling time instant.

[0110] By introducing an incremental model, that is, the system input is changed to incremental control, which can make the system input smoother. By defining new state variables:

[0111] Substitute the new state variables into formula (11) to obtain the state equation of the discretized incremental prediction model:

[0112]

[0113] Where

[0114]

[0115] where C = [I O], I is the identity matrix and O is the all-zero matrix.

[0116] Through the recurrence relation, iterate the values at times k+1, k+2,..., k+N p to obtain the final incremental prediction model as:

[0117] Y = Ψξ(k) + ΘΔU(18)

[0118] Where,

[0119] N c is the control time domain, and N p is the prediction time domain.

[0120] Step 3.3: Since the control inputs are the increment of the centroid velocity v c and the increment of the yaw rate ω, according to the formula

[0121]

[0122] where v1 and v2 are the output rotational speeds of the left and right drive wheels respectively, and l icr is the radius of curvature of the robot centroid rotating around the instantaneous center.

[0123] Further decoupling gives the speeds v1 and v2 acting on each drive wheel as follows:

[0124]

[0125] According to formula (8), by changing the duty ratio of the hub motor drive signal, the system input, i.e., the speeds of the two drive wheels, is changed, enabling the robot to achieve two-wheel differential operation during operation.

[0126] Step 3.4: Based on the error prediction state model of the discretized spraying robot, establish the objective function of the linear model predictive controller MPC.

[0127] Based on the kinematic characteristics of the robot and considering the minimization of both the tracking error and the control input increment, the following MPC path tracking controller objective function is constructed:

[0128]

[0129] where t = 1, 2, …, N c , …, N p , y t is the value of the state variable at time t, is the value of the target reference state variable at time t, u t is the control input at time t, is the control input at time t - 1, ε is the slack variable introduced to ensure a feasible solution, and Q, P, F, ρ are the weight matrices of the system state tracking error, incremental control input, system terminal control input, and slack factor respectively. N p is the MPC prediction horizon length, and N c is the MPC control horizon length.

[0130] Step 3.5: Based on the target navigation path, solve the objective function to determine the input control variable of the MPC at the next moment; the input control variable is used to minimize the error between the robot and the target navigation path with the minimum control energy to control the motion trajectory of the robot in real time.

[0131] Based on the above objective function, the MPC control problem can be summarized as follows:

[0132]

[0133] where u * represents the optimal input control quantity obtained by solving the optimization objective function; and u are the upper and lower limits of the control input respectively; Δu min and Δu max are the maximum and minimum values of the calculated control increment respectively; The above constraint critical parameters can be obtained by constructing extreme cases when the spraying robot actually operates on the ridges of the strawberry field.

[0134] The following is extreme case 1 when the spraying robot actually operates on the ridges of the strawberry field:

[0135] In this case, since the rotation center of the robot is at the suspension center of the front-wheel drive wheels and the lateral displacement of the robot does not deviate significantly from the trajectory center line during driving, the rear-wheel driven wheels are more likely to touch the strawberry plant ridges during rotation. Therefore, in this extreme case, the maximum yaw angle of the vehicle, denoted as

[0136] The following is extreme case 2 when the spraying robot actually operates on the ridges of the strawberry field:

[0137] In this case, since the lateral displacement of the robot deviates significantly from the trajectory center line during driving, the front-wheel drive wheels are more likely to touch the strawberry plant ridges during rotation. Therefore, in this extreme case, the maximum yaw angle of the vehicle is denoted as

[0138] Based on the discretized error prediction state model of the spraying robot, a linear MPC controller is constructed to solve the constrained optimization problem. According to the optimal control quantity obtained by optimizing and solving the objective function, through further decoupling calculation, the control drive signal acting on the bottom hub motor is obtained, realizing the control objective of integrating path planning control and tracking.

