Three-degree-of-freedom grape harvesting robot and harvesting method

Abstract

The invention provides a three-degree-of-freedom grape harvesting robot, and relates to the field of agricultural intelligent machinery, the three-degree-of-freedom grape harvesting robot comprises a chassis (10), a walking mechanism (20), a rotating plate (30), a three-coordinate mechanism, an end execution mechanism, a visual assembly, a navigation system and an electric control box, the bottom surface of the chassis (10) is provided with the walking mechanism (20), the end surface of the chassis (10) is rotatably provided with the rotating plate (30), and the end surface of the rotating plate (30) is provided with the three-coordinate mechanism; the three-coordinate mechanism comprises a Y-axis assembly, an X-axis assembly and a Z-axis assembly, the tail end executing mechanism is arranged at the end, away from the Y-axis assembly, of the X-axis assembly, the visual assembly is located on the upper side of the tail end executing mechanism, and the navigation system is located on the side face of the chassis. According to the harvesting robot, the three-degree-of-freedom rectangular coordinate design is adopted, the mechanical structure is simplified, environmental interference in the harvesting process is avoided, the flexibility and accuracy of the harvesting process are guaranteed, and therefore rapid, efficient and low-cost grape harvesting is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of agricultural intelligent machinery, and in particular to a three-degree-of-freedom grape harvesting robot and a harvesting method. Background Art

[0002] Grapes are woody vines of the genus Vitis in the family Vitaceae. According to their uses, they can be divided into fresh food varieties, wine-making varieties, and dried varieties. Among them, the planting methods and harvesting quality requirements of wine-making varieties and fresh food varieties are different. Wine-making varieties have a large planting area and each bunch of fruits grows densely, so they need to be harvested quickly, efficiently, and at low cost after maturity. At present, grape harvesting is mainly based on manual picking and mechanical picking. Manual picking requires one hand to hold the grape bunch and the other hand to cut the fruit stems, so as to achieve complete and accurate picking of grapes. However, this picking method not only greatly wastes labor resources and increases harvesting costs, but also increases harvesting time and reduces harvesting efficiency, and cannot achieve the purpose of fast, efficient, and low-cost harvesting of wine grapes. Mechanical picking mainly completes the picking and collection of grapes by vibrating the grape plants to make the grape fruits fall off. This method is easy to cause damage to the grape fruits. It is suitable for picking raw materials for raisins or grape juice, but not for wine grapes that need to be processed as a whole bunch. In addition, with the current development of intelligent agriculture, there are still some robots in the existing technology that use machine vision to locate grapes and realize automatic picking. However, the existing picking robots have problems such as difficulty in identifying picking points, low picking success rate, and low harvesting efficiency. In addition, most of the existing picking robots use multi-joint, multi-degree-of-freedom robotic arms, which not only have high manufacturing costs and difficult to control the robotic arms, but the multi-degree-of-freedom robotic arms are also prone to contact with grape fruits, vines, etc. during the control process, resulting in fruit damage, vines entangled with robotic arms, etc., thereby increasing the fruit damage rate and even causing problems such as robotic arm jamming, affecting the normal progress of grape harvesting. Summary of the invention

[0003] In view of the problems existing in the above-mentioned prior art, the object of the present invention is to provide a three-degree-of-freedom grape harvesting robot. The harvesting robot adopts a three-degree-of-freedom rectangular coordinate design, which not only simplifies the mechanical structure and avoids interference with the target fruits around or interference with the grape vines during the harvesting process, but also effectively reduces the complexity and cost of the system, ensures flexibility and accuracy in the harvesting process, thereby realizing fast, efficient and low-cost grape harvesting.

[0004] Another object of the present invention is to provide a harvesting method of the above-mentioned grape harvesting robot for harvesting wine grapes.

[0005] The purpose of the present invention is achieved through the following technical solutions: A three-degree-of-freedom grape harvesting robot comprises a chassis, a walking mechanism, a rotating plate, a three-coordinate mechanism, an end actuator, a visual component, a navigation system and an electric control box. The walking mechanism is arranged on the bottom surface of the chassis and the walking mechanism adopts a crawler-type walking component for controlling the operation of the entire harvesting robot. The end surface of the chassis is rotatably arranged with a rotating plate and the end surface of the rotating plate is arranged with a three-coordinate mechanism. The three-coordinate mechanism comprises a Y-axis component, an X-axis component and a Z-axis component, and the Z-axis component is arranged on the end surface of the rotating plate, the Y-axis component is arranged on the Z-axis component, and the X-axis component is arranged between the Y-axis components. The end actuator is arranged at one end of the X-axis component away from the Y-axis component, and the visual component is arranged at one end of the X-axis component corresponding to the end actuator and is located on the upper side of the end actuator. The navigation system is located on the side of the chassis and the electric control box is arranged on the end surface of the chassis. The electric control box is out of position with the three-coordinate mechanism and is electrically connected to the walking mechanism, the control device of the rotating plate, the three-coordinate mechanism, the end actuator, the navigation system and the visual component respectively.

