Ridge-crossing self-propelled broccoli selective harvesting robot and operation method thereof

Abstract

A ridge-crossing self-propelled broccoli selective harvesting robot comprises a harvesting robot moving platform, a harvesting mechanical arm, an end effector, a navigation control system, a harvesting control system and a power supply distribution system, and the harvesting robot moving platform is provided with the harvesting mechanical arm, the navigation control system, the harvesting control system and the power supply distribution system. The harvesting robot moving platform is provided with a controlled end, the harvesting mechanical arm is provided with an end effector, the navigation control system is connected with the controlled end of the harvesting robot moving platform, the harvesting control system is connected with the controlled end of the end effector, and the power supply distribution system is connected with the harvesting robot moving platform, the navigation control system and the harvesting control system. The invention further provides an operation method of the ridge-crossing self-propelled broccoli selective harvesting robot. Through visual identification, autonomous navigation and the like, broccoli balls with proper sizes can be automatically and selectively harvested without manual operation in the whole process, time and labor are saved, and safety and intelligence are achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of agricultural machinery, specifically involving image recognition, navigation and positioning, and intelligent control technologies, and particularly relates to a cross-ridge self-propelled broccoli selective harvesting robot and its operation method. Background Art

[0002] As a vegetable crop with relatively high economic value, the planting area of broccoli has been expanding year by year, and the market demand is continuously increasing. The traditional broccoli harvesting method mainly relies on manual operation, which not only has a large labor intensity and low efficiency, but also has problems such as inaccurate human judgment and fruit damage during the harvesting process, and it is difficult to meet the production requirements of high efficiency and high quality in modern agriculture.

[0003] In recent years, with the gradual maturity of the application of technologies such as artificial intelligence, automatic control, image recognition, and navigation and positioning in the agricultural field, vegetable harvesting robots based on machine vision and intelligent control have become a research hotspot. Existing broccoli harvesting robots mostly operate with fixed walking trajectories or single-ridge operations, unable to meet the cross-ridge operation requirements in complex field environments. At the same time, their ability to identify the maturity of broccoli and selective harvesting is also relatively limited, with problems such as a small operation range and low recognition accuracy.

[0004] The growth environment of broccoli is complex, the plant distribution is dense, and its harvesting has obvious selective characteristics, that is, it is necessary to judge according to the size and maturity of the flower head, and only harvest the flower heads that meet the commercial standards, which puts higher requirements on the recognition, judgment, and operation capabilities of the robot system. In addition, multi-ridge planting is often used in broccoli fields. To improve the harvesting efficiency, the robot system should have the ability to cross ridges, and at the same time, it should also have precise positioning navigation and path planning functions to ensure autonomous navigation between different ridges and complete efficient harvesting operations.

[0005] Therefore, there is an urgent need to develop a self-propelled broccoli harvesting robot that can achieve autonomous navigation, precise recognition, and selective harvesting in complex field environments, and has the ability to cross ridges, so as to improve the harvesting efficiency, reduce the labor cost, and promote the intelligent and automated process of broccoli harvesting operations. Summary of the Invention

[0006] In order to overcome the deficiencies of the prior art, the present invention provides a cross-ridge self-propelled broccoli selective harvesting robot and its operation method. The broccoli selective harvesting robot of the present invention can, through visual recognition and autonomous navigation, etc., completely operate without manual assistance, selectively harvest broccoli flower heads of appropriate size by itself, and does not require manual operation throughout the process, saving time and effort, and being safe and intelligent.

[0007] The technical solution adopted by the present invention to solve its technical problems is:

[0008] A cross-ridge self-propelled broccoli selective harvesting robot, comprising a harvesting robot mobile platform, a harvesting robotic arm, an end effector, a navigation control system, a harvesting control system, and a power distribution system. The harvesting robotic arm, navigation control system, harvesting control system, and power distribution system are installed on the harvesting robot mobile platform. The end effector is installed on the harvesting robotic arm. The navigation control system is connected to the controlled end of the harvesting robot mobile platform. The harvesting control system is connected to the controlled end of the end effector. The power distribution system is connected to the harvesting robot mobile platform, navigation control system, and harvesting control system.

[0009] Furthermore, the harvesting robot mobile platform has a four-wheel independent drive and four-wheel independent steering structure, including a vehicle frame body, four drive hub motors, four steering brushless motors, spring shock absorbers, an industrial computer, a Jetson AGX Orin embedded computing platform, a vehicle-mounted controller, a router, motor drivers, a battery pack, a waterproof control box, a storage basket, a switch button, and two emergency stop knobs.

[0010] The vehicle frame body is an integrally welded structure made of lightweight high-strength steel, with good structural rigidity and load capacity. The frame size is designed according to the broccoli ridge spacing, enabling the platform to operate across two ridges.

[0011] The four drive hub motors are respectively installed at the positions of the four wheels, with independent drive capabilities. They can flexibly control the wheel speed and direction according to navigation instructions to achieve actions such as forward, backward, and spinning in place. The four steering brushless motors enable independent wheel steering. Through an electric steering structure combined with a path tracking algorithm, it supports automatic adjustment of the driving direction, turning in place at the field head, and diagonal alignment, improving the path tracking accuracy.

