CN113057154A

CN113057154A - Greenhouse liquid medicine spraying robot

Info

Publication number: CN113057154A
Application number: CN202110305635.2A
Authority: CN
Inventors: 王珂; 鲁力群; 王佳欣; 王煜凯; 张琦; 郝杰君; 穴丽颖
Original assignee: Shandong University of Technology
Current assignee: Shandong University of Technology
Priority date: 2021-03-23
Filing date: 2021-03-23
Publication date: 2021-07-02

Abstract

The application discloses greenhouse liquid medicine spraying robot based on visual monitoring includes: the device comprises a driving module, a lifting module, a spraying module, a visual monitoring module and the like. The driving module is responsible for controlling the walking of the robot; the lifting module is responsible for assisting the spraying module to spray the liquid medicine under different conditions; the spraying module is used for conveying the liquid medicine from the medicine box to the spray head for spraying; gather through vision monitoring module robot and wait to spray crops information and then upload to the singlechip, the singlechip is handled according to the information that the collection obtained and is sent to spraying the module and spray the order, accomplishes spraying of liquid medicine under drive module and lift module's assistance. The robot is mainly applied to liquid medicine spraying of small and medium-sized greenhouses, relieves labor force, reduces spraying cost, and solves the problem of manual operation of liquid medicine spraying of the existing small and medium-sized greenhouses. The robot has low manufacturing cost, small volume and strong environmental adaptability.

Description

Greenhouse liquid medicine spraying robot

Technical Field

The invention relates to a spraying robot, and belongs to the field of agricultural plant protection.

Background

The current agricultural spraying equipment commonly used in China comprises a manual type and ultra-low-volume sprayer, a stretcher type motor-driven spray rod type and aviation sprayer matched with a tractor, a backpack type motor-driven and orchard air-assisted mist sprayer and the like. The spraying machines produced by domestic enterprises have over ten types and over 50 types. However, the small and medium-sized spraying equipment in China still mainly comprises a knapsack manual sprayer and a knapsack mechanical mist sprayer. Wherein, the control area of the former is 60-70%; the control area of the latter accounts for 15-20%. Most of the products of the sprayers have simple structures, serious environmental pollution, easy poisoning of operators and low control efficiency, and the crops effectively utilized by the pesticides only account for 20-30% of the total crops. The boom sprayer in China has various problems of simple structure, backward technical performance, few product types, long product updating period and the like, so that the boom sprayer has low working efficiency, overproof pesticide residues and serious environmental pollution, and the human health is seriously harmed due to the contact of people and medicines.

The developed countries have high industrialization degree and large greenhouse scale, are dedicated to the research of large-scale spraying equipment, have more advanced control technology and management method, mostly adopt unmanned aerial vehicles to spray at night, and spray systems with complete structures spray pesticides to promote pesticide sedimentation, wherein the newly researched unmanned aerial vehicle electrostatic spraying system is more perfect. But because unmanned aerial vehicle sprays the cost higher, the duration is worse makes its long time operation receive the limitation in warmhouse booth.

The concept of intelligent vegetable greenhouse spraying robot is provided according to foreign experience, but the robot needs to be laid in a greenhouse with complex terrain conditions, the production cost is high, the design of a walking track limits the freedom of movement of a spraying robot, the robot cannot flexibly meet the complex environment in the vegetable greenhouse, the robot is easy to have short circuit and other problems in the humid greenhouse environment, the safety and the economy of the robot are not suitable for farmers with general economic conditions, the automatic dispensing can be realized, the size of the robot can be correspondingly increased, the robot is not suitable for vegetable greenhouses with narrow spaces, and the design limitation is too large and the robot cannot be popularized in large scale in China.

Disclosure of Invention

Along with the adjustment of planting industry structure, the greenhouse industry development is rapid, and long-term investigation finds that in the greenhouse liquid medicine spraying process, traditional manual spraying efficiency is low, the effect is general and the liquid medicine is heated in narrow and small space and volatilizes and can produce harm to the human body. The invention provides a liquid medicine spraying robot based on visual monitoring. The robot realizes the accurate spraying of the pesticide through the intelligent visual monitoring system; through PC intelligence control, convenient and fast just realizes people and liquid medicine zero contact, the effectual staff that has protected. The production and use of the robot can greatly liberate labor force, reduce spraying cost and the like, and increase economic benefit.

The liquid medicine spraying robot based on visual monitoring is composed of five modules including a common driving module, a lifting module, a spraying module, a control module and a visual module. To PID motion control algorithms, visual recognition, single-chip control, data analysis, and the like.