[0139] Specifically for extreme case 1, such as Figure 5 the maximum yaw angle shown The calculation process is as follows:

[0140] First, simplify the mechanical structure model into a simplified schematic diagram with only the chassis suspension, as shown in Figure 6 Since the rear-wheel size is small, ignore the radius of the rear-wheel driven wheels and only judge that when the hub axis touches the ridge boundary, extreme case 1 is triggered;

[0141] Assume that the wheel travels along the center line of the ridge, and draw PF perpendicular to the extension of ED at F. From the geometric relationship, it can be obtained that BO is equal to half of the width of the robot chassis, OD = BP is equal to the length of the robot chassis. According to ΔOAB~ΔOCD and According to the mathematical relationship, AE is equal to the length from the center point of the robot to the outermost boundary of the ridge channel, that is, AE = l + d / 2, where l is the distance from the center line to the boundary of the ridge channel and d is the width of the ridge channel. Denote BP = L0 as the length of the robot chassis and BO = D0 as half of the width of the robot chassis.

[0142] To obtain Then the specific formula is:

[0143]

[0144] Furthermore, solve the variable At The analytical solution on.

[0145]

[0146] The maximum yaw angle of the specific extreme case 2 The calculation process is as follows:

[0147] Since the size of the front wheels is relatively large and cannot be ignored, considering the radius r of the front-wheel drive wheels, the mechanical structure model is simplified to a simplified schematic diagram with only the chassis suspension. It is judged that when the front-wheel drive wheels touch the boundary of the ridge channel, extreme case 2 is triggered, as Figure 7 Shown.

[0148] According to the geometric relationship as Figure 8 Shown, the maximum yaw angle Is the angle between the vehicle body movement direction and the axis parallel to the ridge channel The angle between the contact point of the front-wheel drive wheels of the chassis touching the inner wall of the ridge channel and the front axle of the front-wheel drive wheels is ∠QPN = β, the angle between the contact point of the front-wheel drive wheels of the chassis touching the inner wall of the ridge channel and the direction perpendicular to the ridge channel is ∠MPN = γ, and the distance from the center of the front-wheel drive wheel suspension to the center line is d c , and the distance from point P to the boundary of the ridge channel is PM = l + d.

[0149] Thus, the following relational expressions can be obtained:

[0150]

[0151] Furthermore, the maximum yaw angle is obtained

[0152]

[0153] Denote the maximum value of the robot yaw angle as The minimum value of the robot yaw angle is

[0154] The range of the robot yaw angle is finally determined as:

[0155]

[0156] Among them and are respectively solved according to formula (24) and formula (26) to obtain

[0157] By solving its differential within a sampling period T, the range of the angular velocity is as follows:

[0158]

[0159] Among them, the maximum value of the angular velocity of the robot's center of mass is denoted as The minimum value of the angular velocity of the robot's center of mass is denoted as

[0160] The range of the input velocity is artificially given. By setting a constant tracking linear velocity of the center of mass, a set of actual control variables u t is obtained. The constraints are as follows:

[0161]

[0162] Among them is the minimum value of the increment of the linear velocity of the robot's center of mass; is the maximum value of the increment of the linear velocity of the robot's center of mass.

[0163] Finally, through the transformation of the incremental control variables, a set of control variable increments Δu t is obtained. The constraints are as follows:

[0164]

[0165] Among them is the minimum value of the increment of the linear velocity of the robot's center of mass; the maximum value of the increment of the linear velocity of the robot's center of mass; is the minimum value of the increment of the angular velocity of the robot's center of mass; is the minimum value of the increment of the angular velocity of the robot's center of mass.

[0166] By constraining the input according to the special environmental conditions of the strawberry field ridges, the phenomenon of "hitting the ridges" caused by large adjustments is avoided. Solving the above MPC problem can obtain the system input control variables at the next moment, so as to minimize the tracking distance error; the input is as small as possible; the change rate of the input is as small as possible, achieving the purpose of reducing power consumption.

