A vision-guided intelligent lightweight flat tomato picking robot and its control method

Abstract

This invention discloses a vision-guided, intelligent, lightweight, flattened tomato-picking robot and its control method. The robot's main frame is constructed from aluminum profiles. Four flexible cables are attached at one end to the four corners of a servo-electric cylinder mounting bracket, while the other ends wind around four fixed pulleys and onto corresponding reels. A motor drives the reels on the coupling, winding the cables and achieving planar motion of the end effector. A camera module visually guides and identifies the three-dimensional coordinates of the tomato fruit, enabling the control system to drive the robot to harvest the tomatoes. The end effector, a servo-electric cylinder, achieves Z-axis motion in space, while a flexible manipulator grasps the fruit. A movable platform at the bottom allows the robot to locate and navigate within the tomato orchard. This invention features a novel design, a compact and lightweight structure, a large workspace, low cost, and intelligent intelligence.

Description

Technical Field

[0001] The present invention belongs to the technical field of agricultural picking equipment and relates to a vision-guided intelligent lightweight flat tomato picking robot, and in particular to the picking robot and a control method. Background Art

[0002] Tomatoes, with their sweet flesh, are widely cultivated worldwide. However, because tomato picking is a phased process, labor shortages often occur during the harvest season. Therefore, to address the mismatch between growing tomato production and a shrinking agricultural workforce, it is essential to promote the transition from manual to autonomous robotic harvesting, which can improve picking efficiency and promote the development of modern agriculture.

[0003] The picking mechanism is a fundamental component of the harvesting operation, and its quality determines the performance of the harvesting robot. Currently, articulated robotic arms are widely used, but their ability to reach any position in space is limited. They are also bulky and heavy, making control difficult, slow, and expensive to manufacture. Tomato plants grow in a hierarchical structure and at high heights, requiring an articulated robotic arm to assist with picking, further increasing the cost and control complexity of the equipment.

[0004] Therefore, there is a need for a harvesting robot capable of stable harvesting in unstructured agricultural environments. This invention designs a vision-guided, intelligent, lightweight, and flat tomato-picking robot. This robot uses flexible cable traction to drive the end effector and achieve vision-guided harvesting, replacing the rigid drive of traditional articulated robotic arms. This robot features a large workspace, fast response, low inertia, and low cost, making it flexible and adaptable to tomato picking in unstructured agricultural environments. Summary of the Invention

[0005] In order to overcome the shortcomings of the above-mentioned prior art, the present invention provides a vision-guided intelligent lightweight flat tomato picking robot and control method. The robot is characterized in that point cloud data fusion is performed through tomato fruit images to obtain the three-dimensional coordinate points of the tomato fruit, and visual guidance is used to identify and locate the tomatoes. Four flexible cables and an end effector are driven to pick the tomato fruits, and the robot can realize autonomous positioning and navigation in the tomato orchard.

[0006] The present invention is implemented by the following technical solution: a vision-guided intelligent lightweight flat tomato picking robot, comprising:

[0007] The main frame is made of aluminum profiles to build the framework of the picking robot. The size of the aluminum profiles can be customized according to the size of the tomato plants and tomato orchards.

[0008] The flexible cable drive device includes a motor driving a reel on a coupling to reel in the flexible cable to achieve planar motion of the end actuator;

[0009] The flexible cable group includes four flexible cables, one end of which is fixed to the four corners of the servo cylinder fixing frame, and the other end is wound around four fixed pulleys and wound on the corresponding reels;

[0010] The end effector includes a servo cylinder and a flexible manipulator. The servo cylinder realizes the Z-axis movement in space. The flexible manipulator is connected to the end of the servo cylinder and is used to grab the fruit.

[0011] The bottom movable platform includes a rubber track chassis assembly and a laser radar;

[0012] A camera module for visually guiding the recognition of the three-dimensional coordinates of tomato fruits for picking;

[0013] The control system is used to drive the corresponding motor to rotate based on the three-dimensional coordinates of the tomato fruit to be picked through feedback from the camera module, so that the four cables in the cable group can be retracted or extended to the corresponding lengths, realizing the planar motion of the end effector to the plane coordinates where the tomato is located, and then driving the servo electric cylinder to move in the Z-axis direction to the Z-axis coordinates where the tomato is located, and finally driving the flexible manipulator to grasp the tomato.

