Full-automatic intelligent transplanting machine and method

Abstract

The invention provides a full-automatic intelligent transplanting machine and method, and the method comprises the steps: carrying out the evaluation of the planting adaptability of a target transplanting plot through ground surface environment data, screening out a target to-be-planted point comprising a loose structure region, a stable humidity region and a safe ground surface slope region, and generating a transplanting path suitable for the mechanical operation condition of the target transplanting plot; identifying the adaptability information of the soil penetration depth of the transplanting execution mechanism of the intelligent transplanter and the attitude stability of the seedling body in the transplanting process based on the multi-modal sensing data; performing real-time correction on a mechanical arm execution path of the intelligent transplanter according to the adaptability information of the soil penetration depth and the attitude stability of the seedling body, and generating an execution control sequence; and performing adaptive correction on the transplanting action of the transplanter by taking the execution control sequence as a reference according to the soil resistance fluctuation fed back in real time in the transplanting process. By means of the scheme, real-time sensing and linkage control can be conducted on the working environment and the working state in the transplanting process so as to achieve dynamic optimization of the transplanting path.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of agricultural machinery control, and more particularly to a full-automatic intelligent transplanting machine and method. BACKGROUND

[0002] Transplanting operation is one of the key links in the process of agricultural mechanization planting, and its purpose is to transplant seedlings to the target land according to the set row spacing, plant spacing and depth, so as to ensure the survival rate and growth consistency of seedlings. In modern agricultural large-scale planting, the full-automatic intelligent transplanting machine has become the core equipment to replace manual transplanting, reduce labor intensity and ensure planting timeliness, and its technical core revolves around transplanting path planning, actuator control and operation state feedback.

[0003] The current automatic or semi-automatic transplanting machine usually relies on fixed walking track and pre-set soil penetration depth for mechanical operation, and can realize basic transplanting function in specific environment; however, in complex surface environment, such as uneven surface humidity, obvious soil structure difference, slope, etc., the traditional transplanting machine generally has poor environmental adaptability, low soil penetration depth control precision, easy seedling posture deflection and uneven compaction strength, etc. These problems lead to unstable seedling hole morphology, insufficient root and soil contact in the transplanting process, thereby affecting the transplanting survival rate and subsequent growth quality. Therefore, how to realize real-time sensing and linkage control of the operation environment and operation state in the transplanting process to realize dynamic optimization of the transplanting path, so that the full-automatic transplanting machine can stably and efficiently complete the operation task in the complex farmland environment has become a difficult problem in the industry. SUMMARY

[0004] The present application provides a full-automatic intelligent transplanting machine and method, which can realize real-time sensing and linkage control of the operation environment and operation state in the transplanting process to realize dynamic optimization of the transplanting path.

[0005] In a first aspect, the present application provides a control method of a full-automatic intelligent transplanting machine, comprising the following steps: obtaining surface environment data of a target transplanting land; evaluating the suitability of the target transplanting land according to the surface environment data, screening out target planting points including structure loose area, humidity stable area and surface slope safety area, and generating a transplanting path suitable for the mechanical operation conditions of the target transplanting land from the spatial distribution characteristics of all target planting points; when the intelligent transplanting machine travels along the transplanting path, real-time collection of multi-modal sensing data in the operation process, identification of the adaptability information of the soil penetration depth of the transplanting actuator of the intelligent transplanting machine and the posture stability of the seedlings in the transplanting process based on the multi-modal sensing data; The adaptive information of the depth of the intelligent transplanting machine and the posture stability of the seedling body are used to perform real-time correction on the mechanical arm execution path of the intelligent transplanting machine, and an execution control sequence including a depth adjustment instruction, a seedling body righting instruction, and a seed hole covering and compaction instruction is generated. The execution control sequence is used as a reference, and the transplanting action of the transplanting machine is adaptively corrected according to the real-time feedback of the soil resistance fluctuation in the transplanting process.

[0006] In some embodiments, the adaptability of the target transplanting land is evaluated according to the ground surface environment data, and target planting points including a loose structure area, a humidity stable area, and a ground surface slope safe area are screened out, which specifically includes: Key evaluation parameters representing soil structure, humidity characteristics, and terrain slope are extracted from the ground surface environment data; The key evaluation parameters are weighted and fused to obtain an adaptability comprehensive score; According to the adaptability comprehensive score and a preset threshold, target planting points including a loose structure area, a humidity stable area, and a ground surface slope safe area are screened out.

[0007] In some embodiments, the transplanting path suitable for the mechanical operation conditions of the target transplanting land is generated from the spatial distribution characteristics of all target planting points, which specifically includes: The coordinates and regional type attributes of all target planting points are obtained, and the spatial distribution characteristics of all target planting points are determined; An initial work path covering all target planting points is generated based on the spatial distribution characteristics; The initial work path is smoothed and optimized according to the minimum turning radius and the maximum climbing angle of the intelligent transplanting machine, and a transplanting path suitable for the mechanical operation conditions of the target transplanting land is generated.

[0008] In some embodiments, the adaptability information of the depth of the intelligent transplanting machine and the posture stability of the seedling body are identified based on the multi-modal perception data, which specifically includes: The depth feature parameters and the seedling body posture motion parameters are extracted from the multi-modal perception data, respectively; The depth feature parameters are compared with a preset ideal depth range to obtain the adaptability information of the depth of the intelligent transplanting machine; The posture stability of the seedling body during the transplanting process is output based on the seedling body posture motion parameters.

[0009] In some embodiments, the adaptive information of the depth and the posture stability of the seedling body are used to perform real-time correction on the mechanical arm execution path of the intelligent transplanting machine, and an execution control sequence including a depth adjustment instruction, a seedling body righting instruction, and a seed hole covering and compaction instruction is generated, which specifically includes: adjusting the pressing path of the mechanical arm of the intelligent transplanting machine according to the adaptability information of the soil penetration depth, and generating a pressing depth adjustment instruction; correcting the gripper path of the mechanical arm based on the attitude stability of the seedling body, and generating a seedling body righting instruction for controlling the gripper of the mechanical arm; generating a seedling hole covering and compacting instruction from the current seedling hole state after the depth adjustment and righting operation are completed; integrating the pressing depth adjustment instruction, the seedling body righting instruction, and the seedling hole covering and compacting instruction into an execution control sequence according to the operation sequence.