[0167] Step 5: Implement closed-loop control. According to the motion state of the robot and the feedback information of the spraying system, adjust the spraying parameters in real time to ensure the uniformity and accuracy of spraying. When the robot turns or encounters special terrain, automatically adjust the spraying and mode to avoid waste and uneven distribution of the liquid medicine. The control box system (6) converts the PWM signal into a driving signal through the TTL-RS485 signal conversion module, and precisely controls the operating states of the water pump motor and the hub motor to ensure the stable operation of the robot in a complex environment.

Claims

1. A cross-ridge strawberry variable spray width spraying robot structure, characterized in that, It includes a cross-ridge variable-height mobile chassis (1), a variable spray width spraying system (2), a composite sensor group (3), a battery pack (4), an image sensing module (5), a control box system (6), and a support platform (7); the cross-ridge variable-height mobile chassis (1) is used to achieve stable driving and position adjustment of the robot between the ridges in the strawberry orchard; the variable spray width spraying system (2) is installed on the support platform (7), and the variable adjustment of the spray width is achieved through a carbon fiber folding rod; the image sensing module (5), mainly a depth camera, is used to collect image information of the environment between the ridges in the strawberry orchard; the composite sensor group (3) is used to monitor the operating state of the robot and the working environment; the control box system (6) is used to control the movement and spraying of the robot; the battery pack (4) provides power support for the robot; the support platform (7) is used to carry the above components and connect to the chassis support (103).

2. The structure of a cross-ridge strawberry variable spray-width spraying robot according to claim 1, characterized in that The cross-ridge variable-height mobile chassis (1) is composed of a hub motor drive wheel set (101), a universal wheel set (102), a chassis support (103), and a height adjustment device (104). The support platform (7) is provided on the chassis support (103). The height adjustment device (104) is connected to the chassis support (103). One end of the chassis support (103) is movably connected to the left front and right front hub motor drive wheel sets (101), and the other end is movably connected to the left rear and right rear universal wheel sets (102).

3. The structure of a cross-ridge strawberry variable spray-width spraying robot according to claim 1, characterized in that, The carbon fiber rod (201) of the variable spray width spraying system (2) is fixedly connected through a hinge structure composed of multiple folding joints 1 (207) and folding joint 2 (208), realizing the unfolding and folding of the three-section carbon fiber rod (201). The overall carbon fiber rod (201) is rigidly connected to the support platform (7) through a connecting bracket (204), thus forming a complete mechanical system with the main structure of the robot. The three-way pipe fitting (209) with a fixed ring is fixed on the carbon fiber rod (201) to form a modular assembly structure; the nozzle (205) adopts an atomizing nozzle, and the nozzle (205) is connected to the variable-spacing water pump (203) and the water tank (202) through a PVC transparent hose (206). The variable-spacing water pump (203) is a variable-frequency centrifugal pump, and the flow rate is adjusted in real time. The spraying flow rate and spraying volume are adjusted in real time according to the traveling speed of the robot.

4. The structure of a cross-ridge strawberry variable spray-width spraying robot according to claim 1, characterized in that The composite sensor group (3) mainly includes ultrasonic sensors, liquid level monitoring sensors, spray flow sensors, motor encoders, and GNSS navigation modules. The ultrasonic sensors are installed at the chassis brackets (103), with two installed on each side of the left and right drive wheels and their heights adjustable to adapt to different ridge height ranges. Additionally, one ultrasonic sensor is vertically installed below the support platform (7) to detect the relative height between the platform and the ridge horizontal plane and the distance between the vehicle tires and ridge obstacles, assisting in path planning and obstacle avoidance. The liquid level monitoring sensor is installed inside the water tank (202) to detect the liquid level of the liquid medicine and feed back to the control system. The spray flow sensor is installed at the liquid outlet of the nozzle (205) to detect the spraying flow rate and feed back to the control system. The motor encoder is embedded in the hub motor to detect the rotational speed and feed back to the control system to achieve speed closed-loop control.

