Intelligent control system and method for tea tender shoot picking robot

Abstract

The invention discloses an intelligent control system and method for a tea tender shoot picking robot, and the method comprises the steps: obtaining scene image information through a visual analysis module, and accurately recognizing a to-be-picked tea tender shoot region based on a preset neural network model; a task decomposition module is used for dividing a tea tender shoot area, each task in a large number of picking tasks is actually distributed to a corresponding mechanical arm, and a shortest safe working route of the mechanical arm is formulated; through the motion control module, it is ensured that the picking range of each mechanical arm of the robot can completely cover the area where the tea tender shoots suitable for being picked are located, and the missing picking rate of the tea tender shoots is effectively reduced. Multiple mechanical arms of the picking robot can be coordinated to pick tea tender shoots, mechanical picking has the advantages of being controllable in cost, accurate in collection and excellent in efficiency, data calculation and control of the whole system are completely conducted locally, smoothness and high efficiency of the system can be guaranteed, and popularization of the tea picking robot technology is facilitated.

Description

Technical Field

[0001] The present application relates to the technical field of tea picking robot control, and in particular to an intelligent control system and method for a tea bud picking robot. Background Art

[0002] The tea industry is a labor-intensive industry, in which fresh leaf picking and pruning is highly seasonal, with the largest number of workers, the most concentrated labor, and the highest cost. Manual tea picking and pruning accounts for more than 60% of the cost of tea production. As the area of ​​tea gardens continues to expand, the gap in labor required for tea production will continue to widen. Traditional manual tea picking alone can no longer meet market demand. Mechanized tea picking is the trend of future tea production development. Solving the problem of "difficult tea picking" through mechanized and intelligent picking is of great significance to the development of the tea industry.

[0003] However, unlike the picking of other crops, the automated picking of tea buds in the field still faces many challenges: the lighting in the field tea garden changes dramatically, it is difficult to segment images with a background similar to the color of the buds, and the occlusion and overlap between leaves cause unsatisfactory recognition results. In addition, the different heights and high growth density of tea shoots have put forward requirements for the efficiency of the harvesting movement of the robotic arm. At present, the end-point picking actuators of the picking robots, especially those for medium-height crops, mostly use a single robotic arm, which has a slow picking speed, and the buds are easily damaged by the robotic arm, affecting the quality of the tea.

[0004] Therefore, there is an urgent need for an intelligent control system and method for a tea bud picking robot that can improve the robot's picking efficiency and accuracy for tea buds and reduce the cost of tea picking. Summary of the invention

[0005] The present application provides an intelligent control system and method for a tea bud picking robot, which is used to solve the technical problems in the prior art that the picking robot is controlled by a cloud system, is overly dependent on the network environment, and only uses a single robotic arm, resulting in slow picking speed and low efficiency.

[0006] The present application provides an intelligent control system for a tea bud picking robot, comprising a visual analysis module, a task decomposition module and a motion control module connected in sequence; A visual analysis module is used to obtain scene image information, identify the tea bud area to be picked in the scene image information based on a preset neural network model, and establish a spatial coordinate system to determine the relative position of the picking robot and the tea bud area; The task decomposition module is used to divide the tea bud area into a number of picking sub-areas using a preset division method, determine a number of target picking areas corresponding to each robotic arm according to the spatial positions of the multiple picking sub-areas and the multiple robotic arms of the picking robot, and formulate a planning path based on the relative positions of each robotic arm and the multiple target picking areas; The motion control module is used to determine the joint activity data of each robotic arm according to the planned path, and control each robotic arm to pick and suck the tea buds in the target picking area according to the planned path according to the joint activity data.

[0007] Further, the visual analysis module includes an image acquisition module, a recognition module and a coordinate establishment module; An image acquisition module is used to obtain scene image information of the current position of the picking robot through a visual sensor; A recognition module is used to identify the tea bud area to be picked in the scene image information based on the YOLO neural network model; The coordinate establishment module is used to establish a spatial coordinate system, determine the spatial position of the tea bud area and the spatial position of multiple mechanical arms of the picking robot, and establish the positional relationship between the tea bud area and the mechanical arms of the picking robot and the picking robot body.