[0010] In some embodiments, the adaptive correction of the transplanting action of the transplanting machine is based on the execution control sequence and the real-time feedback of the soil resistance fluctuation during the transplanting process, and specifically includes: real-time monitoring and collecting soil resistance data during the transplanting process through a force sensor installed on the transplanting execution mechanism; analyzing the fluctuation characteristics of the collected soil resistance data, and then determining the soil resistance fluctuation through the analyzed abnormal changes in resistance; based on the soil resistance fluctuation, dynamically adjusting the related parameters in the execution control sequence to complete the adaptive correction of the transplanting action.

[0011] In some embodiments, the multi-modal perception data refers to a fused data set representing the state of the transplanting execution mechanism and the dynamic attitude of the seedling during the execution of the intelligent transplanting machine.

[0012] In a second aspect, the present application provides a full-automatic intelligent transplanting machine, comprising a control unit, wherein the control unit comprises: an acquisition module for acquiring surface environment data of a target transplanting plot; a processing module for evaluating the suitability of the target transplanting plot according to the surface environment data, screening out target planting points including a loose structure area, a humidity stable area, and a surface slope safe area, and generating a transplanting path suitable for the mechanical operation conditions of the target transplanting plot from the spatial distribution characteristics of all target planting points; The processing module is also used to collect multi-modal perception data in the operation process in real time when the intelligent transplanting machine travels along the transplanting path, and identify adaptability information of the soil penetration depth of the transplanting execution mechanism of the intelligent transplanting machine and the attitude stability of the seedling during the transplanting process based on the multi-modal perception data; The processing module is also used to instantaneously correct the mechanical arm execution path of the intelligent transplanting machine through the adaptability information of the soil penetration depth and the attitude stability of the seedling, and generate an execution control sequence including a pressing depth adjustment instruction, a seedling body righting instruction, and a seedling hole covering and compacting instruction. An execution module is configured to adaptively correct the transplanting action of the transplanting machine according to the real-time feedback of the soil resistance fluctuation in the transplanting process, with the execution control sequence as a reference.

[0013] In a third aspect, the present application provides a computer device, which comprises a memory and a processor, the memory is configured to store a computer program, and the processor is configured to call and run the computer program from the memory, so that the computer device executes the control method of the full-automatic intelligent transplanting machine.

[0014] In a fourth aspect, the present application provides a computer readable storage medium, which stores instructions or codes, when the instructions or codes are run on a computer, the computer executes the control method of the full-automatic intelligent transplanting machine.

[0015] The technical scheme provided by the embodiments of the present application has the following beneficial effects: In the present application, the ground surface environment data of the target transplanting plot is acquired, the adaptability of the target transplanting plot is evaluated according to the ground surface environment data, the target to-be-planted points including the loose structure region, the humidity stable region and the ground surface slope safety region are screened out, and the transplanting path suitable for the mechanical operation condition of the target transplanting plot is generated from the spatial distribution characteristics of all the target to-be-planted points; when the intelligent transplanting machine travels along the transplanting path, the multi-modal perception data in the operation process is collected in real time, the adaptability information of the intelligent transplanting machine transplanting execution mechanism earth-penetrating depth and the attitude stability of the seedling body in the transplanting process are identified based on the multi-modal perception data; the mechanical arm execution path of the intelligent transplanting machine is instantaneously corrected through the adaptability information of the earth-penetrating depth and the attitude stability of the seedling body, and the execution control sequence containing the down-pressing depth adjustment instruction, the seedling body righting instruction and the seedling hole covering and compacting instruction is generated; the execution control sequence is taken as a reference, and the transplanting action of the transplanting machine is adaptively corrected according to the real-time feedback of the soil resistance fluctuation in the transplanting process.

[0016] It can be seen that in this application, first, a transplanting path suitable for the mechanical operation conditions of the target transplanting plot is generated based on the spatial distribution characteristics of all target planting points, which can achieve the linkage matching of path planning and environmental characteristics, and can avoid problems such as excessive soil penetration resistance or seedling body inclination caused by unreasonable paths during mechanical operation, enabling the intelligent transplanting machine to maintain stable operation under different plot structures, thereby enhancing the terrain adaptability and path dynamic optimization level; second, the adaptability information of the soil penetration depth of the transplanting actuator of the intelligent transplanting machine and the attitude stability of the seedling body during the transplanting process are identified based on the multi-modal perception data. Through this multi-source perception fusion method, the intelligent transplanting machine can dynamically judge the changes in the operating environment and operating state, realize the transformation from preset control to real-time perception control, and provide data support for the linkage optimization of the path and operating actions, thereby significantly enhancing the environmental adaptability and operating accuracy; then, the mechanical arm execution path of the intelligent transplanting machine is immediately corrected based on the adaptability information of the soil penetration depth and the attitude stability of the seedling body, which can achieve the dynamic feedback and action coordination of the mechanical arm during operation, enabling the intelligent transplanting machine to maintain the balance of the soil penetration depth and the stability of the seedling body verticality under environmental disturbances, so as to enhance the response sensitivity of the operation execution layer; finally, based on the execution control sequence and according to the soil resistance fluctuation feedback in real time during the transplanting process, the transplanting action of the transplanting machine is adaptively corrected, which can avoid overpressure or cavity phenomena caused by sudden changes in soil structure, thereby maintaining the stability of the soil penetration depth and uniform compaction. This adaptive correction mechanism enables the transplanting machine to have continuous environmental response and action adjustment capabilities, realizes the closed-loop optimization of the operation state, and further provides core support for stable and efficient transplanting operations in complex farmland environments; in summary, this solution can perform real-time perception and linkage control on the operation environment and operation state during the transplanting process to achieve the dynamic optimization of the transplanting path, so that the full-automatic transplanting machine can complete the operation task stably and efficiently in the complex farmland environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0018] Figure 1 is an exemplary flowchart of a control method for a full-automatic intelligent transplanting machine shown in some embodiments of the present application; Figure 2 is an exemplary flowchart of screening target planting points shown in some embodiments of the present application; Figure 3This is an exemplary flowchart illustrating the adaptive information for determining the soil penetration depth and the posture stability of the seedling, according to some embodiments of this application. Figure 4 This is a schematic diagram of the structure of a control unit according to some embodiments of this application; Figure 5 This is a schematic diagram of the structure of a computer device for implementing a control method for a fully automatic intelligent transplanter according to some embodiments of this application. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0020] refer to Figure 1 The figure is an exemplary flowchart of a control method for a fully automatic intelligent transplanter according to some embodiments of this application. The control method for the fully automatic intelligent transplanter mainly includes the following steps: In step 101, the surface environment data of the target transplanting plot is obtained.