In the moving process of the robot, the motor drives the high-ground-clearance wheel to rotate to realize walking, and a four-wheel drive mode is adopted. The differential motion of the high ground clearance wheels is realized when the rotating speeds of the motors on the two sides are different, and the turning, the deviation correction and other motion states are realized. The driving motor is provided with a power source by the storage battery, the driving motor transmits power to the 4 driving wheels through the transmission device, and differential motion of the driving wheels on the left side and the right side is realized by changing the transmission ratio of the transmission device, so that the robot is driven to complete the functions of advancing, retreating and steering. Ultrasonic sensors are respectively arranged at 4 directions of front, back, left and right on a base frame of the robot, the distance between the robot and an obstacle is measured and calculated by transmitting and receiving ultrasonic signals, and distance data is transmitted to a main controller for calculation. Photoelectric encoders are mounted on the left main driving wheel and the right main driving wheel and used for monitoring the movement speed and the geographical position information of the robot.

The spraying module comprises a sprayer, a water tank, a pesticide box, a pesticide liquid conveying pipe, a pressure station and the like, and pesticide liquid spraying on crops is completed; the control module is the brain of the robot and is responsible for controlling the cooperative operation among all the modules, the robot adopts an stm32f407ZGT6 single chip microcomputer as a control core and can control the robot to work through a PC end and a mobile phone end; the vision module is the eyes of the robot, the information such as varieties and growth conditions of crops is automatically identified through Opencv computer vision, the acquired information is fed back to the single chip microcomputer, and commands are sent to other modules after being processed by the single chip microcomputer, so that the robot operation is completed. Meanwhile, the image-transfer information is fed back to the PC end or the mobile phone end so that the working condition of the robot can be observed by the working personnel in real time.

In order to realize a wireless control scheme of the platform, the control of the visual positioning system adopts an embedded system design, and a raspberry pi 3b + module is introduced to process optical information so as to realize an intelligent information processing scheme; the circuit of the battery replacement system is driven by 32 minimum system boards and a stepper motor driver; the circuit board of will independently researching and developing is as the master control, controls the chassis motion, and above three electrical system carry out serial communication, and whole design retrencies intelligence, makes the volume quality of platform reach minimum steady under reliable prerequisite as far as possible.

In addition, a WiFi function is developed on the machine body, and schemes such as remote control, remote updating and firmware upgrading are achieved. The project is more suitable for large-scale point arrangement at professional level, and the whole system is convenient to control and manage. The visual control system interface is written by adopting Python language, an API interface is easy for third-end development, custom development can be carried out on Labview, WeChat small programs, Java and the like, and the visual control system interface is suitable for individual users and experiences the product in a single point and small range.

The robot obstacle avoidance path planning structure comprises an input part, a motion platform and an output part. The input part mainly comprises the steps of inputting barrier distances (comprising front, rear, left side and right side), environment types and vehicle body rotation angles to the robot motion platform. Before planning an obstacle avoidance path, the robot must generally judge the external environment, that is, judge the relative position of an obstacle and the robot. If there is no obstacle in front of the robot, or the obstacle and the target position are respectively located at two sides of the robot, the robot can move to the target position through advancing and rotating, and if there is an obstacle in front of the robot, or the obstacle and the target position are located at the same side of the robot, the robot must complete obstacle avoidance path selection through advancing, retreating and rotating. And classifying and numbering the environment types according to the analysis, and determining that the robot adopts corresponding movement modes (forward, backward and rotation) when moving according to the numbers. The robot motion platform is mainly responsible for collecting data transmitted by the input part, and sending the data to the output part after analysis and processing. The output section is mainly composed of two parameters, i.e., a steering control amount and a speed desired value. The method comprises the following steps that a computer motion platform firstly judges the environment type of a robot, drives corresponding ultrasonic sensors to start working after the environment type is determined, and after receiving obstacle distance information of 4 directions, combines a mathematical model to carry out kinematic analysis to determine that the robot needs to adopt corresponding speed and steering control quantity in the path selection process so as to ensure that the robot does not collide in the obstacle avoidance process; the invention mainly has the following three points in innovative characteristics.

1. Visual monitoring: the robot adopts a mode of processing image information by visual monitoring to provide data information of crops to be sprayed for the robot, and the data information is compared with spraying information provided according to the spraying experience of workers.