[0008] Furthermore, the task decomposition module includes a partitioning module, an allocation module, and a path planning module; A partitioning module is used to divide the tea bud area into several picking sub-areas based on the K-Means algorithm and determine the regional picking task parameters of each picking sub-area; An allocation module, used to determine a target picking interval corresponding to each robotic arm according to regional picking task parameters and characteristic parameters of each robotic arm based on an auction algorithm; The path planning module is used to preset a fusion algorithm to plan the activity path information of each robotic arm according to the spatial position of each robotic arm and the target picking interval to obtain a picking planning path.

[0009] Furthermore, the preset fusion algorithm is constructed based on Q-learning and optimized genetic algorithm, Q-learning is used to obtain a preliminary picking path, and the genetic algorithm is used to optimize the preliminary picking path to obtain a picking planning path.

[0010] Further, the motion control module includes a model building module, a motion analysis module and a command output module; A model building module is used to build a robot kinematics model of the picking robot; The motion analysis module is used to perform forward and reverse position analysis of the picking robot based on the planned path and the robot kinematic model to obtain the joint activity data of each mechanical arm of the picking robot; The command output module is used to generate control commands according to the joint activity data of the robot arm, and control the robot to pick and suck the tea buds in the target picking area according to the planned path.

[0011] Furthermore, the motion model is a Delta parallel robot kinematic model.

[0012] Furthermore, the motion analysis module is also used to perform singularity analysis and workspace analysis on the parallel robot and optimize the joint activity data so that the range of motion of the robot arm covers the target picking area.

[0013] Furthermore, the system is built based on the ROS platform.

[0014] On the other hand, the present application also provides an intelligent control method for a tea bud picking robot, which is implemented by using any one of the intelligent control systems for a tea bud picking robot described in the above technical solution, and includes: The visual analysis module obtains scene image information, identifies the tea bud area to be picked in the scene image information based on a preset neural network model, and establishes a spatial coordinate system to determine the relative position of the picking robot and the tea bud area; The task decomposition module uses a preset division method to divide the tea bud area into several picking sub-areas, and according to the spatial positions of the multiple picking sub-areas and the multiple robotic arms of the picking robot, multiple target picking areas corresponding to each robotic arm are determined, and a planning path is formulated based on the relative positions of each robotic arm and the multiple target picking areas; The motion control module determines the joint activity data of each robotic arm according to the planned path, and controls each robotic arm according to the joint activity data to pick and absorb the tea buds in the target picking area according to the planned path.

[0015] The beneficial effects of this application are: The present application provides an intelligent control system for a tea bud picking robot. The visual analysis module uses common visual sensors to obtain scene image information, and accurately identifies the tea bud area to be picked, as well as the relative position of the picking robot and the tea bud area based on a preset neural network model; the tea bud area is divided by a task decomposition module, each of a large number of picking tasks is actually assigned to a corresponding robotic arm, and the shortest safe working route for the robotic arm is formulated; the motion control module is used to ensure that the picking range of each robotic arm of the robot can completely cover the area where the tea buds are suitable for picking, effectively reducing the missed picking rate of tea buds and improving the picking efficiency. The present application coordinates a variety of highly independent and complex technologies to serve the same robot body together, and operates multiple robotic arms to perform picking work, so that the tea picking robot has the characteristics of controllable cost, accurate collection, and excellent efficiency. The data calculation and control of the entire system are completely performed locally, which can ensure the smoothness and efficiency of the system, and is more conducive to the commercial promotion of tea picking robots. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a system architecture diagram of an embodiment of an intelligent control system for a tea bud picking robot provided in this application; Figure 2 A schematic diagram of the structure of the visual analysis module provided in the present application in one embodiment; Figure 3 A schematic diagram of the structure of the task decomposition module provided in the present application in one embodiment. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0018] Before introducing the embodiments of the present application, the relevant terms of the present application are explained first.

[0019] YOLO: You Only Look Once, is a real-time object detection algorithm based on deep learning. The main idea of ​​YOLO is to use the entire image as the input of the network and directly regress the position and category of the bounding box in the output layer.

[0020] K-Means: K-means is an unsupervised machine learning algorithm mainly used for data clustering. In K-Means, K represents the number of categories and Mean represents the average value of the data.

[0021] Auction algorithm: It is essentially a process that simulates human auction activities. Its algorithmic process is like an auction. Unassigned bidders bid for the target object at the same time, thereby increasing the price of the target object. The auction house obtains the bidding information of all bidders and sells the target object to the bidder with the highest bid. The final result should maximize the interests of all bidders and the auction house as much as possible.