[0006] Based on the further optimization of the above scheme, the Z-axis assembly includes two groups of horizontal guide rails, two horizontal lead screws, two horizontal sliders and two horizontal slides. The two groups of horizontal guide rails are arranged in parallel on both sides of the end surface of the rotating plate, and a rotating horizontal lead screw is respectively arranged on the side of the two groups of horizontal guide rails close to each other. The outer walls of the horizontal lead screws are respectively threaded with the horizontal sliders, and the horizontal slides are slidably arranged on the horizontal guide rails, and the horizontal slides are fixedly connected to the corresponding horizontal sliders; the Y-axis assembly includes a connecting cross bar, two groups of vertical guide rails, two vertical lead screws, two vertical sliders and two vertical slides. The bottom surface of the connecting cross bar is respectively fixed to the end surfaces of the two horizontal sliders and the two horizontal slides. Two vertical guide rails parallel to each other are fixedly connected and connected on both sides of the end faces of the cross bar, a rotating vertical lead screw is respectively arranged on the side faces of the two sets of vertical guide rails away from each other, and the outer walls of the vertical lead screws are respectively threadedly sleeved with vertical sliders, and vertical slides are slidably arranged on the outer walls of the vertical guide rails, and the vertical slides are fixedly connected to the corresponding vertical sliders; the X-axis assembly includes a positioning cross plate, a moving lead screw and a linkage bracket, the two ends of the positioning cross plate are respectively fixedly connected to the inner side walls of the corresponding vertical slides, and a rotating moving lead screw is arranged on one side of the positioning cross plate, the linkage bracket is perpendicular to the positioning cross plate, and the bottom surface of the linkage bracket is threadedly connected to the moving lead screw and is slidably connected to the positioning cross plate.

[0007] Based on the further optimization of the above scheme, the end actuator includes two servo motors, two adjustment ropes, a support plate, a connecting hose, a universal ball cage, a high-speed motor and a three-sided blade tool. The support plate is set on the bottom surface of the end of the linkage bracket away from the positioning cross plate through a connecting hose. There are multiple connecting hoses and they are evenly distributed on the outer circle of the end surface of the support plate; two servo motors are fixedly set on the end surface of the linkage bracket away from the positioning cross plate, and an adjustment rope is wrapped around the output ends of the two servo motors respectively. The end of the adjustment rope away from the servo motor passes through the fixed pulley set on the end surface of the linkage bracket and penetrates the linkage support. The bracket is connected to the end face of the support plate, the adjusting ropes are symmetrically distributed on both sides of the support plate, and the adjusting ropes and the connecting hoses are arranged in a different position; a universal ball cage coaxial with the support plate is arranged between the linkage bracket and the support plate and at the inner circle of the connecting hose; a high-speed motor is fixedly arranged on the bottom surface of the support plate, and the output end of the high-speed motor is fixedly connected to a three-sided blade tool through a coupling. The main body of the three-sided blade tool is a disc-shaped structure, and three main cutting surfaces are extended from the base (that is, three notches are evenly arranged on the outer circle of the disc-shaped structure, dividing the disc-shaped outer circle into three cutting surfaces), and each cutting surface is provided with a bionic angular groove similar to a dinosaur tooth.

[0008] Based on further optimization of the above solution, arc grooves are respectively provided on the inner ring of the three-sided cutting tool corresponding to the three cutting surfaces.

[0009] Based on further optimization of the above solution, the visual component includes a camera bracket and a visual camera, and the visual camera is fixedly arranged on the end face of the linkage bracket through the camera bracket.

[0010] Based on further optimization of the above solution, the navigation system includes two sets of antenna brackets and two sets of Beidou antennas. The two sets of antenna brackets are respectively arranged at the front and rear ends of the chassis and the Beidou antennas are fixed on the antenna brackets.

[0011] A harvesting method of a three-degree-of-freedom grape harvesting robot, comprising: Step 1: The harvesting robot moves to the area to be harvested, starts the visual camera to capture images and videos, and completes target recognition; Step 2: extract the navigation path in real time through the Beidou antenna, and start the harvesting robot to move toward the target identified in step 1 according to the navigation path; during the navigation process, the start, stop and steering control of the harvesting robot are realized by identifying obstacles between rows; Step 3: Start the three-coordinate mechanism and the end-effector to cut and harvest the target in real time.

[0012] Based on the further optimization of the above scheme, the improved YOLOv10 model is used in step 1 to realize target recognition, specifically: First, we obtain an image dataset D containing grape main vines, fruiting branches and fruit stalks. I 1 ,I 2,…,I N}, I i (i=1,2,…,N) represents a single image; and the image dataset is divided into training set, validation set and test set; Then, use the annotation tool to add bounding boxes and category labels to each bunch of fruit; After that, an improved YOLOv10 model is constructed. The YOLOv10 network model includes a deep learning network structure of Backbone, Neck and Head. The network model starts with image input, extracts features through the Backbone part, performs feature fusion through the Neck part, and finally performs regression and classification prediction through the Head part. Therefore, the FasterBlock module is used in the improved YOLOv10 model to replace the Bottleneck module in the C2f of the Backbone part to optimize the computational efficiency and improve the accuracy of small target detection:

[0013] Where: H.W. represents the size of the feature map, k represents the convolution and size, C p Indicates the number of channels calculated; In order to meet the detection needs of targets of different scales such as grape main vines and fruiting branches, the LSKA attention mechanism is used to replace the SPPF module of the Backbone part. LSKA reduces the computational complexity and improves the long-distance dependency characteristics and multi-scale adaptability of the model by decomposing the two-dimensional convolution kernel into horizontal and vertical 1D convolution kernels. LSKA outputs each channel Z c for:

