A rubber tree targeted fertilization robot
By using a rubber tree fixed-point fertilization robot, which employs a tracked walking mechanism and multi-source information fusion technology, the problems of terrain adaptability and accuracy in rubber tree fertilization operations have been solved. This has enabled precise quantitative fertilization in complex terrains, improving the quality and efficiency of fertilization in rubber plantations.
Patent Information
- Application Number
- CN202611090526.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-22
- Publication Date
- 2026-11-17
AI Technical Summary
Existing rubber tree fertilization practices suffer from problems such as poor terrain adaptability, inaccurate fertilization location, inconsistent drilling depth, low fertilizer utilization rate, low efficiency of process separation, and fertilizer bridging blockage, which cannot meet the requirements of modern precision agriculture in rubber plantations.
A rubber tree fixed-point fertilization robot was designed, which adopts a tracked walking mechanism, environmental perception and autonomous navigation, soil resistance adaptive fixed-depth drilling, anti-bridging closed-loop quantitative fertilization and soil covering functions to achieve precise fixed-point fertilization. It combines a binocular depth camera and LiDAR for environmental perception, uses a drill bit rotary motor and a screw lifting motor for drilling and fertilization control, and achieves autonomous navigation and obstacle avoidance through multi-source information fusion.
It improves the quality and efficiency of rubber tree fertilization operations, increases fertilizer utilization, reduces labor intensity, adapts to complex terrain, and enables precise quantitative fertilization, while being compatible with both solid and liquid fertilizer application modes.
Smart Images

Figure CN123100353A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of agricultural intelligent equipment technology, specifically relating to a rubber tree fixed-point fertilization robot, which is suitable for precise, efficient and low-damage fertilization operations for high-value economic forests such as rubber plantations. Background Technology
[0002] Natural rubber is an important industrial raw material and strategic resource, and its yield and latex quality are highly dependent on the nutritional status of the rubber trees. Scientific and precise fertilization is a key measure to ensure the healthy growth of rubber trees and improve latex production. However, current rubber tree fertilization is still mainly done manually, which presents the following prominent problems: Poor terrain adaptability: Rubber trees are mostly planted on hills and slopes, where forest roads are narrow and muddy, making it difficult for traditional wheeled machinery to enter or prone to slipping and sinking, hindering stable operation. Inaccurate fertilization location: When fertilizing manually or with simple machinery, the fertilization point is often off-center from the drip line of the tree canopy or the root concentration area, resulting in low fertilizer utilization and easily causing root burn or insufficient fertility. Arbitrary drilling depth and fertilizer amount: Lack of real-time perception of soil hardness leads to inconsistent drilling depths, resulting in fertilizer being applied too shallowly and easily volatilized and lost; the amount of fertilizer relies on manual estimation, which has large deviations. Low efficiency due to process separation: Drilling holes first and then manually or separately applying fertilizer requires secondary alignment, which is not only inefficient but also often results in misalignment of the fertilizer in the hole, with fertilizer not reaching the bottom, leading to significant waste. Fertilizer bridging and blockage: Granular compound fertilizer is prone to bridging and arching within the feed hopper, leading to poor or even interrupted flow, severely affecting the uniformity of fertilization. Existing mechanical fertilization devices, such as suspended trenching fertilizer applicators and hand-push seeders, while mechanized, generally lack features such as an adaptive chassis for forest use, soil resistance sensing, anti-bridging quantitative fertilizer dispensing, and drilling-fertilizing-covering functions. Their overall performance cannot meet the requirements of precision agriculture in modern rubber plantations. Summary of the Invention
[0003] To address the problems existing in the prior art, this invention provides a rubber tree fixed-point fertilization robot, designed to achieve stable passage and precise operation in complex forest terrain, including slopes, narrow paths, and muddy surfaces. Through environmental perception and autonomous navigation, soil resistance-adaptive fixed-depth drilling, anti-bridging closed-loop quantitative fertilization, and automatic soil covering after fertilization, it enables fixed-point drilling and fertilization of solid fertilizer, fixed-point surface fertilization of liquid fertilizer, and soil backfilling operations, thereby significantly improving the quality, efficiency, and fertilizer utilization rate of rubber tree fertilization.