[0012] The spring shock absorbers are arranged between the wheels and the vehicle frame body to buffer the vibrations and impacts generated during farm operations, ensuring the stability of the platform when driving on uneven ground and improving the clarity of image recognition and the accuracy of positioning.

[0013] The waterproof control box is welded on both sides of the vehicle frame to install core electronic control components such as an industrial computer, a Jetson AGX Orin embedded computing platform, a vehicle-mounted controller, a router, motor drivers, and a battery pack. The control box adopts a dustproof and waterproof structure to adapt to the harsh field environment of high humidity and high dust. The storage basket is installed in the bottom area of the vehicle tail to temporarily store the harvested broccoli flower heads. The switch button and the two emergency stop knobs are respectively set above the waterproof control box. In case of an emergency, the operator can quickly trigger the emergency stop device to cut off the power supply, ensuring the safety of the entire vehicle equipment and personnel.

[0014] Furthermore, the harvesting robotic arm is designed with a gantry structure, including two sets of Cartesian robotic arms, drive motors, motion control cards, motor drivers, groove-shaped photoelectric switches, cable carriers, cable carrier brackets, and connecting plates, etc.;

[0015] The two sets of Cartesian robotic arms are symmetrically installed on the left and right sides of the frame of the mobile platform of the harvesting robot, and have the ability of X, Y, and Z-axis linkage movement. Among them, both the X-axis and the Y-axis adopt high-precision screw module structures, and the Z-axis is a push-rod type screw slide structure. The X-axis module is fixed at the threaded holes on the frame by bolt connection. The Y-axis module is installed on the slider of the X-axis module, and the Z-axis is fixed on the slider of the Y-axis module through a connecting plate. The X, Y, and Z axes respectively correspond to lateral movement, longitudinal telescoping, and lifting movement, realizing precise positioning and motion control of the end effector in three-dimensional space;

[0016] The drive motors are respectively arranged at the power output ends of each motion axis, and servo motors with encoders are selected and electrically connected to the motion control card and the motor driver, which are used to realize real-time dynamic control and speed regulation of each degree of freedom of the robotic arm;

[0017] The groove-shaped photoelectric switches are respectively arranged at the limit positions of the X, Y, and Z axes, which are used to detect the position states of each axis in real time, provide origin reset signals and travel end protection signals, and effectively avoid over-travel operations;

[0018] The cable carrier bracket is arranged on one side of the X-axis module, which is used to support the cable carrier and maintain its stability during the movement process. The cable carrier is placed above the cable carrier bracket, which is used for the orderly winding and unwinding of the cable to avoid entanglement or wear during the movement of the robotic arm.

[0019] The end effector is of a sleeve structure, including a cylindrical shell (inner wall diameter 20 cm), an upper plate, a lower plate, a fixed connection flange, a multi-blade cutting assembly, a bottom plate, a spacer block, a servo motor for driving the blade to rotate, a proximity switch sensor, a motor bracket, a sensor mounting plate, a rack, and a rack bottom support, etc.;

[0020] The cylindrical shell is connected to the fixed connection flange through the upper plate and is installed at the end of the Z-axis of the Cartesian robotic arm, which is used to surround and press down from the top of the broccoli floret during the harvesting process. The shell material is made of lightweight and high-strength aluminum alloy to ensure both strength and reduce the load of the whole machine;

[0021] The multi-blade cutting assembly is arranged at the bottom of the sleeve, between the lower plate and the bottom plate, and is fixed and separated by a plurality of spacer blocks. The cutting edges of the blades are evenly arranged along the circumference, facing the center of the sleeve, forming an annular surrounding structure. During operation, when the Z-axis of the robotic arm precisely presses the end effector down to the target harvesting position, the servo motor is activated, and drives the blades to rotate synchronously inward through the drive mechanism, achieving efficient circumferential cutting of the broccoli flower ball stem, and completing the stem detachment operation;

[0022] The servo motor is fixed to the outer side wall of the housing through a motor bracket, and its output end is meshed and connected with a rack arranged at the bottom, and transmits rotational power to the cutting blades through a gear structure, ensuring stable power and controllable rotational speed during the cutting process, and meeting the operation requirements for different hardness of the flower ball stem;

[0023] The rack bottom support is fixedly connected above the bottom plate by bolts, and is used to carry and limit the rotational movement of the arc-shaped rack. It is internally provided with an arc-shaped guide groove structure matching the shape of the arc rack, which is used to precisely guide and support the rotation path of the rack. This structure can not only effectively prevent the rack from shaking or shifting during the cutting operation, but also improve the movement stability when the blades rotate synchronously.