2. The spraying mechanism: breakthrough innovation on spray mechanism abandons the working method of current push rod direct-push type, chooses the lead screw over-and-under type that the stroke is higher for use, compromises the spraying scope and is bigger when accomplishing spray mechanism miniaturization.

3. Unmanned intelligent control: through visual monitoring and other a great deal of peripheral hardware module, this robot can realize unmanned intelligent control, compares with traditional manual control, and this robot all has great promotion in the aspect of spraying efficiency, spraying effect etc. and the visual monitoring greenhouse liquid medicine that possess automatic control moreover sprays the robot and can realize a plurality of robots collaborative work to single just can operate, also improved the efficiency of spraying when practicing thrift the manpower.

Drawings

Fig. 1 is a schematic structural diagram of a greenhouse liquid medicine spraying robot based on visual monitoring.

Fig. 2 is a flow chart of greenhouse liquid medicine spraying robot work based on visual monitoring.

Fig. 3 is a schematic diagram of a robot running track.

Fig. 4 shows a diagram of an obstacle avoidance path planning structure of the robot.

FIG. 5 is a simplified computer vision diagram of a robot dynamic recognition and intelligent spraying system.

Figure 6 robot test analysis chart.

Detailed Description

In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application. Embodiments of the present invention are further described below with reference to the accompanying drawings.

The embodiment of the application provides a greenhouse liquid medicine spraying robot based on visual monitoring, and fig. 1 is a schematic structural diagram of the greenhouse liquid medicine spraying robot based on visual monitoring provided by the embodiment of the application.

As shown in fig. 2, the operation of a vision robot relies mainly on five modules: the device comprises a driving module, a lifting module, a spraying module, a control module and a vision module. The robot vision monitoring system identifies, analyzes and processes the target crops with sprays by utilizing vision processing such as Opencv, Tensorflow and the like. And the data are sent to the singlechip, the information is compared and processed in the database after being calculated and processed by the singlechip, the processed information is fed back to the walking module and the automatic lifting module, and other modules receive the information fed back by the singlechip and then are finely adjusted.

After the robot starts to enter a working mode, the driving motor receives a moving signal and starts to walk according to a specified route, the motor in the lifting mechanism drives the screw rod to realize the up-and-down movement of the spraying module while walking, and meanwhile, the spraying rod driven by the stepping motor realizes the left-and-right movement of the spraying module.

The information that sprays the in-process vision module and constantly will gather sends the singlechip to, for spraying the module and provide sufficient data support, guarantee to spray going on smoothly. In addition, the vision module sends the acquired image information and the acquired image to the collection APP end or the PC end, so that the working state of the liquid medicine spraying robot can be monitored by a worker.

The premise of the robot for automatically avoiding the obstacle in the unknown environment is to acquire the positioning information of the robot, including the geographical position information and the motion state information. The agricultural robot has two types of motion modes, namely linear motion and rotary motion, because the left main driving wheel and the right main driving wheel of the agricultural robot can realize differential motion. A rectangular coordinate system is established with any point in the plane as the center, and the movement locus of the robot from the starting position to the target position is shown in fig. 3.

Assuming that the diameter of a main driving wheel of the robot is 2R, the number of lines of the photoelectric encoder is R, and the encoder outputs N pulses in total in time T, the calculation formula of the movement distance S of the robot in the time period is as follows:

(1)

and calculating the target position of the robot according to the initial position and the moving distance of the robot. Since a certain directional error exists when performing position positioning, a directional increment needs to be corrected, and the correction model is:

(2)

wherein M is_compass(k+1)As a number of photoelectric encodersReading of a word compass at the target location, M_compass(θ)The reading of the digital compass of the photoelectric encoder at the starting position is obtained.

Assuming that the motion track of the robot is a circular arc or a straight line, using a group (V)_t，W_t) And representing the track of the circular motion of the robot, the motion model of the circular motion or linear motion of the robot is derived as follows:

the radius of the arc motion of the robot is g, the rotating speed is w, and the relation between the radius and the rotating speed is as follows:

(3)

because the robot makes circular motion, the rotation speed w of the robot is not equal to 0, and then the coordinates of the circular motion of the robot are:

(4)

(5)

the robot has limited conditions, and the moving speed must be kept within a certain range, namely:

(6)

the robot obstacle avoidance path planning structure comprises an input part, a motion platform and an output part. The input part mainly comprises the steps of inputting barrier distances (comprising front, rear, left side and right side), environment types and vehicle body rotation angles to the robot motion platform. Before planning an obstacle avoidance path, the robot must generally judge the external environment, that is, judge the relative position of an obstacle and the robot. If there is no obstacle in front of the robot, or the obstacle and the target position are respectively located at two sides of the robot, the robot can move to the target position through advancing and rotating, and if there is an obstacle in front of the robot, or the obstacle and the target position are located at the same side of the robot, the robot must complete obstacle avoidance path selection through advancing, retreating and rotating. And classifying and numbering the environment types according to the analysis, and determining that the robot adopts corresponding movement modes (forward, backward and rotation) when moving according to the numbers.