5. The structure of a cross-ridge strawberry variable spray width spraying robot according to claim 1, characterized in that, The battery pack (4) includes lithium batteries, which are fixed in the support structure of the cross-ridge variable-height mobile chassis (1). It is equipped with an intelligent battery management system to monitor power, voltage, and current parameters in real time, achieve charge and discharge protection and optimized management, extend the battery life, and stably supply power to the variable-height mobile chassis (1), the image sensing module (5), and the control box system (6).

6. The structure of a cross-ridge strawberry variable spray width spraying robot according to claim 1, characterized in that, The image sensing module (5) includes a depth camera, which can obtain image information of the strawberry field ridges under different spectra for ridge positioning, tracking control, and autonomous navigation.

7. The structure of a cross-ridge strawberry variable spray width spraying robot according to claim 1, characterized in that, The control box system (6) includes a Jetson Nano B01 central processing module, an STM32F407VGT6 single-chip microcomputer bottom-layer MCU main control board, and a TTL-RS485 signal conversion module inside.

8. The structure of a cross-ridge strawberry variable spray-width spraying robot according to claim 1, characterized in that, The support platform (7) is formed by connecting aluminum profiles and aluminum plates through metal corner pieces, and corresponding installation holes are set according to design requirements to tightly fix the variable spray width spray system (2), the battery pack (4), the image perception module (5), and the control box system (6) on the support platform (7). While providing support and load-bearing for the entire system, the support platform (7) minimizes the platform weight and is connected to the variable-height mobile chassis (1) through metal corner pieces, playing a role of connecting the upper and lower parts.

