Weeding path planning method for agricultural weeding robot

Abstract

The invention discloses a weeding path planning method for an agricultural weeding robot. The weeding path planning method comprises the following steps that 1, the weeding robot is started, and whether all sensors and cameras work normally or not is checked; 2, establishing a basic three-dimensional map of the farmland; 3, generating a real-time weed distribution map; 4, identifying the types and forms of the weeds, and marking on the farmland map; 5, generating an optimal weeding path; 6, in the advancing process of the robot, the path is dynamically adjusted to avoid crops; 7, according to the weed density and type, the robot executes a corresponding weeding strategy; 8, optimizing a weeding strategy and a path planning algorithm; and 9, after the weeding task is completed, the robot returns to the initial position or the charging station along the shortest path to prepare for next operation. The weeding efficiency and accuracy are improved, unnecessary repeated weeding and herbicide waste are reduced, and the weeding effect of the whole farmland is optimized.

Description

Technical Field

[0001] The present invention relates to the technical field of robot weeding path planning, and in particular to a weeding path planning method for an agricultural weeding robot. Background Art

[0002] With the rapid development of modern agriculture, farmland weeding is an important part of the crop growth process, and its efficiency and accuracy directly affect the yield and quality of crops. Traditional weeding methods mainly rely on manual or mechanical spraying of herbicides, but this method has many disadvantages. First, manual weeding is labor-intensive and inefficient, and it is difficult to meet the weeding needs of large-scale farmland; second, although mechanical spraying of herbicides improves efficiency, it often has problems such as excessive use, accidental injury to non-target areas, and environmental pollution.

[0003] In order to overcome the shortcomings of these traditional weeding methods, agricultural robot technology came into being and gradually became a research hotspot in the field of modern agriculture. Among them, agricultural weeding robots have shown great potential in farmland weeding operations with their high efficiency, precision and environmental protection. However, to achieve efficient operation of weeding robots, their path planning problem must be solved. Path planning not only affects the energy consumption and operation time of weeding robots, but is also directly related to weeding effects and operating costs.

[0004] Existing path planning methods for agricultural weeding robots are mostly based on simple geometric algorithms or random search algorithms. These methods often find it difficult to generate optimal paths in complex farmland environments, resulting in low weeding efficiency, high energy consumption, and easy accidental damage to non-target areas. Summary of the invention

[0005] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a weeding path planning method for an agricultural weeding robot, comprising the following steps:

[0007] Step 1: Start the lawn mower robot and check if all sensors and cameras are working properly;

[0008] Step 2: Use LiDAR and ultrasonic sensors to conduct a preliminary scan of the farmland environment and build a basic 3D map of the farmland;

[0009] Step 3: Use the multispectral camera and sensor on the robot to collect the spectral characteristics and density information of ground weeds and generate a real-time weed distribution map;

[0010] Step 4: Use a high-resolution camera to capture images of the farmland to identify the types and forms of weeds, classify the identified weeds by type, size, and form, and mark them on the farmland map;

[0011] Step 5: Based on the generated weed distribution map and classification information, the dynamic programming algorithm is used to generate the optimal weeding path, and the weeding robot tasks are intelligently assigned according to the weed density and type in different areas;

[0012] Step 6: As the robot moves, it continues to use sensors and machine vision systems to monitor the status of weeds and crops in real time and dynamically adjusts its path to avoid crops.

[0013] Step 7: During the robot's movement, the robot executes corresponding weeding strategies according to the density and type of weeds;

[0014] Step 8: During the weeding process, the weeding path, herbicide usage, and weeding effect are recorded in real time. The weeding effect and herbicide usage efficiency are evaluated based on these data, and the weeding strategy and path planning algorithm are optimized to provide an improvement basis for the next weeding operation;

[0015] Step 9: After the weeding task is completed, the robot returns to the initial position or charging station along the shortest path to prepare for the next operation.

[0016] As a preferred solution of the weeding path planning method for the agricultural weeding robot described in the present invention, in step five, when generating the optimal weeding path, weed density, weed type, farmland topography, and the power and operating capacity of the weeding robot itself should be considered.

