本篇論文主要是研究四足機器人在行走時，因為受到外在因素等影響，導致行走的軌跡有不規則的方向偏移時，如何動態的偵測出它的偏移方向。我們以SONY公司開發的娛樂用機器狗AIBO為實驗平台，提出一個有效的方法，僅需要利用AIBO本身低解析度的感測器及有限的運算能力即可達成有效的偵測結果，也因此很容易應用於其他的四足機器人身上。
我們對步行中感測器傳回的加速度值藉由Matlab進行處理和分析，並且透過觀察AIBO行走時的步態，比較兩者之間的關係，提出一個理想的單步偵測偏移演算法。這個演算法是以步態為單位，去偵測每步的方向偏移。另外我們也提出了多步伐的偵測偏移策略，將每步偏移加總起來，用以改善步與步之間的誤差。透過實驗，我們驗證了依照本篇的方法，可以偵測五步內的方向偏移，而且單步的正確率可高達90%以上。

In this thesis, we proposed an effective approach to detect the directional bias of walking four-legged robots. We found that four-legged robots tend to have unstable directional bias while walking, thus this approach must be able to dynamically and continuously detect the directional bias. We use AIBO, a four-legged robot made by SONY, as the experiment platform. One of the advantage of our method is that it only needs to use low sensitive sensors and limited computing power, so it is considered easy to apply the method to other four-legged robots.

Our approach is developed by analyzing the acceleration sensor data from AIBO robot with Matlab software, and by observing the feature of AIBO`s walking gaits. By utilizing the relation between the accelerometer data and the gaits, we developed an ideal gait based approach to detect directional bias. Also we extend this method to a multi-gait detecting strategy which is used to compensate the bias between every gait. In order to make sure the correctness of our approach, several experiments were designed and executed. As a result, we can correctly detect the direction of bias within five gaits, and the correct rate of single gait can be up to 90%.

[1] Anton Wijbenga ,Mart van de Sanden ,“How To Make AIBO Do Tricks”, University of Groningen,Netherlands,2004.
[2] Bernhard Hengst,Darren Ibbotson,Son Bao Pham,Claude Sammut “The UNSW United 2000 Sony Legged Robot Software System”, University of New South Wales，Australia,2000
[3] Douglas Vail ,Manuela Veloso, “Learning from Accelerometer Data on a Legged Robot”, Computer Science Department,Carnegie Mellon University,2004.
[4] Franois Serra, Jean-Christophe Baillie,“Aibo programming Using OPEN-R SDK. Tutorial", ENSTA, France ,2003.
[5] Francisco Martín Rico, Rafaela González-Careaga, Jose María Canas Plaza,Vicente Matellán Olivera,“Programming Model Based on Concurrent Objects for the AIBO Robot”, Universidad Rey Juan Carlos, 2004.
[6] George A. Bekey, “Autonomous Robots From Biological Inspiration fo Implementation and Control”,The MIT Press , Cambridge, Massachusetts, London, England,2005.
[7] H. Liu, G. Pang, “Accelerometer for Mobile Robot Positioning,” Proc. Of the 1999 IEEE Industry Applications Society Conference, Oct 3~7, 1999.
[8] J. P. Schmiedeler, K. J. Waldron, “The Mechanics of Quadrupedal Galloping and the Future of Legged Vehicles”, The International Journal of Robotics Research, Dec 1999, ,Vol.18, No.12 pp.1224-1234
[9] Makoto Toyomasu , Ayumi Shinohara, “DEVELOPING DYNAMIC GAITS FOR FOUR LEGGED ROBOTS”, The International Symposium on Information Science and Electrical Engineering 2003, 2003, pp. 577-580
[10] Seong Yun Cho, Chan Gook Park and Gyu In Jee,“Measurement System of Walking Distance Using Low-cost Accelerometers”,The 4th Asian Control Conference,Singapore,Sep 2002,pp.1799-1803.
[11] Sony Corporation, “OPEN-R SDK”,2004.
[12] Sven Westermark, “Programming tools for AIBO”,2005.
[13] Thomas B.Sheridan, “Telerobotics,Automation,and Human Supervisory Control”,London,England,1992.
[14] 王亮元, “希伯特轉換於時變系統識別之應用”,國立成功大學土木工程研究所碩士論文,2003.
[15] 李青峰, “四足步行機器人之設計與動態分析”,國立中央大學機械工程研究所碩士論文,2004.
[16] 林真真,實用統計學,東華書局,台北,2003, pp.44-74.
[17] 林軒立、張祥傑、陳永裕、陳其亮,「微慣性感測元件於運動定位之訊號分析與處理」,微奈米機械系統技術專輯,2005,pp.89-93.
[18] 郭信宏, “從關節角度變化觀察四足機器人之步行動作矯正”, 國立台灣科技大學資訊工程系碩士論文,2005.
[19] 黃坤祥, “GPS虛擬距離與加速儀之卡門濾波整合之定位”,國立台灣科技大學機械工程系碩士論文,2005.