[0021] In specific implementation, obtaining the surface environment data of the target transplanting plot can be achieved in the following way: Aerial scanning of the target transplanting plot can be performed using a multispectral imager and lidar mounted on a drone platform, simultaneously acquiring multispectral images and lidar point cloud data of the surface. Simultaneously, a soil sensor network pre-deployed within the plot collects real-time data on soil moisture, soil compaction, and surface slope. Subsequently, radiometric and geometric corrections are performed on the multispectral images to eliminate atmospheric and topographic distortions. Noise removal and classification processing are performed on the lidar point cloud data to generate a high-precision digital elevation model (DEM). Time synchronization and outlier removal are performed on the soil sensor data. Next, the processed multispectral image data, DEM data, and soil sensor data are spatiotemporally registered and fused to generate standardized surface environment data of the target transplanting plot, including surface spectral features, three-dimensional topographic features, and real-time physical parameters, which serves as input for planting suitability evaluation. In a preferred embodiment, the drone platform can be equipped with real-time dynamic (Real-time) technology. To ensure spatial positioning accuracy, models with kinematic (RTK) positioning capabilities can be used. In other embodiments, satellite remote sensing imagery combined with a ground-based mobile measurement system can be used to replace the UAV platform for data acquisition, which is not specifically limited here.

[0022] It should be noted that the surface environment data in this application refers to a set of multi-dimensional parameters characterizing the physical structure, moisture status, and topographic features of the topsoil of the target transplanting site.

[0023] In step 102, the suitability of the target transplanting plots is evaluated based on the surface environment data. Target planting points are selected, including areas with loose structure, stable humidity, and safe surface slope. Transplanting paths suitable for mechanical operation conditions of the target transplanting plots are generated based on the spatial distribution characteristics of all target planting points.

[0024] In some embodiments, reference Figure 2 As shown, this figure is an exemplary flowchart for screening target planting sites in some embodiments of this application. In this embodiment, the suitability evaluation of the target transplanting plots based on the surface environment data, and the screening of target planting sites including areas with loose structure, stable humidity, and safe surface slope, can be achieved by the following steps: First, in step 1021, key evaluation parameters characterizing soil structure, moisture properties, and topographic slope are extracted from the surface environment data. Secondly, in step 1022, the key evaluation parameters are weighted and fused to obtain a comprehensive plantability score; Finally, in step 1023, target planting sites are selected based on the comprehensive suitability score and preset thresholds, including areas with loose structure, stable humidity, and safe surface slope.

[0025] In specific implementation, the key evaluation parameters characterizing soil structure, moisture characteristics, and topographic slope can be extracted from the surface environment data in the following manner: First, digital elevation model data, multispectral image data, and soil sensor data are retrieved from the surface environment data. Then, based on the digital elevation model, the surface slope angle of each grid cell is calculated using neighborhood analysis as the topographic slope parameter. The surface soil volumetric water content is obtained from the multispectral image data by calculating the ratio of the normalized vegetation index to the soil index as the moisture characteristic parameter. The pressure value directly read from the soil compaction sensor is converted to a unit and used as the soil structure compaction parameter. Finally, the topographic slope parameter, moisture characteristic parameter, and soil structure compaction parameter corresponding to each grid cell are correlated and integrated to output structured key evaluation parameters for subsequent fusion calculations. In a preferred embodiment, the soil structure compaction parameter can be further filtered by moving average through historical sensor readings to eliminate instantaneous fluctuations. In other embodiments, soil texture type can also be introduced to standardize and correct the compaction parameter; this is not limited here.

[0026] In specific implementation, the weighted fusion of the key evaluation parameters to obtain a comprehensive plantability score can be achieved in the following way: First, preset weight coefficients corresponding to the three evaluation dimensions of loose structure, stable humidity, and safe surface slope. For example, soil structure compaction has the highest weight, humidity characteristics have the second highest weight, and topographic slope has the lowest weight. Then, normalize the set of key evaluation parameters, converting the three parameters with different dimensions—surface slope angle, surface soil volumetric water content, and soil structure compaction—to dimensionless values ​​of 0-1. Next, a linear weighted summation model can be used to multiply the normalized parameter value of each grid cell by its corresponding weight coefficient and sum them to calculate a comprehensive plantability score within the range of 0-1. The higher the score, the better the comprehensive plantability conditions of the area. In a preferred embodiment, the weight coefficients can be determined based on historical transplant survival rate data through multiple regression analysis. In other embodiments, the analytic hierarchy process (AHP) can be used to determine the weight coefficients through scoring by domain experts. This is not limited here.

[0027] In practice, the selection of target planting sites, including areas with loose structure, areas with stable humidity, and areas with safe slope, based on the comprehensive suitability score and preset thresholds can be achieved as follows: First, independent comprehensive suitability score thresholds are set for each of the three area types: loose structure, stable humidity, and safe slope. For example, the score for loose structure areas should be greater than 0.7, for stable humidity areas greater than 0.6, and for safe slope areas greater than 0.8. Then, the grid cells of the target transplanting plot are traversed, and their comprehensive suitability scores are compared with the preset thresholds. If the comprehensive suitability score of a certain grid cell simultaneously meets the threshold requirements of a certain area type, the planting site is selected. If a region is identified as a candidate region of that type, then a spatial clustering algorithm is used to aggregate the identified discrete candidate regions into contiguous regional blocks, and the coordinates of the geometric center point of each regional block are calculated. Finally, the obtained geometric center point coordinates and their respective regional type attributes are used as the final target planting points for the corresponding regional type. In a preferred embodiment, the spatial clustering algorithm can be the density-based DBSCAN algorithm to effectively identify distribution areas of arbitrary shapes. In other embodiments, the K-means algorithm can also be used to cluster within a specified regional type to control the number of planting points. This is not limited here.

[0028] It should be noted that the key evaluation parameters in this application refer to characteristic indicators that characterize the core conditions affecting the survival and growth of transplanted plants; the comprehensive suitability score in this application refers to a quantitative value that reflects the overall planting suitability of a specific location in the target transplanting plot; the preset threshold in this application refers to the critical value of the comprehensive suitability score set in advance to classify different types of suitable planting areas; and the target planting point in this application refers to the specific location coordinates of the target transplanting plot that have the best comprehensive suitability conditions. It is used to provide precise navigation endpoints for transplanting path planning to ensure that the operation of the intelligent transplanter can accurately cover all optimized and screened highly suitable areas.