[0017] As a preferred solution of the weeding path planning method of the agricultural weeding robot of the present invention, the formula for generating the optimal weeding path is as follows:

[0018]

[0019] in:

[0020] P opt is the optimal path, which means the path that saves the most energy and has the best effect is selected by the weeding robot after optimizing various parameters;

[0021] Ω is the integral range, which indicates the area where the weeding robot can work, that is, the range of the farmland;

[0022] D(x,y) is the weed distribution density function, which represents the weed density at the coordinate (x, y);

[0023] T(x, y) is the weed type function, which represents the weed type at the coordinate (x, y);

[0024] E(x, y) is the energy consumption function, which represents the energy required by the robot to reach the coordinates (x, y);

[0025] S(x, y) is the real-time perception information, which represents the environmental information perceived by the robot at the coordinate (x, y);

[0026] α is a normalization coefficient, which is used to adjust the impact of real-time perception information on path selection. By adjusting the value of α, the degree of optimization of the weeding path by the perception information can be controlled;

[0027] exp(-S(x, y)) is an exponential function, which is used to represent the inhibitory effect of real-time perception information on path selection. When S(x, y) is large, the value of the exponential function approaches 0, which inhibits path selection. Otherwise, the path selection priority is higher.

[0028] dx dy represents the double integral of x and y, covering the entire farmland area Ω, and the comprehensive path optimization effect of the entire farmland is calculated by integration;

[0029] D(x, y)·T(x, y) represents the weed control priority under the influence of weed density D(x, y) and weed type T(x, y), that is, areas with higher weed density and more complex weed types will have higher priority;

[0030] E(x, y)+αexp(-S(x, y)) means that the energy consumption E(x, y) and the impact of real-time perception information αexp(-S(x, y)) need to be considered comprehensively when selecting a path. The addition of these two factors can be used to balance the security and energy efficiency of path selection.

[0031] Value range meaning:

[0032] P opt The value range is all path combinations that the robot can reach.

[0033] As a preferred solution of the weeding path planning method for the agricultural weeding robot of the present invention, obtaining the optimal weeding path includes the following steps:

[0034] Step 51: Data initialization and collection

[0035] Step 511: farmland modeling, i.e., defining the farmland region Ω as a region in a two-dimensional rectangular coordinate system, with a range of [0, L]×[0, W], where L and W are the length and width of the farmland respectively;

[0036] Step 512: Place the weeding robot at a certain starting position (x0 ,y 0 );

[0037] Step 513: The robot uses sensors to scan the entire farmland area and collects weed density D(x, y), weed type T(x, y), energy consumption E(x, y) and environmental perception information S(x, y);

[0038] Step 52: Construction of objective function and generation of initial path

[0039] Step 521: Define the objective function:

[0040]

[0041] This function is used to evaluate the pros and cons of each path, with the influence of weed density and type as the numerator and the combined influence of energy consumption and environmental perception factors as the denominator, representing the ratio of the path's weed control priority to energy consumption;

[0042] Step 522: Generate multiple initial paths P using random generation or based on a greedy algorithm i ,These paths constitute the initial solution population of the optimization algorithm;

[0043] Step 53: Path Optimization Process

[0044] Step 531: For each candidate path P i , according to the objective function f(P i ) for assessment;

[0045] Step 532: Based on the calculated objective function value f(P i ), select the path with a lower objective function value as the parent of the next generation population, and then generate a new path combination through crossover operation to simulate the gene recombination in natural selection and form a new candidate path;

[0046] Step 533: Randomly perform mutation operations on the newly generated paths to increase the diversity of the population, and then update the paths of the population, that is, replace the paths with lower adaptability with these mutated paths to form a new path set;

[0047] Step 54: Determination of the optimal path

[0048] Step 541: Determine whether the path optimization algorithm has reached the convergence condition. If the convergence condition is met, the algorithm terminates and selects the path P with the smallest objective function value in the current population. opt As the optimal path;

[0049] Step 542: The optimal path P opt As the path executed by the weeding robot, it is transmitted to the robot control system, and the robot follows the optimal path Popt The robot moves forward and performs weeding tasks, while making fine adjustments based on real-time environmental data to ensure the best weeding effect.

