An intelligent film removal system and method based on peanut touch film detection

Abstract

The present invention discloses an intelligent film-picking system and method based on peanut touch film detection, including the following steps: S1, obtaining the vehicle speed v and the rotation angles of each wheel; S2, matching the driving distance L = vt with the plant spacing Z to obtain the target point A; S3, obtaining the peanut touch-seedling target point B based on image processing technology; S4, obtaining the ground surface point cloud data, defining the target area, and extracting the target point C; S5, performing multi-sensor seedling condition information fusion, and taking the data point with the highest fitting degree as the detected final target point coordinates; S6, moving the robotic arm in the X and Y directions, compensating for the differences ΔX and ΔY, and changing the size of the film-picking claw according to the size of the target object; S7, performing the film-picking action, rotating the mechanical claw clockwise and then counterclockwise to roll up and then release the film. The present invention solves the problems of timely and efficient film-breaking during peanut emergence, preventing phenomena such as seedling burning and scalding, and effectively improves the situation where farmers have no film to use.

Description

Technical Field

[0001] The present invention relates to the technical field of agricultural machinery, and particularly to an intelligent film-picking system and method based on peanut film-touch detection. Background Art

[0002] Planting peanuts with film mulching can not only increase soil temperature, maintain soil humidity, but also effectively inhibit weed growth, thereby improving yield and quality. However, during the field management of film-mulched peanuts, film-picking treatment needs to be carried out when peanut seedlings gradually emerge and grow in the initial stage, that is, the film material covering the buds is peeled off, so that the buds can grow and emerge quickly, ensuring that peanut plants can fully absorb sunlight and nutrients, and avoiding the phenomenon of high-temperature roasting of buds. Traditional film-picking operations usually rely on manual labor. However, manual film-picking operations are not only labor-intensive but also have low efficiency, making it difficult to meet the needs of large-scale production. During the film-picking process, it is also necessary to accurately judge the growth state of peanut plants and the position of buds, resulting in a large labor intensity. To solve the above problems, the present invention proposes an intelligent film-picking system and method based on peanut film-touch detection. Summary of the Invention

[0003] To solve the technical problems in the background art, the present invention proposes an intelligent film-picking system and method based on peanut film-touch detection.

[0004] An intelligent film-picking system based on peanut film-touch detection proposed by the present invention includes a self-propelled film-picking machine body. The self-propelled film-picking machine body includes a vehicle frame, a four-wheel independent steering system, a peanut film-touch detection system, and a film-picking execution device. The four-wheel independent steering system is fixedly installed at the four corners of the vehicle frame. The four-wheel independent steering system includes a steering servo and a chassis walking drive motor. The power output shaft of the chassis walking drive motor is fixedly connected to a wheel. The peanut film-touch detection system includes a solid-state radar, a depth camera, and a speed encoder. The solid-state radar and the depth camera module are fixedly installed at the front end of the vehicle frame, and the speed encoder is fixedly installed on the output shaft of the wheel. The film-picking execution device is fixedly installed at the center position of the vehicle frame. The film-picking execution device includes a connecting rod frame, a position correction mechanism, a rotary stepping motor, an electric push rod, a mechanical claw opening adjustment push rod, and an execution component - a bionic mechanical claw. The position correction mechanism is fixedly installed on the connecting rod frame. The rotary stepping motor is fixedly installed at the upper end of the position correction mechanism. The electric push rod is fixedly connected to the output shaft of the rotary stepping motor. The bionic mechanical claw is fixedly connected to the tail end of the electric push rod. The mechanical claw opening adjustment push rod is fixedly installed on the bionic mechanical claw.

