Full-hydraulic crawler-type orchard plant protection robot and control method thereof

Abstract

The invention discloses a full-hydraulic crawler-type orchard plant protection robot and a control method thereof. The robot comprises a crawler chassis system, a plant protection device, a power system, a hydraulic transmission system and a control system, and the crawler chassis system is used for walking; the plant protection device is used for protecting plants; the power system is used for providing power for the crawler chassis system and the plant protection device; the hydraulic transmission system is used for transmitting power of the power system to the crawler chassis system and the plant protection device; and the control system is used for monitoring the environment, controlling the hydraulic transmission system, driving the crawler chassis system to walk according to a planned path and driving the plant protection device to perform plant protection according to a preset target. By using the robot for operation, the labor intensity of agricultural operation personnel is greatly reduced, effective separation of people and pesticide is realized, and the body health of the operation personnel is guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of agricultural machinery, and more specifically, to a fully hydraulic crawler type orchard plant protection robot and a control method thereof. Background Art

[0002] With the continuous improvement of social science and technology, the mechanization level of orchard operations has improved, but it still faces many challenges, especially in the spraying process. Although some of the orchard spraying machinery on the market has achieved human-driven or remote-controlled operation, which can reduce the intensity of work, the problem of difficulty in separating people and pesticides is still prominent. On the one hand, operators are exposed to pesticide environments for a long time, facing health risks such as skin absorption and respiratory inhalation; on the other hand, problems such as accidental spraying and excessive use of pesticides are also difficult to avoid, posing a potential threat to the environment and ecology.

[0003] The above shortcomings need to be improved. Summary of the invention

[0004] In order to solve or alleviate the problem in the prior art that it is difficult to separate robots and medicines in orchard operations, which affects the health of operators, the present invention provides a fully hydraulic crawler orchard plant protection robot and a control method thereof.

[0005] The technical solution of the present invention is as follows:

[0006] A fully hydraulic crawler orchard plant protection robot comprises a crawler chassis system, a plant protection device, a power system, a hydraulic transmission system, and a control system, wherein the crawler chassis system is used for walking; the plant protection device is used for plant protection; the power system is used for providing power to the crawler chassis system and the plant protection device; the hydraulic transmission system is used for transmitting the power of the power system to the crawler chassis system and the plant protection device; the control system is used for monitoring the environment and controlling the hydraulic transmission system, driving the crawler chassis system to walk along a planned path, and driving the plant protection device to perform plant protection according to a preset target.

[0007] A control method for a fully hydraulic crawler type orchard plant protection robot is applied to the above-mentioned fully hydraulic crawler type orchard plant protection robot, comprising the following steps:

[0008] S1. Dot recording: The staff uses a handheld dot recorder to dot in the work area, and the control system records the GPS-RTK coordinates of each key point;

[0009] S2, trajectory planning: the control system generates the robot's travel route through the GPS-RTK coordinate data recorded by the dotting device;

[0010] S3, robot navigation: the robot receives real-time location information through GPS-RTK and navigates according to the pre-set route;

[0011] S4. Execute tasks: During the navigation process, the robot performs specific operations at the location of each trajectory line according to the set work tasks.

[0012] The present invention according to the above scheme has the following beneficial effects:

[0013] 1. The present invention greatly reduces the labor intensity of agricultural workers by using robots, realizes the effective separation of people and medicines, and ensures the health of workers. Secondly, the crawler chassis system has excellent terrain adaptability, which enables the robot to flexibly shuttle in the complex and changeable orchard environment, ensuring full coverage and efficient completion of plant protection operations. In addition, the application of the hydraulic transmission system improves the efficiency and accuracy of power transmission, allowing the robot to walk and operate flexibly according to work requirements. In addition, through the intelligent control system, the robot can realize real-time monitoring and response to environmental changes, effectively avoiding the accidental spraying and excessive use of pesticides, and reducing pollution to the environment.

[0014] 2. The present invention uses an engine to drive a hydraulic pump to provide energy for the travel system and the operating system, thereby improving the wire control characteristics of the system. At the same time, the hydraulic pressure is not subject to the spatial constraints of the mechanical layout, thereby improving the overall flexibility of the solution.

