High-precision mulching film positioning method and system applied to rice weeding mulching film

Abstract

The invention discloses a high-precision mulching film positioning method and system applied to a rice weeding mulching film, and the method comprises the steps: obtaining the basic information of a rice planting region and the width information of the weeding mulching film, determining a full-coverage path of the weeding mulching film, and representing the full-coverage path through a grating method; obtaining UWB positioning data and LiDAR positioning data in the weeding and plastic film covering process of the rice planting area, and fusing the UWB positioning data and the LiDAR positioning data to generate fused positioning data; segmenting the positioning sample according to the gridding processed full-coverage path, constructing a normal film mulching positioning model by using a multivariate state estimation method, importing the fusion positioning data into the normal film mulching positioning model, comparing the calculated residual error with the reference residual error, judging the coverage deviation in the weeding film mulching process, and adjusting the coverage position. According to the method, the two positioning methods are fused for positioning, the precision in the weeding plastic film covering process is improved, the covering deviation is evaluated based on the positioning result, and the rice field film covering quality is guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of smart agricultural technology, and more specifically, to a high-precision film covering positioning method and system for rice weeding film. Background Art

[0002] Rice weed control mulch is a fully biodegradable mulch that can effectively inhibit the growth of weeds and reduce the use of herbicides. It also has the functions of water retention and drought resistance. It can be divided into two types: mulch containing herbicides and mulch containing sunscreens. This type of mulch is usually made of biological materials such as crop straw and can be naturally degraded within a few months. It will not pollute the soil and can be used as an organic fertilizer. By covering the rice fields, the fully biodegradable mulch can inhibit the growth of more than 90% of weeds without the use of herbicides, thereby reducing the input of chemical substances and protecting the ecological environment. In addition, this mulch can also reduce water evaporation, save irrigation water, and increase the yield and quality of rice.

[0003] The occurrence of barnyard grass in rice seedling fields is large, posing a serious threat to rice seedling cultivation. Therefore, the application of weed control film based on rice seedling trays is an effective way to control weeds in centralized seedling fields. The use of fully biodegradable mulch film can not only suppress weeds in rice fields, but also increase the surface temperature and promote the early growth of rice. However, rice mulching technology can only be done by manual or semi-mechanized operations, which is not only slow and labor-intensive, but also has poor mulching accuracy. It is impossible to ensure high-quality coverage of the weed control film, and the mulch film cannot be evenly attached to the surface of the mud field. Therefore, the credibility and effectiveness of the traceability system are reduced. Therefore, how to achieve high-precision mulching positioning in the automatic mulching scenario to improve the mulching quality is a problem that needs to be solved. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a high-precision film covering positioning method and system for rice weeding film, which aims to achieve high-precision covering operation of rice weeding film and improve the quality and efficiency of rice field film covering.

[0005] A first aspect of the present invention provides a high-precision film covering positioning method for rice weed control film, comprising:

[0006] Obtain basic information of the current rice planting area and weeding film width information, obtain a full coverage path of the weeding film based on the two-dimensional map of the rice planting area and the weeding film width information, and represent the full coverage path;

[0007] Performing weed control and film covering in the rice planting area based on the full coverage path, acquiring UWB positioning data and LiDAR positioning data during the covering process, and fusing the acquired positioning data to generate fused positioning data;

[0008] Segment the positioning samples according to the full coverage path after gridding, build a normal coverage positioning model using a multivariate state estimation method based on the training data corresponding to the positioning samples, and obtain the benchmark residual corresponding to the model;

[0009] The fused positioning data is imported into the normal film covering positioning model, the estimated vector is read to calculate the residual, the calculated residual is compared with the reference residual, the coverage deviation in the weeding film covering process is determined, positioning feedback is performed according to the coverage deviation, and the coverage position is adjusted.