[0014] Where: W kx1 and W 1xk Represent the convolution kernels in the horizontal and vertical directions respectively; F c Represents the convolution kernel weights generated based on content; In order to improve the detection accuracy of small targets such as grape stems, the CARAFE module is introduced in the Neck part and the upsampling operation is optimized. , and its corresponding up-sampled feature is :

[0015] Where: represents the convolution kernel weights generated based on content, r Represents the radius of the neighborhood area; It means that after the convolution operation, it is located at ( i+n,j+m )’s output feature map; Finally, the improved YOLOv10 model was used to train the labeled training set. After the training, the validation set was used to evaluate the model performance. The trained improved YOLOv10 model was deployed to the grape harvesting robot to achieve the purpose of quickly and accurately detecting the main vines, fruiting branches, and fruit stalks of grapes.

[0016] Based on the further optimization of the above solution, the specific method of extracting the navigation path in step 2 is: First, the yolov8-seg model is used to identify the roads between grape rows and extract the left and right edge points of the roads ( x l ,L l )、( x r ,L r ), and fitting the left and right grapevine lines by the least squares method:

[0017] Where: a l 、b l represents the slope and intercept of the left straight line; a r 、b r represents the slope and intercept of the right straight line; Then, based on the grapevine routes on the left and right, extract the reference points of the navigation line P i :( x i ,y i ), i=1,2,…,n :

[0018] Where: x li ,L li ） represents the coordinates of any point on the left line, ( x ri ,L ri ） Indicates that ( x li,L li ） The coordinates of the corresponding points on the right row of the same horizontal line; β Indicates the navigation line position parameters; Setting the position parameter threshold ,like , then the navigation line is considered to be in the center of the grape row; if , then the navigation line is on the left row, if , the navigation line is located on the right side, and the harvesting robot can move to the left or right by controlling the position parameters, thereby realizing the grape harvesting on the left or right side.

[0019] Based on the further optimization of the above scheme, the specific method of realizing the start, stop and steering control of the harvesting robot by identifying obstacles between rows in step 2 is: First, the least squares method is used to fit the orchard navigation line:

[0020] Then, m points of the orchard navigation line are selected equidistantly, that is, from the bottom of the image to the top, with H The width is intercepted by m horizontal lines in sequence on the selected orchard navigation line. m The points are recorded as P navj :( x navj ,y navj ); Afterwards, P navj As a benchmark, obtain the left boundary points for the harvesting robot to drive safely. P slj :( x slj ,y navj ) and the right boundary point P srj :( x srj ,y navj ):

[0021] Where: R a Indicates the ratio between the actual width of the road and the actual width of the harvesting robot; W proad express RoI The width of the lowest road in the pixel coordinate system; Finally, the obstacles on the road are identified through the yolov8-seg model, and the obstacles are obtained. RoI The circumscribed rectangle inside; and obtain the coordinates of the leftmost lower vertex of the circumscribed rectangle P obl :( x obl ,y obl ) and the coordinates of the rightmost lower vertex P obr :( x obl ,y obr ), if satisfied or , it is determined that the obstacle is within the safe driving area of ​​the harvesting robot, and the robot is started to turn left or right to avoid the obstacle; otherwise, the robot keeps moving forward.

[0022] The following are the technical effects of the solution of the present invention: The present application effectively simplifies the structure of the robotic arm through the structural design of the three-coordinate mechanism of the Y-axis component, the X-axis component and the Z-axis component, and avoids the multi-degree-of-freedom robotic arm structure being entangled by grape vines or causing damage or falling of surrounding grape bunches during the picking process, thereby causing the entire harvesting robot to get stuck and the picking efficiency to be reduced. The end effector composed of a servo motor, an adjustment rope, a support plate, a connecting hose, a universal ball cage, a high-speed motor and a three-sided blade tool not only utilizes the coordination of the adjustment rope, the connecting hose and the universal ball cage to improve the flexibility and adaptability of the tool, ensures the efficient operation of the harvesting robot in a complex field environment and reduces the damage of the end effector to the fruit and vines, but also further cooperates with the setting of the three-sided blade tool to improve the fault tolerance of cutting, ensures that accurate and effective cutting can be performed in the cutting process for the attitude angle of the fruit axis or the spatial surface error, and also helps to discharge debris during the cutting process, avoids the debris causing the three-sided blade tool to get stuck and the resulting reduction in cutting efficiency, and further improves the harvesting efficiency.

[0023] At the same time, this application uses an improved YOLOv10 model to identify grapes, significantly improving the detection accuracy of the main vines, fruiting branches and stalks of grapes, so that the harvesting robot can quickly and accurately identify and locate grapes, improve the accuracy and efficiency of harvesting, and reduce missed and wrong picking. Through the combination of visual components and navigation systems, a composite navigation method is formed, which effectively improves the navigation accuracy and stability of the harvesting robot in complex field environments, ensuring that it can effectively avoid obstacles and complete precise picking positioning, thereby improving the automation level and operational safety in the grape harvesting process. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1Schematic diagram of the overall structure of the harvesting robot in an embodiment of the present invention.

[0025] Figure 2 Schematic diagram of the structure of the Z-axis assembly of the harvesting robot in an embodiment of the present invention.

[0026] Figure 3 It is a schematic diagram of the structure of the Y-axis component and the X-axis component of the harvesting robot in an embodiment of the present invention.