[0024] The proximity switch sensor is installed above the rack bottom support structure, fixed by a sensor mounting plate, and the sensor is facing the corresponding rack structure, and is used to detect the zero position information of the servo motor, ensuring that the cutting action is accurately executed at the initial position, thereby improving the reliability and safety of the system control;

[0025] Furthermore, the navigation control system includes a navigation depth camera, a dual RTK real-time differential positioning device, an edge computing unit, etc.;

[0026] The navigation depth camera is fixedly installed at the center in front of the mobile platform of the harvesting robot, and the optical axis of the lens forms an angle of 30° with the horizontal line, and is used to collect images of the crop row in front, providing visual data support for path recognition and navigation line extraction;

[0027] The dual RTK devices are respectively fixed on the tops of the waterproof control boxes on the left and right sides of the harvesting robot, and achieve centimeter-level high-precision positioning through differential positioning, and calculate the heading angle of the robot mobile platform in real time through the baseline vector, which is used for auxiliary correction of path planning and heading control;

[0028] The edge computing unit integrates image processing algorithms and path tracking algorithms. First, it identifies the passable area through image segmentation algorithms, then generates the center line path of the current operation row by using navigation line extraction algorithms, and combines the dual RTK positioning data with pure tracking control algorithms to generate steering angle and target speed control commands for the wheels, achieving autonomous and stable driving of the harvesting robot between crop rows.

[0029] The harvesting control system includes a harvesting depth camera, a proximity switch sensor, a harvesting target recognition module, a robotic arm trajectory planning module, an end effector control module, etc.;

[0030] The harvesting depth camera is installed at the flange part of the end of the Z-axis, with the lens pointing vertically downward, used for real-time detection and recognition of broccoli flower heads, and outputting the center coordinates and height information of the target flower heads;

[0031] The target recognition module uses the method of fusing depth maps and RGB maps, combines the YOLOv8 model to recognize the target flower heads, and post-processes the recognition results to eliminate flower heads that do not meet the harvesting conditions such as overlapping and immature ones;

[0032] The robotic arm trajectory planning module performs three-axis coordinated control according to the downward pressing path of the Z-axis to achieve precise wrapping and positioning of the target flower heads. After completing the cutting operation, the robotic arm can smoothly move the cut flower heads above the storage basket according to the preset path, and control the Z-axis to slowly lower to achieve precise placement and orderly collection of the flower heads;

[0033] The end effector control module controls the servo motor to perform rotary cutting after receiving the trajectory command, and at the same time combines the feedback of the proximity switch sensor to achieve precise control and automatic zeroing;

[0034] The power distribution system adopts a dual-battery configuration, including a 72V lithium battery and a 48V lithium battery respectively. Among them, the 72V battery is used as the high-voltage power source to provide the power required for walking and steering for the four drive hub motors and four steering servo motors on the mobile platform of the harvesting robot, ensuring that the whole machine has strong mobility; the 48V battery is used as the control and operation power source. After being regulated and transformed by the power inverter, it supplies power to the drive motors, motor drivers, limit sensors and other auxiliary equipment in the robotic arm system to ensure the stability and continuity of the harvesting operation;

[0035] An operation method of a cross-ridge self-propelled broccoli selective harvesting robot includes the following steps:

[0036] S1. Start the robot. In the remote control mode, the operator uses the remote control to move the broccoli harvesting robot to the entrance of the field to be harvested, and inputs the boundary information of the current operation field, the broccoli planting mode (ridge spacing, drainage ditch width, etc.) and the maturity parameters of commercial-grade broccoli into the control terminal. The information is transmitted to the control system of the harvesting robot through wireless transmission. The robot control system automatically generates a global navigation path based on the received field boundary and planting mode information, and switches the operation mode to the automatic operation mode;

[0037] After the robot enters the automatic operation state, the navigation depth camera continuously obtains the image of the ridge path environment in front. The collected image is processed by the edge computing unit to extract the local navigation line information of the current ridge path, which is used by the navigation control system for path tracking to guide the robot to move forward along the ridge. During the movement, the harvesting depth camera synchronously collects the crop image, real-time detects the position and size of the broccoli head, and judges whether it meets the commercial harvesting standard based on the preset maturity judgment model.

[0038] When the harvesting control system identifies a broccoli head that meets the commercial standard based on the image data of the harvesting depth camera, it immediately sends a stop driving instruction to the navigation control system, and the robot stops moving. Subsequently, the harvesting control system controls the robotic arm to move above the target head, completes the cutting operation of the head through the cutting device, and places the harvested head in the storage basket. After the harvesting is completed, the harvesting control system sends a continue driving instruction to the navigation control system again, and the robot continues to move forward along the navigation path.

[0039] The harvesting robot repeats the actions in S2 and S3 until the harvesting task of all broccoli that meet the harvesting standards in the entire field area is completed. During the entire operation process, no manual intervention is required, and the system can achieve fully autonomous navigation, autonomous recognition, and selective harvesting operations, significantly improving the harvesting efficiency and automation level.

[0040] Further, in step S2, the process of extracting the local navigation line information of the current ridge path is as follows: Use the improved DeepLabV3+ semantic segmentation network model to process the image obtained by the navigation depth camera to generate the semantic segmentation image of the ridge path area. According to the segmentation result, use the method of circumscribed rectangle in sub-regions to extract the navigation points in the passable area between the ridges. Use the fourth-order cubic polynomial to fit the extracted navigation points to generate a continuous and smooth navigation path curve. Through spatial coordinate transformation, convert the fitted navigation line in the image coordinate system to the robot coordinate system to obtain the local navigation line information of the current ridge path for guiding the robot.