[0010] In this solution, the basic information of the current rice planting area and the width information of the weeding film are obtained, and the full coverage path of the weeding film is obtained based on the two-dimensional map of the rice planting area and the width information of the weeding film. Specifically:

[0011] Acquire the size information and geographical location information of the current rice planting area as basic information, extract the high-resolution remote sensing image of the area closest to the current time, and pre-process the high-resolution image;

[0012] Using basic information of the rice-growing area to crop the high-resolution remote sensing image, constructing a two-dimensional map of the rice-growing area based on the cropped regional remote sensing image, obtaining width information of the weed-removing film, and determining grid side length data based on the width information;

[0013] The two-dimensional map is divided into grids, and the total number of grids required to be covered by the path for full coverage of the weeding film is determined, the coverage direction constraint of the weeding film is determined, the grids of the two-dimensional map are encoded, the number of covered grids in the rice planting area is obtained, and the path coverage rate is calculated;

[0014] When the coverage rate reaches the preset threshold, the full coverage path is output according to the grid map matrix representation, and the endpoints of the straight line segments are connected at the U-turn to output the grid-represented full coverage path of the weed removal film.

[0015] In this solution, UWB positioning data and LiDAR positioning data are obtained during the coverage process, and the acquired positioning data are fused to generate fused positioning data, specifically:

[0016] The edge positioning measurement values ​​of the weeding and mulching process are obtained by using UWB and LiDAR sensors, and the measurement values ​​of the two heterogeneous sensors are unified in time stamp and converted in spatial coordinate system to achieve time synchronization and space synchronization;

[0017] The UWB measurement value is calculated by the signal two-way propagation time method to generate UWB positioning data, the LiDAR inter-frame point cloud is used for feature matching, the LiDAR posture information is obtained to generate LiDAR positioning data, and the state equation and measurement equation are constructed according to the UWB positioning data and the LiDAR positioning data;

[0018] The EKF filter is used to fuse the UWB positioning data and the LiDAR positioning data, and the fused positioning data is obtained according to the updated solution of the state equation and the measurement equation.

[0019] In this solution, the samples are segmented and located according to the full coverage path after gridding, specifically:

[0020] The full coverage path is divided according to the grid, the sub-path in each grid block is read, the morphological characteristics of the sub-path are analyzed, and the positioning sample is constructed according to the sub-path and its morphological characteristics;

[0021] The positioning samples are used to retrieve rice film covering examples, and rice film covering examples with satisfactory film covering accuracy are screened. The planned film covering edge position sequence and the actual film covering edge position sequence in the screened rice film covering examples are read, and the film covering edge position sequence and the actual film covering edge position sequence corresponding to each positioning sample are aggregated to generate a training data set.

[0022] In this solution, a normal film positioning model is constructed based on the training data corresponding to the positioning sample using a multivariate state estimation method, and the reference residual corresponding to the model is obtained, specifically:

[0023] Obtain a training data set corresponding to the positioning sample, sort the training data set according to the sampling time, use Mahalanobis distance as a metric function, expand the mRMR algorithm through the metric function, and screen the offline training data based on the principle that the distance between the training data and the positioning sample is the minimum and the distance between the training data is the maximum;

[0024] A normal film positioning model is constructed using a multivariate state estimation method, a memory matrix is ​​constructed using the screened offline training data, and an estimation vector of an observation vector is calculated using the memory matrix, where the observation vector is generated by a positioning sample of an input model;

[0025] The residuals connecting the observation vector and the estimation are analyzed, and the baseline residuals of the normal film positioning model are obtained according to the validation set corresponding to the training data.

[0026] In this solution, the fused positioning data is imported into the normal film positioning model, and the estimated vector is read to calculate the residual, specifically:

[0027] Importing the fused positioning data into the normal film positioning model, matching and obtaining the current positioning sample, constructing a memory matrix based on the positioning sample, and calculating the estimated vector of the current positioning sample using the constructed memory matrix;

[0028] The residual is obtained based on the observation vector and estimated vector of the current positioning sample, and the obtained residual is compared with the benchmark residual. When the residual difference is greater than the preset difference threshold, a coverage accuracy abnormality warning is issued.

[0029] In this solution, the coverage deviation during the weeding film covering process is determined, positioning feedback is performed based on the coverage deviation, and the coverage position is adjusted. Specifically:

[0030] Obtain the residual difference value obtained by the normal film covering positioning model as the covering deviation in the weeding film covering process, extract the covering speed of the weeding film, and make a coverage deviation prediction value for the next timestamp based on the covering speed and the covering deviation;

[0031] Positioning feedback information is generated based on the coverage deviation prediction value, and the coverage accuracy of the weeding film is compensated based on the positioning feedback information.