[0022] Robot kinematics: Robot kinematics is the study of the motion of robots in space, including mathematical models that describe and analyze the motion, velocity, acceleration, and position of the robot's end effector (such as a robotic arm or a mobile platform). Kinematics focuses mainly on the robot's geometric characteristics and motion laws, and does not involve the robot's mechanics and dynamics.

[0023] Task partitioning: Task partitioning is a method for allocating tasks in a multi-agent system, where a task is divided into multiple parts (or "partitions"), which are then assigned to different agents (such as robots) to perform.

[0024] ROS: The full name is Robot Operating System, which is a framework for writing robot software. It is a flexible framework for writing robot software. It is a set of operating system concepts - providing services such as hardware abstraction, low-level device control, implementation of common functions, inter-process messaging, and package management. It also provides a concept of computational graphs, which makes it a distributed computing environment that can run on different machines. The goal of ROS is to make it easy to create complex and powerful robots.

[0025] The invention concept of the present application is introduced below. The existing automated picking of tea buds has the following shortcomings: One is that the visual recognition effect is poor. The recognition of tea buds is greatly affected by lighting, it is difficult to segment images with a background similar in color to the buds, and the occlusion and overlap between leaves will cause unsatisfactory recognition effects. Compared with traditional machine learning, the current bud and leaf recognition method based on deep learning has good application prospects, but it requires a large number of labeled samples for training, and as the complexity of the network increases, hardware system upgrades are a problem that needs to be solved. In some methods, in order to improve the recognition accuracy of buds, it is necessary to equip a variety of different types of sensors, not only cameras, but even laser radars are needed to build 3D point cloud models, which are expensive to use and cannot be promoted and applied on a large scale; The second is the low efficiency of robotic arm picking. This is because the end-point picking actuators of existing picking robots mostly use a single robotic arm, which has a slow picking speed and can only pick a few kilograms of tea a day, far from the efficiency of manual labor. In addition, most control systems are based on cloud computing platforms. This system only deploys sensors and robot kinematic control components on the robot, and recognition and collection calculations are completely dependent on cloud servers connected to wireless networks. However, the actual production environment cannot guarantee the stability of network signals. Placing key computing processes in the cloud may cause the machine to fall into an endless loop or stop running, seriously hindering work efficiency; The third problem is that it is not practical. The existing harvesting robot control system based on visual recognition technology cannot combine visual recognition, actual scene calculation, and robotic arm harvesting implementation control well to build a complete system for actual production and cannot be put into practical application. Fourth, the picking effect is poor. Because the tender buds of famous and high-quality tea are soft and scattered, the traditional mechanical picking method is prone to damage by the mechanical arm due to low control accuracy, affecting the quality of the tea.

[0026] In response to the above problems, the present application improves the visual recognition part, uses ordinary visual sensors configured in production operations to acquire image data, accurately identifies the area of ​​tea buds to be picked in the image data based on a preset neural network model, and allocates corresponding picking sub-areas to each end picking actuator of a picking robot with multiple robotic arms, and plans the picking path of each robotic arm. Finally, the tea buds are picked according to the picking path, which greatly improves the picking efficiency and picking accuracy of the robot.

[0027] Combination Figure 1 As shown, a specific embodiment of the present application discloses an intelligent control system for a tea bud picking robot, comprising a visual analysis module 101, a task decomposition module 102 and a motion control module 103 connected in sequence; The visual analysis module 101 is used to obtain scene image information, identify the tea bud area to be picked in the scene image information based on a preset neural network model, and establish a spatial coordinate system to determine the relative position of the picking robot and the tea bud area; The task decomposition module 102 is used to divide the tea bud area into a plurality of picking sub-areas using a preset division method, determine a plurality of target picking areas corresponding to each robotic arm according to the spatial positions of the plurality of picking sub-areas and the plurality of robotic arms of the picking robot, and formulate a planning path based on the relative positions of each robotic arm and the plurality of target picking areas; The motion control module 103 is used to determine the joint activity data of each robotic arm according to the planned path, and control each robotic arm to pick and suck the tea buds in the target picking area according to the planned path according to the joint activity data.