[0014] As a further improvement of the above scheme, the main frame also includes two vertical beam aluminum profiles and two horizontal beam aluminum profiles connected in sequence at right angles to form a rectangular frame; the midpoints of the two lower beam aluminum profiles are connected to the end of a horizontal beam aluminum profile in the rectangular frame; the ends of the two oblique beam aluminum profiles are respectively connected to the midpoint of the vertical beam aluminum profile and the end of the lower beam aluminum profile; and four fixed pulleys are respectively installed on the four corners of the rectangular frame.

[0015] As a further improvement of the above scheme, the flexible cable drive device also includes a first drive motor, a coupling, a reel mechanism, and a reel fixing plate; the reel mechanism includes a bearing, a reel and a fixed bracket; the coupling is connected to the end of the first drive motor and one end of the reel, and the two ends of the reel are respectively installed on the bearings; the first drive motor and the fixed bracket are installed on the reel fixing plate; the four flexible cable drive devices are respectively fixed at corresponding positions of the rectangular frame.

[0016] As a further improvement to the above solution, the end effector further includes a servo electric cylinder having a middle portion connected to a servo electric cylinder mounting bracket; the flexible manipulator comprises a second drive motor and two flexible grippers; the second drive motor of the flexible manipulator is connected to the front end shaft of the servo electric cylinder via a fixing plate; the servo electric cylinder controls the movement of the flexible manipulator in the spatial Z-axis direction; and the second drive motor of the flexible manipulator controls the closing of its grippers to achieve the picking of tomatoes.

[0017] As a further improvement to the above solution, the camera module is fixed to a fixed plate connecting the servo electric cylinder and the flexible manipulator; the camera module captures the position of the target tomato and the position of the flexible manipulator in real time; a deep learning algorithm model is used to identify the tomatoes in the image, and a point cloud cluster of the target tomato is separated by establishing a coordinate transformation matrix relationship between a normalized image pixel coordinate system and a base coordinate system of the picking robot. The mean coordinates of the separated point cloud clusters are calculated, and the spatial position of the target tomato is described based on the mean coordinates.

[0018] Furthermore, the control system includes a motor controller and a main controller; the main controller calculates the three-dimensional coordinates of the target tomato and the relative position of the flexible manipulator fed back in real time by the camera module to obtain the corresponding rotation steps and speeds of the motors of the four flexible cable drive devices, and then rotates the first drive motor through the motor controller at the corresponding steps and speeds. At the same time, the motor controller obtains the corresponding rotation steps and speeds fed back by the encoders of the four first drive motors in real time and sends them to the main controller. The main controller compares the values ​​fed back by the encoders with the spatial position information of the flexible manipulator obtained by the current camera module, and adjusts the steps and speeds of each motor in real time.

[0019] Furthermore, force sensors are included at the connections between the ends of the four flexible cables and the four corners of the servo fixed frame to detect the tensioning force of the four flexible cables in real time and feed it back to the main controller; the main controller adjusts the number of steps and speed of the first drive motor through the motor controller in real time to optimize the force distribution of the four flexible cables.

[0020] Furthermore, the robot transforms the tomato picking path planning problem into a single traveling salesman problem based on the principle of the shortest tomato picking path. It uses a genetic algorithm to solve the single traveling salesman problem and achieves the optimal tomato picking sequence planning through iterative calculations of the algorithm.

[0021] Furthermore, the laser radar in the bottom movable platform performs a three-dimensional scan of the scene in the tomato orchard during movement, and then realizes three-dimensional reconstruction of the scene through the main controller to construct an incremental global environment map. Based on the obtained global environment map, when the rubber track chassis assembly automatically operates in the tomato orchard, it scans and generates three local contour maps according to the current position of the laser radar, and compares them with the global environment map to obtain the current position of the bottom movable platform in the global map, thereby realizing positioning and navigation of the bottom movable platform.

[0022] Furthermore, the present invention also provides a control method, which is applied to any of the above-mentioned vision-guided intelligent lightweight flat tomato picking robots, comprising the following steps:

[0023] Step 1: The control system performs a self-check and initialization, resetting each drive motor and force sensor. The rubber track chassis assembly on the movable platform at the bottom is manually operated to move from a starting point within the tomato orchard. The lidar scans the surrounding environment and returns to the starting point to obtain an incremental global environmental map.

