利用電腦視覺與泛晰導航技術作室內環境自動車航行之研究

Abstract

本論文提出了一套基於電腦視覺及泛晰控制理論的導航技術，可讓自動車航行於室內環境中，並且具備避碰障礙物的能力。在實驗中我們利用一可無線遙控且擷取影像的迷你自動車作為研究平台。我們設計了一套人性化、有效而且沒有繁瑣規則與限制的學習方法，能自動分析使用者學習時的操作行為，並自動建立導航路線地圖。在使用此方法得到學習路線資料之後，我們也運用一套完整的導航策略來完成學習路線之航行。此策略主要是建構在泛晰控制導航的技術之下，並且可以搭配學習時記錄的資料，完整地走完所有的學習路線。我們最後亦以成功的學習與導航實驗結果證明本系統的完整性與可行性。

A vision-based fuzzy-guidance approach to autonomous vehicle navigation in indoor environments with a capability of avoiding obstacles appearing in planned paths is proposed. A small vehicle with wireless control and image grabbing capabilities is used as a test bed. A simple learning strategy is designed for flexible and effective non-visual learning of reachable spots in rooms without rules or restrictions. Through this simple learning strategy, a planned path is obtained by analyzing user-driving commands and odometer data. And following the learned path, the vehicle can accomplish specified navigation sessions by proposed collision-avoidance strategies based on the use of fuzzy-control guidance techniques and input images acquired by an onboard vision system. And some strategies for navigation accuracy maintenance are also proposed. Finally, experimental results showing flexibility of the proposed approach for navigations in indoor environments are also included.