[0015] 3. The present invention performs navigation through GPS-RTK or vision and laser sensors, provides multiple navigation solutions, and has good scene adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 It is a structural schematic diagram of the hydraulic transmission system of the present invention;

[0018] Figure 2 This is a schematic diagram of the first hydraulic travel control oil circuit structure of the present invention;

[0019] Figure 3 It is a schematic diagram of the structure of the hydraulic control oil circuit of the fan and the hydraulic control oil circuit of the diaphragm pump of the present invention;

[0020] Figure 4 It is a structural schematic diagram of the present invention;

[0021] Figure 5 A trajectory schematic diagram of the present invention;

[0022] Figure 6 It is a control flow chart of the control system of the present invention;

[0023] Figure 7 is a motion model diagram of the present invention;

[0024] Figure 8 A trajectory adjustment model diagram of the present invention;

[0025] Fig. 9 An orchard image taken by the perception module of the present invention;

[0026] Fig.10 A neural network framework for application of the present invention;

[0027] Fig.11 A navigation line model fitted by the control system of the present invention;

[0028] Fig.12 This is the process of controlling the obstacle judgment of the system of the present invention.

[0029] Among them, the reference numerals in the figure are: 1. crawler chassis system; 101. frame; 102. crawler wheel system; 103. reducer; 104. travel hydraulic motor; 2. plant protection device; 201. atomization mechanism; 202. air delivery mechanism; 203. diaphragm pump; 204. medicine box; 3. power system; 4. hydraulic transmission system; 401. power output oil circuit; 402. first hydraulic travel control oil circuit; 403. second hydraulic travel control oil circuit; 404. fan hydraulic control oil circuit; 405. diaphragm pump hydraulic control oil circuit; 4 06. Filter; 407. Variable triple pump; 408. First one-way valve; 409. Electronically controlled proportional reversing valve; 410. First overflow valve; 411. Second one-way valve; 412. First two-way overflow valve; 413. Second two-way overflow valve; 414. Third one-way valve; 415. Fourth one-way valve; 416. Second overflow valve; 417. First electronically controlled throttle valve; 418. Second electronically controlled throttle valve; 419. Fuel tank; 5. Control system; 501. Perception module; 502. Navigation controller; 503. Chassis controller. DETAILED DESCRIPTION

[0030] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0031] It should be noted that when a component is referred to as being "fixed" or "set" or "connected" to another component, it may be located directly or indirectly on the other component. The directions or positions indicated by the terms "upper", "lower", "left", "right", "front", "back", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc. are based on the directions or positions shown in the accompanying drawings and are only for the convenience of description and should not be construed as limitations on the present technical solution. The terms "first", "second", etc. are only used for the convenience of description and should not be construed as indicating or implying relative importance or implicitly indicating the number of technical features. "Multiple" means two or more, unless otherwise clearly and specifically defined. "Several" means one or more, unless otherwise clearly and specifically defined.

[0032] like Figure 4 As shown, a fully hydraulic crawler orchard plant protection robot described in one embodiment of the present invention includes a crawler chassis system 1, a plant protection device 2, a power system 3, a hydraulic transmission system 4, and a control system 5. The crawler chassis system 1 is used for walking; the plant protection device 2 is used for plant protection; the power system 3 is used to provide power to the crawler chassis system 1 and the plant protection device 2; the hydraulic transmission system 4 is used to transmit the power of the power system 3 to the crawler chassis system 1 and the plant protection device 2; the control system 5 is used to monitor the environment and control the hydraulic transmission system 4, drive the crawler chassis system 1 to walk according to the planned path, and drive the plant protection device 2 to perform plant protection according to the preset target.

[0033] When the robot is put into use, it first performs environmental perception and path planning through its control system 5. Using sensors such as GPS, laser radar and cameras, the robot can obtain the terrain, plant distribution and obstacle information of the orchard in real time. After obtaining this information, the control system 5 will automatically calculate the optimal walking path, and accurately transmit the power to the crawler chassis system 1 through the hydraulic transmission system 4, driving the robot to move forward steadily along the planned path. At the same time, according to the preset plant protection goals and operating parameters, the control system 5 will intelligently control the plant protection device 2, such as a sprayer, through the hydraulic transmission system 4 to ensure that the pesticide or fertilizer can be accurately and evenly applied to the target plant.

[0034] In this embodiment, the use of the robot greatly reduces the labor intensity of agricultural workers, realizes the effective separation of people and medicines, and ensures the health of workers. Secondly, the crawler chassis system 1 has excellent terrain adaptability, which enables the robot to flexibly shuttle in the complex and changeable orchard environment, ensuring full coverage and efficient completion of plant protection operations. In addition, the application of the hydraulic transmission system 4 improves the efficiency and accuracy of power transmission, allowing the robot to flexibly walk and apply according to work requirements. In addition, through the intelligent control system 5, the robot can realize real-time monitoring and response to environmental changes, effectively avoid the accidental spraying and excessive use of pesticides, and reduce pollution to the environment.