[0027] Figure 4 Schematic diagram of the structure of the X-axis assembly and the end effector of the harvesting robot in an embodiment of the present invention.

[0028] Figure 5 It is a schematic diagram of the structure of a three-sided cutting tool of a harvesting robot in an embodiment of the present invention.

[0029] Figure 6 2 is a network structure diagram of the improved YOLOv10 in an embodiment of the present invention.

[0030] Figure 7 The flowchart of extracting the navigation path in the embodiment of the present invention is shown in FIG.

[0031] Figure 8 It is a schematic diagram of the analysis of the recognition of obstacles between rows in an embodiment of the present invention.

[0032] Fig. 9 This is a schematic diagram of a fruit stem low collision ROI in an embodiment of the present invention.

[0033] Among them, 10, chassis; 20, walking mechanism; 30, rotating plate; 411, horizontal guide rail; 412, horizontal screw; 413, horizontal slider; 414, horizontal slide; 421, connecting cross bar; 422, vertical guide rail; 423, vertical screw; 424, vertical slider; 425, vertical slide; 431, positioning cross plate; 432, moving screw; 433, linkage bracket; 4330, fixed pulley; 51, servo motor; 52, adjustment rope; 53, support plate; 54, connecting hose; 55, universal ball cage; 56, high-speed motor; 57, three-edged tool; 61, camera bracket; 62, visual camera; 71, antenna bracket; 72, Beidou antenna. DETAILED DESCRIPTION

[0034] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0035] Embodiment 1: A three-degree-of-freedom grape harvesting robot comprises a chassis 10, a walking mechanism 20, a rotating plate 30, a three-coordinate mechanism, an end-actuating mechanism, a visual component, a navigation system and an electric control box. The walking mechanism 20 is arranged on the bottom surface of the chassis 10 and the walking mechanism 20 adopts a crawler-type walking component (such as Figure 1 As shown in FIG. 1 ), used to control the operation of the entire harvesting robot, the end surface of the chassis 10 is rotatably provided with a rotating plate 30 and the end surface of the rotating plate 30 is provided with a three-coordinate mechanism (as shown in FIG. Figure 1 As shown); the three-coordinate mechanism includes a Y-axis assembly, an X-axis assembly and a Z-axis assembly, and the Z-axis assembly is arranged on the end surface of the rotating plate, the Y-axis assembly is arranged on the Z-axis assembly, and the X-axis assembly is arranged between the Y-axis assemblies. Specifically, the Z-axis assembly includes two sets of horizontal guide rails 411, two horizontal lead screws 412, two horizontal sliders 413 and two horizontal slide seats 414. The two sets of horizontal guide rails 411 are arranged in parallel on both sides of the end surface of the rotating plate 30 (as shown in FIG. Figure 2 A rotating horizontal lead screw 412 (as shown) is respectively provided on the side surfaces of the two sets of horizontal guide rails 411 close to each other. Figure 2 As shown, both ends of the horizontal lead screw 412 are arranged on the end surface of the rotating plate 30 on the side surface of the horizontal guide rail 411 close to each other through a rotating bracket, and a motor for controlling its rotation is arranged at one end of the horizontal lead screw 412). The outer wall of the horizontal lead screw 412 is respectively threadedly sleeved with a horizontal slider 413. A horizontal slide seat 414 is slidably arranged on the horizontal guide rail 411, and the horizontal slide seat 414 is fixedly connected with the corresponding horizontal slider 413 (thereby driving the horizontal slide seat 414 to slide on the horizontal guide rail 411 through the horizontal slide seat 413). The Y-axis assembly includes a connecting cross bar 421, two groups of vertical guide rails 422, two vertical lead screws 423, two vertical sliders 424 and two vertical sliders 425. The bottom surface of the connecting cross bar 421 is respectively fixedly connected to the end surfaces of the two horizontal sliders 413 and the two horizontal sliders 414, and two vertical guide rails 422 parallel to each other are fixedly arranged on both sides of the end surface of the connecting cross bar 421 (as shown in FIG. Figure 3 As shown in FIG. 1 ), a rotating vertical lead screw 423 (as shown in FIG. 1 ) is disposed on each side of the two sets of vertical guide rails 422 that are away from each other. Figure 3 As shown, the two ends of the vertical lead screw 423 are arranged on the side wall of the vertical guide rail 422 away from each other through a rotating bracket, and the end face of the connecting cross bar 421 at the bottom of the vertical lead screw 423 is provided with a motor for controlling its rotation) and the outer wall of the vertical lead screw 523 is respectively threadedly sleeved with the vertical slider 424, and the vertical slide seat 425 is slidably arranged on the outer wall of the vertical guide rail 422, and the vertical slide seat 425 is fixedly connected to the corresponding vertical slider 424 (thereby driving the vertical slide seat 425 to rise and fall on the vertical guide rail 422 through the vertical slider 424); the X-axis assembly includes a positioning cross plate 431, a movable lead screw 432 and a linkage bracket 433, and the two ends of the positioning cross plate 431 are respectively fixedly connected to the inner side wall of the corresponding vertical slider 425 (as shown Figure 3A rotating moving screw 432 (as shown) is provided on one side of the positioning horizontal plate 431. Figure 3 As shown: both ends of the moving screw 432 are arranged on the back of the positioning horizontal plate 431 through a rotating bracket and a motor for controlling its rotation is arranged at one end of the moving screw 432), the linkage bracket 433 is perpendicular to the positioning horizontal plate 431 and the bottom surface of the linkage bracket 433 is threadedly connected to the moving screw 432 and slidingly connected to the positioning horizontal plate 431.