[0041] Furthermore, in step S2, the navigation control system for path tracking includes row operation path tracking and turning operation path tracking at the end of the field. Among them, the row operation path tracking is based on the preset global path planning, combined with the real-time navigation line information for local path correction. The control algorithm used is a pure path tracking algorithm that integrates the fuzzy control strategy, which can effectively improve the path tracking stability and responsiveness in the complex field environment. The turning operation path tracking at the end of the field takes the global path as the reference. When the robot travels to the end of the ridge area, the system automatically triggers the turning action. After the turning is completed, it performs row alignment operation through the navigation depth camera to ensure that the robot accurately enters the next operation row and completes the continuous automatic operation process.

[0042] Furthermore, in the step S3, when the broccoli flower heads meeting the commercial standards are recognized based on the image data of the harvesting depth camera, specifically, the harvesting control system collects the field image data through the harvesting depth camera. First, all possible target regions of broccoli flower heads within the field of view are extracted by using the registration information of the depth map and the color map. Then, the extracted image regions are input into the trained object detection and instance segmentation model to accurately segment and extract the contours of the broccoli targets in the image. Combining the depth information, the three-dimensional spatial position and size parameters of the flower heads are further estimated, including indexes such as the diameter and height of the flower heads, providing data support for the judgment of commercial-grade broccoli and the path planning of the robotic arm for picking.

[0043] The beneficial effects of the present invention are mainly manifested in:

[0044] 1. The robot of the present invention adopts a cross-ridge structure design, which can realize parallel harvesting operations across two ridges, significantly improve the field coverage efficiency, reduce the complexity of path planning, adapt to various ridge tillage planting modes, and is especially suitable for large-scale field harvesting operations of row-planted crops such as broccoli.

[0045] 2. The navigation control system of the present invention integrates Beidou RTK high-precision positioning and depth vision navigation information. Through multi-sensor data fusion technology, real-time local path correction is achieved on the basis of global path planning, which can effectively cope with uncertain factors such as changes in field lighting, uneven crop growth, and complex terrain, and improve the robustness and accuracy of the navigation system.

[0046] 3. The present invention can, through the harvesting depth camera combined with the image recognition model, recognize the position and maturity of broccoli flower heads in real time, automatically judge whether to harvest according to the set commercial standards, and has a highly intelligent selective harvesting ability, avoiding the mis-harvesting of immature or over-ripe broccoli, reducing human intervention, and improving the harvesting quality and commercial rate.

[0047] 4. The end effector adopted by the present invention is of a sleeve type structure, and is designed with a multi-blade rotary cutting mechanism, which has stronger cutting force and effectively avoids the problem of incomplete cutting caused by insufficient force of a single blade. Compared with the traditional gripper type end effector, the sleeve type structure can perform a surrounding and pressing cutting from the top of the flower head, effectively wrapping and stably fixing the flower head during the harvesting process, ensuring accurate and stable cutting actions, and thus reducing the problems of flower head damage or shedding caused by factors such as mis-grasping by the gripper and uneven force. This structure significantly improves the success rate of single harvesting and the integrity of commercial flower heads. Description of the Drawings

[0048] Figure 1 It is a schematic structural diagram of a cross-ridge self-propelled broccoli selective harvesting robot.

[0049] Figure 2 It is a schematic diagram of the structure of the mobile platform of the harvesting robot.

[0050] Figure 3 It is a schematic diagram of the structure of the Cartesian robot arm.

[0051] Figure 4 It is a schematic diagram of the structure of the end effector.

[0052] Figure 5 It is a schematic diagram of the control principle of the cross-ridge self-propelled broccoli selective harvesting robot.

[0053] Figure 6 It is a flowchart of the operation method of the cross-ridge self-propelled broccoli selective harvesting robot.

[0054] The reference numerals are: 1, mobile platform of the harvesting robot; 2, left harvesting robot arm; 3, right harvesting robot arm; 4, left end effector; 5, right end effector; 100, dual RTK real-time differential positioning device; 101, switch button; 102, emergency stop button; 103, vehicle-mounted controller; 104, motor driver; 105, industrial control computer; 106, battery pack; 107, Jetson AGX Orin embedded computing platform; 108, router; 109, storage basket; 110, navigation depth camera; 111, vehicle frame; 112, drive hub motor; 113, steering brushless motor; 114, spring shock absorber; 115, waterproof control box; 200, X-axis screw module; 201, X-direction drive motor; 202, drag chain bracket; 203, drag chain; 204, drag chain connecting plate; 205, Y-direction drive motor; 206, grooved photoelectric sensor; 207, slider; 208, Z-axis push rod type screw; 209, harvesting depth camera; 210, Z-direction drive motor; 211, connecting plate; 212, Y-axis screw module; 300, proximity switch sensor; 301, sensor mounting plate; 302, rack bottom support; 303, servo motor; 304, motor bracket; 305, multi-blade cutting assembly; 306, spacer block; 307, lower plate; 308, bottom plate; 309, housing; 310, upper plate. Specific embodiments

[0055] The present invention will be further described below with reference to the accompanying drawings.