[0035] like Figure 1 , Figure 2 and Figure 4 As shown, in a preferred embodiment, the power system 3 includes an engine, the crawler chassis system 1 includes a frame 101, the engine is arranged on the frame 101, and crawler wheel trains 102 are respectively arranged on both sides of the frame 101. The crawler wheel trains 102 on each side are respectively connected to a reducer 103 and a travel hydraulic motor 104 in sequence, the reducer 103 and the travel hydraulic motor 104 are connected through a coupling, and the travel hydraulic motor 104 is connected to the hydraulic transmission system 4.

[0036] The track wheel system 102 includes a driving wheel, a road wheel, a tensioning wheel, and a towing wheel. The driving wheel is responsible for transmitting power, the road wheel is responsible for bearing weight, the tensioning wheel is responsible for tightening the track and maintaining a certain shape, and the towing wheel is responsible for supporting the track to prevent it from sagging. For the left track, the left travel hydraulic motor 104 is connected to the left driving wheel through the planetary reducer 103, and the input hydraulic oil power is used to drive the travel hydraulic motor 104, thereby driving the driving wheel and the track, so that the frame 101 moves. The right travel hydraulic motor 104 is on the other side of the chassis, and its transmission relationship is the same as that of the left travel hydraulic motor 104.

[0037] The travel hydraulic motor 104 is a variable piston motor, which converts hydraulic energy back into mechanical energy to drive the driving wheel of the crawler wheel system 102 to move.

[0038] like Figure 1 and Figure 2 As shown, in a preferred embodiment, the hydraulic transmission system 4 includes a power output oil circuit 401, a first hydraulic travel control oil circuit 402 and a second hydraulic travel control oil circuit 403, the power output oil circuit 401 is connected to the engine transmission, and the first hydraulic travel control oil circuit 402 and the second hydraulic travel control oil circuit 403 are respectively used to drive the crawler wheel systems 102 on both sides.

[0039] The power output oil circuit 401 includes an oil tank 419, a filter 406, and a variable triple pump 407 which are connected in sequence. The variable triple pump 407 is connected to the engine in a transmission manner. The first connection of the variable triple pump 407 is connected to the first hydraulic travel control oil circuit 402 to control the operation of the left track wheel system 102. The second connection of the variable triple pump 407 is connected to the second hydraulic travel control oil circuit 403 to control the operation of the right track wheel system 102. The first hydraulic travel control oil circuit 402 includes a first check valve 408 connected to the first connection of the variable triple pump 407. The first check valve 408 is connected to the electronically controlled proportional reversing valve 409 and the first overflow valve 410. The first overflow valve 410 is connected to the oil tank 419 through the filter 406. The electric control proportional reversing valve 409 is a three-position four-way valve, the first oil hole (T port) of the electric control proportional reversing valve 409 is connected to the first check valve 408, the fourth oil hole (P port) on the same side of the first oil hole is connected to the oil tank 419, the second oil hole (B port) is connected to the oil inlet hole of the second check valve 411, the oil inlet hole of the first two-way relief valve 412 and the oil drain hole of the second two-way relief valve 413, the oil outlet holes of the second check valve 411 and the first two-way relief valve 412 are connected to the oil inlet hole of the left travel hydraulic motor 104, the oil inlet hole and the oil outlet hole of the left travel hydraulic motor 104 are connected to the oil inlet hole of the second two-way relief valve 413 and the oil outlet hole of the third check valve 414, the oil outlet hole of the second two-way relief valve 413 and the oil inlet hole of the third check valve 414 are connected to the third oil hole (A port) of the electric control proportional reversing valve 409. Similarly, the second hydraulic travel control oil circuit 403 is set.

[0040] The hydraulic oil enters the electronically controlled proportional reversing valve 409 through the first one-way valve 408 , and the first one-way valve 408 prevents the hydraulic oil from flowing back to the variable triple pump 407 and causing impact on the variable triple pump 407 .