[0028] Compared with the prior art, this application uses common visual sensors to obtain scene image information through a visual analysis module, and accurately identifies the tea bud area to be picked, as well as the relative position of the picking robot and the tea bud area based on a preset neural network model; divides the tea bud area through a task decomposition module, and actually assigns each of the large number of picking tasks to the corresponding robotic arm, and formulates the shortest safe working route for the robotic arm; through a motion control module, it ensures that the picking range of each robotic arm of the robot can completely cover the area where the tea buds are suitable for picking, effectively reducing the missed picking rate of tea buds and improving the picking efficiency. This application coordinates a variety of highly independent and complex technologies to serve the same robot body together, and operates multiple robotic arms to perform picking work, so that the tea picking robot has the characteristics of controllable cost, accurate collection, and excellent efficiency. The data calculation and control of the entire system are completely performed locally, which can ensure the smoothness and efficiency of the system, and is more conducive to the commercialization and promotion of tea picking robots.

[0029] In a preferred embodiment, the system is built on the ROS platform. The ROS platform is a flexible framework for developing robot software. The software provides a series of tools, libraries, and conventions that can help developers write robot applications. Through ROS, robot applications can be quickly built, and the resources and tools of the ROS community can be used to accelerate the development process. At the same time, ROS also facilitates the cross-platform development of robots, allowing robots on different hardware platforms to share and reuse software components.

[0030] It should be noted that the ROS platform is a preferred control platform, and other types of control platforms can also be used, such as Gazebo, V-REP, Webots, etc., which provide real-time control and robot application development platforms. In addition, control platforms based on integration between different platforms are also applicable.

[0031] Combination Figure 2 As shown, in a preferred embodiment, the visual analysis module 101 includes an image acquisition module 201, a recognition module 202 and a coordinate establishment module 203; An image acquisition module 201 is used to obtain scene image information of the current position of the picking robot through a visual sensor; The recognition module 202 is used to recognize the tea bud area to be picked in the scene image information based on the YOLO neural network model; The coordinate establishment module 203 is used to establish a spatial coordinate system, determine the spatial position of the tea bud area and the spatial positions of multiple mechanical arms of the picking robot, and establish the positional relationship between the tea bud area and the mechanical arms of the picking robot and the picking robot body.

[0032] In a specific embodiment, the image acquisition module is a commonly used ordinary camera, which identifies the tea bud area in the image information through a preset visual analysis network model. It can achieve accurate identification of the picking target without adding or upgrading the hardware, thereby reducing hardware expenditure and maintenance costs.

[0033] In a specific embodiment, using the YOLO neural network model for target detection can achieve real-time detection and rapid positioning of the tea bud area, improve the data processing speed of the system, and further improve production efficiency. In addition to the recognition network based on the YOLO neural network model, other visual modules with visual recognition capabilities can also be used for the recognition model of the tea bud area to be picked, such as a visual recognition model based on a convolutional neural network, etc.

[0034] In a preferred embodiment, Figure 3 As shown, the task decomposition module 102 includes a partitioning module 301, an allocation module 302 and a path planning module 303; A partitioning module 301 is used to divide the tea bud area into a plurality of picking sub-areas based on a K-Means algorithm, and determine the regional picking task parameters of each picking sub-area; An allocation module 302 is used to determine a target picking interval corresponding to each robotic arm according to regional picking task parameters and characteristic parameters of each robotic arm based on an auction algorithm; The path planning module 303 is used to plan the picking path of each robotic arm according to the spatial position of each robotic arm and the target picking interval by using a preset fusion algorithm to obtain a picking planning path.

[0035] In a specific embodiment, the regional picking task parameters include the type of task, difficulty, estimated working time, etc., and the characteristic parameters of the robotic arm include the working state, load capacity, spatial position, etc. of the robotic arm.

[0036] The specific method of using the auction algorithm based on the regional picking task parameters and each robotic arm is as follows: each robotic arm expresses its interest in the picking task in the form of bidding according to its own conditions and preferences. i Robotic arms j The bid price is bij . Sort all the robotic arms from high to low according to the bidding value, and get the bidding ranking vector of the robotic arms r =[ r 1,r 2,…, rN ],in ri Indicates ranking i The number of the robot arm. According to the bidding ranking, each robot arm is asked in turn whether to accept the task. If a robot arm does not accept the task, the next robot arm is asked until a robot arm accepts the task.