[0004] The rubber tree fixed-point fertilization robot of the present invention includes a tracked walking mechanism, a chassis frame, an angle sensor, a binocular depth camera, a lidar, a fertilizer storage tank, a stirrer, a base plate, a fertilizer discharge screw motor, a sliding plate, a sliding plate hole, a fertilizer discharge inclined pipe, a soil covering plate, a spiral drill bit, a drill bit rotation motor, a screw lifting motor, a lifting guide rail, and a displacement sensor. The tracked walking mechanism is installed on both sides of the chassis frame. The fertilizer storage tank is installed on the chassis frame, and the fertilizer storage tank cover is located on the top of the fertilizer storage tank. An ultrasonic level sensor is installed inside the fertilizer storage tank, and a fertilizer filling port is opened on the fertilizer storage tank cover for adding fertilizer to the fertilizer storage tank. One end of the agitator is fixed to the fertilizer storage tank cover, and the other end is located inside the fertilizer storage tank. The base plate is fixed to the middle of the front end of the chassis frame, the screw lifting motor is fixed to one side of the upper part of the base plate, two lifting guide rails are longitudinally fixed to the base plate, and the moving plate is movably set on the two lifting guide rails. The top of the moving plate is fixed to the output screw of the screw lifting motor. Under the action of the screw lifting motor, the moving plate moves up and down along the lifting guide rails. Displacement sensor I is installed on the base plate. The drill bit rotary motor is fixed to the moving plate, and the auger drill bit is fixed to the output shaft of the drill bit rotary motor. The binocular depth camera is installed on the top of the base plate, and the lidar is installed on the chassis frame through a support frame and located on one side of the base plate. The tilt sensor is installed on the chassis frame. The soil covering plate is fixed to the lower part of the chassis frame and located on one side of the base plate. The auger is located behind the drill bit; one end of the fertilizer discharge inclined tube is located below the fertilizer storage tank outlet, and the other end is located on one side of the auger drill bit; the fertilizer discharge screw motor is mounted on the chassis frame via a bracket and is located behind the fertilizer discharge inclined tube. The output screw of the fertilizer discharge screw motor is fixedly connected to one end of the slide plate. The slide plate is located between the fertilizer storage tank outlet and the upper inlet of the fertilizer discharge inclined tube. The other end of the slide plate has a circular hole. Through the action of the fertilizer discharge screw motor, the fertilizer storage tank outlet is connected or closed to the fertilizer discharge inclined tube inlet via the slide plate to achieve quantitative fertilizer discharge. Displacement sensor II is installed on the right side of the slide plate; the battery pack and electrical control box are installed inside the chassis frame; Hall current sensors are installed on the power supply circuits of the drill bit rotary motor and the screw lifting motor; wheel speed odometers are installed on the tracked walking mechanism for vehicle positioning and speed feedback; The electrical control box integrates the whole machine control unit. The upper-level computing unit carries a program developed based on the ROS Noetic distributed communication framework, including an environmental perception module, a navigation module, a drilling control module, a quantitative fertilization control module, a rubber tree identification and fertilization point determination module, and a communication and storage module. Each module completes data interaction and collaborative decision-making through ROS topics. The lower level uses an Arduino Mega2560 motion controller as the execution unit. The environmental perception module is used to collect point cloud data of rubber trees and other obstacles obtained by LiDAR, and to collect image information obtained by the binocular camera; to identify obstacles using the point cloud data obtained by LiDAR; and to collect robot operating status information obtained by the wheel speed odometer and tilt sensor. The rubber tree identification and fertilization point determination module identifies the rubber tree and its location, the canopy boundary of the rubber tree, and the target fertilization area based on the images and point cloud data obtained by the environmental perception module. It converts the rubber tree coordinates into three-dimensional coordinates in the robot coordinate system and determines the fertilization point coordinates based on the rubber tree trunk coordinates, canopy coverage, and pre-fertilization distance. The three-dimensional coordinates of the fertilization point are calculated using the following formula: ; in, The three-dimensional coordinates of the target fertilization point. The three-dimensional coordinates of the rubber tree trunk in the robot's coordinate system. This is the offset distance for fertilization, with a value of 0.4-0.5m. Let be a unit vector pointing from the tree trunk to the row spacing, and satisfy . .
[0005] The drilling control module controls the screw lifting motor to drive the auger drill bit to complete the lifting and lowering motion according to the target drilling depth. At the same time, it controls the drill bit rotation motor to drive the auger drill bit to rotate and drill. Based on the drilling depth fed back by displacement sensor I and the current data of the drill bit rotation motor collected by Hall current sensor, it calculates the soil resistance value and drilling feed speed, and adjusts the drilling feed speed of the drill bit rotation motor to achieve adaptive fixed-depth drilling. After drilling is completed, it controls the auger drill bit to rise a preset distance to provide fertilization space for quantitative fertilization. The drilling feed rate is calculated using the following formula: ; In the formula: This refers to the drilling feed rate; This is the maximum allowable feed rate for drilling; The critical value for soil resistance is obtained through field pre-calibration and represents the upper limit of resistance to prevent the drill motor from stalling and overloading. To collect and filter soil resistance data in real time, when hour, ; in, N is the length of the filter window, i = 0, 1, 2, ..., N-1; To control the cycle; The soil resistance is the soil resistance during the i-th control cycle prior to the current moment; ; ; Where: Where: This is the rated current of the drill bit rotary motor. For maximum thrust, , These are the weighting coefficients. ,and This refers to the current of the drill bit's rotating motor; This refers to the feed thrust of the lead screw jack motor; This refers to the current of the lead screw jack motor; This refers to the rated current of the screw jack motor. The mechanical efficiency of the lead screw drive is taken as 0.8-0.95; T is the output torque of the motor; P is the distance the nut moves forward when the motor rotates one revolution. To avoid vibration of the drilling mechanism caused by sudden changes in feed rate, the speed command is subjected to first-order smoothing filtering in each control cycle: ; in: The current smoothed feed rate; This represents the current feed rate of the borehole. The feed rate is the smooth output of the previous control cycle; To control the cycle; Let be the speed smoothing coefficient, satisfying 0. ; When the actual drilling depth approaches the target drilling depth, a deceleration control phase is initiated to prevent the drill bit from impacting the bottom of the hole. ; ; in This refers to the final feed rate; The smoothed feed rate; The target drilling depth was set at 30cm based on research. This represents the current drilling depth. The deceleration buffer distance is determined by field borehole calibration tests combined with the screw braking hysteresis distance, where when At that time, the control system sets the feed rate to 0.
[0006] The quantitative fertilization control module calculates fertilization control parameters based on the target fertilization amount and predicted fertilization amount. It controls the fertilization screw motor to drive the sliding plate, adjusting the effective opening area and opening time between the sliding plate's circular hole and the fertilizer storage tank's outlet, thus achieving precise quantitative fertilization of solid or liquid fertilizer. The following formula is used for calculation: + ; ; In the formula: To initially set the opening area, To control the amount of correction, It is the maximum effective opening area when the round hole of the sliding plate and the discharge port of the fertilizer storage box are completely overlapped; The control system calculates the control correction amount based on the magnitude of the error: ; In the formula: This is the proportionality coefficient; The integral coefficient; These are the differential coefficients; This represents the cumulative error in fertilizer application from the start of fertilizer application to the current time t. This indicates the rate of change in fertilizer application error; ; ; In the formula: The error in fertilizer application at time t; Target fertilizer application rate; The cumulative predicted fertilizer discharge from the start of fertilizer discharge to time t is expressed in grams. For integration time; for The material layer height is constantly measured by an ultrasonic level sensor; g is the acceleration due to gravity; k is the flow correction coefficient; for The real-time effective opening area between the circular hole of the sliding plate and the discharge port of the fertilizer storage box is fed back in real time by displacement sensor II, with the unit being mm. 2 ; The fertilizer storage tank has a circular discharge port at the bottom, and a circular through-hole is also formed on the sliding baffle. By sliding the baffle left and right, the size of the overlapping area of the two holes is changed, thereby controlling the fertilizer outflow area. The baffle is controlled to slide gradually from fully closed to fully open. At regular intervals, the overlapping area of the two holes is recorded and compiled into a table showing the displacement value and corresponding opening area, which is stored in the controller. During operation, the displacement sensor detects the movement distance of the baffle in real time, and the controller directly retrieves the table to obtain the current actual fertilizer flow area.
