室內步行聲源定位

Abstract

目前室內定位技術多半為必須攜帶定位器的主動式系統，只能針對身上有定位識別器的活動者，時常攜帶在身上會造成生活上的麻煩，而被動式的定位系統又常常受到家俱與空間佈置或者其它雜訊的影響，所以本篇論文以改良上述的室內定位缺點為研究目標。 為了解決主動式定位器的困擾，決定採用步行最直接產生的訊號進行利用，於地板所產生的訊號可分為振動與聲音，經過初期研究後判定聲音訊號處理容易、儀器實用性高，因此將研究重點專注於步行聲源定位，不同於傳統麥克風陣列演算法多半以波源幾何方向角度進行定位，於本文中所推導的掃描最佳化法，在有限空間內掃描出誤差函數的最低點來尋找波源位置，巧妙避開了最佳化怕落入局部最小值的隱憂，也確定最佳化的計算量，誤差函數計算的優點允許感測器以散佈式排列進而增加室內家俱擺設的變動性，於文中也探討了感測器佈置方式對定位精準度的影響，進而做出感測器最佳擺法的建議。實驗中除了證實研究理論的可行性，並且找出會影響定位誤差的原因，例如受到生活中其他聲音雜訊干擾或者聲源傳遞路徑改變等。最後於未來展望將分成三個階段目標前進，最終目標為於智慧家庭中應用定位偵測結果。

Most of the current indoor positioning technology is the active systems which need to bring portable identification locators all the time. It will cause trouble in life. However, passive position systems have often been interfaced by furniture and space layout. The purpose of this research is to improve the above-mentioned disadvantages of indoor positioning technology. Deciding to position the signal caused by footsteps .The signal induced by footsteps can be divided into vibration and sound, after the primary stage to study, we must judge sound is a better signal for analysis because processing easy and high equipment availability. Therefore, research priorities will focus on sound localization. Unlike most traditional microphone array algorithm positioned by the angle of wave source geometry, in which derived scanning optimization by scanning the graph of the error function and find the minimum region, cleverly avoiding falling into local minimum. The advantages of this method allow for the distributed sensor and thus increase the volatility of indoor furniture placement. This paper also discusses the layout of the sensors position influence the accuracy of positioning. Finally in the experiment to find out the reasons will affect the positioning error and provide improved methods.