| Autonomous Unmanned Aerial Vehicles(UAVs)are increasingly popular as efficient agricultural pesticide spraying devices.These vehicles offer several advantages,including improved safety,energy efficiency,and adaptability in plant protection mechanization.During the spraying operation,plant protection UAVs encounter non-movable obstacles that threaten flight safety and carry a substantial liquid load,which can lead to sloshing during sudden velocity transitions.Currently,many of these UAVs plant protection UAVs rely on single-type environmental sensors for detection and navigate around obstacles by following preset paths.Some use advanced visual sensors for these operations,which can be further hindered by fog clouds caused by moisture and spray droplets and can be expensive.Also,existing obstacle avoidance methods for UAVs often focus on avoiding obstacles by targeting the next way-point,potentially resulting in missed spray areas in the target zone.Furthermore,there are no solutions for mitigating liquid sloshing inside the pesticide tank for existing plant protection UAVs.These challenges,such as the lack of practical obstacle avoidance methods and the presence of liquid sloshing,continue to affect flight stability and lead to non-uniform spray droplet coverage during autonomous spray operations,and unstable liquid slosh during autonomous spray operations,compromising the effectiveness of pesticide application.To enhance the autonomous operation capability and the effectiveness of plant protection operations for UAVs in obstacle-filled environments and to reduce liquid slosh inside the tank,there is a need for a real-time obstacle avoidance method and a solution for liquid slosh for existing tanks and future construction.This thesis proposes a novel obstacle avoidance system,a physical solution for slosh reduction,and a flight stability method to improve the autonomous operation capabilities and the effective spraying of plant protection UAVs in obstacle-filled environments while reducing liquid sloshing.The key contributions and findings of this research are:(1)To develop a novel dynamic data-driven feedback-controlled OA method for liquid-carrying plant protection quadcopter UAVs.This method uses a multisensory architecture of millimeter-wave radar and single-point laser sensors to avoid obstacles.Unlike previous methods that focused on yaw-based forward-moving OA without addressing coverage concerns,the proposed method simplifies the obstacle shape by identifying three outer points of the object and considers parameters such as liquid weight and operational velocity for dynamic avoidance maneuvers.The algorithm determines avoidance velocity and triggers transitional changes based on changing mass,velocity,and obstacle information.The proposed OA method aims to avoid obstacles at safer distances while reducing avoidance time.Extensive evaluations were performed under various obstacle positions and Monte Carlo analysis.The results obtained from the simulations demonstrate the effectiveness of the proposed method in reducing flight time,altitude error,and avoidance distance for both fixed and varied obstacle positions.In particular,for the fixed-position obstacle avoidance case,the algorithm achieved an average time reduction of 23.75%and a 117-127.1 cm y-axis displacement during a maximum maneuver time of 43.3 seconds,along with an average altitude error of 10.9 cm at a3 m flying altitude for 10%～100%liquid load.When the obstacle positions varied in spray mission mode,the method achieved an average coverage of 98.77%,98.46%,and 98.15%for4 m,3 m,and 2 m path gap missions,respectively,in a 400-meter square mission area,with an average altitude loss of 1.66 cm.Thus,the proposed algorithm demonstrated dynamic suitability for liquid-carrying plant protection UAVs and scalability based on selected physical parameters.(2)To validate the performance of the developed data-driven feedback control OA system for plant protection UAVs.While simulations have validated the method’s theoretical effectiveness,physical disturbances such as liquid slosh and vehicle vibration can affect its performance during low-altitude operations.To address this issue,the next objective was to conduct field experiments to demonstrate the flight variance of the method and identify potential shortcomings.The experiment involved using a parallelepiped obstacle to examine three avoidance parameters:minimum distance,mean distance deviation,and avoidance time.The feedback controller was designed as a two-layer control system based on an 8-bit airborne microcontroller(feedback generator)and Pixhawk flight controller(flight controller).The UAV successfully detected the obstacle,planned appropriate paths,and adjusted its trajectory based on real-time sensor data,achieving deviations ranging from 1.3 to 2.71 meters during obstacle avoidance.The actual avoidance time for 0%,50%,and 100%liquid loads were 7.11,7.37,and 8.11 seconds,respectively.Therefore,the experiment validates the practical applicability and dynamic behavior of the proposed feedback-controlled OA method in real-world agricultural settings,contributing to a more efficient and safer plant protection UAV system.