[0029] In some embodiments, generating a suitable transplanting path for mechanical operation of the target transplanting plot based on the spatial distribution characteristics of all target planting points can be achieved through the following steps: Obtain the coordinates and region type attributes of all target planting points, and then determine the spatial distribution characteristics of all target planting points; Based on the spatial distribution characteristics, an initial operation path covering all target planting points is generated; The initial working path is smoothed and optimized based on the minimum turning radius and maximum climbing angle of the intelligent transplanter to generate a transplanting path suitable for the mechanical operating conditions of the target transplanting plot.

[0030] In specific implementation, obtaining the coordinates and regional type attributes of all target planting points, and then determining the spatial distribution characteristics of all target planting points, can be achieved in the following way: First, obtain the coordinates of the geometric center point of each target planting point and its corresponding regional type attribute; then, based on the coordinate data of all target planting points, calculate their distribution density, aggregation degree, and orientation relative to the boundary of the target transplanting plot through spatial statistical analysis, and identify sparse and dense distribution areas; finally, combined with the regional type attribute, analyze the spatial correlation and overlap of different types of planting points (e.g., loosely structured points and humidity-stable points), thereby comprehensively determining the spatial distribution characteristics of all target planting points. This characteristic is used to characterize the global layout pattern of planting points and provide a topological basis for path planning. In a preferred embodiment, the distribution density can be visualized and analyzed using a kernel density estimation algorithm. In other embodiments, the Moran index can also be used for spatial autocorrelation quantification, which is not limited here.

[0031] In specific implementation, generating an initial work path covering all target planting points based on the spatial distribution characteristics can be achieved in the following way: First, the coordinates of the target planting points in the spatial distribution characteristics are taken as the necessary nodes of the path, and the path search strategy is determined according to the principle of prioritizing connectivity in densely distributed areas; then, the Traveling Salesman Problem optimization algorithm can be used to calculate the shortest access sequence to traverse all necessary nodes of the path with the goal of minimizing the total travel distance, thereby generating an initial work path; next, known obstacle areas, such as rocks or ditches, are marked on the initial work path according to the digital elevation model data, and the A* search algorithm is used to replan local detours to ensure the feasibility of the path; finally, a collision-free continuous coordinate point sequence covering all target planting points is output as the initial work path; wherein, as a preferred embodiment, the Traveling Salesman Problem optimization algorithm can be a genetic algorithm to efficiently process large-scale point sets, and in other embodiments, the ant colony algorithm can also be used to find an approximate optimal solution in complex distribution patterns, which is not limited here.

[0032] In specific implementation, the initial working path is smoothed and optimized based on the minimum turning radius and maximum climbing angle of the intelligent transplanter to generate a transplanting path suitable for the mechanical operation conditions of the target transplanting plot. This can be achieved in the following way: First, the minimum turning radius and maximum climbing angle of the intelligent transplanter can be obtained from its mechanical parameter library and used as hard constraints for path optimization; then, a B-spline curve fitting algorithm can be used to smooth the sharp bends in the initial working path to ensure that the radius of curvature at all turns is greater than the minimum turning radius; finally, the slope of each segment of the path is calculated using elevation data provided by the digital elevation model. The system optimizes the slope of the transplanter by raising or rerouting sections with gradients exceeding the maximum climbing angle to ensure that the slope of the entire path meets the mechanical climbing capacity. Finally, the optimized path is discretized and sampled to output a set of waypoint sequences with speed and direction suggestions. This sequence serves as the final transplanting path that can be directly executed by the intelligent transplanter control system, i.e., a transplanting path suitable for the mechanical operation conditions of the target transplanting plot. In a preferred embodiment, the B-spline curve fitting can incorporate gradient descent to optimize the control point positions with minimal path length variation. In other embodiments, a spiral curve can be used to design transition curves that conform to vehicle dynamics; this is not a limitation.

[0033] It should be noted that the spatial distribution characteristics in this application refer to the comprehensive characteristics describing the macroscopic layout and clustering state of all target planting points within the target transplanting plot; the initial operation path in this application refers to the continuous coordinate sequence initially planned with the primary goal of traversing all target planting points; and the transplanting path in this application refers to the action trajectory that can be directly driven by the transplanting machine control system after smoothing and optimizing the initial operation path in conjunction with the specific mechanical motion constraints of the intelligent transplanter.

[0034] In step 103, as the intelligent transplanter moves along the transplanting path, multimodal sensing data is collected in real time during the operation. Based on the multimodal sensing data, the adaptability information of the soil penetration depth of the intelligent transplanter's transplanting actuator and the posture stability of the seedling during the transplanting process are identified.

[0035] In specific implementation, as the intelligent transplanter moves along the transplanting path, the real-time acquisition of multimodal perception data during the operation can be achieved in the following way: data can be collected synchronously by multiple sensors integrated into the end of the transplanting actuator of the intelligent transplanter. Among them, a laser displacement sensor is used to continuously measure the real-time distance between the robotic arm bucket and the soil surface; an embedded vision module captures the visual image sequence of the seedling-soil contact area; an inertial measurement unit monitors the changes in the triaxial acceleration and angular velocity of the seedling during the transplanting process; and a pressure sensor measures the soil reaction force when the actuator presses down. Subsequently, the original sensor data obtained above is time-stamped and spatially coordinate unified, and the laser data is filtered by moving average, the image data is denoised by Gaussian, the inertial data is zero-biased, and the pressure data is standardized by unit, finally generating multimodal perception data with time alignment and uniform dimensions during the operation. In a preferred embodiment, the embedded vision module can use a global shutter camera to eliminate motion blur. In other embodiments, a ring light can also be added to ensure imaging quality in dark environments, which is not limited here.

[0036] It should be noted that the multimodal sensing data in this application refers to a fusion data set that characterizes the soil entry state of the transplanting actuator and the dynamic posture of the seedling during the transplanting operation of the intelligent transplanter.