[0050] As a preferred solution of the agricultural weeding robot weeding path planning method of the present invention, wherein: for each candidate path P i , according to the objective function f(P i The specific calculation method for evaluation is as follows:

[0051] Calculate the path P i The product of the weed density D(x, y) and the weed type T(x, y) corresponding to each position (x, y) on the graph;

[0052] Calculate the energy consumption E(x, y) required to reach these locations and the impact of environmental perception information S(x, y) on the path;

[0053] Combining these data, the objective function value f(P i ).

[0054] As a preferred solution of the weeding path planning method of the agricultural weeding robot of the present invention, the weeding strategy includes the following contents:

[0055] In the case of high weed density, a high-intensity weeding strategy is adopted. The robot will choose to perform mechanical weeding and spraying herbicides simultaneously in the area with high weed density D(x, y) to ensure that the weeds are completely removed;

[0056] In the case of low weed density, a strategy is adopted. For areas with low weed density D(x, y), the robot only performs mechanical weeding operations to avoid unnecessary use of herbicides, thereby saving resources and reducing the impact on the environment;

[0057] When encountering weed type A, the robot will adopt a fast passing strategy. In path planning, the robot will prioritize areas with weed type A and pass through them quickly to reduce energy consumption while maintaining a high operating speed.

[0058] In the case of weed type B, a moderate treatment strategy is adopted, where the robot will slow down the operation speed during path planning, ensuring that mechanical weeding is combined with the appropriate amount of herbicide spraying to improve the weed control effect;

[0059] In the case of weed type C, the heavy treatment strategy is adopted. The robot will prioritize and stay in these areas in path planning, perform multiple mechanical weeding and high-concentration herbicide spraying to ensure that these difficult weeds are completely removed;

[0060] When obstacles are detected in the farmland, the robot will adopt the obstacle avoidance priority strategy, dynamically adjust the path to avoid obstacles and ensure the safety of the operation, and select the appropriate weeding strategy according to the position after the path adjustment;

[0061] When energy reserves are low, an energy-saving weeding strategy is adopted. The robot will prioritize areas with lower energy consumption in path planning and minimize the use of herbicides to extend the operation time.

[0062] In the case of complex farmland terrain and changing environment, a real-time adaptation strategy is adopted. The robot will dynamically adjust the weeding path and strategy based on the real-time perceived terrain changes and environmental information to ensure the continuity and effectiveness of the operation.

[0063] As a preferred solution of the weeding path planning method of the agricultural weeding robot described in the present invention, the weeding strategy also includes an adaptive learning algorithm, which can continuously adjust and optimize the working parameters of the weeding equipment and the formula of the herbicide according to the effect of each weeding operation, so as to improve the weeding efficiency and accuracy of the next operation.

[0064] As a preferred solution of the weeding path planning method for the agricultural weeding robot described in the present invention, after the weeding task is completed, the robot needs to be cleaned, maintained and the sensor calibrated regularly to ensure the normal operation of the equipment and the weeding accuracy.

[0065] Beneficial effects of the present invention:

[0066] 1. The present invention uses high-precision sensors such as laser radar, ultrasonic sensors, and multispectral cameras to enable the robot to construct a three-dimensional map of the farmland in real time and identify the distribution, type, and density of weeds. Based on this information, a dynamic programming algorithm is used to generate the optimal weeding path, so that the weeding robot can select the optimal path with the lowest energy consumption, which not only improves the weeding efficiency and accuracy, but also reduces unnecessary repeated weeding and the waste of herbicides, and optimizes the overall weeding effect of the farmland.

[0067] 2. According to different weed densities and types, the robot can intelligently select weeding strategies, such as full coverage, selectivity, concentration, zoning, precision and environmental adaptation strategies, effectively reducing accidental damage to non-target areas and improving the accuracy and efficiency of weeding.

[0068] 3. Through accurate weed identification and distribution mapping, the present invention enables the robot to apply herbicides only where needed, avoiding environmental pollution and increased costs caused by the large-scale use of herbicides in traditional weed control methods. This not only reduces farmers' input costs, but also conforms to the green and environmentally friendly concept of modern agriculture. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0070] Figure 1 The present invention is a flowchart of a weeding path planning method for an agricultural weeding robot. DETAILED DESCRIPTION

[0071] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0072] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0073] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0074] Secondly, the present invention is described in detail with reference to the schematic diagram. When describing the embodiments of the present invention in detail, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0075] Example 1

[0076] Reference Figure 1 , which is the first embodiment of the present invention, provides a weeding path planning method for an agricultural weeding robot, comprising the following steps:

[0077] Step 1: Start the weeding robot and check whether all sensors and cameras are working properly. Sensors include multispectral cameras, lidar, ultrasonic sensors, etc.