[0007] The opening time is calculated using the following formula. : ; ; In the formula: Let t be the fertilizer discharge flow rate; This represents the remaining amount of fertilizer to be discharged, and ; t represents the material layer height measured by the ultrasonic level sensor at time t; g represents the gravitational acceleration; and k represents the flow correction coefficient. The effective opening area between the sliding plate hole and the fertilizer storage box outlet at time t is the real-time feedback value, expressed in mm, provided by displacement sensor II. 2 .
[0008] The navigation module performs path planning based on the fertilization point coordinates, robot operating status information, and environmental information determined by the above modules. It generates a navigation trajectory by combining the obstacle detection results of the lidar and controls the tracked walking mechanism to achieve autonomous navigation, path tracking, obstacle avoidance, and precise target point positioning. The communication and storage module is used to realize data communication between the host computer and the slave computer, record the robot's operating status, navigation trajectory, target fertilization point coordinates, drilling depth, fertilizer discharge volume and fault information, and support operation data storage, parameter configuration and remote monitoring.
[0009] The Arduino Mega2560 motion controller is connected to the lead screw lifting motor, drill bit rotary motor, fertilizer discharge lead screw motor, mixer, and tracked walking mechanism to execute motion control commands issued by the host computer.
[0010] The tracked walking mechanism of this invention is a rubber tracked walking mechanism. The rubber tracks have a large ground contact area and a low ground contact specific voltage, which can effectively reduce the risk of the robot sinking into soft, muddy surfaces and improve its ability to traverse complex terrain and its driving stability. The chassis frame supports functional components such as the fertilizer storage tank, the drilling and fertilization mechanism, and the control system, and ensures that the entire machine has sufficient structural strength and load-bearing capacity. The tracked walking mechanism adopts an independent left and right drive mode. By adjusting the rotation speed of the two side tracks, the robot can move forward, backward, turn, and turn in place to meet the mobile operation needs in the narrow rows and complex terrain environment of rubber plantations.
[0011] A binocular depth camera is mounted on the front of the robot to identify the location of rubber tree trunks, canopy boundaries, and target fertilization areas; a lidar is mounted on the top of the robot's upper support frame for environmental perception, map building, and obstacle detection; a wheel speed odometer is used to acquire information about the robot's motion status; and a tilt sensor is used to detect the robot's posture in real time to ensure stability when working on slopes. The lidar is mounted at a higher height than the binocular depth camera, which expands the environmental scanning range, reduces the obstruction of laser signals by plant branches and leaves and the robot's structure, and improves the robot's mapping accuracy, obstacle detection capability, and navigation stability in the rubber plantation environment.
[0012] By fusing information from multiple sources such as depth vision, lidar, and wheel speed odometer, the robot can achieve autonomous navigation between rows of rubber trees, obstacle avoidance, and precise positioning of target fertilization points. This enables the robot to automatically reach the pre-fertilization area around the trunk and complete drilling, fertilization, and soil covering operations.
[0013] When using the above device: S1: The system is powered on, and the control unit completes the initialization of each sensor, establishing an environmental perception and positioning system based on a binocular depth camera, LiDAR, and wheel speed odometer. S2: The robot travels along the rows of rubber trees, acquiring image information from a binocular depth camera; the lidar acquires point cloud data of the rubber trees and other obstacles, and uses the acquired point cloud data to identify obstacles; the acquired images and point cloud data are used to identify the rubber trees and their locations, the boundaries of the rubber tree canopy, and the target fertilization area, converting the rubber tree coordinates into three-dimensional coordinates in the robot's coordinate system, and determining the coordinates of the fertilization point based on the rubber tree coordinates, the canopy coverage area, and the distance to the pre-fertilization facility; S3: The robot moves to the vicinity of the fertilization point and selects either solid fertilizer fertilization mode or liquid fertilizer fertilization mode according to the type of fertilizer. S4: When using solid fertilizer application mode, drilling is performed. During the operation, the screw lifting motor drives the auger drill bit downward, and the drill bit rotation drive motor drives the auger drill bit to rotate, thus drilling the soil. After the drilling reaches the target depth, the screw lifting motor is controlled to reverse, raising the auger drill bit a preset distance, causing the auger drill bit to disengage from the hole, providing fertilization space for the fertilizer drop pipe. At the same time, the agitator is started (to break down the structure between fertilizer particles, improve fertilizer flowability, and prevent clogging). The fertilizer discharge screw motor is started to drive the slide plate to move and open the effective opening area and opening time of the slide plate's round hole and the fertilizer storage box's outlet. The fertilizer falls directly into the bottom of the borehole through the fertilizer drop pipe, achieving quantitative fertilization. S5: When using liquid fertilizer application mode, the fertilizer is applied directly to the surface of the target fertilization area through the fertilizer drop tube to achieve quantitative fertilization; after fertilization, it directly proceeds to the next fertilization point. S6: In solid fertilizer mode, the soil covering operation is performed. After the fertilizer is discharged, the sliding plate is driven to move and close the fertilizer storage box outlet. The drilling mechanism is controlled to be raised to a safe height and the robot is driven to continue to move forward. The soil covering plate backfills the fertilizer hole with soil during the forward movement. S7: After the soil covering is completed, control the drilling mechanism to fully reset, and the robot continues to move to the next fertilization point to repeat steps S2 to S6.