[0037] In some embodiments, reference Figure 3 As shown, this figure is an exemplary flowchart of determining the adaptive information of the soil penetration depth and the posture stability of the seedling in some embodiments of this application. In this embodiment, the adaptive information of the soil penetration depth of the transplanting actuator of the intelligent transplanter and the posture stability of the seedling during transplanting based on the multimodal sensing data can be achieved by the following steps: First, in step 1031, the soil penetration depth feature parameters and seedling posture motion parameters are extracted from the multimodal sensing data, respectively. Secondly, in step 1032, the soil penetration depth characteristic parameters are compared with the preset ideal depth range to obtain the adaptability information of the soil penetration depth of the transplanting actuator of the intelligent transplanter; Finally, in step 1033, the posture stability of the seedling during transplanting is output based on the seedling posture motion parameters.

[0038] In specific implementation, the soil penetration depth feature parameters and seedling posture motion parameters extracted from the multimodal sensing data can be achieved in the following way: First, the multimodal sensing data is analyzed. The real-time distance value between the robotic arm bucket and the soil surface is extracted from the laser displacement sensor data as the initial depth reading. The location of the junction between the base of the seedling stem and the soil surface is identified from the visual image sequence captured by the embedded vision module through background subtraction and contour extraction algorithms. The visual depth compensation value is calculated in combination with the kinematic model of the robotic arm. The soil penetration depth feature parameters are obtained by weighted fusion of the laser reading and the visual depth compensation value. At the same time, the three-axis angular velocity variance and the synthetic acceleration modulus of the seedling are extracted from the inertial measurement unit data as the sway intensity index. The pixel displacement variance of the key points at the top of the seedling is calculated from the visual image sequence using the optical flow method as the tilt fluctuation index. The sway intensity index and the tilt fluctuation index are then combined to form the seedling posture motion parameters. In a preferred embodiment, the visual depth compensation can introduce a stereo vision matching algorithm to improve the three-dimensional positioning accuracy. In other embodiments, Kalman filtering can also be used to fuse the laser and visual data. This is not limited here.

[0039] In specific implementation, the soil penetration depth characteristic parameter is compared with a preset ideal depth range to obtain the adaptability information of the soil penetration depth of the intelligent transplanter's transplanting actuator. This can be achieved in the following way: First, a corresponding ideal depth range can be preset according to the agronomic requirements of the current transplanted crop. This ideal depth range is stored in the intelligent transplanter control system in the form of intervals. Then, the soil penetration depth characteristic parameter is compared with the ideal depth range. If the soil penetration depth characteristic parameter falls within the ideal depth range, it is marked as "suitable depth". If it is higher than the upper limit of the ideal depth range, it is marked as "too deep". If it is lower than the lower limit of the ideal depth range, it is marked as "too shallow". Next, adaptability information containing qualitative status and quantitative deviation value is generated according to the comparison result. For example, the output is "too shallow, deviation value is -1.5 cm", which can obtain the adaptability information of the soil penetration depth of the intelligent transplanter's transplanting actuator. In a preferred embodiment, the ideal depth range can be set with different tolerance intervals according to the soil texture type. In other embodiments, the range can also be dynamically adjusted based on the crop growth stage. This is not limited here.

[0040] In specific implementation, the output of the seedling's posture stability during transplantation based on the seedling's posture motion parameters can be achieved in the following way: First, the angular velocity variance, acceleration magnitude, and pixel displacement variance of the seedling's posture motion parameters are input into a pre-trained posture stability evaluation model, which can be a classifier based on a support vector machine or a lightweight neural network; then, the model outputs a level label characterizing the seedling's stability by analyzing the comprehensive pattern of the input motion parameters, the level including at least three levels: "stable," "slightly unstable," and "unstable"; next, the current posture stability level of the seedling, along with its main influencing factors, is output, for example, "unstable, mainly due to excessive lateral sway"; thus, the posture stability of the seedling during transplantation can be obtained, serving as the basis for correcting the robotic arm's execution path; wherein, as a preferred embodiment, the posture stability evaluation model can be pre-trained using a publicly available posture dataset through transfer learning, and in other embodiments, ensemble learning can be used to combine multiple weak classifiers to improve judgment robustness, which is not limited here.

[0041] It should be noted that the soil penetration depth characteristic parameter in this application refers to the parameter characterizing the actual soil penetration depth of the transplanting execution mechanism, which is used to determine whether the soil penetration depth is appropriate; the seedling posture motion parameter in this application refers to the parameter describing the degree of swaying and tilting of the seedling during transplanting, which is used to reflect the instantaneous changes and instability of the seedling posture; the ideal depth range in this application refers to the upper and lower limit range of the optimal value of the transplanting depth preset according to the specific agronomic requirements of the current transplanted crop, which is used to measure whether the actual soil penetration depth characteristic parameter is within the optimal range; the soil penetration depth adaptability information in this application refers to the evaluation result reflecting the degree of matching between the pressing action of the intelligent transplanting machine's transplanting execution mechanism and the soil resistance, which includes qualitative status and quantitative deviation values, and can be used to indicate the execution quality of the current transplanting operation in the depth dimension; the seedling posture stability in this application refers to the graded evaluation characterizing the ability of the seedling to maintain a vertically stable state during transplanting, which is used to reflect the upright quality of the seedling planting.

[0042] In step 104, the execution path of the robotic arm of the intelligent transplanter is corrected in real time by using the adaptive information of the soil penetration depth and the posture stability of the seedling, and an execution control sequence including the pressing depth adjustment command, the seedling straightening command, and the seedling hole covering and compaction command is generated.

[0043] In some embodiments, the execution path of the robotic arm of the intelligent transplanter is modified in real time based on the adaptive information of the soil penetration depth and the posture stability of the seedling. The generation of an execution control sequence including instructions for adjusting the pressing depth, straightening the seedling, and covering and compacting the seedling hole can be achieved through the following steps: Based on the adaptability information of the soil penetration depth, the pressing path of the intelligent transplanter's robotic arm is adjusted, and a pressing depth adjustment command is generated. Based on the posture stability of the seedling, the gripper path of the robotic arm is straightened and corrected to generate a seedling straightening command for controlling the gripper of the robotic arm. After the depth adjustment and straightening operations are completed, a seedling hole covering and compaction command is generated based on the current seedling hole status; The pressing depth adjustment command, the seedling uprighting command, and the seedling hole covering and compaction command are integrated into an execution control sequence according to the operation sequence.

