| Target spraying is an effective technique for reducing pesticide usage and minimizing environmental pollution.It allows for precise and targeted delivery based on the distribution of target areas(pests and diseases).However,the stability of the target spraying system is poor due to issues such as poor target recognition,drug deposition deviation,and uneven distribution of drug deposits,leading to over-dosing in certain areas and resulting in target residue,the evolution of resistance,and decreased crop production.Therefore,based on the specificity of the droplet group and target space-time distribution in the target spraying process,developing a stable and reliable field target spraying strategy and method will contribute to the sustainability of agriculture.This study analyzed comprehensive management measures for target(weeds as an example)both domestically and abroad,taking into account the characteristics of weed space-time distribution.The research objects include weeds,plant protection equipment,and deposited droplets,with the research objectives being weed stability recognition,precise output of operation prescription maps,precise positioning of plant protection equipment nozzles,uniform deposition of droplets,and sustainability of farmland ecosystems.Through systematic conceptual design,numerical simulation,experimental validation and other means,this study seeks the optimal weed detection algorithm and optimizes the design of the target spraying strategy.It clarifies the impact of the structural characteristics of plant protection equipment on nozzle positioning error and establishes an equivalent model of structural coupling error;it explores the factors affecting droplet deposition deviation due to the movement properties of plant protection equipment and establishes a nozzle droplet deposition prediction and control model;it investigates the influence of weed space-time distribution characteristics on pipeline pressure fluctuation and establishes a storage and pressure stabilization pipeline system.Further optimization of equipment field operation parameters is sought to achieve sustainable farmland ecology.The main research content includes:(1)Construction of Target Spraying Model for Air-Ground CooperationIn view of the problems of poor stability of target spraying equipment,high system costs,and short effective operating time,this research analyzes the growth and distribution characteristics of crops and targets(weeds as an example)and finds that the distribution of crops has continuity and the distribution of associated weeds and newly generated weeds has specificity.Based on the development of drone technology and the improvement of the accuracy of the global positioning system,this research proposes the use of drones to obtain near-ground images of farmland,further uses the improved YOLO v5 model,and conducts weed detection using a binary discriminant method to obtain a precise map of weed distribution at the scale of farmland and to verify the map accuracy.The results show that the drone can efficiently obtain information of the farmland image to quickly understand the distribution of farmland weeds,and the accuracy of the improved weed detection method is 98.87%.The location accuracy error average of the dot-based operation prescription map is 0.9 cm,which is enough to support the target spraying operation of a robot aimed at the size level of weeds.(2)Robot Nozzle Position Analysis and CorrectionThis research addresses the problem of low accuracy of target spraying operations in unstructured farmland environments by the robot system.The study adopts a spatial error compensation technique and carries out the calculation and correction of nozzle position based on the large-scale target spraying robot designed.By using an auxiliary coordinate position association method,a nozzle position calculation model is established based on the robot coordinates,orientation,and structural parameters.The error equivalent transformation method is used to quantify the coupled error decomposition terms,and finally,various correction methods are compared and the mean correction method is selected to correct the reconstructed nozzle position calculation model.The results from both the flat field target spraying simulation and the field verification experiments show that the Gaussian regression modeling method can achieve an accurate estimate of the robot structural parameters,with the relative height and relative distance between the nozzle and the reference point having average deviations of 4.3 and 1.3 mm,respectively.The longer the response distance between the nozzle and the target center(the higher the spray resolution),the higher the target spraying accuracy and the better the system stability.When the driving speed is 1 m/s,the samples under the 0,15,and 30 cm target guiding distance have a target spraying accuracy of less than or equal to 30 mm,with 94.4%,96.7%,and 99.4% of the samples,respectively.The coefficient of variation of the target spraying hit probability for samples under 0,15,and 30 cm target guiding distance is 0.010,0.017,and 0.010,respectively.(3)Droplet Deposition Location Estimation and Nozzle Control MethodTo solve the problem of reduced accuracy in large-scale targeted medication due to droplet deposition deviation in plant protection machinery,this study conducted research on the droplet directional deposition control method based on the large-scale targeted medication robot system.Taking the targeted spraying droplet group as the research object,the working principle and operation characteristics of targeted medication were explained,and the causes of droplet group deposition deviation were analyzed.Based on the spatial structure position relationship of the robot,a nozzle motion state perception model and deposition location estimation model were established based on GNSS/IMU Kalman filtering signals,and the nozzle response control criteria were determined.Based on this,flat ground targeted medication simulation experiments were carried out with the operating speed and nozzle control method as the experimental factors,and the accuracy of targeted medication and the deposition deviation distance as the experimental indicators.The experimental results show that the average accuracy of targeted medication on a flat ground surface at operating speeds of 0.5 m/s,1.0 m/s,1.5 m/s,and 2 m/s was 99.8%,98.4%,95.9%,and 76.5%,respectively,and the deposition deviation distances were 3.8 cm,5.4 cm,7.5cm,and 10.0 cm,respectively.The accuracy of targeted medication in the field at operating speeds of 0.5 m/s,1.0 m/s,and 1.5 m/s was 98.7%,96.7%,and 95.3%,respectively.The results show that the droplet deposition location estimation and nozzle control method based on GNSS/IMU Kalman filtering meets the requirements of large-scale targeted medication operations.