Position fingerprint localization method based on linear interpolation in robot auditory system

Abstract

Localization method based on position fingerprint can effectively solve the problems such as high dependence of model and low positioning accuracy in robot auditory system. However, this method requires a large number of reference points to achieve high precision positioning. To solve the problem of low precision under low fingerprint density, a position fingerprint localization method based on linear interpolation is proposed. We use the K Nearest Neighbors (KNN) algorithm to the coarse localization. And then we update the fingerprint database to increase the density of the position fingerprint in the credible area with linear interpolation method. Finally, we use the Weighted K Nearest Neighbors (WKNN) algorithm to accurately estimate the location of the sound source. The experimental results show that under low fingerprint density and high noise environment, the proposed method can improve the localization efficiency effectively and has a strong practicality compared with the conventional robot auditory localization methods.