[0037] It should be noted that if multiple manipulators accept tasks at the same time, the best manipulator is selected based on the matching degree between the manipulator and the task. The matching degree can be calculated using the function f ( j , m ) to indicate that j and m Represent the numbers of the robot arm and the object respectively. The matching evaluation may involve calculating the distance between the robot arm and the object, whether the load capacity meets the requirements of the object, etc. Specifically, a weighted average function can be defined g ( j , m ), as follows: g ( j , m )= αf ( j , m )+(1− α ) brjj in, α is the decay coefficient, which is used to control the relative weight of matching and bidding. Generally, it can be set to 0.5, which balances the contribution of bidding and matching; brjj Represents a robotic arm j The competitive value is weighted according to the bidding ranking.

[0038] According to the weighted average function g ( j , m ) value, select the robot arm corresponding to the largest value to perform the picking task. In the optimal allocation decision, an exponentially weighted moving average can be used to comprehensively consider the bidding and matching degree.

[0039] In a preferred embodiment, the preset fusion algorithm is constructed based on Q-learning and an optimized genetic algorithm. A preliminary picking path is obtained using Q-learning, and the preliminary picking path is optimized using a genetic algorithm to obtain a picking planning path.

[0040] Specifically, the Q-learning algorithm is first used to train the robot arm to select the best action in different states to obtain a preliminary activity path planning. Then, the optimization genetic algorithm is used to optimize and improve these paths to improve the quality and efficiency of the paths. By iteratively alternating the Q-learning and optimization genetic algorithms, the activity path of the robot arm is gradually optimized until the optimal picking planning path is obtained. The main steps include: Step S11: define the problem space: define the problem space of the robot arm activity path planning, including the joint angles of the robot arm, the position and posture of the end effector, etc.; Step S12: Initialize the Q value table and population: initialize an initial Q value for each state-action pair, and randomly generate an initial population, each individual represents a possible path planning solution; Step S13: Q-learning phase, including the following sub-steps: Step a1: Based on the current state and the Q-value table, select an action to perform. You can use the ε-greedy strategy to randomly select actions with a certain probability in order to explore new possibilities; in other cases, select the action with the highest Q-value; Step b1: Execute actions and observe rewards: Execute the selected actions and observe the feedback rewards given by the environment, that is, calculate the fitness of the current path planning scheme; Step c1: Update Q value: Update the Q value of the previously selected action based on the observed reward. The update formula is as follows: Q(s, a) = Q(s, a) + α * (r + γ * maxQ(s', a') - Q(s, a)) Where Q(s, a) is the Q value of executing action a in state s, r is the reward obtained, s' is the new state after executing the action, a' is the action selected in the new state, α is the learning rate, and γ is the discount factor; Step d1: Iterative update, repeating steps a1 to c1 until convergence or reaching a predetermined number of iterations; Step S14: Optimizing the genetic algorithm phase, including the following sub-steps: Step a2: Selection operation: According to the fitness of each individual, a selection operation is used to select a part of the better individuals from the current population; Step b2: Crossover operation: Through the crossover operation, the selected individuals are paired and new individuals are generated; Step c2: Mutation operation: Perform mutation operation on the newly generated individuals to introduce a certain amount of randomness to increase the diversity of the population; Step d2: Replacement operation: replace some individuals in the original population with newly generated individuals; Step e2: Iterative update: Repeat steps a2 to d2 until a predetermined number of iterations is reached; Step S15: Combining Q-learning and optimized genetic algorithm: In each iteration of the genetic algorithm, the Q-learning algorithm is used to guide the selection and fitness evaluation of individuals. Specifically, the best fitness value in each state can be stored in the Q-value table of Q-learning and used to calculate the fitness of the individual. After the iteration of Q-learning and optimized genetic algorithm is completed, the individual with the highest fitness is selected as the final path planning solution.

[0041] In a preferred embodiment, the motion control module includes a model building module, a motion analysis module and an instruction output module; A model building module is used to build a robot kinematics model of the picking robot; The motion analysis module is used to perform forward and reverse position analysis of the picking robot based on the planned path and the robot kinematic model to obtain the joint activity data of each mechanical arm of the picking robot; The command output module is used to generate control commands according to the joint activity data of the robot arm, and control the robot to pick and suck the tea buds in the target picking area according to the planned path.