[0044] In specific implementation, adjusting the pressing path of the intelligent transplanter's robotic arm based on the adaptive information of the soil penetration depth and generating a pressing depth adjustment command can be achieved in the following way: First, analyze the qualitative state and quantitative deviation value in the adaptive information of the soil penetration depth. If the state is "too shallow," calculate the positive depth compensation amount; if it is "too deep," calculate the negative retraction amount, i.e., the negative depth compensation amount. Then, based on the quantitative deviation value, generate the vertical position correction amount of the robotic arm through a PID controller and convert this position correction amount into a pulse control signal for the servo motor. Next, adjust the pressing speed according to the current motion state and dynamic constraints of the intelligent transplanter's robotic arm to avoid impact, and finally generate a pressing depth adjustment command containing the target depth, motion speed, and acceleration parameters. In a preferred embodiment, the calculation of the depth compensation amount can incorporate soil resistance fluctuation data for feedforward compensation. In other embodiments, a fuzzy control strategy can also be used to handle nonlinear soil response, which is not limited here.

[0045] In specific implementation, the gripper path of the robotic arm is straightened and corrected based on the posture stability of the seedling. The seedling straightening command for controlling the gripper can be generated in the following way: First, the straightening intensity is determined according to the level corresponding to the posture stability of the seedling. If it is "unstable", a large-angle rapid correction strategy is adopted; if it is "slightly unstable", a small-angle gradual correction strategy is adopted. Then, the target posture angle to be achieved by the gripper is calculated through the inverse kinematics model of the robotic arm, and a smooth rotation trajectory from the current posture to the target posture is planned. Next, the clamping force is adjusted in real time with the force feedback sensor data to ensure the straightening effect while avoiding damage to the seedling. Finally, a seedling straightening command containing rotation angle, angular velocity and clamping force parameters is generated. In a preferred embodiment, the rotation trajectory planning can use quaternion interpolation to avoid the gimbal lock problem. In other embodiments, an impedance control strategy can also be introduced to achieve compliant straightening. This is not limited here.

[0046] In specific implementation, after the depth adjustment and straightening operations are completed, the seedling hole covering and compaction command generated from the current seedling hole status can be achieved in the following way: First, the integrity of the soil covering around the seedling hole is detected by machine vision to identify uncovered areas and loose soil areas; then, the optimal compaction intensity is determined based on soil moisture data and seedling characteristics. When the moisture content is high, a light compaction multiple times strategy is adopted, and when the moisture content is low, a heavy compaction fewer times strategy is adopted; next, the movement path of the compaction tool at the end of the robotic arm is planned, and a spiral asymptotic approach can be used to uniformly compact the soil around the seedling hole, finally generating a seedling hole covering and compaction command that includes the compaction path, compaction intensity, and number of compaction times; wherein, as a preferred embodiment, the determination of the optimal compaction intensity can be optimized based on the soil plasticity model. In other embodiments, closed-loop compaction control can also be achieved through pressure sensor feedback, which is not limited here.

[0047] In specific implementation, the integration of the pressing depth adjustment command, the seedling straightening command, and the seedling hole covering and compaction command into an execution control sequence according to the operation time sequence can be achieved in the following way: First, establish a timestamp-based command scheduling mechanism to ensure that pressing depth adjustment is executed first, seedling straightening second, and seedling hole covering and compaction last; then, check the motion conflicts between each command and insert smooth transition segments between command transitions through trajectory interpolation algorithms; next, add execution condition judgment logic to each command, and trigger the next command only when the previous command is completed and the status verification is passed; finally, encode all commands and their timing relationships into standardized control sequence messages, i.e., execution control sequences, and output them to the robotic arm controller of the intelligent transplanter; wherein, as a preferred embodiment, the command scheduling can adopt the time slice round-robin mechanism of a real-time operating system, and in other embodiments, a finite state machine model can also be used to manage the command flow, which is not limited here.

[0048] It should be noted that the pressing depth adjustment command in this application refers to the specific operation command used to control the vertical movement of the intelligent transplanter's robotic arm, ensuring that the transplanting actuator reaches the ideal soil penetration depth that meets agronomic requirements; the seedling straightening command in this application refers to the operation command used to control the posture adjustment of the robotic arm's gripper, ensuring that the seedling maintains an ideal vertical and stable posture after transplanting; the seedling hole covering and compaction command in this application refers to the operation command used to control the soil covering and compaction operation, ensuring that the seedling roots are in full contact with the soil, providing a stable soil environment for seedling growth; the execution control sequence in this application refers to the standardized control message formed by integrating the pressing depth adjustment command, the seedling straightening command, and the seedling hole covering and compaction command according to the operation sequence, which is used to ensure that the three commands can be executed sequentially by the robotic arm controller according to strict timing logic and condition judgment, so as to realize the automation, coordination, and integrity of the transplanting operation process.

[0049] In step 105, the transplanting action of the transplanter is adaptively corrected based on the execution control sequence and the real-time feedback of soil resistance fluctuations during the transplanting process.

[0050] In some embodiments, the adaptive correction of the transplanting action of the transplanter based on the execution control sequence and according to the real-time feedback of soil resistance fluctuations during the transplanting process can be achieved by the following steps: The soil resistance data during the transplanting process is monitored and collected in real time by force sensors installed on the transplanting actuator; The collected soil resistance data were analyzed for fluctuation characteristics, and the fluctuation of soil resistance was determined by the analyzed anomaly changes in resistance. Based on the soil resistance fluctuations, the relevant parameters in the execution control sequence are dynamically adjusted to complete the adaptive correction of the transplanting action.

[0051] In specific implementation, the real-time monitoring and collection of soil resistance data during the transplanting process by force sensors installed on the transplanting actuator can be achieved in the following way: First, a strain gauge-type force sensor is installed at the connection between the bucket and the drive shaft of the transplanting actuator to continuously collect the soil reaction force on the transplanting actuator in the vertical direction at a sampling frequency of not less than 100Hz; then, the collected raw force signal is hardware filtered to remove high-frequency noise and converted into a digital signal by an ADC module; next, the digital signal is transmitted to the main control system of the transplanter and time-stamped with the robotic arm pose data to form soil resistance data with time-series markers; wherein, as a preferred embodiment, the force sensor can be an S-type tension sensor based on the Wheatstone bridge principle, and in other embodiments, a torque sensor can be installed at the joint of the robotic arm to indirectly calculate the soil resistance, which is not limited here.