[0078] Step 2: Use lidar and ultrasonic sensors to conduct a preliminary scan of the farmland environment, build a basic three-dimensional map of the farmland, and identify the general farmland boundaries, crop rows, and preliminary obstacle information.

[0079] Step 3: Use the multispectral camera and sensor on the robot to collect the spectral characteristics and density information of ground weeds and generate a real-time weed distribution map. Use the sensor data analysis algorithm to process the collected data and generate a weed density heat map in different areas of the farmland to determine high-density weed areas and low-density weed areas.

[0080] Step 4: Use a high-resolution camera to capture images of farmland, and use deep learning algorithms (such as convolutional neural networks) to process the images to identify weed species and forms, classify the identified weeds by type, size, and form, and mark them on the farmland map. Based on the characteristics of different weeds, this provides a reference for subsequent path planning and weed control strategies.

[0081] Step 5: Based on the generated weed distribution map and classification information, use the dynamic programming algorithm to generate the optimal weeding path. The path planning should take into account the weed density, weed type, farmland topography, and the power and operating capacity of the weeding robot itself. Intelligently assign weeding robot tasks according to the weed density and type in different areas.

[0082] Step 6: As the robot moves, it continues to use sensors and machine vision systems to monitor the status of weeds and crops in real time, dynamically adjusting the path to avoid crops and further improving weeding accuracy.

[0083] Step 7: During the robot's movement, the robot executes corresponding weeding strategies according to the density and type of weeds.

[0084] Step 8: During the weeding process, the weeding path, herbicide usage, and weeding effect are recorded in real time. Based on these data, the weeding effect and herbicide usage efficiency are evaluated, and the weeding strategy and path planning algorithm are optimized to provide a basis for improvement for the next weeding operation.

[0085] Step 9: After the weeding task is completed, the robot returns to the initial position or charging station along the shortest path to prepare for the next operation. After the weeding task is completed, the robot needs to be cleaned, maintained and the sensor calibrated regularly to ensure the normal operation of the equipment and the weeding accuracy.

[0086] Specifically, the formula for generating the optimal weeding path is as follows:

[0087]

[0088] in:

[0089] P opt is the optimal path, which means the path that saves the most energy and has the best effect is selected by the weeding robot after optimizing various parameters;

[0090] Ω is the integral range, which indicates the area where the weeding robot can work, that is, the range of the farmland;

[0091] D(x,y) is the weed distribution density function, which represents the weed density at the coordinate (x, y);

[0092] T(x, y) is the weed type function, which represents the weed type at the coordinate (x, y);

[0093] E(x, y) is the energy consumption function, which represents the energy required by the robot to reach the coordinates (x, y);

[0094] S(x, y) is the real-time perception information, which represents the environmental information perceived by the robot at the coordinate (x, y);

[0095] α is a normalization coefficient, which is used to adjust the impact of real-time perception information on path selection. By adjusting the value of α, the degree of optimization of the weeding path by the perception information can be controlled;

[0096] exp(-S(x, y)) is an exponential function, which is used to represent the inhibitory effect of real-time perception information on path selection. When S(x, y) is large, the value of the exponential function approaches 0, which inhibits path selection. Otherwise, the path selection priority is higher.

[0097] dx dy represents the double integral of x and y, covering the entire farmland area Ω, and the comprehensive path optimization effect of the entire farmland is calculated by integration;

[0098] D(x, y)·T(x, y) represents the weed control priority under the influence of weed density D(x, y) and weed type T(x, y), that is, areas with higher weed density and more complex weed types will have higher priority;

[0099] E(x, y)+αexp(-S(x, y)) means that the energy consumption E(x, y) and the impact of real-time perception information αexp(-S(x, y)) need to be considered comprehensively when selecting a path. The addition of these two factors can be used to balance the security and energy efficiency of path selection.