[0041] In the first hydraulic travel control oil circuit 402, the proportional reversing valve usually controls the direction of the fluid flow according to the change of the polarity (positive / negative) of the input signal, and controls the valve opening according to the change of the size of the input signal, thereby controlling the flow size. The size and flow direction of the oil entering the travel hydraulic motor 104 are adjusted, thereby adjusting the speed and direction of the left crawler track; under normal circumstances, the hydraulic oil flows from the first oil hole of the electronically controlled proportional reversing valve 409 to the second oil hole, and the oil enters the left travel hydraulic motor 104 through the second one-way valve 411, driving the crawler wheel system 102 to rotate, and the hydraulic oil torque is transmitted to the crawler wheel system 102 and then returns to the oil tank 419 through the second two-way overflow valve 413; when the crawler needs to rotate in the opposite direction, the electronically controlled proportional reversing valve 409 is controlled to operate, and the hydraulic oil flows from the first oil hole to the third oil hole, and the hydraulic oil enters the left travel hydraulic motor 104 through the one-way valve below, driving the entire wheel system to reverse, and the hydraulic oil torque is transmitted to the wheel system and then returns to the oil tank 419 through the upper overflow valve. Similarly, the second hydraulic travel control oil circuit 403 is controlled.

[0042] By controlling the travel speed and direction of the left and right track wheel systems 102, forward, backward and turning movements can be achieved. A two-way overflow brake valve is provided in the hydraulic system. When the valve core of the electronically controlled proportional reversing valve 409 is in the middle position, the hydraulic oil supplied by the variable triple pump 407 directly flows back to the oil tank 419 through the electronically controlled proportional reversing valve 409. The travel hydraulic motor 104 and the two-way overflow brake valve are in a locked state because there is no oil pressure supply and the overflow valve in the two-way overflow brake valve cannot be opened. The travel hydraulic motor 104 stops rotating to achieve the purpose of braking.

[0043] like Figure 6 As shown, the robot's travel modes include remote control mode and automatic driving mode.

[0044] In remote control mode, v and w are proportional to the travel of the forward and backward joystick and the steering joystick, respectively, that is:

[0045] v rc =k v d v

[0046] ω rc =k ω d ω

[0047] v rc and ω rc are the target linear velocity and target angular velocity in remote control mode, respectively, v and k ω are the proportional coefficients of the joystick, normalizing the joystick signal to the range of [0, 1], which is the calibration quantity; d v and d ω is the actual joystick travel.

[0048] The remote control is carried out through a 2.4G remote control. The 2.4G remote control transmits data through wireless signals. After the remote control and the receiver mounted on the robot are matched, they can transmit data to each other. The 2.4G remote control outputs digital signals, which can be encoded through different communication protocols, such as SBUS signals. The SBUS signal is a serial signal that can transmit 16 channels of data through a signal line. The receiver is connected to the serial port of the chassis controller 503 and parsed according to the protocol. The control amount of the joystick on the remote control will be collected by the remote control and converted into a digital signal, which is encoded and transmitted by the 2.4G wireless signal carrier through the SBUS protocol and transmitted to the receiver. The receiver and the chassis controller 503 are connected through the serial port, and the signal is transmitted to the chassis controller 503 through the serial port. The chassis controller 503 can parse the channel value of the remote control according to the SBUS protocol to obtain the joystick stroke.

[0049] In the autonomous driving mode, v and w are calculated, and w is calculated by tracking algorithms such as PID, pure pursuit, LQR, etc.

[0050] On a tracked differential robot, the kinematic model is as follows Figure 7 and Figure 8 shown.

[0051]

[0052]

[0053] In the Pure Pursuit (PP) algorithm, g x , g y is the next waypoint to be tracked, which is located on the planned path. Control the rear axle of the robot to pass through this waypoint. d represents the distance from the current position of the vehicle (i.e. the rear axle position) to the target waypoint, x = α, represents the angle between the current vehicle posture and the target waypoint, then according to the sine theorem, the following conversion formula can be derived:

[0054]

[0055] but:

[0056]

[0057] Assume that the current lateral error from the target is e y ,but:

[0058]

[0059] so:

[0060]

[0061] Combined with the kinematic model, we can know that:

[0062]

[0063] Usually, v auto The rules can be set, such as setting the robot to run at full speed in the straight section and reducing the speed to half of the full speed when turning; then, in any case, the lateral error between the current positioning point and the target position is calculated as e y , and the distance l from the vehicle's current position (i.e., rear axle position) to the target waypoint d , we can get the target angular velocity ω auto .

[0064] Calculate left and right wheel speeds:

[0065] According to the kinematic model of the above tracked differential robot:

[0066]

[0067]

[0068] When the target linear velocity v auto and the target angular velocity ω auto After obtaining through the above steps, calculate:

[0069]

[0070] but:

[0071]

[0072] q l and q r V is the required flow of the left and right hydraulic motors. m is the motor displacement, r l and r r is the radius of the driving wheel of the left and right crawler wheel system 102, i is the speed ratio of the reducer, η m It is the volumetric efficiency of the motor, usually around 95%.

