Control method and control system of tree planting robot

Abstract

The invention discloses a control method and system of a tree planting robot, and belongs to the technical field of tree planting robot control, and the method comprises the steps: adjusting or planning a motion path according to terrain parameters; according to the environment parameters, the execution action of the tree planting robot is adjusted; generating a control strategy of the robot according to the motion path and the execution action; and controlling the tree planting robot to work according to the control strategy. According to terrain parameters and environment parameters, planning of a motion path and adjustment of planting execution actions are respectively carried out, the operation strategy of the tree planting robot is adjusted in combination with planning and adjustment, automatic monitoring and control can be achieved, the efficiency of tree planting operation is remarkably improved, the operation quality is improved, and the labor cost is reduced. And the device can adapt to tree planting operation in a severe environment.

Description

Technical Field

[0001] The invention relates to the technical field of tree planting robot control, and in particular to a control method and a control system for a tree planting robot. Background Art

[0002] Traditional tree planting methods face problems such as low efficiency, low survival rate, and high labor costs in deserts and desertified areas. With the development of science and technology, tree planting activities are gradually moving from manual operations to mechanization, intelligence, and automation.

[0003] At present, a large number of tree planting robots have been developed, which usually use human driving control or close-range remote control to operate. They cannot meet the needs of large-scale and efficient tree planting. Therefore, it is necessary to develop a new control method and control system for intelligent tree planting robots to automatically control the operation of tree planting robots and tree planting operations. Summary of the invention

[0004] In view of the above-mentioned technical problems existing in the prior art, the present invention provides a control method and control system for a tree planting robot, which automatically controls the operation rules of the tree planting robot and the tree planting operation to improve the planting efficiency.

[0005] The present invention discloses a control method for a tree planting robot, comprising the following steps: adjusting or planning a motion path according to terrain parameters; adjusting the execution action of the tree planting robot according to environmental parameters; generating a control strategy for the robot according to the motion path and the execution action; and controlling the operation of the tree planting robot according to the control strategy.

[0006] Preferably, the method of adjusting or regularizing the motion path includes:

[0007] Based on the Hybrid A* method, global path planning is obtained;

[0008] Based on Teb's method, the local path of the global path planning is planned to obtain the local path;

[0009] Determining whether the local path includes a first path; the first path includes a U-shaped bend or a herringbone bend;

[0010] If included, the first path is tracked and controlled based on the PID method.

[0011] Preferably, the method for planting path planning includes:

[0012] Obtain a planting area, and set planting points and interval information between planting points in the planting area;

[0013] Plan multiple first trackings according to the rows or columns of planting points;

[0014] Planning a turning point between two adjacent first tracking points, wherein the turning point adopts G2 curve fitting;

[0015] Based on the first tracking and the turning points, a planting path is generated.

[0016] Preferably, the method for identifying the position of a sapling includes:

[0017] Collect 3D point cloud data in the environment;

[0018] Segmenting the point cloud data according to the parking space height or the first threshold height to obtain a first point cloud data set;

[0019] The first point cloud data set is screened according to the geometric information of the sapling to obtain a second point cloud data set;

[0020] Matching the second point cloud data set based on the size of the geometry to obtain the location of the sapling; or

[0021] Based on the sapling recognition model, the second point cloud data set is recognized to obtain the position of the sapling.

[0022] Preferably, the saplings are managed and maintained according to the positions of the saplings;

[0023] Methods for adjusting the actions performed by the tree planting robot include:

[0024] Obtaining environmental parameter values, the environmental parameters including: soil temperature, air temperature, wind speed and light intensity;

[0025] According to the environmental parameter value, matching the corresponding behavior strategy;

[0026] According to the behavior strategy, the operating parameters of the tree planting robot are adjusted, and the operating parameters include: watering amount, watering frequency or hole depth of the tree planting point.

[0027] Preferably, the method for identifying landforms or dynamic objects includes:

[0028] Segmenting the point cloud data according to a third threshold height to obtain a third point cloud data set;

[0029] According to the third point cloud data set, identifying obstacle features, the obstacle features including: potholes, obstacles or dynamic objects;

[0030] Determining whether the tree-planting robot is passable according to the size of the obstacle feature on the path;

[0031] If it is impassable, the path is adjusted or the dynamic object is waited for to pass according to the size of the obstacle feature.