[0036] The end actuator is arranged at one end of the X-axis assembly away from the Y-axis assembly (specifically, the end of the linkage bracket 433 away from the positioning horizontal plate 431, such as Figure 4 As shown in FIG. 1 ), the end actuator includes two servo motors 51, two adjustment ropes 52, a support plate 53, a connecting hose 54, a universal ball cage 55, a high-speed motor 56 and a three-sided blade cutter 57. The bottom surface of the end of the linkage bracket 433 away from the positioning horizontal plate 431 is provided with a support plate 53 through a connecting hose 54. The connecting hose 54 is multiple and evenly distributed on the outer ring of the end surface of the support plate 53 (the number of the connecting hose 54 is generally 3 to 8, and is set according to the actual situation, such as Figure 4 As shown, in this embodiment, four evenly distributed connecting hoses 54 are used); two servo motors 51 are fixedly arranged on the end face of the linkage bracket 433 away from the positioning horizontal plate 431, and the output ends of the two servo motors 51 are respectively wound with an adjusting rope 52, and the end of the adjusting rope 52 away from the servo motor 51 passes through the fixed pulley 4330 arranged on the end face of the linkage bracket 433, passes through the linkage bracket 433 and is connected to the end face of the support plate 53, the adjusting rope 52 is symmetrically distributed on both sides of the support plate 53, and the adjusting rope 53 and the connecting hose 54 are arranged in different positions; a universal ball cage 55 coaxial with the support plate 53 is arranged between the linkage bracket 433 and the support plate 53 and located on the inner circle of the connecting hose 54; a high-speed motor 56 is fixedly arranged on the bottom surface of the support plate 53, and the output end of the high-speed motor 56 is fixedly sleeved with a three-sided blade tool 57 through a coupling, and the main body of the three-sided blade tool 57 is a disc-shaped structure (such as Figure 5 As shown in the figure, three main cutting surfaces are extended from the base (i.e., three notches are evenly opened on the outer ring of the disc-shaped structure, dividing the outer ring of the disc-shaped structure into three cutting surfaces, as shown in the figure). Figure 5 As shown) and each cutting surface is provided with bionic angular grooves similar to dinosaur teeth (as shown Figure 5 As shown in the partial enlarged view of the three-sided blade tool 57, arc grooves are respectively provided on the inner circle corresponding to the three cutting surfaces. At the same time, in order to reduce the weight of the three-sided blade tool 57 and ensure its cutting performance, the three-sided blade tool 57 is made of three different materials, among which the cutting edge part is made of high-hardness and high-wear-resistant materials such as high-speed steel, the tool body is made of lightweight materials such as alloy steel or aluminum alloy, and the internal part of the tool is made of low-density materials such as carbon steel or its composite materials.

[0037] The visual component is arranged at one end of the X-axis component corresponding to the end effector and is located on the upper side of the end effector; the visual component includes a camera bracket 61 and a visual camera 62, and the visual camera 62 is fixedly arranged on the end surface of the linkage bracket 433 through the camera bracket 61. The navigation system is located on the side of the chassis 10, and the navigation system includes two sets of antenna brackets 71 and two sets of Beidou antennas 72. The two sets of antenna brackets 71 are respectively arranged at the front and rear ends of the chassis 10, and the Beidou antenna 72 (such as Figure 1 The electric control box is arranged on the end face of the chassis 10, and the electric control box is dislocated from the three-coordinate mechanism and is electrically connected to the walking mechanism 20, the control device of the rotating plate 30, the three-coordinate mechanism, the end actuator, the navigation system, and the visual component.

[0038] Embodiment 2: As a preferred embodiment of the present invention, on the basis of the scheme of Example 1, in order to protect the three-sided blade cutter 57 and prevent the three-sided blade cutter 57 from damaging fruits, vines, etc. during movement, a protective shell is provided on the bottom surface of the support plate 53 and located at the outer ring of the high-speed motor 56, and a cutting shell with a front end opening is provided on the bottom surface of the protective shell corresponding to the three-sided blade cutter 57, and the three-sided blade cutter 57 is located in the cutting shell and its front end protrudes from the opening of the cutting shell.

[0039] Embodiment 3: As a preferred embodiment of the present invention, on the basis of the scheme of Example 1, a collection frame is provided on the linkage bracket 433 away from the positioning horizontal plate 431 and located at the lower side of the end actuator (the collection frame can be a net bag type collection device or a hard collection device, which can be set according to actual conditions) for collecting the grape bunches after being cut by the three-edged knife 57.

