實現自閉症孩童之互動式機器人教學系統

Realization of the Interactive Robotic Education System for Autistic Children

蔡政宏(Cheng-Hung Tsai)

指導教授 ： 徐保羅

國立交通大學/電機學院/電機與控制工程系所/碩士(2011年)

https://doi.org/10.6842/NCTU.2011.00698

摘要

由於自閉症孩童的獨特缺陷，會造成特定的行為與情感問題，目前復健輔助教育的方法包括：遊戲、音樂與藝術等結構性教學。本研究有別於傳統靜態平面教學，透過簡易之網路攝影機，建立起人形機器人的介面，以影像處理技術，得到身體特徵點之姿態擷取；並使用低成本加速度感測器，以無線感測網路ZigBee回傳至電腦，搭配動態規劃演算法，以最長共同子序列演算法(longest common subsequence; LCS)獲得相似度的比率，完成動作辨識的效能。本研究結合遠端電腦核心的處理，機器人不須藉由人為的操控，完成自主式之無線感測系統。

透過整個無線感測整合系統，本研究所設計之軟硬體實現於建立人形機器人與自閉症孩童之間的(1)靜態互動教學、(2)伴隨互動教學、(3)動態互動教學之三項教學，搭配音效等回饋系統，建構起整個迴路互動教學；並將設計之三項教學實現在自閉症孩童的日常教學中。實地測試研究對象為三位自閉症孩童，分別為十二歲男童、十一歲男童與十一歲女童，經實地觀察後，適當調整動作辨識率之門檻值為75%，並量化統計及做質性分析，整合實地測試的結果，由測試前後的差異性，經Likert scale彙整，得到此教學之可行性與自閉症孩童反應的評估。本研究成果驗證所設計之機器人回饋互動教學，能明顯的改善自閉症孩童的專注力與行為能力。

關鍵字

自閉症孩童 ； 人形機器人 ； 影像處理 ； 姿態擷取 ； 加速度計 ； 動作追蹤

並列摘要

Certain distinctive mental defects cause specific behaviors and emotional problems for autistic children. Recently, contemporary rehabilitation education for autistic kids mainly employs games, music, pictures, or virtual reality. In this study, the proposed approach is developed mainly on a humanoid robot interface to capture static gestures of the autistic children by applying image processing technologies. Furthermore, the system is integrated with data measured by the tri-axial accelerometer with the longest common subsequence (LCS) algorithm to determine their dynamic motion. Finally, the developed interactive system integrated with a ZigBee wireless system with a star topology network to communicate with a humanoid robot.

In this thesis, three types of interactive education programs are developed for autistic kids as: (1) the static education for autistic children to imitate the robot, (2) the accompanying education for the robot to imitate the autistic children, and (3) the dynamic education for the autistic children to present a repetitive swinging motion. There were three kids participated in the present experiments: a 12-year boy, an 11-year boy, and an 11-year girl with different levels of the autism. The threshold ratio of motion recognition is suitably adjusted to 75% during the evaluation. Moreover, the Likert scale is adopted to compare the difference between the data before the test and those after test for the kids. Through the experiments, feasibility of the present designed education system has been proven and the developed robotic interactive education system can significantly improve concentration and behavior capacity of autistic children.

並列關鍵字

autistic children ； humanoid robot ； image process ； gesture capture ； accelerometer ； motion tracking

參考文獻

[2] 張豑勻，“模仿自閉症幼兒對其社會性趨向行為成效之研究”。國立台北教育大學特殊教育學系碩士班論文，台北，中華民國97年。

[3] 高玉馨，“比較介入模仿與情感交流活動對提升自閉症幼兒社會性趨向行為之差異”。國立台北教育大學特殊教育學系碩士班論文，台北，中華民國98年。

[5] 劉嫚妮，“應用體感互動遊戲於自閉症兒童認知學習之研究” 國立台北科技大學創新設計研究所碩士論文，台北，中華民國97年。

[6] Meltzoff, A. N. & Moore, M. K, (1994). “Imitation, memory, and the representation of persons.” Infant Behavior and Development, 17, 83-99.

[7] Carpenter. M., Nagell, K., & Tomasello, M. (1998). “Social cognition, joint attention, and communicative competence from 9 to 15 months of age.” Monographs of the Society for Research in Child Development, 63, 1-142