A chrysanthemum picking robot and a flower picking method thereof

Abstract

The application relates to a chrysanthemum picking robot and a flower picking method thereof, which comprises a wheel hub motor, a rack, a battery, a collecting box, a negative pressure fan, an industrial control box, a parallel connection mechanical arm, a hose, a terminal picking mechanism, a binocular depth camera, a box cover and a buckle; the parallel connection mechanical arm comprises a static platform, a servo motor, a driving arm, a driven arm and a dynamic platform; the terminal picking mechanism comprises a horn mouth, an electric push rod and a clamping jaw; the parallel connection mechanical arm is installed on the upper front end of the rack; the collecting box is installed on the rear part of the rack; the terminal picking mechanism is installed on the dynamic platform of the parallel connection mechanical arm; the small end of the horn mouth of the terminal picking mechanism is connected with one side of the collecting box through the hose; the other side of the collecting box is connected with the negative pressure fan; the binocular depth camera is installed at the upper middle position of the parallel connection mechanical arm; the application integrates a visual perception system and an intelligent identification algorithm, so that the robot can distinguish the position of chrysanthemum flowers, realizes accurate picking and reduces the labor cost.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of picking robots, in particular to a chrysanthemum morifolium picking robot and a flower recognition method thereof. BACKGROUND

[0002] Chrysanthemum morifolium is a traditional Chinese precious medicinal material with a long history and rich cultural connotations, and has the effects of stopping diarrhea, reducing inflammation, improving eyesight, lowering blood pressure, lowering blood lipids, and strengthening the body, and is widely used in medicinal materials, skincare products and food.

[0003] Chrysanthemum morifolium is picked in stages, grows towards the sun, and the flower is mostly located on the surface of the plant, while the bud and the bud are mostly arranged below the flower in a three-dimensional distribution from top to bottom. Chrysanthemum morifolium planting patterns are diverse, plants are uneven, flower spatial arrangement is complex, opening time is different, and picking conditions are complex. Its picking is the most labor-intensive link in production operations, and is also the most difficult link to realize mechanization.

[0004] Chrysanthemum morifolium picking mainly relies on manual work. Since the flower of chrysanthemum morifolium is small and fragile, it needs to be picked manually to maintain the integrity of the flower. The existing equipment is usually comb tooth type, which cannot accurately identify chrysanthemum morifolium, and has the problems of missed picking, flower damage and high impurity rate. Therefore, it is urgent to provide a more efficient, flexible and accurate chrysanthemum morifolium picking robot that can quickly and accurately identify and pick mature flowers. SUMMARY

[0005] In order to overcome the shortcomings of the prior art, a chrysanthemum morifolium picking robot and a flower recognition method thereof are provided, and the specific technical solutions of the present application are as follows:

[0006] A chrysanthemum morifolium picking robot, comprising a hub motor, a rack, a battery, a collection box, a negative pressure fan, an industrial control box, a parallel mechanical arm, a hose, a terminal picking mechanism, a binocular depth camera, a bin cover and a buckle, the parallel mechanical arm comprises a static platform, a servo motor, a driving arm, a driven arm and a moving platform, the hub motor has four, which are respectively installed at the four corners of the bottom of the rack, the parallel mechanical arm is installed on the upper front end of the rack, the driving arm has three, one end of which is hinged to the static platform and the other end is hinged to the driven arm, the driven arm has three, one end of which is hinged to the moving platform and the other end is hinged to the driving arm, the collection box is installed at the rear of the rack, the terminal picking mechanism is installed on the moving platform of the parallel mechanical arm, the small end of the trumpet mouth of the terminal picking mechanism is connected to one side of the collection box through the hose, the other side of the collection box is connected to the negative pressure fan, the bin cover has two, which are respectively hinged to the bottom of the collection box, the buckle has two, which are respectively installed on the two bin covers, the battery has three, which are installed in the middle rear part of the rack below the collection box, the industrial control box is installed above the battery and below the collection box, and the binocular depth camera is installed at the upper middle position of the parallel mechanical arm.

