基於ROS之足球機器人的模糊行為決策設計

ROS-based Fuzzy Behavior Decision Design for Soccer Robot

黃聖博(Sheng-Po Huang)

指導教授 ： 李世安

淡江大學/工學院/電機工程學系碩士班/碩士(2017年)

https://doi.org/10.6846/TKU.2017.00333

摘要

本論文提出一機器人模糊行為決策系統來解決機器人的二元策略樹之判斷方法。使用模糊行為決策系統來取代二元樹的策略判斷，可以讓策略端在做決策時能更圓滑、有彈性，並且提高足球機器人進攻的效率。本論文以FIRA(Federation of International Robot-soccer Association)中RoboSot組的足球機器人競賽為研究平台。本論文將足球機器人模糊行為決策系統實現於ROS(Robot Operation System)內的Gazebo模擬器，並設計四種足球機器人的進攻策略，讓我方機器人能在進攻時，可以躲避敵方機器人，降低被抄球的機率。此四種策略配合本論文提出之模糊行為決策系統，將場地資訊輸入到模糊分類器內運算後會決定出在機器人目前的狀態下最佳的進攻策略。在最後的實驗結果中，本論文將有使用模糊行為決策系統與只使用有限狀態機之行為決策系統及使用單一策略之系統進行對戰。除此之外也將此系統使用在實際機器人中。從對戰結果可證實本方法能使機器人更貼近專家之想法，並增加足球比賽的勝率。

關鍵字

機器人作業系統 ； 足球機器人 ； 行為決策 ； 模糊分類器 ； Gazebo模擬器

並列摘要

This paper proposes a behavior decision fuzzy system to solve the behavior decision based on binary tree method. Using behavior decision fuzzy system to replace binary tree can make the behavior decision smoother and flexible in strategy side, also improve the efficiency when soccer robot is attacking. This paper is based on middle-size robots with the competition of FIRA (Federation of International Robot-soccer Association) RoboSot. Moreover, this study implement the behavior decision fuzzy system in Robot Operating System (ROS) and Gazebo simulator. We design four attack strategies in this behavior decision fuzzy system, so that our robot can avoid opponent and reduce the probability of steals by opponent when robot is attacking. Using these strategies and inputting the information of the soccer field to the behavior decision fuzzy system will determine the optimal attack strategy which in present situation. In the experimental results, we execute the program which is using behavior decision fuzzy system to battle with using finite state machine (FSM) system and four strategies respectively. In addition, we also use in the actual robot and playing on real competition. From the results of the battle, we can prove this system let behavior closer to the thinking of experts and increase the winning percentage of soccer game.

並列關鍵字

Robot Operating System ； ROS ； Soccer Robot ； Behavior Decision ； Fuzzy Classifier ； Gazebo Simulator

參考文獻

[1] 林怡女匀，基於PSO模糊分類器於多目標色彩模型的設計，淡江大學電機工程研究所碩士論文(指導教授:李世安)，2013。

[3] 游翔麟，基於模糊力量感測之六軸機械手臂的直覺式教導，淡江大學電機工程研究所碩士論文(指導教授:翁慶昌)，2015。

[2] 黃文鴻，於ROS之地圖建置與探索系統設計，淡江大學電機工程研究所碩士論文(指導教授:李世安)，2016。

[11] F. Michaud, “Selecting behaviors using fuzzy logic,” IEEE International Conference on Fuzzy Systems, vol. 1, pp. 585-592, 1997.

[14] J. Maitin-Shepard, M. Cusumano-Towner, J. Lei, and P. Abbeel, “Cloth grasp point detection based on multiple-view geometric cues with application to robotic towel folding,” 2010 IEEE International Conference on Robotics and Automation (ICRA), pp. 2308-2315, 2010.