[0056] Refer to Figures 1 to 6, A cross-ridge self-propelled selective harvesting robot for broccoli, comprising a harvesting robot mobile platform 1, a left harvesting robotic arm 2, a right harvesting robotic arm 3, a left end effector 4, and a right end effector 5. The harvesting robot mobile platform 1 is designed according to the ridge spacing of the broccoli field and can perform operations across two ridges; the left harvesting robotic arm 2 and the right harvesting robotic arm 3 are symmetrically installed on the left and right sides of the center of the frame 110 of the harvesting robot mobile platform, and the harvesting area is directly below; the left end effector 4 and the right end effector 5 are connected to the fixed connection flange through the upper plate 301 and are respectively installed at the Z-axis ends of the left harvesting robotic arm 2 and the right harvesting robotic arm 3 by bolts;

[0057] The power of the navigation depth camera 110 and the harvesting depth camera 209 is provided by the Jetson AGX Orin embedded computing platform 108 through a USB cable, and the data transmission function is realized through the USB cable;

[0058] As attached Figure 2As shown in the figure, the harvesting robot mobile platform 1 includes a dual RTK real-time differential positioning device 100, a switch button 101, an emergency stop button 102, a vehicle-mounted controller 103, a motor driver 104, an industrial control computer 105, a battery pack 106, a Jetson AGX Orin embedded computing platform 107, a router 108, a storage basket 109, a navigation depth camera 110, a vehicle frame 111, drive hub motors 112, steering brushless motors 113, spring shock absorbers 114, and a waterproof control box 115. The vehicle frame 111 is an integrally welded structure, and a harvesting area that can pass through the broccoli ridges is formed below it, which is used to support the harvesting robotic arms 2 and 3, the waterproof control box 115, and the storage basket 109; the drive hub motors 112 are installed at the four wheel positions to achieve actions such as forward, backward, and spinning in place; the steering brushless motors 113 are evenly distributed at the four corner positions of the vehicle frame 111 to achieve independent wheel steering; the waterproof control box 115 is symmetrically welded to the left and right sides of the vehicle frame 111 and is used to place electrical control equipment; the storage basket 109 is fixed to the lower rear side of the vehicle frame 111 by bolts, and the installation height is higher than the broccoli plants to avoid collision with the broccoli flower heads during walking, and is used to temporarily store the harvested broccoli flower heads; the spring shock absorbers 114 are arranged between the wheels and the vehicle frame 111 to buffer the vibrations and impacts generated during farm operations; the dual RTK real-time differential positioning device 100 is fixed to the central position at the top of the waterproof control boxes 115 on the left and right sides of the harvesting robot by a strong magnetic suction cup; the navigation depth camera 110 is installed on a camera bracket and is fixedly installed at the central position in front of the vehicle frame 111 of the harvesting robot mobile platform through a hinge, and the optical axis of the lens forms an angle of 30° with the horizontal line; the vehicle-mounted controller 103, the motor driver 104, the industrial control computer 105, and the Jetson AGX Orin embedded computing platform 107 are fixedly installed in the upper area inside the waterproof control box 115 for path recognition and vehicle control; the battery pack 106 is fixed in the lower area inside the waterproof control box 115 to supply power to the robot's electrical control equipment and has a charging interface outside; the switch button 101 is connected to the industrial control computer 105 and the battery pack 106 and is installed above the waterproof control box 115 to cut off the power supply in case of an emergency to ensure the safety of the whole vehicle equipment and personnel. When restarting the equipment, just rotate the emergency stop button 102 clockwise by 45° and then release it, and the button pops up to restore the power supply of the equipment;

[0059] As Figure 3As shown, the harvesting robotic arm structure includes an X-axis screw module 200, an X-direction driving motor 201, a drag chain support 202, a drag chain 203, a drag chain connecting plate 204, a Y-direction driving motor 205, a groove-type photoelectric switch 206, a slider 207, a Z-axis push rod type screw 208, a harvesting depth camera 209, a Z-direction driving motor 210, a connecting plate 211, and a Y-axis screw module 212. The X-axis screw module 200 is fixed at the threaded hole on the vehicle frame 111 by means of bolt connection. The Y-axis screw module 212 is installed on the slider 207 of the X-axis screw module 200. The Z-axis push rod type screw 208 is fixed on the slider 07 of the Y-axis screw module 212 through the connecting plate 211 to achieve three-axis linkage movement. The driving motors 201, 205, and 210 are respectively arranged at the power output ends of each moving axis, and servo motors with encoders are selected and electrically connected to the internal motion control card and the motor driver to realize real-time dynamic control and speed adjustment of each degree of freedom of the robotic arm. The motion control card and the motor driver are fixedly installed inside the waterproof control box 115 by screws. The harvesting depth camera 209 is fixed to the flange part at the end of the Z-axis through a camera support, and the optical axis of the lens is vertically downward. The groove-type photoelectric switches 206 are respectively fixed at the limit position grooves of the X, Y, and Z-axis modules by screws to detect the position states of each axis in real time and provide origin reset signals and stroke end protection signals. The drag chain support 202 is arranged on one side of the X-axis module to support the drag chain 203 and maintain its stability during movement. The drag chain 203 is used for the orderly retraction and extension of the cable to avoid winding or abrasion during the movement of the robotic arm.

