Radius of curvature and location estimation of cylindrical objects with sonar using a multi-sensor con®guration

Although acoustic sensors are inexpensive and convenient to use in time-of-flight (TOF) ranging systems, they have some important limitations one of which is their low angular resolution that makes localization difficult. In this paper, an adaptive multi-sensor sonar system is introduced to compensate for the low angular resolution and improve the localization accuracy. The sensor configuration is used to estimate the radius of curvature and location of cylindrical objects. Sensitivity analysis of the curvature estimate with respect to some measurement errors and certain system parameters is provided. Two methods of TOF estimation are used and compared: thresholding and curve-fitting methods. Theory and simulations are verified by experimental data from a real sonar system. The adaptive configuration improves the estimates by 35-45 % and the curve-fitting method, compared to the thresholding, brings an improvement of about 30 % in the absence of noise and 50 % in the presence of noise. The radius of curvature estimation is shown to be useful for target discrimination. 1