陳美勇Chen, Mei-Yung鍾秉剛Jhong, Bing-Gang2024-12-172026-12-082023https://etds.lib.ntnu.edu.tw/thesis/detail/09d5a2dfb44837cd2d7ba4141feaeb8f/http://rportal.lib.ntnu.edu.tw/handle/20.500.12235/123089本論文改良機器人控制系統中的演算法，主題涵蓋機器人的運動規劃、定位與控制器設計，藉此提升控制系統的運作效率。在運動規劃領域，我們探討或提出對雙向快速探索隨機樹（BRRT）演算法、A*演算法與hybrid A*演算法的改進措施，並且設計剪枝與平滑算法優化路徑品質，最後搭配梯形速度規劃完成運動規劃工作。在定位方面，在使用特徵地圖的場合採用拓展卡曼濾波器，而在網狀地圖使用改良式蒙地卡羅定位法。此改良式蒙地卡羅定位法由本論文提出，藉由重新設計演算法的權重分配與重新採樣的架構提升演算法的搜尋效率。而在控制器設計方面，我們提出了一種自適應控制器，旨在最小化機器人的預定狀態和當前狀態之間的追蹤誤差。透過我們的機器人控制系統，機器人可以順利地從目前位置導航到指定目標。該系統的性能透過模擬和實驗結果的結合得到證實。This dissertation enhances the robot control system algorithm, addressing motion planning, localization, and controller design to improve overall system efficiency. In motion planning, we propose improvements to the Bidirectional Rapid Exploration Random Tree (BRRT), A*, and hybrid A* algorithms. We design pruning and smoothing algorithms to optimize path quality and implement a trapezoidal velocity profile to finalize motion planning. For localization, we utilize the extended Kalman filter (EKF) with feature maps and introduce an improved Monte Carlo localization (IMCL) method for grid maps. This novel Monte Carlo localization method, introduced in this dissertation, enhances algorithm search efficiency by redesigning weight distribution and resampling structures. In controller design, we introduce an adaptive controller to minimize tracking errors between the predetermined and current states of the robot. Our robot control system enables seamless navigation from the current location to the designated target. The performance is validated through a combination of simulation and experimental results.機器人控制系統路徑規劃蒙地卡羅定位適應性控制robotic control systempath planningMonte Carlo localizationadaptive control學術論文