[0005] An intelligent film-picking method based on peanut film-touch detection is proposed, including the following steps:

[0006] S1. Obtain the driving speed v of the vehicle through a speed measurement encoder, perform speed feedback to control the vehicle speed, measure the rotation angles of each wheel through a magnetosensitive angle sensor, perform angle feedback, and use a four-wheel independent steering system to control the vehicle steering;

[0007] S2. Obtain the peanut film-touching situation information in the driving direction based on agronomic conditions, and calculate the travel distance L of the machine according to the obtained driving speed v:

[0008] L = vt

[0009] where t is the travel time of the machine;

[0010] Compare the value of L with the peanut plant spacing Z. When L = Z, the position of the center point A of the target is (x0, y0 + Z, z0), where x0, y0, z0 are the coordinates of the initial film-picking point;

[0011] S3. Obtain the coordinates of the center point B of the peanut seedling-touching target through a deep learning algorithm and image processing technology, denoted as (x1, y1, z1). According to the crop row spacing H, simulate the sowing route to calculate the vehicle trajectory route, and control the steering servos A1, A2, A3, A4 to steer and correct the forward trajectory route;

[0012] S4. Obtain the surface point cloud data through a solid-state radar, traverse the point cloud data of each convex area, sort them according to the z-axis size to obtain the highest and lowest point coordinates, calculate the maximum z-axis spacing between the point clouds in each convex area. When the maximum z-axis spacing between the point clouds is greater than the threshold, define the convex area corresponding to the maximum z-axis spacing between the point clouds as the target area, and extract the center point C (x2, y2, z2) of the target in the target area;

[0013] S5. Perform multi-sensor seedling condition information fusion, perform data matching processing based on agronomic requirements, target information collected by a depth camera, and target information obtained by a solid-state radar, and use the data point with the highest fitting degree as the detected final target point coordinates (x3, y3, z3);

[0014] S6. According to the target point coordinates (x3, y3, z3), calculate the compensation differences ΔX and ΔY required to move the robotic arm along the X and Y directions respectively in combination with the machine driving speed; according to the detected size of the target, control the push rods C1, C2, C3, C4 of the mechanical claw opening to change the opening size of the mechanical claw to adapt to the seedling conditions in different growth periods;

[0015] S7. Control the electric push rods B1, B2, B3, B4 to push the film-picking mechanical claws to extend and retract to perform the film-picking action; while the mechanical claws break the film, control the rotary stepper motors D1, D2, D3, D4 to perform clockwise rotation of the mechanical claws to roll up the film, and then the mechanical claws rotate counterclockwise to loosen the film to complete the film-picking action.

[0016] Preferably, the obtaining of the driving speed v of the vehicle and the steering angles of each wheel in S1 specifically includes the following operations:

[0017] S1.1: Fix the speed measurement encoder on the wheel rotating shaft disc, and fix the rotating shaft of the magnetic - sensitive angle sensor to the wheel bracket.

[0018] S1.2: When the machine is moving, use the speed measurement encoder to obtain the wheel speed in real - time, convert it into the traveling speed v, and the angle sensor detects the rotation angles of each wheel to feedback - control the traveling direction.

[0019] Preferably, the obtaining of the coordinates (x1, y1, z1) of the center point B of the target position and the path planning in S3 specifically includes the following operations:

[0020] S3.1: Establish a space coordinate system with the depth camera as the central origin, the direction of the vehicle body cross - beam as the x - axis, the direction of the vehicle longitudinal beam as the y - axis, and the vertical direction as the z - axis. The coordinates of the tire position in the space coordinate system are (x5, y5, z5).

[0021] S3.2: Obtain the image information of the field in the forward direction through the depth camera and transmit it to the trained YOLO v7 model for target detection, identify the peanut film - touching area in the image, and extract the center point B (x1, y1, z1) of the target position.

[0022] S3.3: During the operation, the film - removing machine operates across 2 ridges and 4 rows of peanuts, matching the operation mode of the 4 - row peanut planter. Calculate the optimal traveling route based on the peanut row spacing H and the film - removing machine wheelbase W, and at this time, the distance between the tire and the crop is X.

[0023] S3.4: Theoretically, the distance where K is the ridge width and J is the ridge spacing; in the actual process, the actual distance value X * = y5 - y1, and use the actual distance value X * = X as the target to maintain path tracking.