[0014] The angle of inclination of the fertilizer dripping tube is greater than the natural angle of repose of the fertilizer, and its inner wall adopts a smooth structure design to ensure that the fertilizer can slide smoothly into the fertilizer hole, reducing fertilizer retention and blockage.
[0015] The beneficial effects of this invention are: (1) Strong forest passability and environmental adaptability: The tracked walking mechanism has a large ground contact area and low ground contact pressure, which can drive stably in complex rubber plantation environments such as slopes, narrow lanes and mud. It has good passability and operational adaptability and can meet the needs of forest fertilization operations. (2) Drilling-fertilizing-covering integrated operation: After the auger drill bit completes the drilling, the drill bit is raised to a preset distance, and the fertilizer falls directly into the bottom of the fertilization hole through the fertilizer drop pipe to achieve precise fixed-point fertilization; after the fertilization is completed, the covering hopper backfills the loose soil around the fertilization hole into the fertilization hole to achieve automatic covering, reduce fertilizer exposure and nutrient loss, and improve fertilizer utilization rate. (3) Precise and consistent drilling depth: Based on the real-time adaptive feed control strategy of soil resistance, the drilling feed speed is dynamically adjusted according to the change of drill bit load. The drilling depth error of the auger bit can be controlled within ±1 cm, while effectively reducing drill bit wear and motor load and improving equipment service life. (4) Anti-bridging and high-precision quantitative fertilization: The mixer and the screw sliding table quantitative fertilizer dispensing mechanism work together to effectively prevent fertilizer bridging, blockage and flow interruption. The fertilizer dispensing volume control error is small, which can realize the precise control of fertilizer amount per plant, improve the uniformity of fertilization and fertilizer utilization. (5) High level of intelligence: It integrates multi-source perception information such as binocular depth camera, lidar and wheel speed odometer to realize functions such as autonomous navigation, tree recognition, target fertilization point positioning, borehole depth determination, quantitative fertilization and automatic soil covering. It can realize unmanned or remote control operation, greatly reduce labor intensity and improve operation efficiency. (6) It is compatible with both solid fertilizer and liquid fertilizer application modes, and the application method can be flexibly selected according to different agronomic requirements, thereby improving the applicability of the equipment and the flexibility of operation. Attached Figure Description
[0016] Figure 1 This is a schematic diagram of the rubber tree fixed-point fertilization robot of the present invention; Figure 2 This is a front view structural diagram of the robot of the present invention; Figure 3 This is a schematic diagram of the drill bit rotary motor section. Figure 4 This is a schematic diagram of a partial structure of the fertilizer discharge section; Figure 5 This is a schematic diagram of the stirrer section. In the diagram: 1-Crawler-type walking mechanism; 2-Chassis frame; 3-Tilt sensor; 4-Depth camera; 5-LiDAR; 6-Fertilizer storage tank cover; 7-Fertilizer storage tank; 8-Agitator; 9-Base plate; 10-Fertilizer discharge screw motor; 11-Slide plate; 12-Slide plate round hole; 13-Fertilizer discharge inclined pipe; 14-Soil covering plate; 15-Auger drill bit; 16-Drill bit rotary motor; 17-Screw lifting motor; 18-Lifting guide rail; 19-Displacement sensor I; 20-Moving plate. Detailed Implementation
[0017] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment takes the implementation of single-plant fixed-point fertilization in a rubber plantation as an example, but the application of the present invention is not limited to this.
[0018] Example 1: As Figure 1-5 As shown, the rubber tree fixed-point fertilization robot includes a tracked walking mechanism 1, a chassis frame 2, an tilt sensor 3, a binocular depth camera 4, a lidar 5, a fertilizer storage tank 7, a stirrer 8, a base plate 9, a fertilizer discharge screw motor 10, a sliding plate 11, a sliding plate round hole 12, a fertilizer discharge inclined pipe 13, a soil covering plate 14, a spiral drill bit 15, a drill bit rotary motor 16, a screw lifting motor 17, a lifting guide rail 18, and a displacement sensor 19. The tracked walking mechanism 1 is installed on both sides of the chassis frame 2. The fertilizer storage tank 7 is installed on the chassis frame 2, and the fertilizer storage tank cover 6 is located at the top of the fertilizer storage tank 7. An ultrasonic level sensor is installed inside the fertilizer storage tank, and a fertilizer filling port is opened on the fertilizer storage tank cover 6 for adding fertilizer to the fertilizer storage tank 7. One end of the agitator 8 is fixed to the fertilizer storage tank cover 6, and the other end is located inside the fertilizer storage tank 7. The base plate 9 is fixed to the middle of the front end of the chassis frame 2, and the screw lifting motor 17 is fixed to one side of the upper part of the base plate 9. Two lifting guide rails 18 are longitudinally fixed on the base plate 9. A movable plate 20 is movably mounted on the two lifting guide rails 18, and the top of the movable plate 20 is fixed to the output screw of the screw lifting motor 17. Under the action of the screw lifting motor, the movable plate 20 moves up and down along the lifting guide rails 18. A displacement sensor I 19 is mounted on the base plate 9. A drill bit rotary motor 16 is fixed on the movable plate 20, and a spiral drill bit 15 is fixed on the output shaft of the drill bit rotary motor 16. A binocular depth camera 4 is mounted on the top of the base plate 9. A lidar 5 is mounted on the chassis frame 2 via a support frame and is located on one side of the base plate. A tilt sensor 3 is mounted on the chassis frame 2. A soil cover plate 14 is fixed to the lower part of the chassis frame 2. Located behind the auger bit 15; one end of the fertilizer discharge pipe 13 is located below the outlet of the fertilizer storage tank 7, and the other end is located on one side of the auger bit 15; the fertilizer discharge screw motor 10 is mounted on the chassis frame 2 via a bracket and is located behind the fertilizer discharge pipe 13. The screw at the output end of the fertilizer discharge screw motor is fixedly connected to one end of the slide plate 11. The slide plate 11 is located between the outlet of the fertilizer storage tank 7 and the upper inlet of the fertilizer discharge pipe 13. The other end of the slide plate 11 has a