[0040] Embodiment 4: A grape harvesting method using a three-degree-of-freedom grape harvesting robot, using any one of the harvesting robots in Embodiments 1 to 3, comprising: Step 1: The harvesting robot moves to the area to be harvested, starts the visual camera to capture images and videos, and completes target recognition; the improved YOLOv10 model is used to achieve target recognition, specifically: First, we obtain an image dataset D containing grape main vines, fruiting branches and fruit stalks. I 1 ,I 2 ,…,I N}, I i (i=1,2,…,N) represents a single image; and the image dataset is divided into a training set, a validation set, and a test set (in this embodiment, the ratio of the training set, the validation set, and the test set is 8:1:1); Then, use the annotation tool to add bounding boxes and category labels to each bunch of fruit; After that, build the improved YOLOv10 model, refer to Figure 6 As shown in the figure: The YOLOv10 network model includes a deep learning network structure of Backbone, Neck and Head; the network model starts with image input, extracts features through the Backbone part, performs feature fusion through the Neck part, and finally performs regression and classification prediction through the Head part; therefore, the improved YOLOv10 model uses the FasterBlock module to replace the Bottleneck module in the C2f of the Backbone part to optimize the computational efficiency and improve the accuracy of small target detection:

[0041] Where: H.W. represents the size of the feature map, k represents the convolution and size, C p Indicates the number of channels calculated; In order to meet the detection needs of targets of different scales such as grape main vines and fruiting branches, the LSKA attention mechanism is used to replace the SPPF module of the Backbone part. LSKA reduces the computational complexity and improves the long-distance dependency characteristics and multi-scale adaptability of the model by decomposing the two-dimensional convolution kernel into horizontal and vertical 1D convolution kernels. LSKA outputs each channel Z c for:

[0042] Where: W kx1 and W 1xk Represent the convolution kernels in the horizontal and vertical directions respectively; F c Represents the convolution kernel weights generated based on content; In order to improve the detection accuracy of small targets such as grape stems, the CARAFE module is introduced in the Neck part and the upsampling operation is optimized. , and its corresponding up-sampled feature is :

[0043] Where: represents the convolution kernel weights generated based on content, r Represents the radius of the neighborhood area; It means that after the convolution operation, it is located at ( i+n,j+m )’s output feature map; Finally, the improved YOLOv10 model was used to train the labeled training set. After the training, the validation set was used to evaluate the model performance. The trained improved YOLOv10 model was deployed to the grape harvesting robot to achieve the purpose of quickly and accurately detecting the main vines, fruiting branches, and fruit stalks of grapes.

[0044] Step 2: Use the Beidou antenna to extract the navigation path in real time, and start the harvesting robot to move toward the target identified in step 1 according to the navigation path. Figure 7 As shown, the specific method for extracting the navigation path is: First, the yolov8-seg model is used to identify the roads between grape rows and extract the left and right edge points of the roads ( x l ,L l )、( x r ,L r ), and fitting the left and right grapevine lines by the least squares method:

[0045] Where: a l 、b l represents the slope and intercept of the left straight line; a r 、b r represents the slope and intercept of the right straight line; Then, based on the grapevine routes on the left and right, extract the reference points of the navigation line P i :( x i ,y i ), i=1,2,…,n :

[0046] Where: x li ,L li ） represents the coordinates of any point on the left line, ( x ri ,L ri ） Indicates that ( x li ,L li ）The coordinates of the corresponding points on the right row of the same horizontal line; β Indicates the navigation line position parameters; Setting the position parameter threshold (Location parameter threshold According to the actual situation, the setting is performed. ),like , then the navigation line is considered to be in the center of the grape row; if , then the navigation line is on the left row, if , the navigation line is located on the right side, and the harvesting robot can move to the left or right by controlling the position parameters, thereby realizing the grape harvesting on the left or right side.

[0047] During the navigation process, the start, stop and steering control of the harvesting robot is realized by identifying obstacles between rows. The specific method is as follows: First, the least squares method is used to fit the orchard navigation line:

[0048] Then, m points of the orchard navigation line are selected equidistantly, that is, from the bottom of the image to the top, with H The width is intercepted by m horizontal lines in sequence on the selected orchard navigation line. m The points are recorded as P navj :( x navj ,y navj ); Afterwards, refer to Figure 8 As shown, P navj As a benchmark, obtain the left boundary points for the harvesting robot to drive safely. P slj :( x slj ,y navj ) and the right boundary point P srj :( x srj ,y navj ):

[0049] Where: R a Indicates the ratio between the actual width of the road and the actual width of the harvesting robot; W proad express RoI The width of the lowest road in the pixel coordinate system; Finally, the obstacles on the road are identified through the yolov8-seg model, and the obstacles are obtained. RoI The bounding rectangle inside (such as Figure 8 As shown, the distance between the bottom of the outer rectangular frame and the bottom of the picture is H s ); and obtain the coordinates of the leftmost lower vertex of the circumscribed rectangle respectively P obl :( x obl ,y obl ) and the coordinates of the rightmost lower vertex P obr :( x obl ,y obr ), if satisfied or , it is determined that the obstacle is within the safe driving area of ​​the harvesting robot, and the robot is started to turn left or right to avoid the obstacle; otherwise, the robot keeps moving forward.