[0052] In specific implementation, the collected soil resistance data is analyzed for fluctuation characteristics, and the soil resistance fluctuations are determined by the analyzed abnormal changes in resistance. This can be achieved in the following way: First, the collected soil resistance data is analyzed in real time using the sliding window method to calculate the mean and standard deviation of resistance within each window; then, wavelet transform is used to detect the frequency domain characteristics of the resistance signal to identify instantaneous impact components and slowly changing trend components; next, based on preset resistance thresholds and rate of change thresholds, signal segments exceeding three times the standard deviation of the mean or exceeding the set threshold for rate of change are marked as abnormal resistance changes; finally, the duration, peak intensity, and gradient of the marked abnormal resistance changes are extracted as characteristic parameters of soil resistance fluctuations. In a preferred embodiment, the resistance threshold can be dynamically adjusted according to different soil types. In other embodiments, machine learning algorithms can also be used to identify complex resistance patterns, which is not limited here.

[0053] In specific implementation, the adaptive correction of the transplanting action by dynamically adjusting the relevant parameters in the execution control sequence based on the soil resistance fluctuation can be achieved in the following way: First, the pressing depth adjustment command and the seedling straightening command in the execution control sequence are parsed, and the characteristic parameters of the soil resistance fluctuation are mapped to specific control parameters; when a short-term high-amplitude resistance peak is detected, the target depth value in the pressing depth adjustment command is reduced and the pressing speed is decreased accordingly; when a continuous medium-amplitude resistance fluctuation is detected, the clamping force in the seedling straightening command is adjusted and the buffer time of the straightening action is increased accordingly; then, the depth, speed and force parameters in the execution control sequence can be updated in real time using an online optimization algorithm, and a smooth transition can be achieved through the command interpolation mechanism of the robotic arm controller; finally, the corrected control parameters are re-encapsulated into an execution control sequence and sent for execution to complete the adaptive correction; wherein, as a preferred embodiment, the characteristic parameter mapping can be implemented using a fuzzy rule base based on expert knowledge, and in other embodiments, a resistance-control parameter transfer function can also be established for feedforward compensation, which is not limited here.

[0054] It should be noted that the soil resistance data in this application refers to the measured value characterizing the soil reaction force experienced by the transplanting actuator during operation, which is used to reflect the intensity of the interaction between the robotic arm and the soil medium; the soil resistance fluctuation in this application refers to the significant abnormal change characteristics relative to the normal resistance pattern identified from the soil resistance data, which is used to reveal potential obstacles such as local hard foreign objects, root entanglement or abrupt changes in texture in the soil, and to provide clear trigger signals and quantitative basis for dynamically adjusting key parameters in the execution control sequence, such as downpressing depth and straightening force, thereby ensuring the adaptability and robustness of transplanting actions in the face of complex soil environments.

[0055] In another aspect, in some embodiments, this application provides a fully automatic intelligent transplanter, which includes a control unit, as referenced. Figure 4 The figure is a schematic diagram of the structure of a control unit according to some embodiments of this application. The control unit 400 includes: an acquisition module 401, a processing module 402, and an execution module 403, which are described below: The acquisition module 401 in this application is mainly used to acquire the surface environment data of the target transplanting site; Processing module 402, in this application, is mainly used to evaluate the suitability of the target transplanting plot based on the surface environment data, screen out the target planting points including areas with loose structure, areas with stable humidity and areas with safe surface slope, and generate a transplanting path suitable for the mechanical operation conditions of the target transplanting plot based on the spatial distribution characteristics of all target planting points. The processing module 402 described in this application is also used to collect multimodal perception data in real time during the operation of the intelligent transplanter while it is moving along the transplanting path, and to identify the adaptability information of the soil penetration depth of the transplanting actuator of the intelligent transplanter and the posture stability of the seedling during the transplanting process based on the multimodal perception data. The processing module 402 described in this application is also used to make real-time corrections to the execution path of the robotic arm of the intelligent transplanter based on the adaptability information of the soil penetration depth and the posture stability of the seedling, and to generate an execution control sequence including the pressing depth adjustment command, the seedling straightening command and the seedling hole covering and compaction command. The execution module 403 in this application is mainly used to adaptively correct the transplanting action of the transplanter based on the execution control sequence and the real-time feedback of soil resistance fluctuations during the transplanting process.

[0056] The foregoing has detailed examples of the fully automatic intelligent transplanter and method provided in the embodiments of this application. It is understood that, in order to achieve the above functions, the corresponding device includes hardware structures and / or software modules corresponding to the execution of each function. Those skilled in the art should readily recognize that, based on the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed by hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0057] In some embodiments, this application also provides a computer device, the computer device including a memory and a processor, the memory for storing a computer program, and the processor for calling and running the computer program from the memory, so that the computer device performs the control method of the fully automatic intelligent transplanter described above.

[0058] In some embodiments, reference Figure 5 The dashed lines in the figure indicate that the unit or module is optional. This figure is a structural schematic diagram of the computer device implementing the control method of the fully automatic intelligent transplanter of this application. The control method of the fully automatic intelligent transplanter in the above embodiments can be achieved through… Figure 5 The computer device 500 shown is used to implement this, and the computer device 500 includes at least one processor 501, a memory 502 and at least one communication unit 505. The computer device 500 may be a terminal device, a server or a chip.

[0059] The processor 501 can be a general-purpose processor or a special-purpose processor. For example, the processor 501 can be a central processing unit (CPU). The CPU can be used to control the computer device 500, execute software programs, and process data from the software programs. The computer device 500 may also include a communication unit 505 for inputting (receiving) and outputting (transmitting) signals.

[0060] For example, computer device 500 may be a chip, communication unit 505 may be the input and / or output circuit of the chip, or communication unit 505 may be the communication interface of the chip, and the chip may be a component of terminal device, network device or other device.

[0061] For example, computer device 500 may be a terminal device or a server, and communication unit 505 may be a transceiver of the terminal device or the server, or communication unit 505 may be a transceiver circuit of the terminal device or the server.

[0062] The computer device 500 may include one or more memories 502 storing a program 504. The program 504 can be executed by a processor 501 to generate instructions 503, causing the processor 501 to perform the methods described in the above method embodiments according to the instructions 503. Optionally, the memory 502 may also store data (such as a target audit model). Optionally, the processor 501 may also read data stored in the memory 502, which may be stored at the same storage address as the program 504, or the data may be stored at a different storage address than the program 504.

[0063] The processor 501 and memory 502 can be configured separately or integrated together, for example, integrated on the system-on-chip (SOC) of the terminal device.

[0064] It should be understood that each step of the above method embodiment can be completed by hardware logic circuits or software instructions in the processor 501. The processor 501 can be a CPU, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, such as discrete gates, transistor logic devices, or discrete hardware components.