slide plate hole 12. Through the action of the fertilizer discharge screw motor 10, the outlet of the fertilizer storage tank 7 is connected to or closed with the inlet of the fertilizer discharge pipe 13 via the slide plate, realizing quantitative fertilizer discharge. The displacement sensor II is installed on the right side of the slide plate 11; the battery pack and the electrical control box are installed inside the chassis frame. Hall current sensors are installed on the power supply circuits of the drill bit rotary motor 16 and the screw lifting motor; wheel speed odometers are installed on the tracked walking mechanism 1 for vehicle positioning and speed feedback. The electrical control box integrates the whole machine control unit. The upper-level computing unit carries a program developed based on the ROS Noetic distributed communication framework, including an environmental perception module, a navigation module, a drilling control module, a quantitative fertilization control module, a rubber tree identification and fertilization point determination module, and a communication and storage module. Each module completes data interaction and collaborative decision-making through ROS topics. The lower level uses an Arduino Mega2560 motion controller as the execution unit. (1) Environmental perception module, used to collect point cloud data of rubber trees and other obstacles obtained by lidar, collect image information obtained by binocular camera; identify obstacles using point cloud data obtained by lidar; collect robot running status information obtained by wheel speed odometer and tilt sensor; During operation in rubber plantations, the robot may roll or pitch due to factors such as uneven ground slopes, depressions, and loose soil. To ensure the robot's stability and drilling and fertilization accuracy, the system uses tilt sensors to monitor the robot's attitude in real time and adjusts its speed and movement accordingly.
[0019] Tilt sensor 3 acquires the robot's roll angle at a sampling frequency of 100 Hz. and pitch angle To reduce the impact of measurement noise on system control, a moving average filtering method is used to process the attitude data:
[0020] In the formula: The angle after filtering; is the k-th sample value; N is the length of the filter window.
[0021] The control unit calculates the vehicle's attitude angle in real time, and when the roll angle... or pitch angle When the pitch exceeds the preset safety threshold θ_max, the robot's speed will be automatically reduced. ; Where: v is the current driving speed; v_max is the maximum driving speed; θ is the current attitude angle; θ_max is the maximum allowable safe tilt angle.
[0022] When the tilt angle of the vehicle body continuously exceeds the safety threshold, the control system stops the robot's movement and issues an alarm signal. The robot will continue to operate after its posture returns to the safe range, so as to ensure the safety of robot operation and the accuracy of drilling, fertilization and soil covering operations.
[0023] (2) Rubber tree identification and fertilization point determination module: Based on the images and point cloud data obtained by the environmental perception module, identify the rubber tree and its location, the boundary of the rubber tree canopy, and the target fertilization area, convert the rubber tree coordinates into three-dimensional coordinates in the robot coordinate system, and determine the fertilization point coordinates based on the rubber tree trunk coordinates, the canopy coverage area and the pre-fertilization distance. During the robot's movement, a binocular depth camera captures real-time images of the rubber plantation ahead and inputs these images into a rubber tree recognition model built on the YOLOv8 target detection network. The YOLOv8 model is trained using a large number of pre-collected rubber plantation images and uses the rubber tree trunk as the detection target. It can quickly identify rubber trees in complex lighting conditions, with weeds obscuring the trees and dense tree environments.
[0024] After the image is input into the model, image features are first extracted through the backbone network, then multi-scale feature maps are generated through the feature fusion network, and finally the detection head outputs the location and confidence information of the rubber tree trunk target. The detection result is represented as follows:
[0025] in: The target bounding box is the tree trunk. The coordinates of the detection frame center; and These are the width and height of the detection frame, respectively.
[0026] The control unit filters target tree trunks based on detection confidence levels and acquires depth information of the corresponding areas using a binocular depth camera. It then calculates the distance D between the tree trunk center and the robot using binocular ranging principles. Combining this with camera calibration parameters, the image coordinates of the rubber tree trunk are converted into three-dimensional coordinates in the robot's coordinate system.
[0027] in: , , These are the spatial coordinates of the tree trunk center in the robot coordinate system.
[0028] The control system then calculates the coordinates of the target fertilization point based on the trunk center location, canopy coverage, and distance from the pre-installed fertilization facility. ; in, The three-dimensional coordinates of the target fertilization point. The three-dimensional coordinates of the rubber tree trunk in the robot's coordinate system. This is the offset distance for fertilization, with a value of 0.4-0.5m. Let be a unit vector pointing from the tree trunk to the row spacing, and satisfy . .
[0029] (3) Drilling control module: According to the target drilling depth, the screw lifting motor drives the auger to complete the lifting motion, and at the same time controls the drill bit rotation motor to drive the auger to rotate and drill. Based on the drilling depth fed back by displacement sensor I and the current data of the drill bit rotation motor collected by Hall current sensor, the soil resistance value and drilling feed speed are calculated, and the drilling feed speed of the drill bit rotation motor is adjusted to achieve adaptive fixed-depth drilling. After drilling is completed, the auger is raised by a preset distance to provide fertilization space for quantitative fertilization. The drilling feed rate is calculated using the following formula: ; In the formula: This refers to the drilling feed rate; This is the maximum allowable feed rate for drilling; The critical value for soil resistance is obtained through field pre-calibration and represents the upper limit of resistance to prevent the drill motor from stalling and overloading. To collect and filter soil resistance data in real time, when hour, .