[0032] Preferably, the training method of the seedling recognition model includes:

[0033] A first training set is obtained by using terrain parameters and environmental parameters through sensors;

[0034] Preprocessing, screening and data annotation of the first training set to obtain a second training set;

[0035] Based on the classification method, the second training set is trained to obtain a seedling recognition model.

[0036] Preferred method for selecting planting area:

[0037] Get weather parameters for multiple regions;

[0038] Determining whether the weather parameters in the area are conducive to the planting and growth of saplings;

[0039] If so, regular movement paths and planting tasks are determined based on the terrain parameters of the area.

[0040] The present invention also provides a control system for implementing the above control method, comprising an intelligent monitoring module, a first path planning module, an action adjustment module and a strategy management module.

[0041] The intelligent monitoring module is used to collect terrain parameters and environmental parameters;

[0042] The first path planning module is used to adjust or plan the movement path according to the terrain parameters;

[0043] The action adjustment module is used to adjust the execution action of the tree planting robot according to the environmental parameters;

[0044] The strategy management module is used to generate a control strategy for the robot according to the motion path and execution action; and the control strategy is executed by the tree planting robot.

[0045] Preferably, the control system further comprises an effect evaluation module and a host computer connected to the tree planting robot, and the tree planting robot is deployed with a second planning module, a third planning module and a recognition module;

[0046] The first path planning module is deployed on the host computer and obtains global path planning based on the Hybrid A* method;

[0047] The second planning module is used to plan the local path of the global path planning based on Teb's method to obtain the local path;

[0048] If the local path includes a U-shaped bend or a herringbone bend, calling a third planning module, the third planning module tracking and controlling the U-shaped bend or the herringbone bend based on a PID method;

[0049] The identification module is used to identify obstacle features, and identify the location and type of saplings;

[0050] The effect evaluation module is used to evaluate the planting effect according to the tree planting effect, economic cost and economic benefit.

[0051] Compared with the prior art, the beneficial effects of the present invention are as follows: the motion path is planned and the planting execution action is adjusted according to the terrain parameters and the environmental parameters respectively, and the operation strategy of the tree planting robot is adjusted in combination with the planning and adjustment, which can realize automatic monitoring and control, significantly improve the efficiency of tree planting operations, improve the operation quality, reduce labor costs, and can adapt to tree planting operations in harsh environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 is a flow chart of the control method of the tree planting robot of the present invention;

[0053] Figure 2 It is a flow chart of the motion path adjustment and planning method;

[0054] Figure 3 It is a schematic diagram of sapling identification and obstacle identification;

[0055] Figure 4 It is a logic block diagram of the control system of the present invention. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0057] The present invention is further described in detail below in conjunction with the accompanying drawings:

[0058] The present invention provides a control method for a tree planting robot. Figure 1 As shown, the following steps are included:

[0059] Step 101: Collect terrain parameters, environmental parameters, historical data, planting data, and management data.

[0060] Historical data include: the latitude and longitude of the planting point, whether it is planted, the type of seedlings, the planting interval, the planting area, etc. The terrain parameters and environmental parameters can be collected by the following sensors: camera, lidar, IMU and RTK, etc.

[0061] Step 102: Adjust or plan the movement path according to the terrain parameters and the planting data.

[0062] Step 103: Adjust the execution action of the tree planting robot according to the environmental parameters and planting data.

[0063] Step 104: Generate a control strategy for the robot according to the motion path and the execution action.

[0064] Step 105: Control the tree planting robot to operate according to the control strategy.

[0065] Step 106: Analyze the operation data, adjust and update some parameters, and execute steps 102 and 103.

[0066] The motion path is planned and the planting execution action is adjusted according to the terrain parameters and environmental parameters respectively. The operation strategy of the tree planting robot is adjusted in combination with planning and adjustment. Automated monitoring and control can be achieved, which can significantly improve the efficiency of tree planting operations, enhance operation quality, reduce labor costs, and can adapt to tree planting operations in harsh environments.