[0007] A chrysanthemum picking robot flower picking method, comprising the following steps:

[0008] S1: Start the binocular depth camera to take pictures of chrysanthemum, and output the image with depth information to the industrial computer box;

[0009] S2: According to the depth information image output in S1, the planar coordinates of each chrysanthemum in the image are identified through Yolo11n deep learning model, and the corresponding depth information is read from the depth information image through the planar coordinates, so as to obtain the three-dimensional space coordinates of each chrysanthemum;

[0010] S3: Curve fitting is performed on the three-dimensional space coordinates of each chrysanthemum to obtain the curve information of the top surface of the chrysanthemum cluster, and then the top surface is divided into a plurality of small curved surfaces, the center point coordinates of each small curved surface are calculated, and a shortest picking path is solved through genetic algorithm according to the center point coordinates of each small curved surface, and the picking path information is sent to the parallel manipulator;

[0011] S4: The parallel manipulator drives the end picking mechanism to move above the first point of the picking path, the clamping jaw is opened, and then it is moved downward by a certain distance to completely envelop the chrysanthemum flowers in the small curved surface area, and then the clamping jaw is closed, and then the parallel manipulator drives the end picking mechanism to move upward, and the chrysanthemum flowers are picked off through the pulling force;

[0012] S5: Start the negative pressure fan, and suck the chrysanthemum flowers into the collecting box through the horn mouth and the hose;

[0013] S6: Repeat the steps of S4-S5 until the flowers in the small curved surface area divided in the step S3 are picked off.

[0014] S7: After the chrysanthemum picking robot finishes picking a region, it continues to walk forward and moves to the next picking region, and repeats the steps of S1-S6, so as to complete the picking of chrysanthemum in the field.

[0015] Compared with the prior art, the present application has the following advantages:

[0016] (1) The end picking mechanism provided by the present application realizes flower picking by the way that the two side comb-shaped clamping jaws act on the receptacle and break the peduncle through pulling force, without force acting on the petals, which can reduce flower damage, and the two sides combs can pick 10-15 flowers at the same time to ensure picking efficiency.

[0017] (2) The chrysanthemum picking robot provided by the present application integrates visual perception system and intelligent recognition algorithm, so that the robot can distinguish the position of chrysanthemum flowers and realize accurate picking, thereby reducing the missed picking rate and impurity rate.

[0018] (3) The Hangzhou white chrysanthemum picking robot provided by the present invention, when it finishes picking Hangzhou white chrysanthemum in one area and moves to the next picking area, acquires the surrounding environment information through a binocular depth camera and transmits the image to the industrial control box. The industrial control box identifies the two side lines of the field ridge image and then calculates the forward direction. Based on the direction information, it controls the rotation speed and speed difference of the hub motor to realize the autonomous forward movement and navigation of the Hangzhou white chrysanthemum picking robot, thereby reducing labor costs. Attached Figure Description

[0019] Fig. 1 This is a three-dimensional structural schematic diagram from the first perspective of the present invention;

[0020] Fig. 2 This is a three-dimensional structural schematic diagram from a second perspective of the present invention;

[0021] Fig. 3 This is a schematic diagram of the open state of the gripper of the end-feeding mechanism of the present invention.

[0022] In the diagram: 1. Hub motor; 2. Frame; 3. Battery; 4. Collection box; 5. Negative pressure fan; 6. Industrial control box; 7. Parallel robotic arm; 8. Hose; 9. End-effector harvesting mechanism; 10. Binocular depth camera; 11. Box cover; 12. Buckle; 13. Static platform; 14. Servo motor; 15. Active arm; 16. Driven arm; 17. Moving platform; 18. Trumpet mouth; 19. Electric push rod; 20. Gripper. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of the embodiments of this invention will be described in more detail below with reference to the accompanying drawings. In the drawings, the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The described embodiments are some, but not all, embodiments of this invention. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention. The embodiments and directional terms described below with reference to the accompanying drawings are exemplary and intended to explain this invention, and should not be construed as limiting this invention.