[0030] in, N is the length of the filter window, i = 0, 1, 2, ..., N-1; To control the cycle; The soil resistance is the soil resistance during the i-th control cycle prior to the current moment; ; ; In the formula: This is the rated current of the drill bit rotary motor. For maximum thrust, , These are the weighting coefficients. ,and This refers to the current of the drill bit's rotating motor; This refers to the feed thrust of the lead screw jack motor; This refers to the current of the lead screw jack motor; This refers to the rated current of the screw jack motor. The mechanical efficiency of the lead screw drive is taken as 0.8-0.95; T is the output torque of the motor; P is the distance the nut moves forward when the motor rotates one revolution. To avoid vibration of the drilling mechanism caused by sudden changes in feed rate, the speed command is subjected to first-order smoothing filtering in each control cycle:
[0031] in: The current smoothed feed rate; This represents the current feed rate of the borehole. The feed rate is the smooth output of the previous control cycle; To control the cycle; Let be the speed smoothing coefficient, satisfying 0. ; When the actual drilling depth approaches the target drilling depth, a deceleration control phase is initiated to prevent the drill bit from impacting the bottom of the hole.
[0032]
[0033] in This represents the final feed rate at the current moment. The smoothed feed rate; The target drilling depth was set at 30cm based on research. This represents the current drilling depth. The deceleration buffer distance is determined by field borehole calibration tests combined with the screw braking hysteresis distance, where when At that time, the control system sets the feed rate to 0.
[0034] (4) Quantitative fertilization control module: Calculates fertilization control parameters based on the target fertilization amount and predicted fertilization amount; controls the fertilization screw motor to drive the sliding plate to adjust the effective opening area and opening time of the sliding plate's circular hole and the fertilizer storage tank's outlet, thereby achieving precise quantitative fertilization of solid or liquid fertilizer; the effective opening area of the sliding plate's circular hole and the fertilizer storage tank's outlet... The following formula is used for calculation: +
[0035]
[0036] In the formula: To initially set the opening area, To control the amount of correction, It is the maximum effective opening area when the round hole of the sliding plate and the discharge port of the fertilizer storage box are completely overlapped; The control system calculates the control correction amount based on the magnitude of the error: ; In the formula: This is the proportionality coefficient; The integral coefficient; These are the differential coefficients; This represents the cumulative error in fertilizer application from the start of fertilizer application to the current time t. This indicates the rate of change in fertilizer application error; ; ; In the formula: The error in fertilizer application at time t; Target fertilizer application rate; The cumulative predicted fertilizer discharge from the start of fertilizer discharge to time t is expressed in grams. For integration time; for The material layer height is constantly measured by an ultrasonic level sensor; g is the acceleration due to gravity; k is the flow correction coefficient; for The real-time effective opening area between the circular hole of the sliding plate and the discharge port of the fertilizer storage box is fed back in real time by displacement sensor II, with the unit being mm. 2 ; The opening time is calculated using the following formula. : ; ; In the formula: Let t be the fertilizer discharge flow rate; This represents the remaining amount of fertilizer to be discharged, and ; t represents the material layer height measured by the ultrasonic level sensor at time t; g represents the gravitational acceleration; and k represents the flow correction coefficient. The effective opening area between the sliding plate hole and the fertilizer storage box outlet at time t is the real-time feedback value, expressed in mm, provided by displacement sensor II. 2 .
[0037] To avoid nutrient deficiency affecting operations, the control unit calculates the amount of fertilizer applied per plant based on the number of plants already treated (L) and the target amount applied per plant. Estimate the amount of remaining fertilizer: ; Chinese: M remain Estimated amount of remaining fertilizer; M init Q represents the initial total amount of fertilizer loaded; L represents the number of rubber trees that have been fertilized; Q represents the total amount of fertilizer loaded initially. target This is the target fertilizer application amount per plant.
[0038] When the estimated remaining fertilizer amount is lower than the preset threshold M min When the time is right, the control unit issues a fertilization prompt and guides the robot back to the preset fertilization area to re-fertilize. After re-fertilization is completed, the robot automatically resumes its original task path and continues to perform the fertilization operation, achieving fully unmanned operation and dynamic scheduling.
[0039] (5) Navigation module: Based on the coordinates of the fertilization point, robot operating status information and environmental information determined by the above module, the path is planned, and the navigation trajectory is generated by combining the obstacle detection results of the laser radar. The tracked walking mechanism is controlled to realize autonomous navigation, path tracking, obstacle avoidance and precise target point positioning. To ensure navigation accuracy, the controller generates a navigation path based on the coordinates of the target fertilization point, and simultaneously integrates obstacle information provided by LiDAR 5 for path planning and obstacle avoidance control. The distance between the robot and the target fertilization point satisfies: At that time, it enters precise positioning mode; in( , ( ) represents the robot's current position coordinates; , ( ) represents the coordinates of the target fertilization point; The positioning threshold is 0.3m.
[0040] (6) Communication and storage module, used to realize data communication between the host computer and the slave computer, record robot running status, navigation trajectory, target fertilization point coordinates, drilling depth, fertilizer discharge volume and fault information, and support operation data storage, parameter configuration and remote monitoring.
[0041] The Arduino Mega2560 motion controller is connected to the lead screw lifting motor, drill bit rotary motor, fertilizer discharge lead screw motor, mixer, and tracked walking mechanism to execute motion control commands issued by the host computer.
