標題: | 智能型舞伴機器人舞步規劃系統之研究與實作 The Research and Construction of Compliant Gait Planning System of Dance Partner Robot |
作者: | 黃仲緣 Huan, Chung-Yuan 鄭璧瑩 Cheng, Pi-Ying 機械工程學系 |
關鍵字: | 舞伴機器人;Dance Partner Robot |
公開日期: | 2008 |
摘要: | 智能型舞伴機器人舞步規劃系統之研究與實作
研究生:黃仲緣 指導教授:鄭璧瑩博士
國立交通大學機械工程學系碩士班
摘 要
本文研究重點是發展能夠伴隨人類一起跳舞的跳舞機器人”NCiDPR09”;因為人類跳舞時,不同的人舞步大小與速度都不一樣,就算是同一個人也沒有每次都跳一樣,因此機器人必須要有容許舞步的誤差與判斷舞步順序是否正確的能力。目前市面上的機器人大部分已經發展成娛樂型機器人或是幫人類治療精神病的機器人,所以在這種狀態下,機器人與人類的互動會變得非常的重要。
本文提出一個具備即時人機互動與舞步協調模組的全方位移動平台控制技術,建構於具有紅外線感知及全向輪的FESTO公司開發之羅勃提諾(Robotino)機器人平台,搭配圖控程式軟體Robotino View編寫程式,使之達到可與人類搭配一起跳舞。
本研究選用舞步優雅的華爾滋舞步做為探討主題項目,研究華爾滋舞步程式轉換時產生誤差的問題,以及人類與舞伴機器人一起跳華爾滋時的舞伴機能模組之建立。當身體互動時,機器人可以伴隨與適應地配合人類的舞步旋律與動作,過程中人類可能會有舞步順序錯誤或是步法韻律不同的變化,機器人能有效地即時判斷這些錯誤,並立即適當地回應調整舞步,有效地執行舞伴的角色。
本研究所探討的機器人舞伴功能主要可分為兩種:
1.教學功能:將實際的舞步位置轉換成機器人的舞步程式,把程式轉換過程之誤差作改良,應用舞步程式跳出標準的華爾滋舞步,讓機器人可以教導人類跳標準的華爾滋舞步,並給予評分。
2.舞伴功能:機器人系統具有標準的舞步程式,由於增加適應型容錯的舞步修正理論與程式模組,能自行配合舞者的腳步移動,跳舞過程中判斷舞者舞步的位置,配合新研發之學習修正功能做為舞步大小的修正。控制系統擴增並採用8051晶片作處理器,控制介面電路增加機器人上半身的手部擺動機能與自由度,並與舞曲節奏副程式架構結合,讓機器人能順利地配合舞者一起跳華爾滋。本研究所發展的即時人機互動控制方法與應用成果,也可延伸推廣應用於其他愈來愈頻繁有關人機互動為基礎的家用輔助與職場支援等機器人控制系統之發展與應用參考。
關鍵字:跳舞機器人、華爾滋舞步、羅勃提諾機器人、全方位移動平台、全向輪、人機互動。 The Research and Construction of Compliant Gait Planning System of Dance Partner Robot Student:Chung-Yuan Huang Advisor: Dr. Pi-Ying Cheng Department of Mechanical Engineering College of Engineering National Chiao Tung University Abstract This paper focus on the compliant gait planning of the dance partner robot. The developed dance partner system is named as ”NCiDPR09”. Most of the time, a human dance with different partner the dance step size and velocity need to be adjusted. The control of the dance robot should be considered and applied for the integrated compliant motion with partner. Currently, many robots have been developed and applied for entertainment and health care etc. It means that the relationship and interaction between the robot and the human is getting closer. This paper proposes an omnidirectional mobile platform control technique includes human-computer interaction and dance gait planning module, by using infrared sensors and omnidirectional wheel etc. The basic robot platform that adopted in the study is Robotino of FESTO with the command editing software “Robotino View”. The study proposes the “Waltz” as the dance key code, i.e. the music temple, dance gaits are all related to the contents of Waltz. The research topics are divided into two topics ”Teaching mode” and “Dance partner mode”. The detail of the two research topics are expressed as follows. 1. Teaching mode: Tracking a standard dance step of Waltz and converting the location to robot gait program. Adopting some innovative methods developed from the study to improve the precision of the dance gait data. Then, robot can be used to teach the human how to play standard Waltz dance, and can also issue a performance score to learner for reference. 2. Dance partner mode: Turning to “Dance partner mode”, the robot plays as a role of human dancer partner. The robot can follow the master dancer perform a standard Waltz gait under the standard music temple. While, the master dancer will show much different dance step size or now standard music temple, a robot can easy to push or collide or even hurt the master dancer. For establishing a smarter dance partner system, the study has developed the gait adjusting and temple following techniques etc to improve the performance of the dance partner robot. And the NCiDPR09 has also proved itself successfully to be a good dance partner when dancing in Waltz in the study. The achievement of the study can also be applied to improve the human-robot interaction tasks in many other situations. Keywords: Dance Robot, Waltz dance gait, Robotino Robot, Omnidirectional Mobile Platform, Omnidirectional Wheels, Human-Machine Interaction. |
URI: | http://140.113.39.130/cdrfb3/record/nctu/#GT079614513 http://hdl.handle.net/11536/42080 |
Appears in Collections: | Thesis |
Files in This Item:
If it is a zip file, please download the file and unzip it, then open index.html in a browser to view the full text content.