[0042] In a preferred embodiment, the robot kinematic model is a Delta parallel robot kinematic model. The Delta parallel robot is a special parallel robot system, and the model describes the relationship between the position and posture of the robot's end effector and the joint angles of each movable arm. In the work of picking young tea buds, the coordinate system of the Delta robot is first defined, and the relationship between each coordinate system is established. Through measurement and calibration, the position and posture information of each movable arm end effector relative to its corresponding local coordinate system can be obtained. The position and posture of the end effector are determined according to the spatial coordinates of the target picking area, and then the joint angles of each movable arm are solved.

[0043] In a preferred embodiment, the motion analysis module is also used to perform singularity analysis and workspace analysis on the parallel robot, and optimize the joint activity data so that the range of motion of the robot arm covers the target picking area.

[0044] Specifically, for singularity analysis, the location of singular points can be determined by calculating the Jacobian matrix and checking whether its determinant is zero. Workspace analysis can use the inverse solution method to calculate the ability of the robot to reach different positions and postures based on the conditions such as the rotation range of the robot joint, the size of the end effector and the posture restriction, thereby determining the workspace of the robot. Based on the joint activity model, the optimization algorithm is used to find the optimal joint activity data so that the range of motion of the robot covers the target picking area. The optimized joint activity data is input into the robot to verify whether its effect meets the requirements. If not, the optimization algorithm can be readjusted or the joint activity model can be modified for iterative optimization to ensure that the range of motion of the robot covers the target picking area.

[0045] Furthermore, when a single target is picked, the next target picking process is entered in the current area. When all targets in the current ceiling camera shooting area are picked, the walking module is used to move to the next area and repeat the steps of image recognition, area division and motion control until the tea bud picking operation is completed.

[0046] The system of this embodiment uses local visual recognition to establish a coordinate system and transmits data to a local task allocation path planning module, which allocates the calculated task data to multiple robotic arm end effectors to pick tea leaves and drives the travel system to move after the task is completed. Therefore, the data calculation and control of the entire system are completely performed locally, which can ensure the smoothness and efficiency of the system.

[0047] As a specific embodiment, the robot arm is modeled using the URDF model (Unified Robot Description Format), which is a universal robot modeling language. The URDF model defines the robot's structure, linkage relationship, sensor and kinematic information, etc., and is stored in XML format. The URDF model is commonly used in applications such as robot simulation, path planning, and control design.

[0048] It should be noted that the URDF model is provided by the picking hand (robotic arm) manufacturer or developed independently. It describes the length, relative position, size and other parameters of each link of the robotic arm and is used to solve the forward and inverse kinematics of the robotic arm. The URDF model is solved by the movelt module of the ROS software, the picking trajectory of the robotic arm is planned, the joint angle of each motion link is inversely solved, and the motion control instructions are generated according to the joint angles. The hardware system performs corresponding picking and suction work according to the control instructions to complete the picking of tea buds. The above method can control the robotic arm to work with high precision and efficiency, so that the picking robot can accurately pick tea buds, including picking under the influence of wind speed, light, slope and other factors.

[0049] The tea bud picking robot intelligent control system provided in this embodiment uses the YOLO model to locally identify and locate tea leaves, locally calculates how to allocate tea collection tasks to each robotic arm in the best way, and then enables each robotic arm to pick and collect tea leaves with a reasonable path and correct posture, and has the advantages of high generalization, strong robustness and fast response speed. The hardware sensor required for visual recognition is an ordinary camera with low cost. By organically combining the recognition, analysis, processing and collection processes, a full-process production is formed, so that the system can complete the picking process by one machine in an actual commercial production environment, and no additional hardware support is required. This system can control the picking of multiple robotic arms, and the picking efficiency in the actual production environment is significantly higher than that of humans, which significantly improves the benefits of tea farmers.

[0050] The embodiment of the present application also provides an intelligent control method for a tea bud picking robot, which is implemented by using any one of the intelligent control systems for a tea bud picking robot described in the above technical solution, and the method includes: The visual analysis module obtains scene image information, identifies the tea bud area to be picked in the scene image information based on a preset neural network model, and establishes a spatial coordinate system to determine the relative position of the picking robot and the tea bud area; The task decomposition module uses a preset division method to divide the tea bud area into several picking sub-areas, and according to the spatial positions of the multiple picking sub-areas and the multiple robotic arms of the picking robot, multiple target picking areas corresponding to each robotic arm are determined, and a planning path is formulated based on the relative positions of each robotic arm and the multiple target picking areas; The motion control module determines the joint activity data of each robotic arm according to the planned path, and controls each robotic arm according to the joint activity data to pick and absorb the tea buds in the target picking area according to the planned path.