[0024] Step 2: The camera module of the picking robot captures and records the initial position of the current flexible manipulator as the basic reference coordinate point to establish the robot coordinate system;

[0025] Step 3: The end effector of the picking robot traverses the entire picking area from its current initial position. The camera module captures an image of the target tomato. The main controller processes the depth map of the target tomato fruit and locates the key points of picking to obtain the three-dimensional coordinates of the key points of picking tomato in the captured image.

[0026] Step 4: The main controller obtains the 3D coordinates of the key tomato picking points from the camera, plans the picking sequence for the tomatoes, and sends the picking sequence instructions to the first drive motor, servo cylinder, and second drive motor of the flexible manipulator. This controls the X and Y movement of the four flexible cables within the rectangular frame, the Z-axis movement of the servo cylinder, and the picking motion of the flexible manipulator.

[0027] Step 5: The camera module captures and records the movement position of the flexible manipulator in real time;

[0028] Step 6: The main controller calculates the current motion position of the flexible manipulator using the motor steps and speed feedback from the encoder of the first drive motor of the flexible cable drive device and the encoder of the servo cylinder drive motor. It compares this with the real-time motion position of the flexible manipulator fed back by the camera, and fine-tunes the steps and speed of the drive motor to adjust the real-time position of the flexible manipulator.

[0029] Step 7: The picking robot's end effector returns to its initial position and traverses the entire picking area again. The camera module captures the target tomato image and feeds it back to the main controller to check whether the tomatoes in the picking area have been picked. If not, the main controller will continue to plan the picking sequence with the 3D coordinates of the unpicked target tomatoes and send the picking instructions to the cable drive device and end effector.

[0030] Step 8: At this point, the picking work in the current area is completed, the end effector returns to the initial position, and the bottom movable platform locates and navigates to the next picking area in the tomato orchard.

[0031] Compared with existing agricultural picking robots, the vision-guided intelligent lightweight flat tomato picking robot of the present invention has the following beneficial effects:

[0032] The present invention designs a vision-guided intelligent lightweight and flat tomato-picking robot, which uses flexible cable traction to drive the end effector and realize vision-guided picking, replacing the rigid drive of traditional articulated robotic arms. It has the characteristics of large workspace, fast response speed, small inertia, and low cost, enabling the robot to flexibly adapt to tomato picking work in unstructured agricultural environments.

[0033] The present invention obtains the three-dimensional coordinate points of tomato fruits by performing point cloud data fusion on tomato fruit images, performs visual guidance to identify and locate tomatoes, drives four flexible cables and an end effector to pick the tomatoes, and realizes autonomous positioning and navigation of the robot in the tomato orchard. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 This is a schematic structural diagram of the tomato picking robot of the present invention;

[0035] Figure 2 It is a structural diagram of the main frame of the present invention;

[0036] Figure 3 Schematic diagram of the structure of the flexible cable drive device of the present invention;

[0037] Figure 4 Schematic diagram of the structure of the flexible cable group of the present invention;

[0038] Figure 5 Schematic diagram of the structure of the camera module of the present invention;

[0039] Figure 6 It is a structural schematic diagram of the servo electric cylinder of the present invention;

[0040] Figure 7 This is a schematic structural diagram of the flexible two-claw manipulator of the present invention;

[0041] Figure 8 This is a schematic structural diagram of the bottom movable platform of the present invention;

[0042] Figure 9 It is the workflow diagram of the present invention.

[0043] Figure numerals: main frame 1, flexible cable drive device 2, camera module 3, flexible manipulator 4, fixed pulley 5, bottom movable platform 6, flexible cable group 7, servo electric cylinder 8, servo electric cylinder fixing frame 9, fixing plate 10, vertical beam aluminum profile 101, horizontal beam aluminum profile 102, lower beam aluminum profile 103, inclined beam aluminum profile 104, first drive motor 201, coupling 202, reel fixing plate 203, bearing 204, reel 205, fixing bracket 206, second drive motor 401, rubber track chassis assembly 601, laser radar 602, servo electric cylinder drive motor 801. DETAILED DESCRIPTION

[0044] The following is a detailed description of the embodiments of the present invention. The embodiments are implemented based on the technical solution of the present invention, and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0045] Implementation Case 1

[0046] Combine Figure 1 A vision-guided intelligent lightweight flat tomato picking robot consists of a main frame 1, a flexible cable drive device 2, a camera module 3, a flexible manipulator 4, a fixed pulley 5, a bottom movable platform 6, a flexible cable group 7, a servo cylinder 8, a servo cylinder fixing frame 9 and a fixing plate 10.