[0050] Step 3: Start the three-coordinate mechanism and the end-effector to cut and harvest the target in real time; Among them, the method for obtaining the picking point is specifically as follows: For the fruit stems and fruiting branches of the target grapes, a low-collision ROI for the fruit stems is constructed (e.g. Fig. 9 As shown in the figure, the width of the rectangular frame is used to limit the horizontal range of the fruit stem low collision ROI; the lowest point of the fruit branch in the current horizontal range is used as the upper limit of the ROI vertical direction, and the highest point of the fruit bunch rectangular frame is used as the lower limit of the ROI area vertical direction (as shown in the figure). Fig. 9 As shown in the figure), the specific ROI area of ​​the fruit stem is:

[0051] Where: Wid_roi Indicates the horizontal range limit of the ROI area; Hei_roi Indicates the horizontal range limit of the ROI area; P_Clutopl.x Indicates the top left vertex of the fruit bunch detection box x value, P_Clutopl.y Indicates the top left vertex of the fruit bunch detection box y value, P_Clutopr.x Indicates the top right vertex of the fruit bunch detection box x value, P_Stembot.y Indicates the lowest point of the fruiting branch in the current horizontal range. y value; If the current grape bunch has an ROI, the circumscribed rectangle of the stalk portion retained by the stalk ROI is obtained, and the centroid of the circumscribed rectangle is used as the picking point of the three-edged tool; if the current grape ROI does not exist, it means that the current grapes cannot be picked.

Claims

1. A three-degree-of-freedom grape harvesting robot, characterized in that: It includes a chassis, a walking mechanism, a rotating plate, a three-coordinate mechanism, an end actuator, a visual component, a navigation system and an electric control box. The walking mechanism is arranged on the bottom surface of the chassis and the walking mechanism adopts a crawler-type walking component. The end surface of the chassis is rotatably arranged with a rotating plate and the end surface of the rotating plate is arranged with a three-coordinate mechanism. The three-coordinate mechanism includes a Y-axis component, an X-axis component and a Z-axis component, and the Z-axis component is arranged on the end surface of the rotating plate, the Y-axis component is arranged on the Z-axis component, and the X-axis component is arranged between the Y-axis components. The end actuator is arranged at one end of the X-axis component away from the Y-axis component, and the visual component is arranged at one end of the X-axis component corresponding to the end actuator and is located on the upper side of the end actuator. The navigation system is located on the side of the chassis and the electric control box is arranged on the end surface of the chassis. The electric control box is out of position with the three-coordinate mechanism and is electrically connected to the walking mechanism, the control device of the rotating plate, the three-coordinate mechanism, the end actuator, the navigation system and the visual component respectively.

2. The three-degree-of-freedom grape harvesting robot according to claim 1, characterized in that: The Z-axis assembly includes two sets of horizontal guide rails, two horizontal lead screws, two horizontal sliders and two horizontal slides. The two sets of horizontal guide rails are arranged in parallel on both sides of the end surface of the rotating plate, and a rotating horizontal lead screw is respectively arranged on the side surface of the two sets of horizontal guide rails close to each other. The outer walls of the horizontal lead screws are respectively threaded with the horizontal sliders. The horizontal sliders are slidably arranged on the horizontal guide rails, and the horizontal sliders are fixedly connected to the corresponding horizontal sliders; the Y-axis assembly includes a connecting cross bar, two sets of vertical guide rails, two vertical lead screws, two vertical sliders and two vertical slides. The bottom surface of the connecting cross bar is respectively fixedly connected to the end surfaces of the two horizontal sliders and the two horizontal slides, and the connecting cross bar is Two vertical guide rails parallel to each other are fixedly arranged on both sides of the rod end surface, a rotating vertical lead screw is respectively arranged on the side surfaces of the two sets of vertical guide rails away from each other, and the outer walls of the vertical lead screws are respectively threadedly sleeved with vertical sliders, a vertical slide is slidably arranged on the outer walls of the vertical guide rails, and the vertical slide is fixedly connected to the corresponding vertical slider; the X-axis assembly includes a positioning cross plate, a moving lead screw and a linkage bracket, the two ends of the positioning cross plate are respectively fixedly connected to the inner side walls of the corresponding vertical slides, and a rotating moving lead screw is arranged on the side surface of one side of the positioning cross plate, the linkage bracket is perpendicular to the positioning cross plate, and the bottom surface of the linkage bracket is threadedly connected to the moving lead screw and is slidably connected to the positioning cross plate.

3. The three-degree-of-freedom grape harvesting robot according to claim 1 or 2, characterized in that: The end actuator includes two servo motors, two adjustment ropes, a support plate, a connecting hose, a universal ball cage, a high-speed motor and a three-sided blade tool. The support plate is set on the bottom surface of the end of the linkage bracket away from the positioning cross plate through a connecting hose, and there are multiple connecting hoses that are evenly distributed on the outer circle of the end surface of the support plate; two servo motors are fixedly set on the end surface of the linkage bracket away from the positioning cross plate, and an adjustment rope is respectively wound around the output ends of the two servo motors. The end of the adjustment rope away from the servo motor passes through the fixed pulley set on the end surface of the linkage bracket, penetrates the linkage bracket and is connected to the end surface of the support plate, the adjustment ropes are symmetrically distributed on both sides of the support plate, and the adjustment ropes and the connecting hoses are arranged in different positions; a universal ball cage coaxial with the support plate is set between the linkage bracket and the support plate and at the inner circle of the connecting hose; a high-speed motor is fixedly set on the bottom surface of the support plate, and the output end of the high-speed motor is fixedly connected to the three-sided blade tool through a coupling, and the three-sided blade tool body is a disc-shaped structure, with three main cutting surfaces extending from the base and each cutting surface is provided with a bionic angular groove similar to a dinosaur tooth.

4. The three-degree-of-freedom grape harvesting robot according to claim 3, characterized in that: The visual component includes a camera bracket and a visual camera, and the visual camera is fixedly arranged on the end surface of the linkage bracket through the camera bracket.