[0065] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0066] For example, in some embodiments, this application also provides a computer-readable storage medium storing instructions or code that, when executed on a computer, cause the computer to implement the control method of the fully automatic intelligent transplanter described above.

[0067] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0068] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A control method for a fully automatic intelligent transplanter, characterized in that, Includes the following steps: Obtain surface environmental data of the target transplant site; Based on the surface environment data, the suitability of the target transplanting plots is evaluated, and target planting points including areas with loose structure, stable humidity, and safe surface slope are selected. Then, the spatial distribution characteristics of all target planting points are used to generate transplanting paths suitable for the mechanical operation conditions of the target transplanting plots. As the intelligent transplanter moves along the transplanting path, multimodal sensing data is collected in real time during the operation. Based on the multimodal sensing data, the adaptability information of the soil penetration depth of the intelligent transplanter's transplanting actuator and the posture stability of the seedling during the transplanting process are identified. The robot arm execution path of the intelligent transplanter is corrected in real time by using the adaptive information of the soil penetration depth and the posture stability of the seedling, and an execution control sequence including the pressing depth adjustment command, the seedling straightening command, and the seedling hole covering and compaction command is generated. Based on the execution control sequence, and according to the real-time feedback of soil resistance fluctuations during the transplanting process, the transplanting action of the transplanter is adaptively corrected.

2. The method as described in claim 1, characterized in that, Based on the aforementioned surface environmental data, the suitability of the target transplanting sites was evaluated, and target planting sites were selected, including areas with loose soil structure, stable humidity, and safe surface slope. Specifically, these included: Key evaluation parameters characterizing soil structure, moisture properties, and topographic slope are extracted from the surface environment data. The key evaluation parameters are weighted and fused to obtain a comprehensive plantability score. Based on the comprehensive suitability score and preset threshold, target planting sites were selected, including areas with loose structure, stable humidity, and safe surface slope.

3. The method as described in claim 1, characterized in that, The transplanting path, which generates suitable mechanical operation conditions for the target transplanting plots based on the spatial distribution characteristics of all target planting sites, specifically includes: Obtain the coordinates and region type attributes of all target planting points, and then determine the spatial distribution characteristics of all target planting points; Based on the spatial distribution characteristics, an initial operation path covering all target planting points is generated; The initial working path is smoothed and optimized based on the minimum turning radius and maximum climbing angle of the intelligent transplanter to generate a transplanting path suitable for the mechanical operating conditions of the target transplanting plot.

4. The method as described in claim 1, characterized in that, The adaptive information of the soil penetration depth of the transplanting actuator of the intelligent transplanter and the posture stability of the seedling during the transplanting process, based on the multimodal sensing data, specifically include: The soil penetration depth feature parameters and seedling posture motion parameters are extracted from the multimodal sensing data, respectively. The soil penetration depth characteristic parameters are compared with the preset ideal depth range to obtain the adaptability information of the soil penetration depth of the transplanting actuator of the intelligent transplanter; The stability of the seedling's posture during transplanting is output based on the seedling's posture motion parameters.

5. The method as described in claim 1, characterized in that, The intelligent transplanter's robotic arm execution path is instantly corrected based on the adaptive information of the soil penetration depth and the posture stability of the seedling. This generates an execution control sequence that includes commands for adjusting the pressing depth, straightening the seedling, and covering and compacting the seedling hole. Specifically, this sequence includes: Based on the adaptability information of the soil penetration depth, the pressing path of the intelligent transplanter's robotic arm is adjusted, and a pressing depth adjustment command is generated. Based on the posture stability of the seedling, the gripper path of the robotic arm is straightened and corrected to generate a seedling straightening command for controlling the gripper of the robotic arm. After the depth adjustment and straightening operations are completed, a seedling hole covering and compaction command is generated based on the current seedling hole status; The pressing depth adjustment command, the seedling uprighting command, and the seedling hole covering and compaction command are integrated into an execution control sequence according to the operation sequence.

6. The method as described in claim 1, characterized in that, Based on the aforementioned execution control sequence, and according to the real-time feedback of soil resistance fluctuations during the transplanting process, the transplanting action of the transplanter is adaptively corrected, specifically including: The soil resistance data during the transplanting process is monitored and collected in real time by force sensors installed on the transplanting actuator; The collected soil resistance data were analyzed for fluctuation characteristics, and the fluctuation of soil resistance was determined by the analyzed anomaly changes in resistance. Based on the soil resistance fluctuations, the relevant parameters in the execution control sequence are dynamically adjusted to complete the adaptive correction of the transplanting action.

7. The method as described in claim 1, characterized in that, The multimodal sensing data refers to a fusion data set that characterizes the soil entry state of the transplanting actuator and the dynamic posture of the seedling during the transplanting operation of the intelligent transplanter.

8. A fully automatic intelligent transplanter, comprising a control unit, characterized in that, The control unit includes: The acquisition module is used to acquire surface environmental data of the target transplant site; The processing module is used to evaluate the suitability of the target transplanting plots based on the surface environment data, screen out the target planting points including areas with loose structure, areas with stable humidity and areas with safe surface slope, and generate a transplanting path suitable for the mechanical operation conditions of the target transplanting plots based on the spatial distribution characteristics of all target planting points. The processing module is also used to collect multimodal perception data in real time during the operation of the intelligent transplanter as it moves along the transplanting path, and to identify the adaptability information of the soil penetration depth of the intelligent transplanter's transplanting actuator and the posture stability of the seedling during the transplanting process based on the multimodal perception data. The processing module is also used to make real-time corrections to the execution path of the robotic arm of the intelligent transplanter based on the adaptability information of the soil penetration depth and the posture stability of the seedling, and to generate an execution control sequence that includes instructions for adjusting the pressing depth, instructions for straightening the seedling, and instructions for covering and compacting the seedling hole. The execution module is used to adaptively correct the transplanting action of the transplanter based on the execution control sequence and the real-time feedback of soil resistance fluctuations during the transplanting process.

9. A computer device, characterized in that, The computer device includes a memory and a processor. The memory is used to store computer programs, and the processor is used to call and run the computer programs from the memory, so that the computer device performs the control method of the fully automatic intelligent transplanter according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions or code that, when executed on a computer, cause the computer to implement the control method for the fully automatic intelligent transplanter as described in any one of claims 1 to 7.

Images