[0042] When using the above device: S1: The system is powered on, and the control unit completes the initialization of each sensor, establishing an environmental perception and positioning system based on a binocular depth camera, LiDAR, and wheel speed odometer. S2: The robot travels along the rows of rubber trees, acquiring image information from a binocular depth camera; the lidar acquires point cloud data of the rubber trees and other obstacles, and uses the acquired point cloud data to identify obstacles; the acquired images and point cloud data are used to identify the rubber trees and their locations, the boundaries of the rubber tree canopy, and the target fertilization area, converting the rubber tree coordinates into three-dimensional coordinates in the robot's coordinate system, and determining the coordinates of the fertilization point based on the rubber tree coordinates, the canopy coverage area, and the distance to the pre-fertilization facility; S3: The robot moves to the vicinity of the fertilization point and selects either solid fertilizer fertilization mode or liquid fertilizer fertilization mode according to the type of fertilizer. S4: When using solid fertilizer application mode, drilling is performed. During the operation, the screw lifting motor drives the auger drill bit downward, and the drill bit rotation drive motor drives the auger drill bit to rotate, thus drilling the soil. After the drilling reaches the target depth, the screw lifting motor is controlled to reverse, raising the auger drill bit a preset distance, causing the auger drill bit to disengage from the hole, providing fertilization space for the fertilizer drop pipe. At the same time, the agitator is started (to break down the structure between fertilizer particles, improve fertilizer flowability, and prevent clogging). The fertilizer discharge screw motor is started to drive the slide plate to move and open the effective opening area and opening time of the slide plate's round hole and the fertilizer storage box's outlet. The fertilizer falls directly into the bottom of the borehole through the fertilizer drop pipe, achieving quantitative fertilization. S5: When using liquid fertilizer application mode, the fertilizer is applied directly to the surface of the target fertilization area through the fertilizer drop tube to achieve quantitative fertilization; after fertilization, it directly proceeds to the next fertilization point. S6: In solid fertilizer mode, the soil covering operation is performed. After the fertilizer is discharged, the sliding plate is driven to move and close the fertilizer storage box outlet. The drilling mechanism is controlled to be raised to a safe height and the robot is driven to continue to move forward. The soil covering plate backfills the fertilizer hole with soil during the forward movement. S7: After the soil covering is completed, control the drilling mechanism to fully reset, and the robot continues to move to the next fertilization point to repeat steps S2 to S6.
Claims
1. A robot for targeted fertilization of rubber trees, characterized in that: Includes tracked walking mechanism (1), chassis frame (2), tilt sensor (3), binocular depth camera (4), lidar (5), fertilizer storage tank (7), mixer (8), base plate (9), fertilizer discharge screw motor (10), slide plate (11), slide plate round hole (12), fertilizer discharge inclined pipe (13), soil covering plate (14), auger drill bit (15), drill bit rotary motor (16), screw lifting motor (17), lifting guide rail (18), displacement sensor (19); The tracked walking mechanism (1) is installed on both sides of the chassis frame (2), the fertilizer storage box (7) is installed on the chassis frame (2), the fertilizer storage box cover (6) is set on the top of the fertilizer storage box (7), the fertilizer storage box is equipped with an ultrasonic level sensor, and the fertilizer storage box cover (6) is provided with a fertilizer filling port for adding fertilizer to the fertilizer storage box (7); one end of the agitator (8) is fixed on the fertilizer storage box cover (6) and the other end is located inside the fertilizer storage box (7); the base plate (9) is fixed in the middle of the front end of the chassis frame (2), and the screw lifting motor (17) is fixed on one side of the upper part of the base plate (9). Two lifting guide rails (18) are longitudinally fixed on the base plate (9). A movable plate (20) is movably mounted on the two lifting guide rails (18), and the top of the movable plate (20) is fixed on the output screw of the screw lifting motor (17). Under the action of the screw lifting motor, the movable plate (20) moves up and down along the lifting guide rails (18). Displacement sensor I (19) is mounted on the base plate (9). The drill bit rotary motor (16) is fixed on the movable plate (20), and the auger drill bit (15) is fixed on the output shaft of the drill bit rotary motor (16). A binocular depth camera (4) is mounted on the top of the base plate (9). A laser radar (5) is mounted on the chassis frame (2) via a support frame and is located on one side of the base plate. An inclination sensor (3) is mounted on the chassis frame (2). A soil cover plate (14) is fixed on the chassis frame (9). 2) The lower part is located behind the auger bit (15); one end of the fertilizer discharge pipe (13) is located below the outlet of the fertilizer storage box (7), and the other end is located on the side of the auger bit (15); the fertilizer discharge screw motor (10) is mounted on the chassis frame (2) through the bracket and located behind the fertilizer discharge pipe (13). The screw at the output end of the fertilizer discharge screw motor is fixedly connected to one end of the slide plate (11). The slide plate (11) is located between the outlet of the fertilizer storage box (7) and the upper inlet of the fertilizer discharge pipe (13). The other end of the slide plate (11) has a slide plate round hole (12). Through the action of the fertilizer discharge screw motor (10), the outlet of the fertilizer storage box (7) is connected to or closed with the inlet of the fertilizer discharge pipe (13) through the slide plate to realize quantitative fertilizer discharge. The displacement sensor II is installed on the right side of the slide plate (11); the battery pack and the electrical control box are installed inside the chassis frame. Hall current sensors are installed on the power supply circuits of the drill bit rotary motor (16) and the screw lifting motor; wheel speed odometers are installed on the tracked walking mechanism (1) for vehicle positioning and speed feedback. The electrical control box integrates the whole machine control unit, which includes an environmental perception module, a navigation module, a drilling control module, a quantitative fertilization control module, a rubber tree identification and fertilization point determination module, and a communication and storage module. Each module completes data interaction and collaborative decision-making through ROS topics. The environmental perception module is used to collect point cloud data of rubber trees and other obstacles obtained by lidar, collect image information obtained by binocular camera; identify obstacles using point cloud data obtained by lidar; and collect robot running status information obtained by wheel speed odometer and tilt sensor (3). The rubber tree identification and fertilization point determination module identifies the rubber