[0067] In the first specific motion path adjustment and planning method, the SOM algorithm is used to optimize the tree planting planning path and execution action. The terrain information is input into the SOM for learning. The SOM is conducive to understanding the topological structure and characteristics of the terrain, so as to plan a path that adapts to the terrain. Once the path planning is completed, the SOM can also be used to adjust the order and method of executing actions. For example, based on the location and terrain conditions of each tree planting point, the SOM can recommend the best tree planting order or the optimal movement path to maximize the tree planting efficiency and survival rate.

[0068] The self-organizing map is an artificial neural network proposed by Finnish scientist Teuvo Kohonen in 1980. The basic principle is to automatically form a low-dimensional representation space (usually two-dimensional) by learning the topological structure of sample data, in which similar input samples are mapped to adjacent areas, thereby retaining the topological structure characteristics of the input data.

[0069] The steps of the SOM algorithm include: (1) Initialize the map: Create a two-dimensional grid (map), each grid cell has a vector with the same dimension as the input data, representing the weight vector of the cell. These weight vectors are initially initialized randomly. (2) Learning process: Iteratively process the input data set, each time selecting an input sample and finding the nearest cell on the map (i.e., the best matching unit, BMU). Then, adjust the weight vectors of the BMU and its neighboring cells to make them closer to the current input sample. (3) Neighborhood function: Define a neighborhood function to determine which cells will be adjusted during the learning process. Usually, in the early stages of training, the neighborhood function is large, allowing extensive weight updates; over time, the neighborhood function gradually decreases, so that the weight updates are gradually concentrated in the local area around the BMU. (4) Convergence: Through multiple iterative learning processes, the self-organizing map eventually converges to a state that can effectively represent the distribution of the input data. At this time, the weight vector of each cell represents a certain structure or feature of the input data space.

[0070] Once the path planning is completed, SOM can also be used to adjust the order and manner of performing actions. For example, based on the location and terrain conditions of each tree planting point, SOM can recommend the best tree planting order or the optimal movement path to maximize tree planting efficiency and survival rate.

[0071] like Figure 2 In the second specific motion path adjustment and planning method, the following steps are adopted:

[0072] Step 201: Based on the Hybrid A* method, obtain global path planning.

[0073] Considering the geometric dimensions of the robot, each point on the generated path is checked for collision with obstacles. When a collision occurs at the path point, it is not satisfied and the search for a new path continues. The Hybrid A* algorithm is optimized, and the calculation of the actual cost value of G is determined by multiple strategies such as the length of the trajectory, the number of turns of the robot, whether the slope of the trajectory changes smoothly, and whether it needs to go backward.

[0074] Step 202: Based on Teb's method, a local path of the global path planning is planned to obtain a local path.

[0075] Consider optimizing the trajectory of a certain distance range in the direction of the robot's driving path. First, convert the dynamic obstacle point cloud into the cost map, calculate the cost value of this part of the obstacle point cloud, and evaluate the cost value using Euclidean distance. Add the obstacle cost to the optimization objective function for solution.

[0076] Step 203: Determine whether the local path includes a first path; the first path includes a U-shaped bend or a herringbone bend.

[0077] If included, execute step 204: track and control the first path based on the PID method.

[0078] The generated local path is used for path tracking control of the robot. The control process can use the adaptive PID algorithm to calculate the PID parameters according to the lateral deviation and angular deviation of the nearest point in the current position path point, and then calculate the angular velocity and linear velocity of the robot. For example, if the linear velocity is 0.2m / s, smooth walking can be achieved without speed sampling. The deviation change is calculated within a certain period of time, and the PID parameters are adjusted in real time according to the deviation change to adapt to the tracking in the U-shaped / zigzag path.

[0079] If not included, no tracking control is performed on the local path.

[0080] Step 205: Planting path planning:

[0081] Step 211: Obtain a planting area, and set planting points and interval information between planting points in the planting area.

[0082] Step 212: Plan multiple first tracking points based on the rows or columns of the planting points and the spacing information. The first tracking points are linear or curved.

[0083] Step 213: planning a turning point between two adjacent first tracking points, wherein the turning point is fitted using a G2 curve.

[0084] Step 214: Generate a planting path based on the first tracking and the turning point.