[0060] As Figure 4As shown, the left end effector 4 and the right end effector 5 have the same structure, both including a proximity switch sensor 300, a sensor mounting plate 301, a rack and a rack base 302, a servo motor 303 for driving the blade to rotate, a motor bracket 304, a multi-blade cutting assembly 305, a spacer block 306, a lower plate 307, a bottom plate 308, a cylindrical shell 309, and an upper plate 310. The cylindrical shell 309 is connected to the fixed connection flange through the upper plate 310 and is installed at the end of the Z-axis of the harvesting robotic arm, and is used to surround and press down from the top of the broccoli flower head during the harvesting process; the multi-blade cutting assembly 305 is arranged at the bottom of the shell 309, between the bottom plate 308 and the lower plate 307, and is fixed and separated by a plurality of spacer blocks 305, and is used for circumferentially cutting the stem of the broccoli flower head to complete the separation of the stem; the servo motor 303 is installed on the motor bracket 304 and is fixed to the outer side wall of the shell 309 by bolts, and its output end is meshed with the rack 302 arranged at the bottom, and transmits the rotational power to the cutting blade through a gear structure, and is used to drive the opening and closing of the multi-blade cutting assembly 305; the proximity switch sensor 300 is installed above the bottom plate 308 and is fixed by the sensor mounting plate 301, and the sensor 300 is directly opposite to the corresponding rack and rack base 302, and is used to detect the zero position information in real time;

[0061] As Figure 6 shown, the present invention provides an operation method for a cross-ridge self-propelled broccoli selective harvesting robot, including the following steps:

[0062] Step S1: Initial configuration of the harvesting robot, the process is as follows:

[0063] Step S1.1: The operator moves the broccoli harvesting robot to the entrance of the field to be harvested through the remote control;

[0064] Step S1.2: Input the boundary information of the current working field, the broccoli planting mode (ridge spacing, drainage ditch width, etc.) and the maturity parameters of commercial-grade broccoli in the control terminal;

[0065] Step S1.3: The robot control system automatically generates a global navigation path based on the received field boundary and planting mode information, and switches the operation mode to the automatic operation mode;

[0066] Step S2: Navigation line extraction and broccoli maturity judgment, the process is as follows:

[0067] Step S2.1: The navigation depth camera 110 continuously obtains the image of the front ridge path environment, and the collected image is processed by the edge computing unit 107 to extract the local navigation line information of the current ridge path, so as to provide path tracking for the navigation control system to guide the robot to move forward along the ridge;

[0068] Step S2.2: During the movement, the harvesting depth camera 209 synchronously collects crop images, detects the position and size of the broccoli florets in real time, and determines whether they meet the commercial harvesting standards based on a preset maturity determination model;

[0069] Step S3: Automatic navigation and target harvesting coordinated control, the process is as follows:

[0070] Step S3.1: When the harvesting control system identifies broccoli florets that meet the commodity standards based on the image data from the harvesting depth camera 209, it immediately sends a stop instruction to the navigation control system, causing the robot to stop moving;

[0071] Step S3.2: The harvesting control system controls the robotic arms 2 and 3 to move to the top of the target flower bulb, and completes the flower bulb cutting operation through the end effector. Specifically, the servo motor 303 drives the multi-blade cutting assembly 305 to rotate synchronously around the center to achieve circumferential cutting of the flower bulb stem; then, the Z-axis of the robotic arm moves upward, driving the end effector to move above the storage basket 109, and then controls the multi-blade cutting assembly 305 to rotate synchronously outward to release the harvested broccoli flower bulb into the storage basket 109. After the harvest is completed, the harvesting control system again sends a continue driving instruction to the navigation control system, and the robot continues to move along the navigation path;

[0072] Step S4: The harvesting robot repeats steps S2 and S3 until it completes the task of harvesting all broccoli that meet the harvesting criteria in the entire field area.

[0073] The embodiments of this specification are merely examples of implementations of the invention and are provided for illustrative purposes only. The scope of protection of the present invention should not be considered limited to the specific embodiments described in these embodiments. The scope of protection of the present invention also extends to equivalent technical means that can be conceived by a person of ordinary skill in the art based on the invention.

Claims

1. A cross-ridge self-propelled broccoli selective harvesting robot, characterized in that, The harvesting robot includes a harvesting robot mobile platform, a harvesting robotic arm, an end effector, a navigation control system, a harvesting control system, and a power distribution system. The harvesting robotic arm, the navigation control system, the harvesting control system, and the power distribution system are installed on the harvesting robot mobile platform. The end effector is installed on the harvesting robotic arm. The navigation control system is connected to the controlled end of the harvesting robot mobile platform. The harvesting control system is connected to the controlled end of the end effector. The power distribution system is connected to the harvesting robot mobile platform, the navigation control system, and the harvesting control system.

2. The cross-ridge self-propelled broccoli selective harvesting robot according to claim 1, wherein The harvesting robot mobile platform has a four-wheel independent drive and four-wheel independent steering structure, and includes a vehicle frame body, four drive hub motors, four steering brushless motors, spring shock absorbers, an industrial computer, a Jetson AGX Orin embedded computing platform, a vehicle-mounted controller, a router, motor drivers, a battery pack, a waterproof control box, a storage basket, a switch button, and two emergency stop knobs; The four drive hub motors are respectively installed at the positions of the four wheels, have independent driving capabilities, and can flexibly control the wheel speed and direction according to navigation instructions to achieve forward, backward, and in-situ rotation movements. The four steering brushless motors enable independent wheel steering; the spring shock absorbers are arranged between the wheels and the vehicle frame body. The waterproof control box is fixed on the left and right sides of the vehicle frame and is used to install the industrial computer, the Jetson AGX Orin embedded computing platform, the vehicle-mounted controller, the router, the motor drivers, and the battery pack; the storage basket is installed in the bottom area of the vehicle tail and is used to temporarily store the harvested broccoli flower heads; the switch button and the two emergency stop knobs are respectively arranged above the waterproof control box.