5. The three-degree-of-freedom grape harvesting robot according to claim 3, characterized in that: The navigation system comprises two groups of antenna brackets and two groups of Beidou antennas. The two groups of antenna brackets are respectively arranged at the front and rear ends of the chassis, and the Beidou antennas are fixedly arranged on the antenna brackets.

6. The harvesting method of a three-degree-of-freedom grape harvesting robot according to claim 3, characterized in that: include: Step 1: The harvesting robot moves to the area to be harvested, starts the visual camera to capture images and videos, and completes target recognition; Step 2: extract the navigation path in real time through the Beidou antenna, and start the harvesting robot to move toward the target identified in step 1 according to the navigation path; during the navigation process, the start, stop and steering control of the harvesting robot are realized by identifying obstacles between rows; Step 3: Start the three-coordinate mechanism and the end-effector to cut and harvest the target in real time.

7. The harvesting method of a three-degree-of-freedom grape harvesting robot according to claim 6, characterized in that: In step 1, an improved YOLOv10 model is used to realize target recognition, specifically: First, we obtain an image dataset D containing grape main vines, fruiting branches and fruit stalks. I 1 ,I 2 ,…,I N }, I i (i=1,2,…,N) represents a single image; and the image dataset is divided into training set, validation set and test set; Then, use the annotation tool to add bounding boxes and category labels to each bunch of fruit; After that, an improved YOLOv10 model is constructed. The YOLOv10 network model includes a deep learning network structure of Backbone, Neck and Head. The network model starts with image input, extracts features through the Backbone part, performs feature fusion through the Neck part, and finally performs regression and classification prediction through the Head part. Therefore, the FasterBlock module is used in the improved YOLOv10 model to replace the Bottleneck module in the C2f of the Backbone part to optimize the computational efficiency and improve the accuracy of small target detection: Where: H.W. represents the size of the feature map, k represents the convolution and size, C p Indicates the number of channels calculated; The LSKA attention mechanism is used to replace the SPPF module in the Backbone part. LSKA is used to output each channel. Z c for: Where: W kx1 and W 1xk Represent the convolution kernels in the horizontal and vertical directions respectively; F c Represents the convolution kernel weights generated based on content; Introduce the CARAFE module in the Neck part and optimize the upsampling operation. For a given target position , and its corresponding up-sampled feature is : Where: represents the convolution kernel weights generated based on content, r Represents the radius of the neighborhood area; It means that after the convolution operation, it is located at ( i+n,j+m )’s output feature map; Finally, the improved YOLOv10 model was used to train the labeled training set. After the training, the validation set was used to evaluate the model performance. The trained improved YOLOv10 model was deployed to the grape harvesting robot to achieve the purpose of quickly and accurately detecting the main vines, fruiting branches, and fruit stalks of grapes.

8. The harvesting method of a three-degree-of-freedom grape harvesting robot according to claim 6, characterized in that: The specific method of extracting the navigation path in step 2 is: First, the yolov8-seg model is used to identify the roads between grape rows and extract the left and right edge points of the roads ( x l , L l )、( x r ,L r ), and fitting the left and right grapevine lines by the least squares method: Where: a l 、b l represents the slope and intercept of the left straight line; a r 、b r represents the slope and intercept of the right straight line; Then, based on the grapevine routes on the left and right, extract the reference points of the navigation line P i :( x i ,y i ), i=1,2,…,n : Where: x li ,L li ） represents the coordinates of any point on the left line, ( x ri ,L ri ） Indicates that x li ,L li ） The coordinates of the corresponding points on the right row of the same horizontal line; β Indicates the navigation line position parameters; Setting the position parameter threshold ,like , then the navigation line is considered to be in the center of the grape row; if , then the navigation line is on the left row, if , the navigation line is located on the right side, and the harvesting robot can move to the left or right by controlling the position parameters, thereby realizing the grape harvesting on the left or right side.

9. The harvesting method of a three-degree-of-freedom grape harvesting robot according to claim 8, characterized in that: The specific method for realizing the start, stop and steering control of the harvesting robot by identifying obstacles between rows in step 2 is: First, the least squares method is used to fit the orchard navigation line: Then, m points of the orchard navigation line are selected equidistantly, that is, from the bottom of the image to the top, with H The width is intercepted by m horizontal lines in sequence on the selected orchard navigation line. m The points are recorded as P navj :( x navj ,y navj ); Afterwards, P navj As a benchmark, obtain the left boundary points for the harvesting robot to drive safely. P slj :( x slj ,y navj ) and the right boundary point P srj :( x srj ,y navj ): Where: R a Indicates the ratio between the actual width of the road and the actual width of the harvesting robot; W proad express RoI The width of the lowest road in the pixel coordinate system; Finally, the obstacles on the road are identified through the yolov8-seg model, and the obstacles are obtained. RoI The circumscribed rectangle inside; and obtain the coordinates of the leftmost lower vertex of the circumscribed rectangle P obl :( x obl ,y obl ) and the coordinates of the rightmost lower vertex P obr :( x obl ,y obr ), if satisfied or , it is determined that the obstacle is within the safe driving area of ​​the harvesting robot, and the robot is started to turn left or right to avoid the obstacle; otherwise, the robot keeps moving forward.

Images