[0024] Preferably, the multi - sensor seedling condition information fusion in S5 specifically includes the following operations:

[0025] S5.1: According to the plant spacing Z of the film - covered peanuts, the vehicle speed v, and the distance L traveled by the vehicle within the unit time t, when L = Z, (x0, y0, z0) is the initial film - removing point coordinate, then preliminarily judge that the coordinate of the next target point A is (x0, y0 + Z, z0).

[0026] S5.2: Extract the coordinates (x1, y1, z1) of the center point B of the peanut seedling and the film - removing radius r from the target box identified in S2.

[0027] S5.3. Subtract the scanned point cloud data in S3 from the basic plane of the ridge surface. The points with the difference height within the threshold range are the target points C(x2, y2, z2).

[0028] S5.4. Fit the coordinates of points A, B, and C through the priority and weighted average method. The priority is A > B > C, and the corresponding weight values are w1, w2, and w2 respectively. When fitting, only consider the plane coordinates and ignore the z-axis coordinate value.

[0029] Preferably, the fitting of the coordinates of points A, B, and C is specifically as follows:

[0030] S5.4.1: Points A, B, and C completely coincide: that is, x0 = x1 = x2 = x3, y0 + Z = y1 = y2 = y3. At this time, the plane coordinates of the target center point are (x3, y3).

[0031] S5.4.2: Points A and B coincide: that is, x0 = x1, y0 + Z = y1. At this time, the plane coordinates of the target center point are (x1, y1).

[0032] S5.4.3: Points A and C coincide: that is, x0 = x2, y0 + Z = y2. At this time, the plane coordinates of the target center point are (x0, y0 + Z).

[0033] S5.4.4: Points B and C coincide: that is, x1 = x2, y1 = y2. At this time, the plane coordinates of the target center point are (x1, y1).

[0034] S5.4.5: Points A, B, and C do not coincide:

[0035]

[0036]

[0037] At this time, the plane coordinates of the target center point are (x, y).

[0038] Preferably, the agronomic requirements in S5 include sowing plant spacing and row spacing information.

[0039] An intelligent film-scraping system and method based on peanut film-touching detection proposed by the present invention have the following beneficial effects:

[0040] The present invention provides an intelligent film-picking system and method based on peanut film-touching detection. Feedback control is carried out by obtaining the vehicle driving speed and peanut film-touching situation information. The center point A of the target position is initially predicted according to the peanut planting mode and agronomic situation combined with the machine's forward distance. The center point B of the target position is determined by using deep learning algorithms and image processing techniques. The surface point cloud data is obtained by a solid-state radar and the target area C is analyzed. Then, the target point coordinates are fitted based on multi-sensor information fusion technology. Finally, the compensation difference is calculated according to the coordinates of the target point D and the machine driving speed, and the execution mechanical claw is controlled to perform film-picking operations, so as to realize the intelligent detection and film-picking of peanut film-touching. It has practical and application values, can solve the problems of labor-intensive and low-efficiency manual film-picking operations, and has the advantages of simple structure, easy implementation and high working efficiency compared with the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 FIG. is a schematic structural diagram of an intelligent film-picking system based on peanut film-touching detection proposed by the present invention;

[0042] Figure 2 FIG. is a schematic flow diagram of an intelligent film-picking method based on peanut film-touching detection proposed by the present invention;

[0043] Figure 3 FIG. is a schematic flow diagram of a fusion strategy proposed by the present invention;

[0044] Figure 4 FIG. is a schematic diagram of the operation process of an intelligent film-picking system and method based on peanut film-touching detection proposed by the present invention.

[0045] Figure 5 FIG. is a schematic diagram of data fusion based on the Dempster-Shafer algorithm proposed by the present invention.

[0046] DESCRIPTION OF THE REFERENCE NUMERALS:

[0047] 1, frame; 2, depth camera; 3, solid-state radar; 4, steering servo; 5, angle sensor; 6, drive motor; 7, rotary stepping motor; 8, electric push rod; 9, mechanical claw opening adjustment push rod; 10, mechanical claw; 11, speed measurement encoder; 12, lithium battery; 13, control box; 14, wheel; 15, connecting rod frame; 16, position correction mechanism. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0048] Referring to Figure 1 , this embodiment provides an intelligent film-picking system based on peanut film-touching detection, and intelligent film-picking for peanut film-touching detection is carried out based on the cooperation of each structure in the intelligent film-picking system Figure 2 and the intelligent film-picking method therein.