[0085] In the continuous planting operation in the planting area, a bow-shaped trajectory is planned to meet the scene requirements and improve planting efficiency.

[0086] The present invention also includes a method for identifying the location of a sapling and managing the sapling:

[0087] Step 301: Collect three-dimensional point cloud data in the environment.

[0088] The preprocessing of cloud point data includes: setting distance and angle thresholds, intensity range to filter point cloud data. Point cloud downsampling processing, setting voxel sampling method and threshold to filter point cloud data.

[0089] Step 302: segment the point cloud data according to the parking space height or the first threshold height to obtain a first point cloud data set.

[0090] Step 303: According to the geometric information of the sapling, the first point cloud data set is screened to obtain a second point cloud data set. Execute step 304 or 305.

[0091] Step 304: Match the second point cloud data set based on the geometric size to obtain the position of the sapling.

[0092] Step 305: Based on the sapling recognition model, identify the second point cloud data set to obtain the position of the sapling.

[0093] Step 306: maintaining the saplings according to the positions of the saplings.

[0094] In path planning, Figure 3 It can also identify and avoid obstacles such as terrain and dynamic objects:

[0095] Step 311: segment the point cloud data according to the third threshold height to obtain a third point cloud data set. The point cloud data of the topography and the dynamic object are segmented by the third threshold.

[0096] Step 312: According to the third point cloud data set, obstacle features are identified, wherein the obstacle features include: potholes, obstacles or dynamic objects, including the size of the obstacle features. The obstacle features may be identified by an obstacle identification model based on a machine learning method.

[0097] Step 313: Determine whether the tree planting robot is passable based on the size of the obstacle features on the path and the size of the tree planting robot.

[0098] If it is not passable, execute step 312: adjust the path according to the size of the obstacle feature. For dynamic objects, wait until the dynamic object passes before passing. A dynamic model can also be established based on a deep learning method to identify the dynamic object, such as animals, people, cars, etc.

[0099] If possible, the path is confirmed.

[0100] Among them, the training method of the recognition model includes:

[0101] Step 321: Using the terrain parameters and environmental parameters through sensors, obtain a first training set.

[0102] Step 322: preprocess, screen and label the first training set to obtain a second training set.

[0103] More specifically, data preprocessing includes: Data cleaning: remove outliers, noise data or incomplete data points to ensure data quality. Data normalization: normalize the data collected by different sensors so that they can be compared at the same level. Feature extraction: extract useful features from the raw data, such as extracting texture features or edge features from images, extracting terrain features from LiDAR data, etc. It is also possible to denoise images from cameras and detect outliers and extract terrain features from LiDAR data.

[0104] Data labeling can be done by: Manual labeling: manually labeling data to provide the machine learning model with the label information required for supervised learning. Automatic labeling: using other sensor data or rules for automatic labeling to reduce labor costs. Information from different sensors can be combined for comprehensive labeling to provide more comprehensive data labels. For example, manually labeling images collected by cameras to mark the location and type of trees, or automatically labeling terrain features using lidar data.

[0105] Step 323: Based on the classification method, the second training set is trained to obtain a seedling recognition model.

[0106] Step 324: preprocess, screen and label the first training set to obtain a third training set.

[0107] Step 325: Based on a classification method, the third training set is trained to obtain an obstacle recognition model.

[0108] In step 103, the method for adjusting the tree planting robot to perform an action includes:

[0109] Step 401: Obtain environmental parameter values, where the environmental parameters include soil temperature, air temperature, wind speed, and light intensity.

[0110] Step 402: Match the corresponding behavior strategy according to the environmental parameter value.

[0111] The behavior strategy may be tested in a preset manner.

[0112] Step 403: According to the behavior strategy, the operating parameters of the tree planting robot are adjusted, and the operating parameters include: watering amount, watering frequency or hole depth of the tree planting point. For example, when the soil is detected to be dry, the upper computer can issue instructions to the robot to adjust the depth of the tree planting point or increase the watering frequency to ensure that the growth environment of each tree reaches the best state. This real-time feedback and adjustment capability enables the tree planting robot to operate efficiently under different environmental conditions and improve the overall tree planting effect.

[0113] Adjusting operating parameters according to environmental parameter values ​​can effectively improve planting efficiency and the survival rate of seedlings.