The robot motion platform is mainly responsible for collecting data transmitted by the input part, and sending the data to the output part after analysis and processing.

The output section is mainly composed of two parameters, i.e., a steering control amount and a speed desired value. The method comprises the steps that a computer motion platform firstly judges the type of the environment where the robot is located, after the environment type is determined, a corresponding ultrasonic sensor is driven to start working, after the motion platform receives obstacle distance information of 4 directions, the motion platform performs kinematic analysis by combining a mathematical model, determines that the robot needs to adopt reasonable speed and reasonable steering control quantity in the path selection process, and ensures that the robot does not collide in the obstacle avoidance process. The obstacle avoidance path planning structure of the robot is shown in fig. 4.

The main task of the computer vision part of the dynamic identification and spraying system is to identify the area and position of the crop to be sprayed and to be able to provide corresponding control signals for the spraying device. Computer vision of the system plays an important role in recognition. The recognition accuracy is not only related to the recognition algorithm, but also has a great relationship with the installation height, the installation angle and the distance of the spray head of the camera, so that the proper adjustment of the installation height, the installation angle and the distance of the spray head of the camera becomes an important influence factor for realizing accurate spraying.

In the test, the camera and the test bed are vertically arranged at 90 degrees, and the distance between the camera and the table board of the test bed is about 0.7 m. The camera viewing angle was adjusted to 41.5 deg. taking into account the defined length and width of the test bed, and the size of the captured image was 320 x 240 pixels, corresponding to a test bed range of 0.212m (0.53 m x 0.40 m). The spray heads are arranged on the spray frame, the interval is 0.2m, and the interval distance is adjustable. The whole image area collected is divided into two parts, the two spray heads respectively correspond to one area of the image, when a certain part of the image contains weeds, a control signal of the area is sent out, and the spray heads start to spray.

There is certain distance between camera and the shower nozzle, usually makes a video recording earlier then sprays. The overall identification and spraying process comprises the steps of 1) enabling the test bed to run, enabling the camera to continuously collect images to a memory, 2) utilizing a weed identification and processing program which is programmed in a computer to identify and process the collected images, wherein the time interval of collecting one frame of image is 1s, 3) obtaining the processed and identified images which are binary images, providing control information according to the area size of weeds in the binary images (the weeds larger than a certain area are preset in the program for spraying), and correspondingly controlling the spray nozzle switch. A computer vision sketch of the robot dynamic recognition and intelligent spraying system is shown in fig. 5.

In the speed of research developments spraying in-process locomotive operation, the locomotive that carries camera and sprinkler is with the certain speed forward operation, and camera and sprinkler's distance is fixed unchangeable, and after the height and the visual angle scope of camera have been adjusted, the area that the camera gathered the image is fixed unchangeable. The variable running speed of the locomotive, how to select the running speed of the locomotive and which quantity have a mutual relation, are one of the key problems to be solved in the image acquisition and spraying process. By utilizing the existing test platform, the distance between the spray head and the camera is fixed as d, the visual angle range of the camera is adjusted, and the area of an image collected by the camera is fixed as w multiplied by l (width multiplied by length). In the dynamic acquisition process, the time interval t for acquiring two continuous frame images is 1s, the time interval t is set to be constant, and the running speed v of the conveyor belt is adjustable. The analysis of the value range of the speed v shows that when the camera collects the 2 nd frame image, the process that the 1 st frame image is processed and identified and the spraying controller sends an instruction to control the spraying of the spraying nozzle happens, the interval time is needed for realizing the process, and the running distances of the spraying nozzle and the camera are S = v × t. The following is divided into 4 cases to discuss, 1) when d-w/2< v < d < w, the camera collects the 2 nd frame image, the sprinkler position stays in the range of the 1 st frame image area, and dynamic sprinkling can be carried out, as shown in fig. 6 (a); 2) when d < v < w, the camera collects the 2 nd frame image, the position of the spray head stays in the range of the 1 st frame image area, and dynamic spraying can also be performed, as shown in fig. 6 (b); 3) When v > w, the images of the 1 st frame and the 2 nd frame are not collected continuously, partial images are not collected, and the recognition and spraying of the images are adversely affected, as shown in fig. 6 (c); 4) when v < = (d-w/2), the camera acquires the 2 nd frame image, the acquired 2 nd frame image is overlapped with the 1 st frame image in most areas, and the position of the spray head stops outside the 1 st frame image area acquired by the camera, so that effective dynamic spraying cannot be carried out, as shown in fig. 6 (d);