[0047] Specifically, the main frame 1 is the framework of the entire tomato picking robot, and various components are directly or indirectly mounted on the main frame 1 in various ways. There are four flexible cable drive devices 2, symmetrically mounted above and below the midpoint of the two vertical aluminum profiles 101 of the main frame. The four first drive motors 201 in the flexible cable drive device 2 drive the reels at the other end of the coupling through forward and reverse rotation, thereby driving the corresponding four flexible cables, respectively, to achieve the X and Y translational degrees of freedom of the servo cylinder mounting frame 9. The four fixed pulleys are respectively mounted on the four corners of the two horizontal aluminum profiles 102 of the main frame. One end of the four flexible cables is respectively wound around the reels on the flexible cable drive device 2 and passes around the four fixed pulleys 5. The other ends of the four flexible cables are respectively fixed to the servo cylinder mounting frame 9. The middle part of the servo electric cylinder 8 is connected to the servo electric cylinder fixing frame 9; the flexible manipulator 4 is composed of a second drive motor 401 and two flexible claws; the second drive motor 401 of the flexible manipulator 4 is connected to the front end shaft of the servo electric cylinder 8 through the fixing plate 10; the servo electric cylinder 8 controls the movement of the flexible manipulator 4 in the Z-axis direction of space; the second drive motor 401 of the flexible manipulator 4 controls the closing of its claws to realize the picking of tomato fruits.

[0048] Combine Figure 2 The main frame 1 is composed of a vertical beam aluminum profile 101, a horizontal beam aluminum profile 102, a lower beam aluminum profile 103, an inclined beam aluminum profile 104 and a fixed pulley 5.

[0049] Furthermore, each profile is connected by using steering connectors, angle codes, hexagon socket bolts and T-nuts; the two vertical beam aluminum profiles 101 and the two horizontal beam aluminum profiles 102 are connected at right angles in sequence to form a rectangular frame, and the rectangular frame is the largest area of ​​activity for the entire picking robot; the midpoint of the two lower beam aluminum profiles 103 is connected to the end of a horizontal beam aluminum profile 102 in the rectangular frame; the ends of the two oblique beam aluminum profiles 104 are respectively connected to the midpoint of the vertical beam and the end of the lower beam aluminum profile 103; four fixed pulleys 5 are respectively installed on the four corners of the rectangular frame to guide the flexible cable.

[0050] Combine Figure 3 、 Figure 4 The flexible cable drive device 2 includes a first flexible cable drive motor 201, a coupling 202, a reel mechanism, and a reel fixing plate 203; the reel mechanism includes a bearing 204, a reel 205, and a fixed bracket 206; the coupling 202 is connected to the end of the first drive motor 201 and one end of the reel 205, and both ends of the reel are respectively installed on the bearing 204; the first drive motor 201 and the fixed bracket 206 are installed on the reel fixing plate 203; the four flexible cable drive devices 2 are respectively fixed at corresponding positions of the rectangular frame; the flexible cable group 7 is four flexible cables, the same end of which is fixed to the four corners of the servo electric cylinder fixing frame 9, and the other end is wrapped around the four fixed pulleys 5 and wound on the corresponding reel 205.

[0051] Furthermore, since the position of the cable drive device 2 on the main frame 1 is fixed, during the movement of the end effector, the positional relationship between them remains constant and the four cables are always kept in a tensioned state. By analyzing the kinematic geometric model of the four cables, the position information of the servo electric cylinder fixing frame 9 in the picking robot is recorded in real time by the camera module 3, and the kinematic inverse solution is performed to calculate the length value of each traction cable. The number of pulses of the first drive motor 201 in the cable drive device 2 can be obtained, thereby realizing the translation of the end effector of the picking robot in the X and Y directions.