[0032] The second aspect of the present invention provides a high-precision film covering positioning system for rice weeding film, including a full coverage path planning module, a weeding film covering positioning module, a coverage accuracy abnormality recognition module, an abnormality alarm module and a film covering accuracy compensation module;

[0033] The full coverage path planning module is responsible for obtaining the full coverage path of the weed film based on the two-dimensional map of the rice planting area and the weed film width information;

[0034] The weeding film covering positioning module is responsible for acquiring UWB positioning data and LiDAR positioning data during the covering process, and fusing the acquired positioning data to generate fused positioning data;

[0035] The coverage accuracy anomaly recognition module is responsible for using the multivariate state estimation method to perform data-driven modeling, construct a normal film covering positioning model, import the fused positioning data into the normal film covering positioning model, and determine the coverage deviation during the weeding film covering process;

[0036] The coverage accuracy abnormality recognition module is responsible for generating a lamination position abnormality warning. When the coverage deviation is greater than a preset deviation threshold, a lamination position abnormality warning will be generated.

[0037] The film covering accuracy compensation module is responsible for generating film covering position compensation data according to covering deviation, using the film covering position compensation data for positioning feedback, and adjusting the subsequent covering position of the weeding film.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] The present invention utilizes two positioning methods for fusion positioning, which solves the problem that a single sensor is difficult to obtain high-precision positioning in a complex environment, improves the accuracy in the weeding film covering process, evaluates the covering deviation based on the positioning results, ensures the uniformity and quality of the rice field film covering, reduces the film covering repetition rate and avoids film waste, and greatly improves the utilization rate of the weeding film. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments or exemplary embodiments of the present invention, the drawings required for use in the embodiments or exemplary descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained according to the drawings without paying creative work.

[0041] Figure 1 A flow chart showing a high-precision film covering positioning method for rice weed control film is shown;

[0042] Figure 2 A flow chart showing generation of fused positioning data according to an embodiment is shown;

[0043] Figure 3 A flowchart of an embodiment of the invention is shown for constructing a normal film positioning model to identify abnormal coverage accuracy;

[0044] Figure 4 The block diagram of a high-precision film covering positioning system for rice weed control is shown. DETAILED DESCRIPTION

[0045] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0046] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0047] Figure 1 The present invention shows a flow chart of a high-precision film covering positioning method applied to rice weed control film.

[0048] like Figure 1 As shown, this embodiment provides a high-precision film covering positioning method for rice weeding film, including:

[0049] S102, obtaining basic information of the current rice planting area and weeding film width information, obtaining a full coverage path of the weeding film based on the two-dimensional map of the rice planting area and the weeding film width information, and representing the full coverage path;

[0050] S104, performing weed control and film covering in the rice planting area based on the full coverage path, acquiring UWB positioning data and LiDAR positioning data during the covering process, and fusing the acquired positioning data to generate fused positioning data;

[0051] S106, segmenting and positioning samples according to the full coverage path after gridding, constructing a normal film coverage positioning model using a multivariate state estimation method based on training data corresponding to the positioning samples, and obtaining a reference residual corresponding to the model;

[0052] S108, importing the fused positioning data into a normal film covering positioning model, reading the estimated vector to calculate the residual, comparing the calculated residual with the reference residual, determining the coverage deviation during the weeding film covering process, performing positioning feedback according to the coverage deviation, and adjusting the coverage position.