[0100] Value range meaning:

[0101] P opt The value range is all path combinations that the robot can reach.

[0102] The above formula defines an optimization goal for path selection by integrating multiple factors such as weed density, weed type, energy consumption, and real-time perception information. The optimal path is the path that minimizes the value of this objective function among all possible paths, thereby achieving efficient, energy-saving, and intelligent weed control.

[0103] Furthermore, obtaining the optimal weeding path includes the following steps:

[0104] Step 51: Data initialization and collection

[0105] Step 511: farmland modeling, i.e., defining the farmland region Ω as a region in a two-dimensional rectangular coordinate system, with a range of [0, L]×[0, W], where L and W are the length and width of the farmland respectively;

[0106] Step 512: Place the weeding robot at a certain starting position (x 0 ,y 0 );

[0107] Step 513: The robot uses sensors to scan the entire farmland area and collects weed density D(x, y), weed type T(x, y), energy consumption E(x, y) and environmental perception information S(x, y) data;

[0108] Step 52: Construction of objective function and generation of initial path

[0109] Step 521: Define the objective function:

[0110]

[0111] This function is used to evaluate the pros and cons of each path, with the influence of weed density and type as the numerator and the combined influence of energy consumption and environmental perception factors as the denominator, representing the ratio of the path's weed control priority to energy consumption;

[0112] Step 522: Generate multiple initial paths P using random generation or based on a greedy algorithm i ,These paths constitute the initial solution population of the optimization algorithm;

[0113] Step 53: Path Optimization Process

[0114] Step 531: For each candidate path P i , according to the objective function f(P i ) is evaluated, and the specific calculation method is as follows:

[0115] Calculate the path P i The product of the weed density D(x, y) and the weed type T(x, y) corresponding to each position (x, y) on the graph;

[0116] Calculate the energy consumption E(x, y) required to reach these locations and the impact of environmental perception information S(x, y) on the path;

[0117] Combining these data, the objective function value f(P i );

[0118] Step 532: Based on the calculated objective function value f(P i ), select the path with a lower objective function value as the parent of the next generation population, and then generate a new path combination through crossover operation to simulate the gene recombination in natural selection and form a new candidate path;

[0119] Step 533: Randomly perform mutation operations on the newly generated paths to increase the diversity of the population, and then update the paths of the population, that is, replace the paths with lower adaptability with these mutated paths to form a new path set;

[0120] Step 54: Determination of the optimal path

[0121] Step 541: Determine whether the path optimization algorithm has reached the convergence condition. If the convergence condition is met, the algorithm terminates and selects the path P with the smallest objective function value in the current population. opt As the optimal path;

[0122] Step 542: The optimal path P opt As the path executed by the weeding robot, it is transmitted to the robot control system, and the robot follows the optimal path P opt The robot moves forward and performs weeding tasks, while making fine adjustments based on real-time environmental data to ensure the best weeding effect.

[0123] Specifically, the weed control strategy includes the following contents:

[0124] In the case of high weed density, a high-intensity weeding strategy is adopted. The robot will choose to perform mechanical weeding and spraying herbicides simultaneously in areas with high weed density D(x, y) (e.g. D(x, y)>0.7) to ensure that the weeds are completely removed;

[0125] In the case of low weed density, a strategy is adopted. For areas with low weed density D(x, y) (e.g. D(x, y) < 0.4), the robot only performs mechanical weeding operations to avoid unnecessary use of herbicides, thereby saving resources and reducing the impact on the environment;

[0126] When encountering weed type A (easy to handle), the robot will adopt a fast passing strategy. In path planning, the robot will prioritize areas with weed type A and pass through them quickly to reduce energy consumption while maintaining a high operating speed.

[0127] In the case of weed type B (medium difficulty), a moderate treatment strategy is adopted, where the robot will slow down the operation speed during path planning to ensure that mechanical weeding is combined with the appropriate amount of herbicide spraying to improve the weeding effect;

[0128] In the case of weed type C (difficult to treat), the heavy treatment strategy is adopted. The robot will prioritize and stay in these areas in path planning, perform multiple mechanical weeding and high-concentration herbicide spraying to ensure that these difficult weeds are completely removed;

[0129] When obstacles are detected in the farmland, the robot will adopt the obstacle avoidance priority strategy, dynamically adjust the path to avoid obstacles and ensure the safety of the operation, and select the appropriate weeding strategy according to the position after the path adjustment;

[0130] When energy reserves are low, an energy-saving weeding strategy is adopted. The robot will prioritize areas with lower energy consumption in path planning and minimize the use of herbicides to extend the operation time.