[0073] The required flow of the left and right hydraulic motors is controlled by the opening of the electronically controlled reversing valve. The control is usually implemented by PID, that is, the deviation is calculated by the required flow and the actual flow (obtained by the flow sensor), and the adjustment amount of the valve opening is obtained by multiplying the deviation by the proportional, differential, and integral coefficients. The proportional, differential, and integral coefficients are usually obtained through calibration.

[0074] like Figure 1 , Figure 3 and Figure 4 As shown, in a preferred embodiment, the plant protection device 2 includes an atomization mechanism 201, and the atomization mechanism 201 includes a diaphragm pump 203. The diaphragm pump 203 is connected to the hydraulic transmission system 4 and the medicine box 204, and the diaphragm pump 203 is used to atomize and spray the liquid medicine in the medicine box 204.

[0075] The plant protection device 2 includes an air delivery mechanism 202, which includes a fan motor. The fan motor output shaft is connected to an impeller. The fan motor is connected to a hydraulic transmission system 4. The fan motor is used to drive the impeller to rotate so that the atomized liquid medicine is blown toward the target area.

[0076] The hydraulic transmission system 4 includes a diaphragm pump pressure control oil circuit 405 and a fan hydraulic control oil circuit 404. The diaphragm pump pressure control oil circuit 405 is used to control the operation of the diaphragm pump 203, and the fan hydraulic control oil circuit 404 is used to control the operation of the fan motor.

[0077] The third oil outlet of the variable triple pump 407 is connected to the oil inlet of the fourth check valve 415, and the oil outlet of the fourth check valve 415 is connected to the oil inlet of the second overflow valve 416. The diaphragm pump pressure control oil circuit 405 includes a first electronically controlled throttle valve 417, the oil inlet of the first electronically controlled throttle valve 417 is connected to the oil outlet of the fourth check valve 415, the oil outlet of the first electronically controlled throttle valve 417 is connected to the oil inlet of the diaphragm pump 203, and the oil outlet of the diaphragm pump 203 is connected to the oil inlet of the third overflow valve. The fan hydraulic control oil circuit 404 includes a second electronically controlled throttle valve 418, the oil inlet of the second electronically controlled throttle valve 418 is connected to the oil outlet of the fourth check valve 415, the oil outlet of the second electronically controlled throttle valve 418 is connected to the oil inlet of the fan motor, and the oil outlet of the fan motor is connected to the oil tank 419.

[0078] One of the three pumps is the power source of the plant protection device 2. In the diaphragm pump pressure control oil circuit 405, the hydraulic oil passes through the first electronically controlled throttle valve 417 and is supplied to the diaphragm pump 203 to drive the diaphragm pump 203 to rotate, so that the liquid medicine is atomized and sprayed out. In the fan drive oil circuit, the hydraulic oil passes through the second electronically controlled throttle valve 418 and is supplied to the fan motor to drive the fan to rotate. The atomized liquid medicine enters the impeller axially from the air inlet. After passing through the impeller, the liquid medicine is further fully atomized and flows faster. A guide plate can be provided so that the axially flowing liquid medicine flows to the left, right, and upper side under the action of the guide plate. During the robot operation (usually the ridge part of the global path), the plant protection device 2 is controlled to open by the chassis controller 503, and the fan motor rotates at high speed to fully atomize and accelerate the liquid medicine pumped by the diaphragm pump 203, so that the liquid medicine reaches the fruit trees; when the robot stops when encountering an obstacle or turns around at the head of the ridge, it is controlled to close by the chassis controller 503, and the hydraulic oil returns to the oil tank 419 through the second overflow valve 416, and the fan motor and diaphragm pump 203 lose power and automatically stop rotating, that is, the plant protection device 2 stops running. Stopping the plant protection device 2 can reduce the waste caused by spraying liquid medicine to non-target areas, and prevent excessive spraying of liquid medicine to affect the growth of fruit trees by staying in the same target area for too long.

[0079] like Figure 4 As shown, in a preferred embodiment, the control system 5 includes a perception module 501, a navigation controller 502, and a chassis controller 503. The perception module 501 is used to perceive the surrounding environment, and the perception module 501 includes one or more of GPS-RTK, a camera, and a lidar; the navigation controller 502 is used to collect data from the perception module 501 for navigation, path planning, and control command calculation; the chassis controller 503 is responsible for converting the control instructions of the navigation controller 502 into the opening of the solenoid valves of each control oil circuit of the hydraulic transmission system 4.