[0053] It should be noted that the size information and geographical location information of the current rice planting area are obtained as basic information, and the satellite data source is selected to extract the high-resolution remote sensing image closest to the current time in the area, and the high-resolution image is preprocessed, including orthophoto processing, image fusion, image mosaic, etc., and the original image is appropriately enhanced to improve the image contrast and clarity. The latitude and longitude data are read using the basic information of the rice planting area to crop the high-resolution remote sensing image, and a two-dimensional map of the rice planting area is constructed according to the cropped regional remote sensing image, and the width information of the weeding film is obtained. The grid side length data and the coverage diameter of the grid side length data during the film covering process are determined by the width information. The two-dimensional map is grid-divided, and the total number of grids required to cover the path of full coverage of weeding film is determined, the coverage direction constraint of the weeding film is determined, and the information such as covered, uncovered and obstacles is defined. The grid of the two-dimensional map is encoded, and the number of covered grids in the rice planting area is obtained, and the path coverage rate is calculated. The path coverage rate is defined as the ratio of the final number of covered grids to the target number of covered grids. When the coverage rate reaches the preset threshold, the full coverage path is output according to the grid map matrix representation, and the endpoints of the straight line segments are connected at the U-turn to output the grid-represented full coverage path of the weed removal film.

[0054] Figure 2 A flow chart of generating fused positioning data according to an embodiment is shown.

[0055] According to an embodiment of the present invention, UWB positioning data and LiDAR positioning data are obtained during the coverage process, and the obtained positioning data are fused to generate fused positioning data, specifically:

[0056] S202, obtaining edge positioning measurement values ​​in the weeding film covering process according to the UWB and LiDAR sensors, unifying the time stamps of the measurement values ​​of the two heterogeneous sensors and converting the spatial coordinate systems to achieve time synchronization and space synchronization;

[0057] S204, calculating the UWB measurement value by a two-way signal propagation time method to generate UWB positioning data, performing feature matching using the LiDAR inter-frame point cloud, obtaining the LiDAR posture information to generate LiDAR positioning data, and constructing a state equation and a measurement equation based on the UWB positioning data and the LiDAR positioning data;

[0058] S206, using an EKF filter to fuse the UWB positioning data and the LiDAR positioning data, and obtaining fused positioning data based on the updated solution of the state equation and the measurement equation.

[0059] It should be noted that the UWB and LiDAR sensors are installed on both sides of the weeding mulch covering device, and the distance between them is the width of the weeding mulch. Ultra-wideband positioning technology (UWB) is a short-range wireless positioning technology that uses pulse signals within the ultra-wideband frequency range for data transmission. It has a wide spectrum range, strong anti-interference ability and security. The signal two-way propagation time method indirectly completes the distance measurement by measuring the two-way flight time of the UWB pulse signal between the base station and the mobile station, and calculates the UWB positioning data using methods such as three-sided positioning method and least squares method. LiDAR detects the distance to the surrounding obstacles by emitting lasers. It has extremely high resolution and can provide rich environmental perception information. The point cloud data of the rice planting area is pre-processed as a reference point cloud library, and the continuously scanned laser point cloud is converted to the same coordinate system for point cloud registration. The point cloud features are used to perform feature matching in the reference point cloud library to solve the relative pose transformation matrix, and the LiDAR positioning data is estimated based on the position information at the previous moment, such as the LOAM feature matching algorithm.

[0060] The state equation and measurement equation are constructed according to the UWB positioning data and the LiDAR positioning data. The state equation is: k =F t X k-1 +W k-1 , where X k , X k-1 is the state vector, which is composed of the coordinate information of UWB positioning data and LiDAR positioning data at time k and time k-1 and the speed deviation in the x direction and y direction, Wk-1 is the process noise, F t is the state transfer matrix; the measurement equation is Y k =H k X k +V k , Y k is the observation vector, which includes the difference between the measured value and the approximate value between the i-th base station and the mobile station at time k in the UWB positioning data and the difference between the measured value and the approximate value of the LiDAR displacement change at adjacent times in the LiDAR positioning data, V k is the linearized noise, H k is the measurement coefficient matrix. The measurement value between the i-th reference station and the mobile station at time k in the UWB positioning data is expressed as represents the coordinate information of the UWB mobile station at time k, Represents the coordinate information of the i-th base station. The measured value of the LiDAR displacement change at adjacent moments in the LiDAR positioning data is expressed as Represents the coordinate information of LiDAR at time k and time k-1. As the EKF filter iterates, the state and measurement are updated through EKF to obtain the fused positioning data.