[0131] In the case of complex farmland terrain and changing environment, a real-time adaptation strategy is adopted. The robot will dynamically adjust the weeding path and strategy based on the real-time perceived terrain changes and environmental information to ensure the continuity and effectiveness of the operation.

[0132] The weeding strategy also includes an adaptive learning algorithm that can continuously adjust and optimize the working parameters of the weeding equipment and the formula of the herbicide according to the effect of each weeding operation to improve the weeding efficiency and accuracy of the next operation.

[0133] In summary, the present invention uses high-precision sensors such as laser radar, ultrasonic sensors, and multispectral cameras, so that the robot can construct a three-dimensional map of the farmland in real time and identify the distribution, type, and density of weeds. Based on this information, the dynamic programming algorithm is used to generate the optimal weeding path, so that the weeding robot can select the optimal path with the lowest energy consumption, which not only improves the weeding efficiency and accuracy, but also reduces unnecessary repeated weeding and waste of herbicides, and optimizes the overall weeding effect of the farmland. According to different weed densities and types, the robot can intelligently select weeding strategies, such as full coverage, selectivity, concentration, zoning, precision, and environmental adaptation strategies, which effectively reduces accidental damage to non-target areas and improves the accuracy and efficiency of weeding. Through accurate weed identification and distribution maps, the present invention enables the robot to apply herbicides only where needed, avoiding environmental pollution and cost increases caused by the large-scale use of herbicides in traditional weeding methods. This can not only reduce farmers' input costs, but also conform to the green and environmentally friendly concepts of modern agriculture.

[0134] Example 2

[0135] This is the second embodiment of the present invention, which is a specific embodiment and shows how to calculate and obtain the optimal weeding path through actual data.

[0136] Farmland Background:

[0137] Assume that a rectangular farmland has an area of ​​10×10 meters and is divided into 100 1×1 meter grids. The weeding robot needs to generate the optimal weeding path in the farmland, mainly considering the following factors:

[0138] I. Weed distribution density D(x, y): Assume that weed density data is obtained through sensor detection. The range of weed density is between [0, 1], where 0 indicates no weeds and 1 indicates the densest weeds.

[0139] II. Weed type T(x, y): Assume there are three types of weeds, each with a different numerical weight, for example:

[0140] Type A: Weight 1.0 (easy to handle)

[0141] Type B: Weight 1.5 (Medium Difficulty)

[0142] Type C: Weight 2.0 (difficult to handle)

[0143] III. Energy consumption E(x, y): Energy consumption varies with distance. It is assumed that the energy consumption is the distance the robot moves on the path, and the unit is 1 energy unit per meter.

[0144] IV. Real-time perception information S(x, y): Assuming that the robot can sense obstacles and terrain changes in the farmland, the grid S(x, y) value where the obstacle is located is 1, and 0 elsewhere.

[0145] V. Normalization coefficient α: Assume α = 0.5, which is used to adjust the influence of perceptual information.

[0146] Data example:

[0147] Assume that the specific data of the following grid points in the farmland are:

[0148]

[0149]

[0150] Calculate the optimal path:

[0151] Using the formula calculation, the selection of the optimal path needs to minimize the following integral expression:

[0152]

[0153] For the data in the table, calculate the objective function value for each grid point:

[0154] 1. For point (1, 1):

[0155]

[0156] 2. For point (2, 1):

[0157]

[0158] 3. For point (3, 1) (where there is an obstacle):

[0159]

[0160] 4. Regarding point (4, 1):

[0161]

[0162] 5. Regarding point (5, 1):

[0163]

[0164] By comparing these objective values, the path with the smallest objective value can be selected:

[0165] Since there is an obstacle at point (3, 1) and the calculated target value is the lowest 0.433, it means that it is part of the better path.

