研究資料首頁-> 研討會論文
研究資料明細
論文名稱 | 四足類生化機器動物之建構及視覺伺服控制 |
研討會開始日期 | 2006-11-05 |
研討會結束日期 | 2006-11-05 |
所有作者 | 蔡清元*;謝東利 |
作者順序 | 第二作者 |
通訊作者 | 否 |
研討會名稱 | 中國機械工程學會第二十三屆全國學術研討會 |
是否具有對外公開徵稿及審稿制度 | 是 |
研討會舉行之國家 | NATTWN-中華民國 |
研討會舉行之城市 | 台南市崑山科技大學 |
發表年份 | 2011 |
所屬計劃案 | 無 |
可公開文檔 |
![]() |
可公開文檔 |
![]() |
可公開文檔 |
![]() |
附件 |
![]() ![]() ![]() ![]() |
[摘要] :
隨著電子科技與控制技術之快速發展,近十年來各種類生化機器動物應運而生。由於其應用範圍包括娛樂、生活援助、家庭、災害救援等領域,因此吸引了各國研究人員積極地投入研究。本論文的目的在於設計ㄧ隻四足類生化機器動物與發展其利用視覺導引去咬拾目標物體的控制策略。
在此所建構的四足類生化機器動物的特徵是每一隻腳有3個自由度,頭部也有3個自由度和1個下顎的自由度,以及ㄧ個嵌入在頭上的CCD攝影機。四足機器動物在執行咬拾的操作時分有兩個階段。第一個階段是機器動物往目標物體做趨進的動作,直到抵達參考位置時停止動作,然後進行第二個階段,機器動物緊鄰物體,並執行咬拾的動作。本論文並針對此一機器動物,提出以行為為基礎之看而後動的視覺導引控制策略。在第一個階段中此控制策略使用三個預先定義的影像特徵,而在第二個階段中,則使用六個預先定義的影像特徵。每一影像特徵透過模糊規則直覺式地決定相對於攝影機座標的一個自由度運動,此即所設計之行為。這些行為經過整合便可執行四足機器動物之趨進及咬拾任務。
[英文摘要] :
Abstract
With advances in electronics and control tecnology, bio-like robots have been designed in the recent decade. They have been increasingly investigated by many researchers, and applied extensively in many fields, such as entertainment, life service, household chores and calamity rescue. The objective of this thesis is to create a quadruped bio-like robot and develop its control strategy for visually guided gripping of material.
The quadruped bio-like robot constructed herein features 3 degrees of freedom for each leg, 3 for the head and another 1 for the mouth, as well as a head-mounted CCD camera. In the execution of gripping operations, the robot moves in two stages. In the first stage of movement, the robot walks toward the material and gets close to it. Then, in the second stage of movement, the robot stands next to the material and continues to perform the gripping manipulation. A vision-guided control strategy with a behavior-based look-and-move structure is proposed. The strategy is based on three predefined image features in the first stage and six predefined image features in the second stage. In the designed neural fuzzy controllers, each image feature is taken to generate intuitively one DOF motion command relative to the camera coordinate frame using fuzzy rules, which define a specific visual behavior. These behaviors are then combined and executed in turns to perform approaching and gripping tasks.