[0049] Referring toFigures 2 - 4 , first, a spatial coordinate system is established with the position point of the depth camera 2 as the central origin. The whole film peeling machine is placed in the coordinate system. The direction of the vehicle body crossbeam is the x-axis, the direction of the vehicle longitudinal beam is the y-axis, and the vertical direction is the z-axis. The speed v and steering angle information of the vehicle in the current driving direction are obtained. The optimal tracking distance is calculated according to the row spacing and wheelbase. The travel distance L of the machine is calculated based on the obtained driving speed v:

[0050] L = vt

[0051] where t is the travel time of the machine;

[0052] The value of L is compared with the peanut plant spacing Z. When L = Z, the position of the center point A of the target is (x0, y0 + Z, z0), where x0, y0, z0 are the coordinates of the initial film peeling point;

[0053] The coordinates of the position where the wheel 14 is located in the spatial coordinate system are (x5, y5, z5);

[0054] Image information of the field in the forward direction is obtained through the depth camera 2 and transmitted to the trained YOLO v7 model for target detection to identify the peanut film-touching area in the image, and the center point B (x1, y1, z1) of the target position is extracted;

[0055] During the operation process, the film peeling machine operates across 2 ridges and 4 rows of peanuts, matching the operation mode of a 4-row peanut planter,

[0056] Refer to Figure 4 , based on the peanut row spacing H and the wheelbase W of the film peeling machine, the optimal travel route is calculated, and at this time, the distance between the wheel 14 and the crop is X;

[0057] Theoretically, the distance where K is the ridge width and J is the ridge spacing. In the actual process, the actual distance value X * = y5 - y1, with the actual distance value X * = X as the target to maintain path tracking.

[0058] The solid-state radar 3 is also used to obtain the ground point cloud data. The point cloud data of each convex area is traversed, and sorted according to the z-axis size to obtain the coordinates of the highest point and the lowest point. The maximum z-axis spacing between the point clouds in each convex area is calculated. When the maximum z-axis spacing between the point clouds is greater than the threshold, the convex area corresponding to the maximum z-axis spacing between the point clouds is defined as the target area, and the center point C (x2, y2, z2) of the target in the target area at this time is extracted;

[0059] Please refer to Figure 5, using the Dempster-Shafer algorithm, data matching processing is carried out based on the agronomic requirements of the seeding plant spacing and row spacing information, the target information collected by the depth camera 2, and the target information obtained by the solid-state radar 3. The data point with the highest fitting degree is used as the detected final target point coordinates. Based on five types of fitting situations, taking the fifth type of fitting as an example: Points A, B, and C do not coincide:

[0060]

[0061]

[0062] At this time, the plane coordinates of the target center point are (x, y).

[0063] Then, according to the driving speed information and the target position data, the displacement compensation values ΔX and ΔY of the mechanical claw 10 are predicted. When the mechanical claw 10 touches the soil during the execution of the action, it starts to rotate clockwise to roll up the film. While the mechanical claw 10 retracts, it rotates counterclockwise to loosen the film, completing the entire film-picking action. Through the parallel operation of three detection methods, the above solution finally fits the target object information, pre-calculates the start time of film-picking and the corresponding displacement compensation amount of the mechanical claw 10, and formulates a control strategy by information fusion and sends it to the film-picking execution system, so that the film-picking mechanical claw 10 adjusts its position in advance before the vehicle arrives and picks the film in real time, thus realizing a truly self-propelled intelligent film-picking. While significantly improving the film-picking accuracy, it also reduces the labor intensity.

[0064] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution of the present invention and its inventive concept, makes equivalent substitutions or changes, and all should be covered by the protection scope of the present invention.