3. The cross-ridge self-propelled broccoli selective harvesting robot according to claim 1 or 2, characterized in that, The harvesting robotic arm is designed with a gantry structure and includes two groups of Cartesian robotic arms, drive motors, motion control cards, motor drivers, slot-type photoelectric switches, cable carriers, cable carrier brackets, and connecting plates; The two groups of Cartesian robotic arms are symmetrically installed on the left and right sides of the vehicle frame of the harvesting robot mobile platform and have the ability of X, Y, and Z-axis linkage movement. Among them, both the X-axis and the Y-axis adopt high-precision screw module structures, and the Z-axis is a push-rod type screw slide table structure. The X-axis module is fixed at the threaded holes on the vehicle frame by bolt connection. The Y-axis module is installed on the slider of the X-axis module, and the Z-axis is fixed on the slider of the Y-axis module through a connecting plate. The X, Y, and Z axes respectively correspond to lateral movement, longitudinal extension, and lifting movement; the drive motors are respectively arranged at the power output ends of each motion axis, and servo motors with encoders are selected and are electrically connected to the motion control card and the motor driver to achieve real-time dynamic control and speed adjustment of each degree of freedom of the robotic arm; the slot-type photoelectric switches are respectively arranged at the limit positions of the X, Y, and Z axes and are used to detect the position states of each axis in real time and provide origin reset signals and travel end protection signals; the cable carrier bracket is arranged on one side of the X-axis module and is used to support the cable carrier and maintain its stability during movement. The cable carrier is placed above the cable carrier bracket and is used for the orderly retraction and extension of the cables.

4. The cross-ridge self-propelled broccoli selective harvesting robot according to claim 3, characterized in that, The end effector is of a sleeve structure, including a cylindrical shell, an upper plate, a lower plate, a fixed connection flange, a multi-blade cutting assembly, a bottom plate, a spacer block, a servo motor for driving the blade to rotate, a proximity switch sensor, a motor bracket, a sensor mounting plate, a rack and a rack base support; The cylindrical shell is connected to the fixed connection flange through the upper plate and is installed at the end of the Z-axis of the Cartesian robot arm, and is used to surround and press down from the top of the broccoli floret during the harvesting process; the multi-blade cutting assembly is arranged at the bottom of the sleeve, between the lower plate and the bottom plate, and is fixed and separated by a plurality of spacer blocks; the cutting edges of the blades are evenly arranged along the circumference, facing the center of the sleeve, forming an annular surrounding structure; the servo motor is fixed to the outer side wall of the shell through the motor bracket, and its output end is meshed and connected with the rack arranged at the bottom, and transmits the rotational power to the cutting blade through a gear structure; The rack base support is fixedly connected above the bottom plate by bolts, and is used to carry and limit the rotational movement of the arc-shaped rack. An arc-shaped guide groove structure matching the shape of the arc rack is arranged inside, and is used to accurately guide and support the rotation path of the rack; the proximity switch sensor is installed above the rack base support structure and is fixed through the sensor mounting plate. The sensor is directly opposite to the corresponding rack structure, and is used to detect the zero position information of the servo motor to ensure that the cutting action is accurately executed at the initial position.

5. The cross-ridge self-propelled broccoli selective harvesting robot according to claim 1 or 2, characterized in that, The navigation control system includes a navigation depth camera, a dual RTK real-time differential positioning device and an edge computing unit; The navigation depth camera is fixedly installed at the center in front of the mobile platform of the harvesting robot, and is used to collect images of the crop row in front, providing visual data support for path recognition and navigation line extraction; the two RTK devices are respectively fixed on the tops of the waterproof control boxes on the left and right sides of the harvesting robot, and achieve centimeter-level high-precision positioning through differential positioning, and calculate the heading angle of the mobile platform of the robot in real time through the baseline vector, which is used for the auxiliary correction of path planning and heading control; the edge computing unit integrates image processing algorithms and path tracking algorithms. First, it identifies the passable area through the image segmentation algorithm, and then uses the navigation line extraction algorithm to generate the central line path of the current operation row. Combining the dual RTK positioning data and the pure tracking control algorithm, it generates the steering angle and target speed control instructions of the wheels, realizing the autonomous and stable driving of the harvesting robot between the crop rows.