[0061] It should be noted that the full coverage path is divided according to the grid, the sub-path in each grid block is read, the morphological characteristics of the sub-path are analyzed, and the positioning samples are constructed according to the sub-path and its morphological characteristics; the positioning samples are used to retrieve rice film covering instances, and rice film covering instances with satisfactory film covering accuracy are screened, the planned film covering edge position sequence and the actual film covering edge position sequence in the screened rice film covering instances are read, the film covering edge position sequence and the actual film covering edge position sequence corresponding to each positioning sample are aggregated to generate a training data set, and the actual film covering edge position sequence is extracted as verification data for the training data.

[0062] Figure 3 A flow chart of an embodiment for constructing a normal coating positioning model to identify abnormal coating accuracy is shown.

[0063] According to an embodiment of the present invention, a normal film coverage positioning model is constructed to identify abnormal coverage accuracy, specifically:

[0064] S302, obtaining a training data set corresponding to the positioning samples, sorting the training data set according to the sampling time, using Mahalanobis distance as a metric function, expanding the mRMR algorithm through the metric function, and filtering offline training data based on the principle that the distance between the training data and the positioning samples is the smallest and the distance between the training data is the largest;

[0065] S304, constructing a normal film coverage positioning model using a multivariate state estimation method, constructing a memory matrix using the screened offline training data, and calculating an estimation vector of an observation vector using the memory matrix, wherein the observation vector is generated by a positioning sample of an input model;

[0066] S306, analyzing the residuals connected between the observation vector and the estimation, and obtaining the reference residuals of the normal film positioning model according to the validation set corresponding to the training data;

[0067] S308, importing the fused positioning data into the normal film positioning model, matching and obtaining the current positioning sample, constructing a memory matrix based on the positioning sample, and calculating the estimated vector of the current positioning sample using the constructed memory matrix;

[0068] S310, obtaining residuals according to the observation vector and the estimated vector of the current positioning sample, comparing the obtained residuals with the reference residuals, and issuing a coverage accuracy abnormality warning when the residual difference is greater than a preset difference threshold.

[0069] It should be noted that the multivariate state estimation method mainly includes a memory matrix formed by the rice film covering example with the film covering accuracy reaching the standard, which stores and memorizes the positioning state data of the film covering when the film covering accuracy reaches the standard. The principle of its monitoring is that if the state of the monitored object is normal, the distance between the input data and the normal operating data as training data is close, and the residual between the output estimate and the input data is small, otherwise, the opposite is true.

[0070] Using the extended mRMR algorithm, the offline training data is screened based on the principle of minimizing the distance between the training data and the positioning samples and maximizing the distance between the training data. The online calculation speed of the memory matrix is ​​improved by clearing the redundant data. When selecting data for training, the comprehensiveness of the training data is taken into account, which is more in line with the distribution of the actual film covering accuracy data. The mRMR algorithm after the expansion of the Mahalanobis distance is expressed as: Among them, D(x i ; y) represents the i-th training data x i The Mahalanobis distance between the current positioning sample y, D(x i ;x j ) represents the i-th training data x i With the jth training data x j The Mahalanobis distance between them, S represents the selected offline training data, |S| represents the selected offline training data, Ω s Represents the training dataset.

[0071] It should be noted that the residual difference obtained by the normal covering positioning model is obtained as the coverage deviation in the weeding film covering process, the covering speed of the weeding film is extracted, and the coverage deviation prediction value of the next timestamp is performed based on the coverage speed and the coverage deviation; positioning feedback information is generated based on the coverage deviation prediction value, and the coverage accuracy of the weeding film is compensated based on the positioning feedback information.

[0072] Figure 4 The block diagram of a high-precision film covering positioning system for rice weed control is shown.