9. The operation control method of a spraying robot structure with variable spraying amount and variable spraying width based on a cross-ridge strawberry, according to claim 1, is characterized in that, An autonomous navigation of the robot in the strawberry field ridges is achieved using a navigation solution that fuses neural network vision and lidar, and variable spray control is performed by combining the robot motion model and image recognition technology. The specific steps are as follows: Step 1: Image recognition and analysis: The image sensing module (5) uses a depth camera combined with an optical filter to collect image data of strawberry plants and the ridge environment. Through the depth camera installed on the support platform (7), multi-spectral image information of the strawberry field ridges is obtained. The image data is processed in real time by the Jetson Nano B01 central processing module, and image recognition is achieved by combining the Fast R-CNN network to identify the position, size, growth status, and pest and disease conditions of strawberry plants. The image recognition model based on the Fast R-CNN network is constructed as: Dataset construction: Use a depth camera to collect the characteristics of strawberry plants and the environment between ridges, obtain RGB images, and use the annotation tool LabelImg to annotate the images of the strawberry plant areas, marking the positions of the strawberry plants and the categories of the strawberries (healthy plants and diseased plants); Enhance the annotated data. Use geometric transformation, brightness adjustment, and noise addition of the images to obtain an enhanced dataset for training to improve the robustness of the model. Divide the constructed dataset into a training set, a validation set, and a test set in a ratio of 7:2:1; Select the backbone network based on ResNet as the feature extraction part of Fast R-CNN. ResNet has strong feature extraction capabilities and can effectively capture the multi-scale features of strawberry plants. Use the constructed dataset to train the Fast R-CNN model, set appropriate hyperparameters, and use a pre-trained model for fine-tuning to accelerate convergence; Divide the width of the camera image into 6 regions, and sequentially set the 1st, 2nd, 3rd, 4th, 5th, and 6th regions from left to right; the middle region (the 3rd region) is narrower and controls 2 nozzles (205), and the other 5 regions (the 1st, 2nd, 4th, 5th, and 6th regions) are wider, each controlling 4 nozzles (205). According to the strawberry plants detected by Fast R-CNN, frame and return their center points. According to the formula: Find the total number of plants N in the k-th interval region,k ; N k represents the total number of the central points of the strawberry plant coordinates in the k-th area; x i represents the abscissa of the image; where the δ(x) function is defined as: If the central coordinates (x c , y c ) of the strawberry plant identified by the FastR-CNN network fall within the k-th region, the value is 1; otherwise, it is 0. x min,k and x max,k represent the lower and upper limits of the coordinates of the k-th region, and the calculation formula is: where W is the field of view width of the camera, then the width of the middle region is expressed as: The widths of the other 5 regions are respectively: According to the division, the physical area A of each region region,k is as follows: where H is the field of view height of the camera; Calculate the plant density of each divided region as: And normalize to obtain the spray importance weight W k : Among which S region,k is the average severity of pest and disease plants in this area, that is, the proportion of pest and disease plants identified by the network in the total number of plants identified; D max represents the maximum plant density in the six divided areas; α is the weight factor of the pest and disease severity, and this weight coefficient is an important indicator for the subsequent preset spray model decision-making; Convert the trained model into the ONNX format supported by the Jetson Nano platform, and use TensorRT to optimize the model to improve the inference speed. Implement real-time inference on the Jetson Nano, combine the image data collected by the depth camera, identify the positions of strawberry plants, and mark healthy plants and diseased plants to provide real-time decision-making support for the intelligent management of strawberry gardens; Step 2: Spray volume calculation: Based on the image recognition results, the central processing module combines the robot's traveling speed, the growth status of strawberry plants, the distribution of pests and diseases, and a preset spray model to calculate the spray volume of each nozzle (205); in the composite sensor group (3): the ultrasonic sensor detects the ridge width and plant spacing, the liquid level monitoring sensor monitors the liquid medicine height in the water tank (202), and the spray flow sensor monitors the spray volume of the nozzle (205) to provide data support; for areas with strong growth and severe pests and diseases, the central processing module commands to increase the spray volume; for sparse or healthy areas, reduce the spray volume; The preset spray model is: Among them, Q k represents the flow rate that each nozzle in the k-th area needs to provide; β is the proportional amplification adjustment gain; W k is the calculated normalized spray importance weight; Q max represents the maximum spray flow rate of each nozzle; n k represents the total number of nozzles covering the k-th area; Step 3: Spray volume control: The central processing module transmits the PWM control signal to the water pump motor drive module through the STM32F407VGT6 single-chip microcomputer bottom-layer MCU main control board to adjust the flow rate and pressure of the water pump output. The liquid medicine is transported to the nozzle (205) through the PVC transparent hose (206) to achieve variable spraying operation; The spray rod structure composed of the carbon fiber rod (201), folding joint 1 (207) and folding joint 2 (208) is unfolded or folded as needed, and the nozzles (205) are evenly distributed through the tee pipe fitting (209) to achieve flexible adjustment of the spray width. At the same time, a new type of atomizing nozzle is adopted to support dynamic adjustment of the spray angle and spray volume to ensure spraying uniformity; Step 4: Chassis dynamic lifting control: Through the hub motor drive wheel set (101) and the universal wheel set (102), the chassis bracket (103) ensures the flexible movement and stable support of the robot between the ridges in the strawberry field; The central processing module uses the MPC model predictive control algorithm to perform differential control on the drive wheel set to achieve precise navigation of the robot under different terrain conditions. The ultrasonic sensor installed under the support platform (7) detects the relative height between the robot and the ridge, and adjusts the height adjustment device (104) in real time to ensure that the distance between the spray rod and the plants is always appropriate; The telescopic motor adjusts the chassis height through the guide rail and the support rod to ensure that the spraying device always maintains the best working height; The central processing module controls the telescopic motor to complete the height adjustment by calculating the error between the current height and the target height to adapt to the terrain changes of different ridges; Step 5: Implement closed-loop control: According to the motion state of the robot and the feedback information of the spraying system, the spraying parameters are adjusted in real time to ensure the uniformity and accuracy of spraying; When the robot turns or encounters special terrain, the spraying mode is automatically adjusted to avoid liquid medicine waste and uneven distribution; The control box system (6) converts the PWM signal into a drive signal through the TTL-RS485 signal conversion module to accurately control the operating states of the water pump motor and the hub motor to ensure the stable operation of the robot in a complex environment.

Images