[0114] The present invention also provides a method based on selecting a planting area:

[0115] Step 501: Obtain weather parameters of multiple areas.

[0116] Step 502: Determine whether the current and / or next 5-day weather parameters in the area are conducive to the planting and growth of saplings.

[0117] If yes, execute step 501: according to the terrain parameters of the area, regular motion paths and planting tasks.

[0118] If not, select from multiple areas the area that is most conducive to the planting and growth of the seedlings.

[0119] Optionally, the weather parameters and the predicted weather parameters can be used to predict whether the weather parameters within a period of time are conducive to the growth of the seedlings. If the weather parameters are conducive to the growth of the seedlings, the area is added to the candidate planting area.

[0120] The present invention also provides a real-time control method:

[0121] Step 601: Obtain first data of tree planting or seedling management.

[0122] Step 602: Perform trend analysis on the first data to generate a data model.

[0123] Step 603: Optimize the motion path and execute the action according to the data model.

[0124] The second aspect of the present invention also provides a control system for implementing the above control method, such as Figure 4 As shown, it includes an intelligent monitoring module 11, a first path planning module 12, an action adjustment module 24 and a strategy management module 23.

[0125] The intelligent monitoring module 11 is used to collect terrain parameters and environmental parameters;

[0126] The first path planning module 12 is used to adjust or plan the movement path according to the terrain parameters;

[0127] The action adjustment module 24 is used to adjust the execution action of the tree planting robot according to the environmental parameters;

[0128] The strategy management module 23 is used to generate a control strategy for the robot according to the motion path and the execution action;

[0129] The control strategy is executed by the actuator 25 of the tree-planting robot 2 .

[0130] More specifically, the intelligent monitoring module integrates high-precision Beidou, RTK and I MU combined navigation systems to ensure stable driving and precise positioning of the robot in complex terrain and environments lacking reference objects. At the same time, advanced machine vision technology is used to identify the tree planting location and terrain features on the ground in real time, providing accurate data support for path planning and tree planting operations. Collect relevant regional information, such as longitude and latitude, altitude, etc.; collect forestry resource data from all over the country, such as forest coverage, tree species distribution, forest type, forest age structure, forest storage, forest growth rate, etc.; collect meteorological data, such as temperature, humidity, precipitation, wind direction, wind speed, air pressure, light intensity and air quality, etc.; collect soil data, such as soil type, soil texture, soil pH, soil nutrient content, soil moisture status, soil carbon storage, carbon emission coefficient, etc.; collect human activity data, such as forest felling, afforestation, transportation distance, etc. Appropriate data processing tools and libraries can be used to clean, organize and standardize the data to ensure the accuracy and consistency of the data. And store it in a suitable data structure.

[0131] In addition to the robot's own sensor data, environmental information can also be obtained from external data sources, such as weather data, map data, etc. These data are crucial for decision-making and path planning.

[0132] The control system of the present invention further comprises a host computer 1 and an effect evaluation module 15 connected to the tree planting robot 2, and a second planning module 21, a third planning module 22 and an identification module 26 are deployed on the tree planting robot 2;

[0133] The first path planning module 11 is deployed on the host computer 1, and obtains global path planning based on the Hybrid A* method;

[0134] The second planning module 21 is used to plan the local path of the global path planning based on Teb's method to obtain the local path;

[0135] If the local path includes a U-shaped bend or a herringbone bend, the third planning module 22 is called, and the third planning module 22 tracks and controls the U-shaped bend or the herringbone bend based on the PID method;

[0136] The identification module 26 is used to identify obstacle features and identify the location and type of saplings;

[0137] The effect evaluation module 15 is used to evaluate the planting effect according to the tree planting effect, economic cost and economic benefit. More specifically, it includes the following evaluations: Tree planting effect evaluation: record the improvement of the ecological environment after tree planting, such as changes in vegetation coverage, desertification prevention and control effects, etc. Economic cost and benefit analysis: including evaluation data such as tree planting cost, saved ecological restoration costs, and future economic benefits.

[0138] The host computer has the ability to transmit and analyze data, and can remotely monitor the operating status and work progress of multiple tree-planting robots.