(4)Mechanism of Multi-Working Condition Droplet Deposition Deviation Error and Compensation MethodThis research aims to enhance the multi-condition use requirements of the target spraying robot in terms of operating speed intervals and spray head height intervals,while taking into account the accuracy of target spraying and the reliability of system deployment.Based on the research findings of Item 2 and Item 3,this study further analyzes the effect of position errors on droplet deposition under different operating conditions in the target spraying process and establishes a compensation model based on the length of the suspension system electric cylinder and the speed of the robot’s movement.Additionally,a spray head pose correction calculation model that integrates static structural errors and dynamic offset errors is developed to improve target spraying accuracy.After the experiment,drone images were used to evaluate droplet deposition and drift.The results of the experiment show that different correction methods were tested in 12 experiments,with operating speeds of 0.5 m/s,1.0 m/s,and 1.5 m/s.After completion of the operation,a pixel-based discrimination statistical method was used to evaluate the spraying effect,with a target spraying accuracy of 95.9-98.8% and droplet deposition offset distances of 3.7cm to 7.2 cm,meeting the target spraying needs of different operating speed intervals.The results of the field test indicate that the weed control rate based on the individual scale of the plant(≤15cm)is 94.2-98.1%.These studies demonstrate that the developed target spraying robot system can perform target spraying operations on an individual scale of the plant,and by compensating for random errors,it improves the accuracy of the target spraying operation.(5)Pipeline Pressure Fluctuation Mechanism and Suppression Method in Target SprayingEnsuring the uniformity of target spraying droplet deposition is a prerequisite for obtaining higher economic benefits and environmental benefits,especially for complexly distributed multi-sized targets that are difficult to ensure uniform droplet deposition.The purpose of this research is to design a pipeline system that can improve the uniformity of target spraying droplet deposition and apply it to large field target spraying operations.The droplet deposition effect is evaluated using a "+" scale plate equipped with water-sensitive paper.By analyzing the distribution characteristics of weeds,the spraying process of the nozzle can be indirectly understood.Through monitoring the pressure of the pipeline system,it is indicated that the fundamental cause of the uneven spatial distribution of target spraying droplet deposition is the pressure fluctuation of the pipeline system,and the use of storage components can alleviate this phenomenon.Compared with the original pipeline system,the newly designed pipeline system reduces the range of pipeline pressure changes,and the use of diameter-thicker average pressure pipes reduces the droplet deposition variability coefficient by 6.5% to 11.5%.The use of a storage tank in the pipeline combination reduces the droplet deposition variability coefficient by 0.1% to5.2% compared to the pipeline combination without a storage tank.The use of these two components increases the uniformity of droplet deposition,proving that the system is capable of handling different sized target objects,avoiding excessive or insufficient application in the target area,although there are still differences compared to indiscriminate spraying effects.This research provides new research directions and methods for further improving the uniformity of target spraying droplet deposition.(6)Comprehensive Evaluation of Joint Air-Ground Target Spraying Mode in Agricultural FieldsThe purpose of this research is to clarify the performance and optimization potential of the joint air-ground target spraying system in agricultural fields.Through the analysis of the multi-source error control architecture of the target spraying system,it is confirmed that the system conforms to the error transfer characteristics of a serial-parallel hybrid system.The information links of UAV and UGV are optimized separately,leading to the generation of precise point prescription maps even without network signals.The robot carried out target spraying at a high speed,with an average accuracy rate of 94.6% and an average deviation distance of 6.5cm when traveling at 15 km/h.The results of field experiments(item 4)are used as the basis for long-term tracking of the target spraying performance in large agricultural fields.The experiment factors are the target spraying resolution(15×6 cm,15×15 cm,30×30 cm,45×45 cm,indiscriminate,blank control),and the dynamic evaluation indicators are plant height,stem diameter,and the width of the first expanded leaf in the crop-weed ecological system.The final evaluation indicator is crop yield.The long-term dynamic tracking of target spraying is carried out to understand the impact of drifting droplets on the crop-weed ecological system during the target spraying process.The results show that the higher the target spraying resolution,the smaller the impact on crop production.Compared to indiscriminate spraying,the use of the above-mentioned target spraying resolutions increased corn yields by 10.4%,4.3%,2.2% and 0.5%,respectively.However,when the target spraying resolution is 45 cm,the advantage of the target spraying operation on crop yield is no longer significant.The high-resolution target spraying system maintains the biodiversity of the agricultural ecological system during the later stage of crop growth and has the potential to reduce the weed occurrence base in the following year’s crop production,but this needs further proof. |