[0080] The robot uses GPS-RTK, laser radar, and visual sensors for integrated navigation. GPS-RTK can obtain differential positioning latitude and longitude data, satellite number, RTK solution status, etc. The visual sensor is installed in front of the vehicle, mainly used to collect actual operation images of the orchard, and used for visual navigation line recognition, so that the robot can perform relative positioning and navigation control within the ridge; the laser radar is installed in front of the vehicle or on the roof to collect point cloud data during the actual operation process, so as to detect obstacles near the vehicle body as input for local planning and obstacle avoidance during navigation.

[0081] When it is detected that the number of satellites searched by GPS-RTK is within a reasonable range (usually more than 25 satellites are needed) and RTK has a fixed solution (under normal conditions, a fixed solution means the positioning accuracy is between 1-3cm), the accuracy is high, and only GPS-RTK is needed for robot positioning and navigation. If the number of satellites searched by GPS-RTK is not within a reasonable range or RTK does not have a fixed solution, it usually means that the signal is poor or the robot antenna is blocked in the dense forest, then the robot automatically switches to the visual sensor or lidar for navigation through relative positioning.

[0082] like Figure 5 As shown, a control method of a fully hydraulic crawler type orchard plant protection robot in one embodiment of the present invention is applied to the above-mentioned fully hydraulic crawler type orchard plant protection robot, comprising the following steps:

[0083] S1. Dot recording: The staff uses a handheld dot recorder to dot in the work area, and the control system 5 records the GPS-RTK coordinates of each key point. These coordinates form the basis of the robot's trajectory, such as Figure 5 As shown in the figure, the thick dotted lines represent the trajectory that the robot will travel, which is recorded by a handheld dotting device;

[0084] S2, trajectory planning: through the GPS-RTK coordinate data recorded by the dotting device, the control system 5 generates the robot's route; the route is Figure 5 The robot moves in the work area according to these pre-recorded trajectories, as indicated by the thick dashed lines in the figure.

[0085] S3. Robot navigation: The robot receives real-time location information through GPS-RTK and navigates according to the pre-set route. The high-precision positioning information provided by GPS-RTK ensures that the robot can accurately follow the route and avoid deviating from the track. During the tracking process, when the laser radar detects an obstacle, the robot can plan the trajectory through algorithms such as A* to generate a local planning trajectory, such as Figure 5As shown by the thin dotted line, obstacles are avoided so that during the operation of the robot, the control system 5 can monitor the position and movement trajectory of the robot in real time and make dynamic adjustments when necessary to ensure the accuracy of the operation.

[0086] Among them, the obstacle recognition process includes: obtaining the surrounding environment point cloud information from the on-board sensor laser radar, and then using the obstacle detection algorithm to detect and obtain the x, y, z distance information and w / l length and width information of the obstacle. The planning algorithm adopts methods such as A* / RRT. The A* algorithm is a heuristic search algorithm. Starting from the starting point, first find the grid adjacent to the starting point to determine whether it is the best position. Based on this best grid, it expands outward to its adjacent grids to find the best position at this time. By approaching the target point step by step, blind searches are reduced, and feasibility and search efficiency are improved; RRT explores the map by constructing a random tree that grows from the starting point. When the random tree expands to the target area, a collision-free path from the starting point to the target area can be obtained by backtracking. It should be noted that only the trajectory planning algorithm is applied here, and this application does not improve the trajectory planning algorithm.

[0087] S4. Execute tasks: During navigation, the robot performs specific tasks at each trajectory line according to the set work tasks. The robot performs plant protection operations in the straight line segment, that is, spraying pesticides through wind-blown atomization equipment; when the robot reaches the turning area, it stops spraying the liquid and performs a turning action. In the figure, the robot is moving along the preset trajectory and performing tasks.

[0088] like Figures 9 to 11 As shown, in S3, the robot senses the surrounding environment through the perception module 501 and dynamically adjusts the route.

[0089] S301, data collection and training: The robot collects image information of the working environment under different environmental conditions through the camera. Specifically, the robot relies on the camera mounted on the machine to collect orchard images under different orchards, different weather and lighting conditions;

[0090] S302, data labeling and model training: manually label the work environment image information; generate navigation lines between orchard tree rows through deep learning models for navigation. This process requires labeling of orchard images. Fig. 9 As shown, during the annotation process, the annotation lines of the left and right tree rows are manually defined. The principle of selecting the annotation lines is to use the tree trunk and the ground contact point as the key point, and the line formed by connecting the key points as the annotation line; if the ground contact point cannot be determined during the annotation process, the edge pixel where the outermost edge of the tree shape intersects with the ground is used as the key point to form the annotation line. In this embodiment, the selected training model is the neural network framework UFLD that considers the global structure, such as Fig.10As shown in the figure, this is a lane detection model that is relatively familiar to technicians in this field. The network structure is divided into two layers, the upper layer is an auxiliary layer, which only provides segmentation loss in the training stage and is not included in the actual operation in the inference stage; the lower layer is the actual operation layer, which performs network prediction, performs convolution operations through the superposition of multiple Res modules, obtains highly extracted features and sends them to the fully connected layer, thereby obtaining structural loss and classification loss.

