Flow fields estimation on the basis of a computational model of bionic-sonar

Research Project

Project/Area Number	06650456
Research Category	Grant-in-Aid for General Scientific Research (C)
Allocation Type	Single-year Grants
Research Field	計測・制御工学
Research Institution	The University of Electro-Communications
Principal Investigator	MITSUHASHI Wataru The University of Electro-Communications, Department of Communications and System Engineering, Lecturer, 電気通信学部, 講師 (40017421)
Project Period (FY)	1994 – 1995
Project Status	Completed (Fiscal Year 1995)
Budget Amount *help	¥2,300,000 (Direct Cost: ¥2,300,000) Fiscal Year 1995: ¥1,400,000 (Direct Cost: ¥1,400,000) Fiscal Year 1994: ¥900,000 (Direct Cost: ¥900,000)
Keywords	Echo location / Bionic sonar / Doppler effect / Binaural reception / Parameter estimation / Downward swept FM / Phase sensitive model / コウモリの生体ソナー / 超音波計測
Research Abstract	A bank of filters with constant Q is adopted as an optimum receiver for target parameter estimation, representing the frequency selective mechanisms in the mammalian auditory periphery. The investigator assumes that the phase value of the output from each filter represents a neural firing locked on a specific phase value of received sounds. On this assumption, an estimate of the propagation delay to a moving target is obtained from the time of the largest firing rate, while the phase value that most of the neurons respond to at the instant of the largest firing rate provides an estimate of target velocity. In the final year of this investigation, experiments for target parameter estimation in two dimensional space were carried out in air by using four receivers and a transmitter. In such the two dimensional case, Jeffress model is known as a mechanism for detecting the interaural time difference (ITD) in binaural reception. A maximum likelihood estimates for range and direction of a moving target are calculated based on the ITD. In velocity vector estimation, a radial velocity component can be estimated by the maximum likelihood method with the phase value obtained from the largest firing rates, whereas an angular velocity component has a relative large error because it is calculated by differentiating the binaural phase values. Further improvements must be made on the algorithm for velocity vector estimation.