[0052] Combine Figure 5 The camera module 3 is fixed to the front end of the servo cylinder 8; the camera module 3 captures the position of the target tomato and the position of the flexible manipulator 4 in real time; the target tomato in the image is identified using a deep learning algorithm model, and the point cloud cluster of the target tomato is separated by establishing a coordinate transformation matrix relationship between the normalized image pixel coordinate system and the base coordinate system of the picking robot. The mean coordinates of the separated point cloud clusters are calculated, and the spatial position of the target tomato is described based on the mean coordinates.

[0053] Combine Figure 6 、 Figure 7 The middle part of the servo electric cylinder 8 is connected to the servo electric cylinder fixing frame 9; the flexible manipulator 4 is composed of a second drive motor 401 and two flexible clamps; the second drive motor 401 of the flexible manipulator is connected to the front end shaft of the servo electric cylinder 8 through the fixing plate 10; the servo electric cylinder 8 realizes the movement of the flexible manipulator 4 in the Z-axis direction of space through the rotation of the drive motor 801; the second drive motor 401 of the flexible manipulator 4 controls the closing of its clamps to realize the picking of tomato fruits.

[0054] Combine Figure 8The bottom movable platform 6 under the main frame 1 is composed of a rubber track chassis assembly 601 and a laser radar 602, which can realize the positioning and navigation of the picking robot in the tomato orchard.

[0055] Combine Figure 9 ,A control method for a vision-guided intelligent lightweight and flat tomato picking robot includes the following steps:

[0056] Step 1: The control system performs a self-check and initialization, and each drive motor and force sensor is reset. The rubber track chassis assembly 601 in the bottom movable platform 6 is manually operated to move from a starting point in the tomato orchard. The laser radar 602 scans the surrounding environment and returns to the starting point to obtain an incremental global environment map.

[0057] Step 2: The camera module 3 of the picking robot captures and records the current initial position of the flexible manipulator 4 as the basic reference coordinate point to establish the robot coordinate system;

[0058] Step 3: The end effector of the picking robot traverses the entire picking area from its current initial position. The camera module 3 captures an image of the target tomato. The main controller processes the depth map of the target tomato fruit and locates the key points of picking to obtain the three-dimensional coordinates of the key points of picking tomato in the captured image.

[0059] Step 4: The main controller obtains the three-dimensional coordinates of the key tomato picking points fed back by the camera, plans the picking sequence of the tomato fruits, and sends the picking sequence instructions to the first drive motor 201 of the flexible cable drive device 2, the servo cylinder 8, and the second drive motor 401 of the flexible manipulator 4, controlling the plane X and Y axis movement of the four flexible cables in the rectangular frame, the Z axis movement of the servo cylinder in space, and the picking action of the flexible manipulator 4;

[0060] Step 5: The camera module 3 captures and records the movement position of the flexible manipulator 4 in real time;

[0061] Step 6: The main controller calculates the current motion position of the flexible manipulator 4 using the motor steps and speed feedback from the encoder of the first drive motor 201 of the flexible cable drive device 2 and the encoder of the drive motor 801 of the servo electric cylinder. It compares this with the real-time motion position of the flexible manipulator 4 feedback from the camera, and fine-tunes the steps and speeds of each drive motor to adjust the real-time position of the flexible manipulator 4.

[0062] Step 7: The end effector of the picking robot returns to its initial position and traverses the entire picking area again. The camera module 3 captures the image of the target tomato and feeds it back to the main controller to detect whether the tomatoes in the picking area have been picked. If not, the main controller will continue to plan the picking sequence with the three-dimensional coordinate points of the unpicked target tomatoes and send the picking instructions to the soft cable drive device 2 and the end effector.

[0063] Step 8: At this point, the picking work in the current area is completed, the end effector returns to the initial position, and the bottom movable platform 6 is positioned and navigated to the next picking area in the tomato orchard.

[0064] Implementation Case 2

[0065] This embodiment provides a vision-guided, intelligent, lightweight, and flattened tomato-picking robot. This robot, based on Example 1, adds a path planning algorithm for the tomato-picking sequence. Based on the principle of the shortest tomato-picking path, the tomato-picking path planning problem is transformed into a single traveling salesman problem. This problem is solved using a genetic algorithm, and the algorithm's iterative calculations achieve optimal tomato-picking sequence planning. Furthermore, the present invention allows for flexible arrangement of the main frame 1 dimensions, the positions of the various cable drive devices 2, and the four fixed pulleys 5, based on the tomato-picking area, optimizing the robot's internal space and expanding its applicability within tomato orchards.