[0024] Reference Figs. 1-3 As shown: A Hangzhou white chrysanthemum harvesting robot includes a hub motor 1, a frame 2, a battery 3, a collection box 4, a negative pressure fan 5, an industrial control box 6, a parallel robotic arm 7, a hose 8, an end-effector harvesting mechanism 9, a binocular depth camera 10, a box cover 11, and a buckle 12. The parallel robotic arm includes a static platform 13, a servo motor 14, an active arm 15, a driven arm 16, and a moving platform 17.

[0025] The hub motor 1 is 4 in total, which is installed at the four corners of the bottom of the rack 2; the parallel manipulator 7 is installed on the upper part of the front end of the rack 2; the driving arm 15 is 3 in total, one end of which is hinged with the static platform 13, and one end of which is hinged with the driven arm 16; the driven arm 16 is 3 in total, one end of which is hinged with the dynamic platform 17, and one end of which is hinged with the driving arm 15, the collecting box 4 is installed at the rear of the rack 2; the end picking mechanism 9 is installed on the dynamic platform 17 of the parallel manipulator 7; the small end of the horn mouth 18 of the end picking mechanism 9 is connected with one side of the collecting box 4 through the hose 8; the other side of the collecting box 4 is connected with the negative pressure fan 5; the box cover 11 is 2 in total, which is hinged with the bottom of the collecting box 4 respectively, the buckle 12 is 2 in total, which is installed on the two box covers 11 respectively; the battery 3 is 3 in total, which is installed on the middle rear part of the rack 2 and below the collecting box 4; the industrial control box 6 is installed above the battery 3 and below the collecting box 4, the binocular depth camera 10 is installed at the middle position of the upper part of the parallel manipulator 7;

[0026] As an embodiment of the application, the end picking mechanism 9 further comprises an electric push rod 19 and a clamping jaw 20, the clamping jaw 20 is 2 in total, which is symmetrically arranged on both sides of the horn mouth 18 and hinged with the side wall of the large end of the horn mouth 18; one end of the electric push rod 19 is hinged with the dynamic platform 17, and the other end is hinged with the clamping jaw 20, and the clamping jaw 20 is uniformly distributed with comb-shaped structure;

[0027] As an embodiment of the application, the hub motor 1 is a whole driving system, the industrial control box 6 is used as the core controller, Kalman filtering algorithm and incremental PID control algorithm are loaded in the industrial control box 6, so as to ensure the stability of the robot advancing.

[0028] As an embodiment of the application, the industrial control box 6 comprises a microcomputer loaded with an embedded Windows system and a motion controller, Python is used as a programming language and a plurality of Python function libraries are called, including OpenCV, Camera, threading and GPIO; wherein the GPIO library is used to control the motion controller pin output high and low level, so as to control the motion of the parallel manipulator 7 and the hub motor 1.

[0029] A chrysanthemum picking robot flower picking method, comprising the following steps:

[0030] A chrysanthemum picking robot moves to a working site;

[0031] S1: start the binocular depth camera 10 to shoot chrysanthemum pictures, and output the images with depth information to the industrial control box 6;

[0032] S2: According to the depth information image output in S1, the planar coordinates of each chrysanthemum in the image are identified through a Yolo11n deep learning model, and the corresponding depth information is read from the depth information image through the planar coordinates, so as to obtain the three-dimensional spatial coordinates of each chrysanthemum;

[0033] S3: The three-dimensional spatial coordinates of each chrysanthemum are subjected to surface fitting to obtain the surface information of the top surface of the chrysanthemum cluster, and then the top surface is divided into a plurality of small curved surfaces, the center point coordinates of each small curved surface are calculated, a picking path is planned according to the center point coordinates of each small curved surface, and the picking path information is sent to the parallel manipulator 7;

[0034] S4: The parallel manipulator 7 drives the end picking mechanism 9 to move above the first point of the picking path, the clamping jaw 20 is opened, and then moved downward by a certain distance to completely envelope the chrysanthemum flowers in the small curved surface region, and then the clamping jaw 20 is closed, and then the parallel manipulator 7 drives the end picking mechanism 9 to move upward, and the chrysanthemum flowers are picked off through the pulling force;

[0035] S5: Start the negative pressure fan 5, and the chrysanthemum flowers are sucked into the collection box 4 through the horn mouth 18 and the hose 8.