[0051] It should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0052] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An intelligent control system for a tea bud picking robot, characterized in that: It includes a visual analysis module, a task decomposition module and a motion control module connected in sequence; A visual analysis module is used to obtain scene image information, identify the tea bud area to be picked in the scene image information based on a preset neural network model, and establish a spatial coordinate system to determine the relative position of the picking robot and the tea bud area; The task decomposition module is used to divide the tea bud area into a number of picking sub-areas using a preset division method, determine a number of target picking areas corresponding to each robotic arm according to the spatial positions of the multiple picking sub-areas and the multiple robotic arms of the picking robot, and formulate a planning path based on the relative positions of each robotic arm and the multiple target picking areas; The task decomposition module includes a partitioning module, an allocation module, and a path planning module; A partitioning module is used to divide the tea bud area into several picking sub-areas based on the K-Means algorithm and determine the regional picking task parameters of each picking sub-area; An allocation module, used to determine a target picking interval corresponding to each robotic arm according to regional picking task parameters and characteristic parameters of each robotic arm based on an auction algorithm; A path planning module is used to preset a fusion algorithm to plan the activity path information of each robotic arm according to the spatial position of each robotic arm and the target picking interval to obtain a picking planning path; The preset fusion algorithm is constructed based on Q-learning and optimized genetic algorithm, Q-learning is used to obtain a preliminary picking path, and the genetic algorithm is used to optimize the preliminary picking path to obtain a picking planning path; The motion control module is used to determine the joint activity data of each robotic arm according to the planned path, and control each robotic arm to pick and suck the tea buds in the target picking area according to the planned path according to the joint activity data.

2. The intelligent control system for tea bud picking robots according to claim 1, characterized in that: The visual analysis module includes an image acquisition module, a recognition module and a coordinate establishment module; An image acquisition module is used to obtain scene image information of the current position of the picking robot through a visual sensor; A recognition module is used to identify the tea bud area to be picked in the scene image information based on the YOLO neural network model; The coordinate establishment module is used to establish a spatial coordinate system, determine the spatial position of the tea bud area and the spatial position of multiple mechanical arms of the picking robot, and establish the positional relationship between the tea bud area and the mechanical arms of the picking robot and the picking robot body.

3. The intelligent control system of the tea bud picking robot according to claim 1, characterized in that: The motion control module includes a model building module, a motion analysis module and a command output module; A model building module is used to build a robot kinematics model of the picking robot; The motion analysis module is used to perform forward and reverse position analysis of the picking robot based on the planned path and the robot kinematic model to obtain the joint activity data of each mechanical arm of the picking robot; The command output module is used to generate control commands according to the joint activity data of the robot arm, and control the robot to pick and suck the tea buds in the target picking area according to the planned path.

4. The intelligent control system for tea bud picking robots according to claim 3, characterized in that: The robot kinematics model is a Delta parallel robot kinematics model.

5. The intelligent control system for tea bud picking robots according to claim 4, characterized in that: The motion analysis module is also used to perform singularity analysis and workspace analysis on the parallel robot and optimize the joint activity data so that the range of motion of the robot arm covers the target picking area.

6. The intelligent control system for tea bud picking robots according to claim 1, characterized in that: The system is built based on the ROS platform.

7. An intelligent control method for a tea bud picking robot, characterized in that: The method is implemented by using any one of the tea bud picking robot intelligent control systems as described in claims 1 to 6, comprising: The visual analysis module obtains scene image information, identifies the tea bud area to be picked in the scene image information based on a preset neural network model, and establishes a spatial coordinate system to determine the relative position of the picking robot and the tea bud area; The task decomposition module uses a preset division method to divide the tea bud area into several picking sub-areas, and according to the spatial positions of the multiple picking sub-areas and the multiple robotic arms of the picking robot, multiple target picking areas corresponding to each robotic arm are determined, and a planning path is formulated based on the relative positions of each robotic arm and the multiple target picking areas; The motion control module determines the joint activity data of each robotic arm according to the planned path, and controls each robotic arm according to the joint activity data to pick and absorb the tea buds in the target picking area according to the planned path.

Images