6. The cross-ridge self-propelled broccoli selective harvesting robot according to claim 1 or 2, characterized in that, The harvesting control system includes a harvesting depth camera, a proximity switch sensor, a harvesting target recognition module, a robotic arm trajectory planning module and an end effector control module; The harvesting depth camera is installed at the flange part of the Z-axis end, and is used to detect and identify the broccoli floret in real time, and output the central coordinates and height information of the target floret; the target recognition module identifies the target floret based on the fusion of the depth map and the RGB map, and combines the YOLOv8 model to post-process the recognition result; the robotic arm trajectory planning module performs three-axis coordinated control according to the Z-axis pressing down path to achieve precise wrapping and positioning of the target floret; After completing the cutting operation, the robotic arm can smoothly move the cut cauliflower head above the storage basket along a preset path, and control the Z-axis to slowly lower it to achieve precise placement and orderly collection of the cauliflower head; after receiving the trajectory command, the end effector control module controls the servo motor to perform rotary cutting, and at the same time combines the feedback of the proximity switch sensor to achieve precise control and automatic zeroing.

7. A cross-ridge self-propelled broccoli selective harvesting robot according to claim 1 or 2, characterized in that, The power distribution system adopts a dual-battery configuration, including a 72V lithium battery and a 48V lithium battery respectively. Among them, the 72V battery is used as the high-voltage power source to provide the power required for walking and steering for the four drive hub motors and four steering servo motors on the mobile platform of the harvesting robot, ensuring that the whole machine has strong mobility; the 48V battery is used as the control and operation power source. After being regulated and transformed by the power inverter, it supplies power to the drive motors, motor drivers, limit sensors and other auxiliary equipment in the robotic arm system to ensure the stability and continuity of the harvesting operation.

8. An operation method of the cross-ridge self-propelled broccoli selective harvesting robot according to claim 1, characterized in that, The method includes the following steps: S1. Start the robot. In the remote control mode, the operator uses the remote control to move the broccoli harvesting robot to the entrance of the field to be harvested, and inputs the boundary information of the current working field, the broccoli planting mode (ridge spacing, drainage ditch width, etc.) and the maturity parameters of commercial-grade broccoli at the control terminal. The information is transmitted to the control system of the harvesting robot through wireless transmission. The robot control system automatically generates a global navigation path based on the received field boundary and planting mode information, and switches the operation mode to the automatic operation mode. S2. After the robot enters the automatic operation state, the navigation depth camera continuously obtains the image of the front ridge path environment. The collected image is processed by the edge computing unit to extract the local navigation line information of the current ridge path for the navigation control system to perform path tracking and guide the robot to move forward along the ridge. During the movement, the harvesting depth camera synchronously collects the crop image, real-time detects the position and size of the broccoli head, and judges whether it meets the commercial harvesting standard based on the preset maturity judgment model. S3. When the harvesting control system identifies a broccoli head that meets the commercial standard based on the image data of the harvesting depth camera, it immediately sends a stop driving instruction to the navigation control system, and the robot stops moving. Subsequently, the harvesting control system controls the robotic arm to move above the target cauliflower head, completes the cutting operation of the cauliflower head through the cutting device, and places the harvested cauliflower head in the storage basket. After the harvesting is completed, the harvesting control system sends a continue driving instruction to the navigation control system again, and the robot continues to move forward along the navigation path. S4. The harvesting robot repeats the actions of S2 and S3 until the harvesting task of all broccoli that meet the harvesting standards in the entire field area is completed. During the whole operation process, no manual intervention is required. The system can achieve fully autonomous navigation, autonomous recognition and selective harvesting operations, significantly improving the harvesting efficiency and automation level.

9. The operation method according to claim 8, characterized in that, In the step S2, the process of extracting the local navigation line information of the current ridge diameter is as follows: Use the improved DeepLabV3+ semantic segmentation network model to process the images obtained by the navigation depth camera to generate the semantic segmentation images of the ridge diameter area; According to the segmentation results, use the method of the circumscribed rectangle of the divided area to extract the navigation points in the passable area between the ridges; Use the fourth-order cubic polynomial to fit the extracted navigation points to generate a continuous and smooth navigation path curve; Through the spatial coordinate transformation, convert the fitted navigation line in the image coordinate system to the robot coordinate system to obtain the local navigation line information of the current ridge diameter for guiding the robot; The path tracking of the navigation control system includes the path tracking of the inter-row operation and the path tracking of the turning operation at the end of the field. Among them, the path tracking of the inter-row operation is based on the preset global path planning, and combines the real-time navigation line information to perform local path correction. The control algorithm used is the pure path tracking algorithm that integrates the fuzzy control strategy, which can effectively improve the path tracking stability and responsiveness in the complex field environment; The path tracking of the turning operation at the end of the field takes the global path as the reference. When the robot travels to the end of the ridge area, the system automatically triggers the turning action. After the turning is completed, the navigation depth camera is used for row alignment operation to ensure that the robot accurately enters the next operation row and completes the continuous automatic operation process.

10. The operation method according to claim 8 or 9, characterized in that, In the step S3, the broccoli flower balls that meet the commercial standards are recognized based on the image data of the harvesting depth camera. Specifically, the harvesting control system collects the field image data through the harvesting depth camera. First, use the registration information of the depth map and the color map to extract all possible broccoli flower ball target areas within the field of view; Then input the extracted image area into the trained target detection and instance segmentation model to accurately segment and extract the contour of the broccoli target in the image; Combine the depth information to further estimate the three-dimensional spatial position and size parameters of the flower ball, including indicators such as the diameter and height of the flower ball, providing data support for the judgment of commercial-grade broccoli and the path planning of the robotic arm for picking.

Images