tree and its location, the canopy boundary of the rubber tree, and the target fertilization area based on the images and point cloud data obtained by the environmental perception module. It converts the rubber tree coordinates into three-dimensional coordinates in the robot coordinate system and determines the fertilization point coordinates based on the rubber tree coordinates, the canopy coverage area, and the pre-fertilization distance. The drilling control module controls the screw lifting motor (17) to drive the auger drill bit (15) to complete the lifting motion according to the target drilling depth. At the same time, it controls the drill bit rotating motor (16) to drive the auger drill bit to rotate and drill. Based on the drilling depth fed back by the displacement sensor I (19), and the current data of the drill bit rotating motor collected by the Hall current sensor, it calculates the soil resistance value and the drilling feed speed, adjusts the drilling feed speed of the drill bit rotating motor, and realizes adaptive fixed-depth drilling. After drilling is completed, the auger bit is raised a preset distance to provide space for quantitative fertilization. The quantitative fertilization control module calculates the fertilization control parameters based on the target fertilization amount and the predicted fertilization amount, controls the fertilization screw motor (10) to drive the slide plate (11) to adjust the effective opening area and opening time of the slide plate hole (12) and the discharge port of the fertilizer storage box (7), so as to realize the precise quantitative fertilization of solid fertilizer or liquid fertilizer. The navigation module performs path planning based on the fertilization point coordinates, robot operating status information, and environmental information determined by the above modules. It generates a navigation trajectory by combining the obstacle detection results of the lidar and controls the tracked walking mechanism to achieve autonomous navigation, path tracking, obstacle avoidance, and precise target point positioning. The communication and storage module is used to realize data communication between the host computer and the slave computer, record the robot's operating status, navigation trajectory, target fertilization point coordinates, drilling depth, fertilizer discharge volume and fault information, and support operation data storage, parameter configuration and remote monitoring. The Arduino Mega2560 motion controller is connected to the lead screw lifting motor (17), the drill bit rotary motor (16), the fertilizer discharge lead screw motor (10), the mixer (8), and the tracked walking mechanism (1) to execute motion control commands issued by the host computer.
2. The rubber tree fixed-point fertilization robot according to claim 1, characterized in that, The drilling feed rate is calculated using the following formula: ; In the formula: This refers to the drilling feed rate; This is the maximum allowable feed rate for drilling; The critical value for soil resistance is obtained through field pre-calibration and represents the upper limit of resistance to prevent the drill motor from stalling and overloading. To collect and filter soil resistance data in real time; when hour, ; in, ; In the formula, N is the length of the filtering window, i = 0, 1, 2, ..., N-1; To control the cycle; The soil resistance is the soil resistance during the i-th control cycle prior to the current moment; ; ; In the formula: This is the rated current of the drill bit rotary motor. For maximum thrust, , These are the weighting coefficients. ,and This refers to the current of the drill bit's rotating motor; This refers to the feed thrust of the lead screw jack motor; This refers to the current of the lead screw jack motor; This refers to the rated current of the screw jack motor. The mechanical efficiency of the lead screw drive is taken as 0.8-0.95; T is the output torque of the motor; P is the distance the nut moves forward when the motor rotates one revolution. To avoid vibration of the drilling mechanism caused by sudden changes in feed rate, the speed command is smoothed in each control cycle: ; in: The current smoothed feed rate; This represents the current feed rate of the borehole. The feed rate is the smooth output of the previous control cycle; To control the cycle; Let be the speed smoothing coefficient, satisfying 0. ; When the actual drilling depth approaches the target drilling depth, a deceleration control phase is initiated to prevent the drill bit from impacting the bottom of the hole. ; ; in This refers to the final feed rate; The smoothed feed rate; The target drilling depth is set to 30cm. This represents the current drilling depth. The deceleration buffer distance is determined by field borehole calibration tests combined with the screw braking hysteresis distance, where when At that time, the control system sets the feed rate to 0.
3. The rubber tree fixed-point fertilization robot according to claim 1, characterized in that, The effective opening area of the circular hole (12) of the sliding plate and the discharge port of the fertilizer storage box (7) The following formula is used for calculation: + ; ; In the formula: To initially set the opening area, To control the amount of correction, This is the maximum effective opening area when the circular hole of the sliding plate and the discharge port of the fertilizer storage box are completely overlapped. ; In the formula: This is the proportionality coefficient; The integral coefficient; These are the differential coefficients; This represents the cumulative error in fertilizer application from the start of fertilizer application to the current time t. This indicates the rate of change in fertilizer application error; ; ; In the formula: The error in fertilizer application at time t; Target fertilizer application rate; The cumulative predicted fertilizer discharge from the start of fertilizer discharge to time t is expressed in grams. For integration time; for The material layer height is constantly measured by an ultrasonic level sensor; g is the acceleration due to gravity; k is the flow correction coefficient; for The real-time effective opening area between the circular hole of the sliding plate and the discharge port of the fertilizer storage box is fed back in real time by displacement sensor II, with the unit being mm. 2 ; The opening time is calculated using the following formula. : ; ; In the formula: Let t be the fertilizer discharge flow rate; This represents the remaining amount of fertilizer to be discharged, and ; t represents the material layer height measured by the ultrasonic level sensor at time t; g represents the gravitational acceleration; and k represents the flow correction coefficient. The effective opening area between the sliding plate hole and the fertilizer storage box outlet at time t is the real-time feedback value, expressed in mm, provided by displacement sensor II. 2 .
4. The rubber tree fixed-point fertilization robot according to claim 1, characterized in that, The coordinates of the fertilization point are calculated using the following formula: ; in, The coordinates of the target fertilization point, The three-dimensional coordinates of the rubber tree trunk in the robot's coordinate system. This is the offset distance for fertilization, with a value of 0.4-0.5m. Let be a unit vector pointing from the tree trunk to the row spacing, and satisfy . .