Claims

1. An intelligent film peeling method based on peanut touch film detection, characterized in that, The method is implemented based on an intelligent film removal system, which includes a self-propelled film removal machine body. The self-propelled film removal machine body includes a vehicle frame (1), a four-wheel independent steering system, a peanut film-touching detection system, and a film removal execution device. The four-wheel independent steering system is fixedly installed at the four corners of the vehicle frame (1). The four-wheel independent steering system includes a steering servo (4) and a chassis walking drive motor (6). The power output shaft of the chassis walking drive motor (6) is fixedly connected to a wheel (14). The peanut film-touching detection system includes a solid-state radar (3), a depth camera (2), and a speed measuring encoder (11). The solid-state radar (3) and the depth camera (2) module are fixedly installed at the front end of the vehicle frame (1). The speed measuring encoder (11) is fixedly installed on the output shaft of the wheel (14). The film removal execution device is fixedly installed at the center position of the vehicle frame (1). The film removal execution device includes a connecting rod frame (15), a position correction mechanism (16), a rotary stepping motor (7), an electric push rod (8), a mechanical claw opening adjustment push rod (9), and a bionic mechanical claw (10). The position correction mechanism (16) is fixedly installed on the connecting rod frame (15). The rotary stepping motor (7) is fixedly installed at the upper end of the position correction mechanism (16). The electric push rod (8) is fixedly connected to the output shaft of the rotary stepping motor (7). The bionic mechanical claw (10) is fixedly connected to the tail end of the electric push rod (8). The mechanical claw opening adjustment push rod (9) is fixedly installed on the bionic mechanical claw (10). The method includes the following steps: S1. Obtain the driving speed v of the vehicle through the speed measuring encoder (11), perform speed feedback to control the vehicle speed, measure the rotation angles of each wheel (14) through a magnetosensitive angle sensor (5), perform angle feedback, and use the four-wheel independent steering system to control the vehicle to turn; S2. Obtain the peanut film-touching situation information in the driving direction based on agronomic conditions, and calculate the machine walking distance L according to the obtained driving speed v: L = vt where t is the machine walking time; Compare the L value with the peanut plant spacing Z. When L = Z, the position of the center point A of the target object is (x0, y0 + Z, z0), where (x0, y0, z0) is the initial film removal point coordinate; S3. Obtain the coordinates of the center point B of the peanut seedling-touching target object through a deep learning algorithm and image processing technology, denoted as (x1, y1, z1). According to the crop row spacing H, simulate the sowing route to calculate the vehicle trajectory route, and control the steering servos (4) A1, A2, A3, A4 to turn to correct the forward trajectory route; S4. Obtain the ground point cloud data through the solid-state radar (3), traverse the point cloud data of each convex area, sort them according to the z-axis size to obtain the highest point and the lowest point coordinates, calculate the maximum z-axis spacing between the point clouds in each convex area. When the maximum z-axis spacing between the point clouds is greater than the threshold, define the convex area corresponding to the maximum z-axis spacing between the point clouds as the target area, and extract the center point C (x2, y2, z2) of the target object in the target area; S5. Perform multi-sensor seedling condition information fusion, and perform data matching processing based on agronomic requirements, target information collected by the depth camera, and target information obtained by the solid-state radar (3). Take the data point with the highest fitting degree as the detected final target point coordinates (x3, y3, z3). S6. According to the target point coordinates (x3, y3, z3), calculate the compensation differences ΔX and ΔY for moving the robotic arm along the X and Y directions respectively in combination with the vehicle traveling speed. According to the detected size of the target object, control the push rods C1, C2, C3, and C4 of the mechanical claw opening adjuster (9) to change the opening size of the mechanical claw (10) to adapt to the seedling conditions in different growth periods. S7. Control the electric push rods (8) B1, B2, B3, and B4 to push the film-picking mechanical claw (10) to expand and contract to perform the film-picking action. While the mechanical claw (10) breaks the film, control the rotary stepping motors (7) D1, D2, D3, and D4 to rotate the mechanical claw (10) clockwise to roll up the film, and then rotate the mechanical claw (10) counterclockwise to loosen the film to complete the film-picking action.