[0166] According to the target values ​​of other points, the robot can choose to weed from (1, 1) → (3, 1) → (5, 1), which can avoid obstacles and minimize the overall target value.

[0167] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A weeding path planning method for an agricultural weeding robot, characterized in that: The following steps are involved: Step 1: Start the lawn mower robot and check if all sensors and cameras are working properly; Step 2: Use LiDAR and ultrasonic sensors to conduct a preliminary scan of the farmland environment and build a basic 3D map of the farmland; Step 3: Use the multispectral camera and sensor on the robot to collect the spectral characteristics and density information of ground weeds and generate a real-time weed distribution map; Step 4: Use a high-resolution camera to capture images of farmland to identify weed species and forms, classify the identified weeds by type, size, and form, and mark them on the farmland map; Step 5: Based on the generated weed distribution map and classification information, the dynamic programming algorithm is used to generate the optimal weeding path, and the weeding robot tasks are intelligently assigned according to the weed density and type in different areas; Step 6: As the robot moves, it continues to use sensors and machine vision systems to monitor the status of weeds and crops in real time and dynamically adjusts its path to avoid crops. Step 7: During the robot's movement, the robot executes corresponding weeding strategies according to the density and type of weeds; Step 8: During the weeding process, the weeding path, herbicide usage, and weeding effect are recorded in real time. The weeding effect and herbicide usage efficiency are evaluated based on these data, and the weeding strategy and path planning algorithm are optimized to provide an improvement basis for the next weeding operation; Step 9: After the weeding task is completed, the robot returns to the initial position or charging station along the shortest path to prepare for the next operation.

2. The weeding path planning method of the agricultural weeding robot according to claim 1, characterized in that: When generating the optimal weeding path in step 5, weed density, weed type, farmland topography, and the power and operating capacity of the weeding robot itself should be considered.

3. The weeding path planning method of the agricultural weeding robot according to claim 1, characterized in that: The formula for generating the optimal weeding path is as follows: in: P opt is the optimal path, which means the path that saves the most energy and has the best effect is selected by the weeding robot after optimizing various parameters; Ω is the integral range, which indicates the area where the weeding robot can work, that is, the range of the farmland; D(x,y) is the weed distribution density function, which represents the weed density at the coordinate (x,y); T(x,y) is the weed type function, which represents the weed type at the coordinate (x,y); E(x,y) is the energy consumption function, which represents the energy required by the robot to reach the coordinate (x,y); S(x,y) is the real-time perception information, which represents the environmental information perceived by the robot at the coordinate (x,y); α is a normalization coefficient, which is used to adjust the impact of real-time perception information on path selection. By adjusting the value of α, the degree of optimization of the weeding path by the perception information can be controlled; exp(-S(x,y)) is an exponential function, which is used to represent the inhibitory effect of real-time perception information on path selection. When S(x,y) is large, the value of the exponential function approaches 0, which inhibits path selection. Otherwise, the path selection priority is higher. dxdy represents the double integral of x and y, covering the entire farmland area Ω, and the comprehensive path optimization effect of the entire farmland is calculated by integration; D(x,y)·T(x,y) represents the weed control priority under the influence of weed density D(x,y) and weed type T(x,y). That is, areas with higher weed density and more complex weed types will have higher priority. E(x,y)+αexp(-S(x,y)) means that when selecting a path, it is necessary to comprehensively consider the energy consumption E(x,y) and the impact of real-time perception information αexp(-S(x,y)). The sum of these two factors can be used to balance the security and energy efficiency of path selection. Value range meaning: P opt The value range is all path combinations that the robot can reach.