(3)To conduct a field experiment to evaluate the droplet coverage of the proposed OA method by comparing it with existing and new methods using coverage parameters such as deposition density,droplet coverage,and droplet distribution.Deposit Scan software was used to quantify spray deposition,deposits,coverage,and droplet diameter.The results showed that the proposed data-driven feedback control method was the most effective in delivering spray material to the target zone,outperforming other methods in terms of spray deposition and coverage percentage.Specifically,the data-driven OA method achieved the highest spray deposition values with 16.5μL/cm~2 and 15.5μL/cm~2,respectively,which were significantly higher than the values for the Ellipsoidal bounding box approach(10.4μL/cm~2 and 15.5μL/cm~2),Sliding circle approach(6.3μL/cm~2 and 7.9μL/cm~2),and Potential field approach(3.1μL/cm~2 and 4.7μL/cm~2).Additionally,the data-driven OA method had the highest coverage percentage(Cov D)in the center column at 43.2%,surpassing the percentages of the Ellipsoidal bounding box approach(32.1%),TA3(29.7%),and Sliding circle approach(27.2%).Overall,the data-driven obstacle avoidance method demonstrated superior spray coverage in the target zone.(4)To find a physical solution to minimize the impact of sloshing inside the liquid tank of the sprayer UAV,especially during challenging situations such as wind gusts and OA maneuvers.A physical form of baffle ball structure is proposed to minimize the slosh effect for the sprayer UAV.The main objective is to conduct comprehensive experimental analyses of the impact of liquid sloshing on the dynamics and performance of the sprayer UAV using different tank configurations.Through the experimental analyses,a novel baffle ball design that serves as a universal solution for mitigating liquid sloshing is proposed.Specifically,the experimental results revealed that the flat-hexagonal with baffle ball configuration,featuring a hexagonal tank with baffle ball,exhibits the highest damping ratio(0.06175)and energy dissipation values(0.00965),making it the most effective configuration for mitigating sloshing effects compared to rectangular and cylindrical configurations.These findings not only aim to provide valuable insights for improving the performance of liquid carrier UAVs but also contribute significantly to the advancement of precision agriculture applications by ensuring more stable and reliable UAV operations.(5)To develop an adaptive control scheme for stabilizing the UAV’s flight in the presence of sloshing effects and external disturbances.An adaptive PI-based sliding mode controller was designed to stabilize the sprayer UAV using an error coordinate representation of the UAV system.The controller consisted of a neural network identifier for unknown sloshing forces and a nonlinear disturbance observer for external disturbances.Lyapunov stability analysis was used to ensure system robustness,and a comparative analysis with the Integral Absolute Error(IAE)method showed that the proposed controller outperformed the traditional PID controller,with a total IAE of 11.8 compared to 20.55.The adaptive robust controller demonstrated more efficient,robust control systems,enhancing performance and reliability in agricultural UAV configurations.Overall,this thesis proposes novel strategies to enhance the performance of UAVs used for agricultural pesticide spraying.The thesis identifies the lack of effective OA method and liquid sloshing as significant challenges to flight stability,resulting in non-uniform spray droplet coverage and unstable flight.The proposed solutions include a data-driven feedback control obstacle avoidance method,field experiments to optimize droplet distribution coverage,and a comprehensive analysis to address liquid disturbance inside the tank caused by sloshing.The proposed solutions were found to be effective in delivering spray material to the target zone,outperforming existing and newly developed methods in terms of spray deposition and coverage percentage.Consequently,this research holds substantial industrial and academic application value.Future research and industrial development can benefit from these innovative ideas and research findings. |
PDF Full Text Request