[0139] The control system of the present invention is also provided with a visualization module and a large-scale tree planting data model. The operator can perform real-time scheduling and issue instructions through the graphical interface of the visualization module, and make instant adjustments and optimization of operation strategies based on real-time feedback to ensure the efficient completion of tree planting tasks and improve the survival rate.

[0140] The visualization module provides a user-friendly graphical interface, allowing users to easily operate, set up and monitor the working status of the intelligent tree planting robot; allowing users to input tree planting instructions through the software, including but not limited to seedling type, planting location, planting depth, etc.; providing a real-time feedback mechanism to inform users of the robot's working status and results. The host computer can receive and process real-time data sent back from the intelligent tree planting robot, including but not limited to soil moisture, soil type, meteorological data, etc. The data can be quickly displayed, allowing users to understand the various data at the tree planting site in real time.

[0141] The visualization module can generate a tree planting map based on the tree planting instructions and real-time data input by the user. The tree planting map should clearly display the planting area, seedling type, planting location, planting time and other information, support dynamic simulation of the tree planting process, and enable users to intuitively understand the tree planting process. Allow users to zoom, pan and other operations on the tree planting map to better observe and adjust the tree planting plan.

[0142] The tree planting data model is based on historical tree planting data, soil data, meteorological data, etc. The model is used to perform data analysis to optimize tree planting plans and improve tree planting efficiency and quality. It provides a data export function that allows users to export analysis results into reports or file formats for further analysis and sharing.

[0143] The control system has high stability and reliability, ensuring normal operation in various environments. It supports multi-language display to meet the needs of users in different countries and regions. It provides regular update and upgrade services to adapt to new tree planting technologies and policy changes. It follows relevant security and privacy protection standards to protect the security and privacy of user data. It is a system that integrates rich interactive functions, large-scale data rapid display, tree planting map display and tree planting data model, with the goal of optimizing and improving the work efficiency and quality of intelligent tree planting robots.

[0144] The tree planting data model includes: Geographic Information System (GIS) data model: Geospatial data: including information such as tree planting location, regional boundaries, terrain elevation, etc. Geographic data attributes: such as soil type, climate conditions, water resources, etc. These attributes have a great impact on tree planting and need to be considered in the model. Tree information data model: Tree species information: the growth habits, adaptability and ecological functions of each tree. Tree health status: including data such as tree age, growth rate, and health status. Environmental monitoring data model: Climate data: including data such as precipitation, temperature, and humidity, which affect tree growth and survival rate. Soil data: soil type, nutrient content, pH, etc. Water resource data: including data such as precipitation and groundwater level. Tree planting operation data: Tree planting time and location: record the specific time and location of each tree planting operation. Tree planting process: including planting methods, tools and equipment used, etc. Tree planting personnel data: record information about personnel involved in tree planting, including skill level and training.

[0145] The present invention can improve the efficiency of tree planting: through automated monitoring and control, the efficiency of tree planting operations can be significantly improved. It can improve the quality of operations: accurate path planning and execution of actions can improve the survival rate of planted trees. It can adapt to harsh environments: stable operations in deserts, desertified areas or urban greening projects. It can reduce labor costs: reduce dependence on labor and reduce labor costs.

[0146] For example, when the robot is working in a desertified area, its speed and path selection can be adjusted according to the terrain data to ensure safety and work efficiency. In an urban environment, the robot's operating time and area can be dynamically adjusted according to the surrounding buildings and crowd activities to avoid interference and optimize work results.

[0147] The control system initialization process of the present invention includes the steps of starting the system, initializing the hardware, loading the configuration file, starting the communication module and connecting the robot.

[0148] When the tree planting task is completed, the shutdown process includes: system shutdown preparation, saving data and logs, shutting down the intelligent monitoring module, and shutting down the host computer after the module is shut down.

[0149] The invention provides an innovative solution for environmental restoration and ecological protection, and demonstrates the potential and application prospects for sustainable development in desertified and inaccessible areas. The scope of application should include but is not limited to forest restoration projects, ecological restoration projects, and landscaping.