[0066] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements or improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A vision-guided intelligent lightweight flat tomato picking robot, characterized by: The invention comprises: a main frame (1), a flexible cable driving device (2), a flexible cable group (7), an end actuator, a bottom movable platform (6), a camera module (3) and a control system; The main frame (1) is a frame of the picking robot built with aluminum profiles and is mounted on a bottom movable platform (6). The size of the aluminum profiles can be customized according to the size of the tomato plants and the tomato orchard. The flexible cable driving device (2) is arranged on the main frame (1), and utilizes a first driving motor (201) to drive a reel (205) on a coupling (202) to reel the flexible cable, thereby achieving planar motion of the end actuator; The same ends of the four flexible cables in the flexible cable group (7) are respectively fixed to the four corners of the servo electric cylinder fixing frame (9), and the other ends are wound around the four fixed pulleys (5) and wound on the corresponding reels (205); The end actuator comprises a servo electric cylinder (8) and a flexible manipulator (4), wherein the servo electric cylinder (8) realizes the movement of the Z axis in space, and the flexible manipulator (4) is connected to the end of the servo electric cylinder (8) and is used for grabbing the fruit; The bottom movable platform (6) includes a rubber track chassis assembly (601) and a laser radar (602); The camera module (3) is used for visually guiding the recognition of the three-dimensional coordinates of the tomato fruit for picking; The control system feeds back the three-dimensional coordinates of the tomato fruit to be picked through the camera module (3), drives the corresponding motor to rotate, and makes the four flexible cables in the flexible cable group (7) retract and release the corresponding lengths, so as to realize the planar movement of the end actuator to the plane X and Y coordinates where the tomato is located, and then drives the servo electric cylinder (8) to move in the Z-axis direction to the Z-axis coordinate where the tomato is located, and finally drives the flexible manipulator (4) to grasp the tomato; The main frame (1) further comprises two vertical beam aluminum profiles (101) and two horizontal beam aluminum profiles (102) connected in sequence at right angles to form a rectangular frame; the midpoints of the two lower beam aluminum profiles (103) are connected to the two ends of a horizontal beam aluminum profile (102) below the rectangular frame; the ends of the two oblique beam aluminum profiles (104) are respectively connected to the midpoints of the two vertical beam aluminum profiles (101) and the ends of the two lower beam aluminum profiles (103); four fixed pulleys (5) are respectively installed at the four corners of the rectangular frame to guide the flexible cable; The flexible cable drive device (2) further comprises a first drive motor (201), a coupling (202), a reel mechanism, and a reel fixing plate (203); wherein the reel mechanism comprises a bearing (204), a reel (205), and a bearing fixing bracket (206); the coupling (202) is connected to the drive shaft of the first drive motor (201) and one end of the reel (205); both ends of the reel (205) are respectively mounted on the bearing (204), and the bearing is fixed by the bearing fixing bracket (206); the first drive motor (201) and the bearing fixing bracket (206) are mounted on the reel fixing plate (203); and the four flexible cable drive devices (2) are respectively fixed at corresponding positions of the rectangular frame.

2. A picking robot according to claim 1, characterized in that: The end actuator further comprises a servo electric cylinder (8) whose middle part is connected to a servo electric cylinder fixing frame (9); the flexible manipulator (4) is composed of a second drive motor (401) and two flexible clamping claws; the rear end of the second drive motor (401) of the flexible manipulator (4) is connected to the end of the main rod of the servo electric cylinder (8); the servo electric cylinder (8) controls the movement of the flexible manipulator (4) in the Z-axis direction of space; and the second drive motor (401) of the flexible manipulator (4) controls the closing of its clamping claws to achieve the picking of tomato fruits.

3. The picking robot according to claim 1, characterized in that: The camera module (3) is fixed on the flexible manipulator (4); the camera module (3) captures the position of the target tomato and the position of the flexible manipulator (4) in real time; a deep learning algorithm model is used to identify the tomato in the image, a point cloud cluster of the target tomato is separated by establishing a coordinate conversion matrix relationship between a normalized image pixel coordinate system and a base coordinate system of the picking robot, the mean coordinates of the separated point cloud clusters are calculated, and the spatial position of the target tomato is described based on the mean coordinates.

