- 會議論文
室內四軸無人機之控制操略分析
Analysis of Various Control Strategies for Indoor Quadcopter Applications
摘要
1.研究背景：四軸無人機具備高機動性、操作簡易等特性，非常適合應用於室內飛行。室內四軸無人機可應用於工廠內檢修、精密量測等；相較於戶外四軸無人機，室內干擾相對較少，但需要更精確的控制，而要完成精確控制的任務必須建立在良好的控制器上。2.研究目的：室內四軸無人機在狹小空間內具備高機動性與穩定性等優勢，為符合其作業環境中的低容錯率，本研究使用Proportional-Integral and Derivative（PID）、模糊邏輯控制器（Fuzzy Logic Controller, FLC）、近端策略優化（Proximal Policy Optimization, PPO）控制器，分析不同控制器之暫態及穩態響應，並選出最合適之控制系統。3.研究方法：本研究使用鸚鵡四軸無人機（Parrot Mambo mini drone）做為模擬之室內四軸無人機，模擬環境及控制器建置則是使用MATLAB及Simulink，以高度控制器做為研究之控制器。先使用Ziegler-Nichols（ZN）方法獲取一組PID數值，經由系統式優化該PID控制器，再根據飛行條件創建FLC，最後基於PPO算法設計出PPO高度控制器。藉由暫態、穩態響應與IAE等相關數據對三種控制器在目標高度為1公尺時的飛行狀態進行分析。4.研究結果：經研究後ZN-PID有過衝及無法收斂等問題；FLC雖穩定並收斂，但暫態時間過長；PPO控制器在上升、暫態及穩態時間都有明顯的改善，且相較ZN-PID過衝問題有明顯改善。5.研究結論：比較三種控制器後，本研究所開發之PPO控制器其暫態及穩態為三種控制器中最優秀，因此比起傳統PID及FLC，室內四軸無人載具更適合使用PPO控制器作為高度控制系統。
並列摘要
1. Background: Quadcopters are highly maneuverable and operable, making them suitable for flying in indoor environment. The indoor quadcopters can be utilized for the factory maintenance, precision measurements, and other applications. Less interference is incurred with indoor quadcopters as compared with that of the outdoor quadcopters relatively. However, the indoor one requires more precise control relying on a well-designed controller and strategy. 2. Purpose: Indoor quadcopters with smaller size have benefits in small spaces and are able to meet low-tolerance requirements when operating in indoor environments. This study adopts Proportional-Integral and Derivative (PID) controller, Fuzzy Logic Controller (FLC), and Proximal Policy Optimization (PPO) controller to analyze the transient and steady-state responses according to different controllers, and then determine the most suitable control strategy. 3. Methods: In this study, a commercial Parrot Mambo mini drone is imitated as a mathematic model to be simulated as an indoor quadcopter. MATLAB and Simulink are used to build and simulate the mathematical model and the corresponding test environment for various design of control strategies. The controller design for altitude control is our main purpose in this study. A Ziegler-Nichols (ZN) method provides a set of PID values and then optimizes the PID controller with fine manual adjustments. Furthermore, a FLC is generated based on the flight conditions. In addition, an PPO altitude controller is designed based on the PPO algorithm. The results of performance of three controllers at a target and desired height of one meter is analyzed the transient and steady-state responses. 4. Results: We found the ZN-PID has problems of overshoot and inability, which results in a diverges altitude control. Although the FLC showed its performance is stable and convergent, the transient and steady-state time were too long. The PPO controller showed significant improvements in the ascent, transient, and steady-state times. In addition, the PPO controller had significant improvements in alleviating the overshoot problem of ZN-PID. 5. Conclusions: After comparing the three controllers, the PPO controller designed for the case in this study performs the best transient and stable responses among the other three controllers.