[0091] S303, real-time reasoning to obtain navigation line: During navigation, the navigation controller 502 obtains the image returned by the camera in real time and adjusts the route. The model UFLD is run in the intelligent navigation controller 502 to obtain the detection result, such as Fig.11 As shown, the detection result is the fitting straight line of two tree rows. Draw two parallel lines at different height layers, of which the first parallel line is the lower part of the image, and the second parallel line is the upper part of the image. The two parallel lines are taken at 9 / 10 and 1 / 3 height from top to bottom respectively. Get the two intersection points p1(x1,y1) and p2(x2,y2) of the first parallel line and the tree row fitting straight line, and the two intersection points p3(x3,y3) and p4(x4,y4) of the second parallel line and the tree row fitting straight line. Find the midpoint p_s of the line connecting p1 and p2 and the midpoint p_e of the line connecting p3 and p4. The line connecting the two midpoints is the navigation line obtained by the plane image.

[0092] S304, the navigation line is projected to the real vehicle coordinates and linear navigation control is performed. The navigation line is a pixel coordinate, which needs to be converted to the metric coordinate system of the vehicle coordinates to have navigation significance. Usually, the conversion process is performed through an inverse perspective projection transformation process, or the pixel coordinates are converted through other precise distance measurement sensors (such as lidar). This process is not the main protection scope of this application and is also a technology well known to those skilled in the art, so it will not be described again. According to the navigation line, the current relative deviation of the vehicle body can be obtained, so as to perform lateral control and keep the robot moving along the straight line navigation.

[0093] S305, navigation in the U-turn area. In the U-turn area, it is necessary to identify the end of the field, and methods such as end of field identification by markers, end of field identification by neural network training, and end of field identification based on physical priors are used. In this embodiment, the third method is adopted, that is, the left and right cameras are used to make a judgment in combination with the length of the straight line fitted by the tree row in the above steps. Specifically, when the length of the detected straight line fitted by the tree row gradually decreases to a certain set threshold, it is considered that the robot is approaching the end of the row, and the left and right cameras are started to identify the fruit trees on both sides. If it is determined that the fruit trees no longer appear in the camera within a certain time threshold (usually about 5s), it is considered that the robot has arrived at the U-turn area and starts automatic navigation in the U-turn area. The navigation actions followed by the U-turn area navigation are pre-set.

[0094] S306, re-perform straight segment navigation. After the U-turn is completed, the robot automatically returns to the straight segment navigation and repeats S303 to S305. It should be noted that the directions of navigation in the U-turn area are alternated, i.e., right turn-left turn-right turn-left turn-right turn, to ensure that the robot operates in a fixed direction.

[0095] like Fig.12 As shown, in S303, the robot senses the surrounding environment through the perception module 501 and avoids obstacles.

[0096] S3031. Data preprocessing: Obtain point cloud data from the laser radar, segment the ground based on the method of geometric description, distinguish the ground background and foreground, and cluster the segmented data; Specifically, data preprocessing obtains point cloud data through the laser radar and performs point cloud filtering. In order to speed up the processing of point clouds, downsampling operations are sometimes required. After the filtering operation is completed, the ground is segmented based on the method of geometric description to distinguish the ground background and foreground targets, avoiding the interference of ground data on subsequent clustering; then the segmented data is clustered using the Euclidean clustering method.

[0097] S3032, target posture estimation: perform posture estimation on the clustered data and fit a 3D bounding box. During posture estimation, project the data in each class onto the ground, and then determine the 2D bounding box using the minimum area enveloping rectangle method, and then determine the size of the 3D bounding box in combination with the elevation information.

[0098] S3033, obstacle determination: Set the robot's area of ​​interest. If the 3D bounding box falls into the area of ​​interest, it is determined to be an obstacle. For example, if the robot is 1.5m wide, the area of ​​interest is 5m in the front and 1m on each side, a total of 2m. When the 3D bounding box falls into the range of 5m×2m in the front, it is considered an obstacle.