4. The picking robot according to claim 1, characterized in that: The control system comprises a motor controller and a main controller; the main controller calculates the three-dimensional coordinates of the target tomato and the relative position of the flexible manipulator (4) fed back in real time by the camera module (3), thereby obtaining the Z-axis motion parameters of the servo electric cylinder (8) and the X, Y plane motion parameters of the four flexible cable drive devices (2), i.e., the corresponding rotation steps and speeds of the first drive motor (201); the first drive motor (201) is then rotated by the motor controller at the corresponding number of steps and speeds; at the same time, the motor controller obtains the corresponding rotation steps and speeds fed back by the encoders of the four first drive motors (201) in real time to the main controller; the main controller compares the values ​​fed back by the encoders with the spatial position information of the flexible manipulator (4) obtained by the current camera module (3), and adjusts the number of steps and speeds of each drive motor in real time.

5. The picking robot according to claim 4, characterized in that: The connection points between the ends of the four flexible cables and the four corners of the servo fixed frame (9) all include force sensors, which detect the tensioning force of the four flexible cables in real time and feed it back to the main controller; the main controller adjusts the number of steps and speed of the first drive motor (201) in real time through the motor controller to optimize the force distribution of the four flexible cables.

6. The picking robot according to claim 1, characterized in that: The laser radar (602) in the bottom movable platform (6) performs a three-dimensional scan of the scene in the tomato orchard during the movement, and then realizes three-dimensional reconstruction of the scene through the main controller to construct an incremental global environment map. Based on the obtained global environment map, when the rubber track chassis assembly (601) automatically operates in the tomato orchard, it scans and generates three local contour maps according to the current location of the laser radar (602). The maps are compared with the global environment map to obtain the current position of the bottom movable platform (6) in the global map, thereby realizing positioning and navigation of the bottom movable platform (6).

7. A control method for a vision-guided intelligent lightweight flat tomato picking robot according to any one of claims 1 to 6, characterized in that: Proceed as follows: Step 1: The control system performs self-checking and initialization, and each drive motor and force sensor is reset; the rubber track chassis assembly (601) in the bottom movable platform (6) is manually operated to move from a starting point in the tomato orchard, and the laser radar (602) scans the surrounding environment and returns to the starting point to obtain an incremental global environment map; Step 2: The camera module (3) of the picking robot captures and records the initial position of the current flexible manipulator (4) as a basic reference coordinate point to establish a robot coordinate system; Step 3: The end effector of the picking robot traverses the entire picking area from the current initial position, the camera module (3) collects the target tomato image, and the main controller processes the target tomato fruit depth map and locates the picking key points to obtain the three-dimensional coordinates of the tomato picking key points in the collected image; Step 4: The main controller obtains the three-dimensional coordinates of the key points of tomato picking fed back by the camera module (3), performs picking sequence planning for the tomato fruits, and sends the order instructions for each picking to the first drive motor (201) of the flexible cable drive device (2), the servo electric cylinder (8) and the second drive motor (401) of the flexible manipulator (4), thereby controlling the movement of the four flexible cables in the plane X and Y axis directions in the rectangular frame, the movement of the servo electric cylinder in the space Z axis and the picking action of the flexible manipulator (4); Step 5: The camera module (3) captures and records the movement position of the flexible manipulator (4) in real time; Step 6: The main controller calculates the current motion position of the flexible manipulator (4) by using the motor steps and speed feedback from the encoder of the first drive motor (201) of the flexible cable drive device (2) and the encoder of the servo electric cylinder drive motor (801), compares the current motion position of the flexible manipulator (4) with the real-time motion position of the flexible manipulator (4) feedback from the camera, and drives the steps and speeds of each drive motor to fine-tune and adjust the real-time position of the flexible manipulator; Step 7: The end effector of the picking robot returns to the initial position and traverses the entire picking area again. The camera module (3) collects the target tomato image and feeds it back to the main controller to detect whether the tomatoes in the picking area have been picked. Otherwise, the main controller will continue to plan the picking sequence with the three-dimensional coordinate points of the unpicked target tomatoes and send the picking instructions to the soft cable drive device (2) and the end effector; Step 8: At this point, the picking work in the current area is completed, the end effector returns to the initial position, and the bottom movable platform (6) is positioned and navigated to the next picking area in the tomato orchard.

Images