[0073] The second aspect of the present invention provides a high-precision film covering positioning system 4 for rice weeding film, including a full coverage path planning module 401, a weeding film covering positioning module 402, a coverage accuracy abnormality recognition module 403, an abnormality alarm module 404 and a film covering accuracy compensation module 405;

[0074] The full coverage path planning module is responsible for obtaining the full coverage path of the weed film based on the two-dimensional map of the rice planting area and the weed film width information;

[0075] The weeding film covering positioning module is responsible for acquiring UWB positioning data and LiDAR positioning data during the covering process, and fusing the acquired positioning data to generate fused positioning data;

[0076] The coverage accuracy anomaly recognition module is responsible for using the multivariate state estimation method to perform data-driven modeling, construct a normal film covering positioning model, import the fused positioning data into the normal film covering positioning model, and determine the coverage deviation during the weeding film covering process;

[0077] The coverage accuracy abnormality recognition module is responsible for generating a lamination position abnormality warning. When the coverage deviation is greater than a preset deviation threshold, a lamination position abnormality warning will be generated.

[0078] The film covering accuracy compensation module is responsible for generating film covering position compensation data according to covering deviation, using the film covering position compensation data for positioning feedback, and adjusting the subsequent covering position of the weeding film.

[0079] A third aspect of the present invention provides a computer-readable storage medium, which includes a high-precision film positioning method program applied to rice weeding film. When the high-precision film positioning method program applied to rice weeding film is executed by a processor, the steps of the high-precision film positioning method applied to rice weeding film are implemented.

[0080] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0081] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0082] In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0083] Those skilled in the art can understand that: all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiments; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), disks or optical disks, and other media that can store program codes.

[0084] Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention can be essentially or partly reflected in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

[0085] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A high-precision film covering positioning method for rice weed control film, characterized in that: The following steps are involved: Obtain basic information of the current rice planting area and weeding film width information, obtain a full coverage path of the weeding film based on the two-dimensional map of the rice planting area and the weeding film width information, and represent the full coverage path; Performing weed control and film covering in the rice planting area based on the full coverage path, acquiring UWB positioning data and LiDAR positioning data during the covering process, and fusing the acquired positioning data to generate fused positioning data; Segment the positioning samples according to the full coverage path after gridding, build a normal coverage positioning model using a multivariate state estimation method based on the training data corresponding to the positioning samples, and obtain the benchmark residual corresponding to the model; The fused positioning data is imported into the normal film covering positioning model, the estimated vector is read to calculate the residual, the calculated residual is compared with the reference residual, the coverage deviation in the weeding film covering process is determined, positioning feedback is performed according to the coverage deviation, and the coverage position is adjusted.

2. A high-precision film covering positioning method for rice weed control according to claim 1, characterized in that: Get the basic information of the current rice planting area and the width of the weeding film, and get the full coverage path of the weeding film based on the two-dimensional map of the rice planting area and the width of the weeding film, specifically: Acquire the size information and geographical location information of the current rice planting area as basic information, extract the high-resolution remote sensing image of the area closest to the current time, and pre-process the high-resolution image; Using basic information of the rice-growing area to crop the high-resolution remote sensing image, constructing a two-dimensional map of the rice-growing area based on the cropped regional remote sensing image, obtaining width information of the weed-removing film, and determining grid side length data based on the width information; The two-dimensional map is divided into grids, and the total number of grids required to be covered by the path for full coverage of the weeding film is determined, the coverage direction constraint of the weeding film is determined, the grids of the two-dimensional map are encoded, the number of covered grids in the rice planting area is obtained, and the path coverage rate is calculated; When the coverage rate reaches the preset threshold, the full coverage path is output according to the grid map matrix representation, and the endpoints of the straight line segments are connected at the U-turn to output the grid-represented full coverage path of the weed removal film.

3. The high-precision film covering positioning method for rice weed control according to claim 1, characterized in that: Acquire UWB positioning data and LiDAR positioning data during coverage, and fuse the acquired positioning data to generate fused positioning data, specifically: The edge positioning measurement values ​​of the weeding film covering process are obtained by using UWB and LiDAR sensors, and the measurement values ​​of the two heterogeneous sensors are unified in time stamp and converted in spatial coordinate system to achieve time synchronization and space synchronization; The UWB measurement value is calculated by the signal two-way propagation time method to generate UWB positioning data, the LiDAR inter-frame point cloud is used for feature matching, the LiDAR posture information is obtained to generate LiDAR positioning data, and the state equation and measurement equation are constructed according to the UWB positioning data and the LiDAR positioning data; The EKF filter is used to fuse the UWB positioning data and the LiDAR positioning data, and the fused positioning data is obtained according to the updated solution of the state equation and the measurement equation.