4. The weeding path planning method of the agricultural weeding robot according to claim 3, characterized in that: Obtaining the optimal weeding path includes the following steps: Step 51: Data initialization and collection Step 511: farmland modeling, i.e., defining the farmland region Ω as a region in a two-dimensional rectangular coordinate system, representing a range of [0, L] × [0, W], where L and W are the length and width of the farmland respectively; Step 512: placing the weeding robot at a certain starting position (x0, y0) in the farmland; Step 513: The robot uses sensors to scan the entire farmland area and collects weed density D(x,y), weed type T(x,y), energy consumption E(x,y) and environmental perception information S(x,y); Step 52: Construction of objective function and generation of initial path Step 521: Define the objective function: This function is used to evaluate the pros and cons of each path, with the influence of weed density and type as the numerator and the combined influence of energy consumption and environmental perception factors as the denominator, representing the ratio of the path's weed control priority to energy consumption; Step 522: Generate multiple initial paths P using random generation or based on a greedy algorithm i ,These paths constitute the initial solution population of the optimization algorithm; Step 53: Path Optimization Process Step 531: For each candidate path P i , according to the objective function f(P i ) for assessment; Step 532: Based on the calculated objective function value f(P i ), select the path with a lower objective function value as the parent of the next generation population, and then generate a new path combination through crossover operation to simulate the gene recombination in natural selection and form a new candidate path; Step 533: Randomly perform mutation operations on the newly generated paths to increase the diversity of the population, and then update the paths of the population, that is, replace the paths with lower adaptability with these mutated paths to form a new path set; Step 54: Determination of the optimal path Step 541: Determine whether the path optimization algorithm has reached the convergence condition. If the convergence condition is met, the algorithm terminates and selects the path P with the smallest objective function value in the current population. opt As the optimal path; Step 542: The optimal path P opt As the path executed by the weeding robot, it is transmitted to the robot control system, and the robot follows the optimal path P opt The robot moves forward and performs weeding tasks, while making fine adjustments based on real-time environmental data to ensure the best weeding effect.

5. The weeding path planning method of the agricultural weeding robot according to claim 4, characterized in that: For each candidate path P i , according to the objective function f(P i The specific calculation method for evaluation is as follows: Calculate the path P i The product of the weed density D(x,y) and the weed type T(x,y) corresponding to each position (x,y) on the map; Calculate the energy consumption E(x,y) required to reach these locations and the impact of environmental perception information S(x,y) on the path; Combining these data, the objective function value f(P i ).

6. The weeding path planning method of the agricultural weeding robot according to claim 1, characterized in that: The weed control strategy includes the following: In the case of high weed density, a high-intensity weeding strategy is adopted. The robot will choose to perform mechanical weeding and spraying herbicides simultaneously in the area with high weed density D(x,y) to ensure that the weeds are completely removed; In the case of low weed density, a strategy is adopted. For areas with low weed density D(x,y), the robot only performs mechanical weeding operations to avoid unnecessary use of herbicides, thereby saving resources and reducing the impact on the environment; When encountering weed type A, the robot will adopt a fast passing strategy. In path planning, the robot will prioritize areas with weed type A and pass through them quickly to reduce energy consumption while maintaining a high operating speed. In the case of weed type B, a moderate treatment strategy is adopted, where the robot will slow down the operation speed during path planning, ensuring that mechanical weeding is combined with the appropriate amount of herbicide spraying to improve the weed control effect; In the case of weed type C, the heavy treatment strategy is adopted. The robot will prioritize and stay in these areas in path planning, perform multiple mechanical weeding and high-concentration herbicide spraying to ensure that these difficult weeds are completely removed; When obstacles are detected in the farmland, the robot will adopt the obstacle avoidance priority strategy, dynamically adjust the path to avoid obstacles and ensure the safety of the operation, and select the appropriate weeding strategy according to the position after the path adjustment; When energy reserves are low, an energy-saving weeding strategy is adopted. The robot will prioritize areas with lower energy consumption in path planning and minimize the use of herbicides to extend the operation time. In the case of complex farmland terrain and changing environment, a real-time adaptation strategy is adopted. The robot will dynamically adjust the weeding path and strategy based on the real-time perceived terrain changes and environmental information to ensure the continuity and effectiveness of the operation.

7. The weeding path planning method of the agricultural weeding robot according to claim 6, characterized in that: The weeding strategy also includes an adaptive learning algorithm that can continuously adjust and optimize the working parameters of the weeding equipment and the formula of the herbicide according to the effect of each weeding operation to improve the weeding efficiency and accuracy of the next operation.

8. The weeding path planning method of the agricultural weeding robot according to claim 6, characterized in that: After the weeding task is completed, the robot needs to be cleaned, maintained and the sensor calibrated regularly to ensure the normal operation of the equipment and the weeding accuracy.

Images