[0150] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A control method for a tree planting robot, characterized in that: The following steps are involved: Adjust or plan movement paths according to terrain parameters; Adjust the execution actions of the tree-planting robot according to environmental parameters; generating a control strategy for the robot according to the motion path and the execution action; The tree planting robot is controlled to operate according to the control strategy.

2. The control method according to claim 1, characterized in that: Methods for adjusting or regularizing motion paths include: Based on the Hybrid A* method, global path planning is obtained; Based on Teb's method, the local path of the global path planning is planned to obtain the local path; Determining whether the local path includes a first path; the first path includes a U-shaped bend or a herringbone bend; If included, the first path is tracked and controlled based on a PID method.

3. The control method according to claim 2, characterized in that: It also includes methods for planting path planning: Obtain a planting area, and set planting points and interval information between planting points in the planting area; Plan multiple first trackings according to the rows or columns of planting points; Planning a turning point between two adjacent first tracking points, wherein the turning point adopts G2 curve fitting; Based on the first tracking and the turning points, a planting path is generated.

4. The control method according to claim 1, characterized in that: It also includes a method for identifying the location of saplings: Collect 3D point cloud data in the environment; Segmenting the point cloud data according to the parking space height or the first threshold height to obtain a first point cloud data set; The first point cloud data set is screened according to the geometric information of the sapling to obtain a second point cloud data set; Matching the second point cloud data set based on the size of the geometry to obtain the location of the sapling; or Based on the sapling recognition model, the second point cloud data set is recognized to obtain the position of the sapling.

5. The control method according to claim 4, characterized in that: Caring for the saplings according to the positions of the saplings; Methods for adjusting the actions performed by the tree planting robot include: Obtaining environmental parameter values, the environmental parameters including: soil temperature, air temperature, wind speed and light intensity; According to the environmental parameter value, matching the corresponding behavior strategy; According to the behavior strategy, the operating parameters of the tree planting robot are adjusted, and the operating parameters include: watering amount, watering frequency or hole depth of the tree planting point.

6. The control method according to claim 4, characterized in that: It also includes methods for obstacle feature recognition: Segmenting the point cloud data according to a third threshold height to obtain a third point cloud data set; According to the third point cloud data set, identifying obstacle features, the obstacle features including: potholes, obstacles or dynamic objects; Determining whether the tree-planting robot is passable according to the size of the obstacle feature on the path; If it is impassable, the path is adjusted or the dynamic object is waited for to pass according to the size of the obstacle feature.

7. The control method according to claim 4, characterized in that: The training method of the sapling recognition model includes: A first training set is obtained by using terrain parameters and environmental parameters through sensors; Preprocessing, screening and data annotation of the first training set to obtain a second training set; Based on the classification method, the second training set is trained to obtain a seedling recognition model.

8. The control method according to claim 1, characterized in that: It also includes methods for selecting planting areas: Get weather parameters for multiple regions; Determining whether the weather parameters in the area are conducive to the planting and growth of saplings; If so, regular movement paths and planting tasks are determined based on the terrain parameters of the area.

9. A control system for a tree planting robot, characterized in that: For implementing the control method according to any one of claims 1 to 8, the control system comprises an intelligent monitoring module, a first path planning module, an action adjustment module and a strategy management module, The intelligent monitoring module is used to collect terrain parameters and environmental parameters; The first path planning module is used to adjust or plan the movement path according to the terrain parameters; The action adjustment module is used to adjust the execution action of the tree planting robot according to the environmental parameters; The strategy management module is used to generate a control strategy for the robot according to the motion path and execution action; and the control strategy is executed by the tree planting robot.

10. The control system according to claim 9, characterized in that: It also includes an effect evaluation module and a host computer connected to the tree planting robot, and the tree planting robot is deployed with a second planning module, a third planning module and a recognition module; The first path planning module is deployed on the host computer, and obtains global path planning based on the Hybrid A* method; The second planning module is used to plan the local path of the global path planning based on Teb's method to obtain the local path; If the local path includes a U-shaped bend or a herringbone bend, calling a third planning module, the third planning module tracking and controlling the U-shaped bend or the herringbone bend based on a PID method; The identification module is used to identify obstacle features, and identify the location and type of saplings; The effect evaluation module is used to evaluate the planting effect according to the tree planting effect, economic cost and economic benefit.

Images