2. The intelligent film removal method based on peanut touch film detection according to claim 1, wherein In S1, obtaining the traveling speed v of the vehicle and the rotation angles of each wheel specifically includes the following operations: S1.1: Fix the speed measurement encoder (11) on the rotating shaft disk of the wheel (14), and fix the rotating shaft of the magnetic-sensitive angle sensor (5) to the wheel (14) bracket. S1.2: When the machine is moving, use the speed measurement encoder (11) to obtain the rotation speed of the wheel (14) in real time, convert it into the traveling speed v, and the angle sensor (5) detects the rotation angles of each wheel for feedback control of the traveling direction.

3. The intelligent film removal method based on peanut touch film detection according to claim 1, wherein In S3, obtaining the coordinates of the center point B of the peanut touching the seedling target object as (x1, y1, z1) and path planning specifically includes the following operations: S3.1: Establish a space coordinate system with the depth camera (2) as the central origin, the direction of the vehicle body cross beam as the x-axis, the direction of the vehicle longitudinal beam as the y-axis, and the vertical direction as the z-axis. The coordinates of the tire position in the space coordinate system are (x5, y5, z5). S3.2: Obtain the image information of the field in the forward direction through the depth camera (2) and transmit it to the trained YOLO v7 model for target detection, identify the peanut film-touching area in the image, and extract the center point B (x1, y1, z1) of the target position. S3.3: During the operation process, the film-picking machine operates across 2 ridges and 4 rows of peanuts, matching the operation mode of the 4-row peanut planter. Calculate the optimal traveling route based on the peanut row spacing H and the wheelbase W of the film-picking machine, and at this time, the distance between the tire and the crop is X. S3.4: Theoretical distance where K is the ridge width and J is the ridge spacing; in the actual process, the actual distance value X * = y5 - y1, and keep path tracking with the actual distance value X * = X as the target.

4. The intelligent film removal method based on peanut touch film detection according to claim 1, wherein In S5, the multi-sensor seedling condition information fusion specifically includes the following operations: S5.

1. Based on the plant spacing Z of the plastic-film-covered peanuts, the vehicle speed v, and the distance L traveled by the vehicle within the unit time t, when L = Z, (x0, y0, z0) is the initial film-picking point coordinates, then preliminarily judge that the coordinates of the next target point A are (x0, y0 + Z, z0). S5.

2. Extract the coordinates (x1, y1, z1) of the center point B of the peanut seedling and the film-picking radius r from the target box identified in S2. S5.

3. Subtract the scanned point cloud data in S3 from the basic plane of the ridge surface. The points with the difference height within the threshold range are the target points C(x2, y2, z2). S5.

4. Fit the coordinates of points A, B, and C through the priority and weighted average method. The priority is A > B > C, and the corresponding weight values are w1, w2, and w3 respectively. When fitting, only consider the plane coordinates and ignore the z-axis coordinate value.

5. The intelligent film removal method based on peanut touch film detection according to claim 4, wherein The fitting of the coordinates of points A, B, and C is specifically as follows: S5.4.1: Points A, B, and C completely coincide: that is, x0 = x1 = x2 = x3, y0 + Z = y1 = y2 = y3. At this time, the plane coordinates of the target center point are (x3, y3). S5.4.2: Points A and B coincide: that is, x0 = x1, y0 + Z = y1. At this time, the plane coordinates of the target center point are (x1, y1). S5.4.3: Points A and C coincide: that is, x0 = x2, y0 + Z = y2. At this time, the plane coordinates of the target center point are (x0, y0 + Z). S5.4.4: Points B and C coincide: that is, x1 = x2, y1 = y2. At this time, the plane coordinates of the target center point are (x1, y1). S5.4.5: Points A, B, and C do not coincide with each other: At this time, the plane coordinates of the target center point are (x, y).

6. The intelligent film removal method based on peanut touch film detection according to claim 1, characterized in that The agronomic requirements in S5 include the information of sowing plant spacing and row spacing.

Images