[0036] S6: Repeat the steps of S4-S5 until the flowers in the small curved surface regions divided in the step S3 are picked off.

[0037] S7: After the chrysanthemum picking robot finishes picking in one region, it continues to walk forward and moves to the next picking region, and repeats the steps of S1-S6, and thus the picking of chrysanthemums in the field is completed.

[0038] As an embodiment of the present application, the Yolo11n deep learning model training in step S2 further includes the following steps:

[0039] S21: Collect a large number of chrysanthemum sample images of different plots and light conditions, and mark the flower center region position;

[0040] S22: The chrysanthemum images of different plots and light conditions are used as input parameters of the depth information model to train the depth neural network model, extract feature maps of different scales, fuse the feature maps of different scales, and obtain a sample library of comprehensive feature maps.

[0041] In one embodiment of the present invention, the deep neural network model for Hangzhou white chrysanthemum in step S2 includes a convolutional neural network, which comprises convolutional layers, pooling layers, and fully connected layers. The convolutional layers extract information from the input Hangzhou white chrysanthemum image; this information is called image features. These features are represented by each pixel in the image through combinations or independent methods, such as texture features, size, and color features. The pooling layers then select from the features extracted by the convolutional layers. Following the pooling layers is a fully connected layer, which transforms all the feature matrices from the pooling layers into a one-dimensional feature vector. The deep neural network model automatically extracts and utilizes the multi-scale features of the Hangzhou white chrysanthemum image.

[0042] In one embodiment of the present invention, in step S3, the projection of the curved surface of the top surface of the Hangzhou white chrysanthemum bush in the vertical direction is a rectangle, and the projection of the small curved surface in the vertical direction is a rectangle, which is identical to the rectangular cross-section of the large end of the trumpet mouth 18 of the end picking mechanism 9.

[0043] In one embodiment of the present invention, the gripper 20 of the end effector 9 in S4 has a comb-like structure, the cross-sectional shape of a single comb tooth is stepped, and the gap between two comb teeth decreases from large to small.

[0044] In one embodiment of the present invention, when the Hangzhou white chrysanthemum picking robot finishes picking in one area and moves to the next picking area in step S7, it acquires the surrounding environment information through the binocular depth camera 10 and transmits the image to the industrial control box 6. The industrial control box 6 identifies the two side lines of the field ridge image and calculates the direction of movement. Based on the direction information, it controls the rotation speed and speed difference of the four hub motors 1 to realize the autonomous movement and navigation of the Hangzhou white chrysanthemum picking robot.

[0045] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for picking flowers of Hangzhou white chrysanthemum using a picking robot, characterized in that, The steps include the following: S1: Start the binocular depth camera (10) to take pictures of Hangzhou white chrysanthemums and output the image with depth information to the industrial control box (6); S2: Based on the depth information image output in S1, the planar coordinates of each Hangzhou white chrysanthemum in the image are identified by the improved YOLO11n deep learning model. Then, the corresponding depth information is read from the depth information image through the planar coordinates to obtain the three-dimensional spatial coordinates of each Hangzhou white chrysanthemum. S3: Perform surface fitting on the three-dimensional spatial coordinates of each Hangzhou white chrysanthemum to obtain the surface information of the top surface of the Hangzhou white chrysanthemum cluster. Then divide it into several small surfaces, calculate the center point coordinates of each small surface, and solve the shortest picking path through the genetic algorithm based on the center point coordinates of each small surface. Send the picking path information to the parallel robotic arm (7). S4: The parallel robotic arm (7) drives the end-picking mechanism (9) to move above the first point of the picking path. The gripper (20) opens and moves down a certain distance to completely enclose the Hangzhou white chrysanthemum flowers in the small curved area. The gripper (20) closes, and then the parallel robotic arm (7) drives the end-picking mechanism (9) to move up and pick the Hangzhou white chrysanthemum flowers by pulling force. S5: Start the negative pressure fan (5) to suck the Hangzhou white chrysanthemum flowers into the collection box (4) through the horn (18) and hose (8); S6: Repeat steps S4-S5 until all the flowers in the several small curved surface areas divided in step S3 have been picked. S7: After harvesting one area, continue walking forward to the next harvesting area and repeat steps S3-S6. Repeat this process to complete the harvesting of Hangzhou white chrysanthemums in the field. The projection of the curved surface of the top surface of the S3 chrysanthemum bush in the vertical direction is a rectangle, and the projection of the small curved surface in the vertical direction is a rectangle, which is identical to the rectangular cross section of the large end of the trumpet mouth (18) of the end picking mechanism (9). The end-harvesting mechanism in S4 has a comb-like structure on its gripper. The cross-sectional shape of a single comb tooth is stepped, and the gap between two comb teeth decreases from large to small.