4. The high-precision film covering positioning method for rice weed control according to claim 1, characterized in that: The samples are segmented and located according to the full coverage path after gridding, specifically: The full coverage path is divided according to the grid, the sub-path in each grid block is read, the morphological characteristics of the sub-path are analyzed, and the positioning sample is constructed according to the sub-path and its morphological characteristics; The positioning samples are used to retrieve rice film covering examples, and rice film covering examples with satisfactory film covering accuracy are screened. The planned film covering edge position sequence and the actual film covering edge position sequence in the screened rice film covering examples are read, and the film covering edge position sequence and the actual film covering edge position sequence corresponding to each positioning sample are aggregated to generate a training data set.

5. The high-precision film covering positioning method for rice weed control according to claim 1, characterized in that: Based on the training data corresponding to the positioning sample, a normal film positioning model is constructed using a multivariate state estimation method, and the reference residual corresponding to the model is obtained, specifically: Obtain a training data set corresponding to the positioning sample, sort the training data set according to the sampling time, use Mahalanobis distance as a metric function, expand the mRMR algorithm through the metric function, and screen the offline training data based on the principle that the distance between the training data and the positioning sample is the minimum and the distance between the training data is the maximum; A normal film positioning model is constructed using a multivariate state estimation method, a memory matrix is ​​constructed using the screened offline training data, and an estimation vector of an observation vector is calculated using the memory matrix, where the observation vector is generated by a positioning sample of an input model; The residuals connecting the observation vector and the estimation are analyzed, and the baseline residuals of the normal film positioning model are obtained according to the validation set corresponding to the training data.

6. The high-precision film covering positioning method for rice weed control according to claim 1, characterized in that: The fused positioning data is imported into the normal film positioning model, and the estimated vector is read to calculate the residual, specifically: Importing the fused positioning data into the normal film positioning model, matching and obtaining the current positioning sample, constructing a memory matrix based on the positioning sample, and calculating the estimated vector of the current positioning sample using the constructed memory matrix; The residual is obtained based on the observation vector and estimated vector of the current positioning sample, and the obtained residual is compared with the benchmark residual. When the residual difference is greater than the preset difference threshold, a coverage accuracy abnormality warning is issued.

7. The high-precision film covering positioning method for rice weed control film according to claim 1, characterized in that: Determine the coverage deviation during the weeding film covering process, provide positioning feedback based on the coverage deviation, and adjust the coverage position. Specifically: Obtain the residual difference value obtained by the normal film covering positioning model as the covering deviation in the weeding film covering process, extract the covering speed of the weeding film, and make a coverage deviation prediction value for the next timestamp based on the covering speed and the covering deviation; Positioning feedback information is generated based on the coverage deviation prediction value, and the coverage accuracy of the weeding film is compensated based on the positioning feedback information.

8. A high-precision film covering positioning system for rice weeding film, characterized in that: A high-precision film covering positioning method for rice weeding film as claimed in any one of claims 1 to 7, comprising a full coverage path planning module, a weeding film covering positioning module, a coverage accuracy abnormality recognition module, an abnormality alarm module and a film covering accuracy compensation module; The full coverage path planning module is responsible for obtaining the full coverage path of the weed film based on the two-dimensional map of the rice planting area and the weed film width information; The weeding film covering positioning module is responsible for acquiring UWB positioning data and LiDAR positioning data during the covering process, and fusing the acquired positioning data to generate fused positioning data; The coverage accuracy anomaly recognition module is responsible for using the multivariate state estimation method to perform data-driven modeling, construct a normal film covering positioning model, import the fused positioning data into the normal film covering positioning model, and determine the coverage deviation during the weeding film covering process; The coverage accuracy abnormality recognition module is responsible for generating a lamination position abnormality warning. When the coverage deviation is greater than a preset deviation threshold, a lamination position abnormality warning will be generated. The film covering accuracy compensation module is responsible for generating film covering position compensation data according to covering deviation, using the film covering position compensation data for positioning feedback, and adjusting the subsequent covering position of the weeding film.

Images