Acoustic Localization of Robotic Fish in a Water Tank

本文描述一個仿生機器魚利用在身體側線放置的水下換能器陣列去量測與放置在水槽中固定聲音信號源相對應的位置藉此訂出機器魚在此水槽中的三維位置的被動定位演算法。由於在實際應用時此演算法需同步接受陣列資訊,且因量測的是200KHz左右的高頻信號(VHF),資料串流量極大故利用FPGA(Field Programmable Gate Array )去同步接受訊號,用廣意交叉相關函數(GCC)去估算直接傳遞及水面反射與陣列元互相之間的抵達時間差(Time difference of arrival),在此研究中利用壓力計(Piezometer)做為深度感測器。 接著我們對定位方程式作線性化得到估計位置的變異數與系統參數及感測器雜訊的的變異數之線性方程式,故可以預先藉由系統參數及感測器雜訊的變異數估算定位誤差。文中假設各感測器互相獨立,因此位置估計的變異數可藉由各個感測器變異數去線性疊加而得到。此研究模擬感測器雜訊及系統參數對定位誤差的影響以及利用簡單的應用實驗去驗證本文的演算法。由實驗及模擬結果可知此定位系統的精確度被麥克風陣列所侷限,感測器雜訊大小對精確度的影響可由感測器變異數直接估計定位誤差範圍。The study describes a passive localization algorithm for a robotic fish by utilization of a transducer array attached on sides of body. Localization is performed in the water tank by measuring the signal from fixed sound sources to estimate the relative position of the robot with respect to the source. The algorithm requires synchronously receiving VHF array signal with large amount of data stream, a FPGA(Field Programmable Gate Array) processor is used to collect from high speed analog to digital converter synchronously. Generalized cross correlation technique is applied to estimate the time difference of arrival from signal transmitted directly and reflected by water surface and each set of array sensors. Piezometer is used for measuring the depth of the robot. In order to find position estimation errors function due to sensors noises, linearization of the localization algorithm equations is performed to derive the correlation matrix between position estimate errors and sensor errors. Assuming sensors errors are independent, the variance of position can be linearly superposed of sensors variances. It is shown in simulations and experiments that the system localization accuracy is limited by sensors noises, position values, and the lengths of array’s baselines