抄録
We propose a method that the relative poses of robot vehicles are estimated by using audible sound wave. It consists of CSP(Cross-power Spectrum Phase analysis) method and least square method. We used assumption that 3 robot vehicles exist in an anechoic room, each vehicle has 2ch synchronized microphones and 2ch speakers of known position parameters, and each speaker can make any distinguishable sound from another speaker at any time. Each robot vehicle can observe the differential time of 2ch synchronized microphone from a speaker. The robot vehicles estimated the relative poses from the diffrential times. And on a set of values of the estimated parameters, we calculated how large errors they include by influences from obsevation errors. In the future, if microphones arrays can synchronize each other, we can treat a set of microphone arrays as a larger scale microphones array.