[0099] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A fully hydraulic crawler orchard plant protection robot, characterized in that: include: A crawler chassis system, wherein the crawler chassis system is used for walking; A plant protection device, wherein the plant protection device is used for plant protection; A power system, the power system is used to provide power to the crawler chassis system and the plant protection device; A hydraulic transmission system, the hydraulic transmission system is used to transmit the power of the power system to the crawler chassis system and the plant protection device; A control system is used to monitor the environment and control the hydraulic transmission system, drive the crawler chassis system to move along a planned path, and drive the plant protection device to protect plants according to a preset target.

2. According to the fully hydraulic crawler orchard plant protection robot described in claim 1, it is characterized in that: The power system includes an engine, and the crawler chassis system includes a frame. The engine is arranged on the frame. Crawler wheel systems are respectively arranged on both sides of the frame. The crawler wheel systems on each side are respectively connected to a reducer and a travel hydraulic motor in sequence. The reducer and the travel hydraulic motor are connected through a coupling, and the travel hydraulic motor is connected to the hydraulic transmission system.

3. According to the fully hydraulic crawler orchard plant protection robot described in claim 1, it is characterized in that: The plant protection device comprises an atomizing mechanism, and the atomizing mechanism comprises a diaphragm pump. The diaphragm pump is connected to the hydraulic transmission system and the medicine box, and the diaphragm pump is used to atomize and spray the medicine liquid in the medicine box.

4. According to the fully hydraulic crawler orchard plant protection robot described in claim 1, it is characterized in that: The plant protection device includes an air delivery mechanism, which includes a fan motor. The fan motor output shaft is connected to an impeller. The fan motor is connected to the hydraulic transmission system. The fan motor is used to drive the impeller to rotate so that the atomized liquid medicine is blown toward the target area.

5. According to the fully hydraulic crawler orchard plant protection robot described in claim 1, it is characterized in that: The hydraulic transmission system includes a power output oil circuit, a first hydraulic travel control oil circuit and a second hydraulic travel control oil circuit. The power output oil circuit is connected to the engine transmission. The first hydraulic travel control oil circuit and the second hydraulic travel control oil circuit are respectively used to drive the crawler wheel systems on both sides.

6. According to the fully hydraulic crawler orchard plant protection robot described in claim 1, it is characterized in that: The hydraulic transmission system comprises a diaphragm pump hydraulic control oil circuit and a fan hydraulic control oil circuit. The diaphragm pump hydraulic control oil circuit is used to control the operation of the diaphragm pump, and the fan hydraulic control oil circuit is used to control the operation of the fan motor.

7. According to the fully hydraulic crawler orchard plant protection robot described in claim 1, it is characterized in that: The control system comprises: A perception module, which is used to perceive the surrounding environment. The perception module includes one or more of GPS-RTK, camera, and laser radar. A navigation controller, the navigation controller is used to collect data from the perception module for navigation, path planning and control instruction calculation; The chassis controller is responsible for converting the control instructions of the navigation controller into the opening of the solenoid valves of the hydraulic transmission system.

8. A control method for a fully hydraulic crawler orchard plant protection robot, characterized in that: A fully hydraulic crawler orchard plant protection robot applied to any one of claims 1 to 7 comprises the following steps: S1. Dot recording: The staff uses a handheld dot recorder to dot in the work area, and the control system records the GPS-RTK coordinates of each key point; S2, trajectory planning: the control system generates the robot's travel route through the GPS-RTK coordinate data recorded by the dotting device; S3, robot navigation: the robot receives real-time location information through GPS-RTK and navigates according to the pre-set route; S4. Execute tasks: During the navigation process, the robot performs specific operations at the location of each trajectory line according to the set work tasks.

9. The control method of a fully hydraulic crawler orchard plant protection robot according to claim 8, characterized in that: In S3, the robot senses the surrounding environment through the perception module and dynamically adjusts the route. S301, data collection and training: the robot collects working environment image information under different environmental conditions through a camera; S302, data labeling and model training: manually labeling the work environment image information; S303, real-time reasoning to obtain a navigation line: During navigation, the navigation controller obtains the image returned by the camera in real time and adjusts the route.

10. The control method of a fully hydraulic crawler orchard plant protection robot according to claim 8, characterized in that: In S303, the robot senses the surrounding environment through the perception module and avoids obstacles. S3031. Data preprocessing: Obtain point cloud data from the laser radar, segment the ground based on the method of geometric description, distinguish the ground background and foreground, and cluster the segmented data; S3032, target pose estimation: perform pose estimation on the clustered data and fit a 3D bounding box; S3033, obstacle determination: set the robot's area of ​​interest. If the 3D bounding box falls into the area of ​​interest, it is determined to be an obstacle.

Images