in the dynamic identification and spraying test of crops, the speed v of the locomotive obtained from the previous 4 test conditions has a certain value range, and the locomotive is not suitable for dynamic spraying by selecting any value. The speed is selected to have a certain relation with the width w of the collected image and the distance d between the camera and the nozzle, and the speed is established when the two conditions of d-w/2< v < d < w and d < v < w are met. Otherwise, the execution result is not accurate enough when the image is collected and the dynamic spraying is carried out.

Claims

1. The utility model provides a greenhouse liquid medicine sprays robot based on visual monitoring which characterized in that, including the drive module of control robot walking, assist and spray the lift module that the module accomplished work, the robot is based on the control system that visual monitoring sprayed.

2. And a communication system between the robot and the PC end and the like jointly complete the working operation of the spraying robot.

3. The robot for spraying greenhouse chemical liquid based on visual monitoring as claimed in claim 1, wherein the driving module is four-wheel driven and is provided with an independent suspension device, and the robot turns according to the differential speed of wheels.

4. The robot for spraying greenhouse chemical liquid based on visual inspection as claimed in claim 2, wherein each wheel has an independent suspension device; the suspension device consists of a shock absorber, a hinge and an aluminum pipe, one end of the aluminum pipe is directly connected to the frame, and the other end of the aluminum pipe is connected to the frame through the shock absorber.

5. The robot for spraying greenhouse chemical liquid as claimed in claim 3, wherein the traveling motor is installed on the suspension device to land four wheels simultaneously to ensure smooth output of power.

6. The robot for spraying greenhouse liquid medicine based on visual monitoring as claimed in claim 1, wherein the lifting module controls the nozzle to lift up and down, stretch left and right, and rotate; the lifting module comprises an aluminum pipe, a rotating motor, a bearing seat, a lead screw, a drawer slide rail, an optical axis, a lifting platform and the like; the basic framework of the lifting module is composed of aluminum pipes, and the lifting mechanism is composed of a rotating motor and a bearing seat lifting platform and is responsible for ascending and descending of the spray rod; a telescopic mechanism consisting of an aluminum pipe, a drawer sliding rail and a rotating motor is arranged on the lifting platform to control the extension of the spray rod.

7. The robot for spraying greenhouse liquid medicine according to claim 1, wherein a plurality of vision cameras are mounted on the robot for collecting environmental information and crop growth conditions, so that the robot can realize full-automatic operation, automatic route planning, walking obstacle avoidance, work positioning, crop type identification, crop growth condition identification, spraying process monitoring and spraying result detection in the working process.

8. The greenhouse liquid medicine spraying robot based on visual monitoring as claimed in claim 1, wherein the spraying robot and the PC end communicate with each other, the operation status and spraying operation information of the robot can be clearly known through the PC end, the operation of the robot is controlled if necessary, the information collected by the robot is also sent to the PC end through a communication system, the PC end sends an instruction to the robot by combining with the working environment information after receiving the information, and the robot works after receiving the instruction.

9. The greenhouse liquid medicine spraying robot based on visual monitoring as claimed in claim 1, wherein the spraying robot adopts a GPS positioning system.

10. The greenhouse liquid medicine spraying robot based on visual monitoring as claimed in claim 1, wherein after the spraying robot enters a working site, workers input site information from a PC (personal computer) end, route planning is performed in combination with visual scanning of the robot, after the initial route planning is completed, the PC end workers send working instructions, the spraying robot moves according to the planned route under the support of GPS positioning, a visual module and an ultrasonic module avoid obstacles during moving to prevent the robot from colliding, meanwhile, the visual module collects crop information and sends the crop information to the PC end, the PC end sends spraying information to the spraying robot after data comparison, the spraying robot performs spraying operation according to the received instructions, the workers can monitor the working condition of the spraying robot in real time through the PC end, the visual module also detects the sprayed crops and feeds the information back to the PC end for the workers to record, all the collected information is stored, and data support is provided for later-stage robot improvement, crop growth observation and the like.