2. The flower picking method of the Hangzhou white chrysanthemum picking robot according to claim 1, characterized in that, The improved YOLO11 deep learning model in S2 is trained through the following steps: S21: Collect a large number of images of Hangzhou white chrysanthemum samples from different plots of land under different lighting conditions, and mark the location of the central area of ​​the flower. S22: Use images of Hangzhou white chrysanthemums under different plots and lighting conditions as input parameters for the deep information model to train the deep neural network model, extract feature maps at different scales, and fuse the feature maps at different scales to obtain a sample library of comprehensive feature maps.

3. The flower picking method of the Hangzhou white chrysanthemum picking robot according to claim 1, characterized in that, In step S7, when the Hangzhou white chrysanthemum picking robot finishes picking in one area and moves to the next picking area, it acquires the surrounding environment information through a binocular depth camera (10) and transmits the image to the industrial control box (6). The industrial control box (6) identifies the two side lines of the field ridge image and calculates the forward direction. Based on the direction information, it controls the rotation speed and speed difference of the four hub motors (1) to realize the autonomous forward movement and navigation of the Hangzhou white chrysanthemum picking robot.

4. A Hangzhou white chrysanthemum harvesting robot, used to implement the flower harvesting method described in claim 1, characterized in that, The system includes a hub motor (1), a frame (2), a battery (3), a collection box (4), a negative pressure fan (5), an industrial control box (6), a parallel robotic arm (7), a hose (8), an end-feeding mechanism (9), a binocular depth camera (10), a box cover (11), and a buckle (12). The parallel robotic arm (7) includes a static platform (13), a servo motor (14), an active arm (15), a driven arm (16), and a moving platform (17). There are four hub motors (1), which are installed at the four corners of the bottom of the frame (2). The parallel robotic arm (7) is installed on the upper front end of the frame (2). There are three active arms (15), one end of which is hinged to the static platform (13), and the other end of which is hinged to the driven arm (16). There are three driven arms (16), one end of which is hinged to the moving platform (17), and the other end of which is hinged to the active arm (15). The collection box (4) is installed at the rear of the frame (2); the end-picking mechanism (9) is installed on the moving platform (17) of the parallel robotic arm (7); the small end of the horn (18) of the end-picking mechanism (9) is connected to one side of the collection box (4) through a hose (8); the other side of the collection box (4) is connected to the negative pressure fan (5); there are two boxes (11), which are hinged to the bottom of the collection box (4) respectively; there are two buckles (12), which are installed on the two boxes (11) respectively; there are three batteries (3), which are installed in the middle and rear of the frame (2) and located below the collection box (4); the industrial control box (6) is installed above the battery (3) and below the collection box (4); the binocular depth camera (10) is installed in the middle position of the three servo motors (14) on the upper part of the parallel robotic arm (7); The end-picking mechanism (9) also includes an electric push rod (19) and a gripper (20). There are two grippers (20), which are symmetrically arranged on both sides of the flared mouth (18) and hinged to the side wall of the large end of the flared mouth (18). One end of the electric push rod (19) is hinged to the moving platform (17), and the other end is hinged to the gripper (20). The gripper (20) has a comb-like structure evenly distributed on it. The industrial control box (6) includes a microcomputer equipped with an embedded Windows 10 IoT system and a motion controller. Python is used as the programming language and multiple Python function libraries are called, including OpenCV, Camera, threading and GPIO. The GPIO library is used to control the high and low level output of the motion controller pins to control the movement of the parallel robotic arm (7) and the hub motor (1).

Images