輪型移動機器人之路徑規劃與學習演算法應用於路徑追蹤之模糊控制器設計

Path planning for a wheeled mobile robot and learning algorithms applied to tracking fuzzy controller design.

陳威任(Chen Wei-Jen)

指導教授 ： 陳美勇

國立臺灣師範大學/科技與工程學院/機電工程學系/碩士(2016年)

https://doi.org/10.6345/NTNU202203668

摘要

本研究主要是輪型機器人的路徑規劃與路徑追蹤。在路徑規劃方面，採A*演算法具有最小花費函數搜尋結果之特色生成輪型移動機器人之最短路徑，然而，此種設計方法應用於路徑規劃之結果將會產生大量轉折點以及過於靠近障礙物之路徑，因此，採用了具有局部改變調整特性的B-Spline 曲線來調整A*演算法不適用於輪型移動機器人追跡之區段路徑。在路徑生成之後，接下來即是考慮路徑追蹤的課題。為設計路徑追蹤的控制器，我們必須先了解輪型移動機器人的運動模型才可進行下一步之路徑追蹤控制器設計。路徑追蹤方面採用輪型移動機器人之兩輪速度控制方法來完成目的，本論文採用了模糊控制系統結合類神經網路之適應性模糊類神經網路技術，以此方式可結合模糊控制之定性分析與類神經網路定量分析能力之特色，並具備自我學習調整之能力，最後再加入遺傳演算法進行最佳化設計提高理想之結果。

關鍵字

A*演算法 ； B-Spline曲線 ； 模糊控制 ； 學習演算法

並列摘要

The primary search in this content are path planning and path tracking of wheel mobile robot. For path planning, adopting A * algorithm with the feature of minimum cost function results to design the shortest path of wheeled mobile robots. However, the result of path planning with this method will generate a lot of turning points and too close to the obstacle on the path. Under this situation, then use B-Spline curve which has local change adjustment feature can adjust the segments of A * algorithm where are not suitable for wheeled mobile robot to track. After the path is generated, the next consideration is path tracking. To design tracking controller, we must know kinematics model of wheeled mobile robot, therefore it is required to derived of kinematics model beforehand. For path tracking, it is accomplished the purpose by adjusting the speed of two wheels of the mobile robot, this paper uses adaptive network-based fuzzy inference system technology which combines with fuzzy and neural network, it contains the feature of qualitative analysis capabilities of fuzzy control and quantitative analysis capabilities of neural network, and has ability of self-learning adjustment , then add genetic algorithm to optimize the ideal result.

並列關鍵字

A* algorithm ； B-Spline Curve ； Fuzzy Control ； Learning algorithms

參考文獻

參考文獻

[1]Ak. Kawamura , Byu. Gang, Mi. Uemura and Sa. Kawamura, “Mechanism and control of robotic arm using rotational counterweights,” IEEE International Conference on Robotics and Automation, pp. 2716–2721, May 2015.

[2]Yun. Lou, Yo. Zhang, Ru. Huang, Xin Chen and Ze. Li Sch, “Algorithms for Kinematically Optimal Design of Parallel Manipulators,” IEEE Robotics and Automation Society, pp. 574–584, April 2014.

[3]S. H. Park and S. I. Han, “Robust-tracking Control for Robot Manipilator with Deadzone and Friction Using Backstepping and RFNN Controller,” IET Control Theory and Applications, vol.5, iss. 12, pp. 1397-1417, 2011.

[4]Jin. Yamaguchi, Eiji Soga, Sa. Inoue and Atsuo Takanishi, “ Development of a Bipedal Humanoid Robot-control Method of Whole Body Cooperative Dynamic Biped Walking,” International Conference on Robotics & Automation, pp. 368-374, May 1999.