Local discriminant bases and optimized wavelet to classify ultrasonic echoes: application to indoor mobile robotics

Abstract

To localize a robot in an indoor environment, ultrasonic sensors are used. Our aim is to show we can classify targets into four classes (edges, corner, plane, small cylinder) using the whole information in the received echo. However, to classify received echoes we need to extract features from raw data. The extracted features must be discriminant to improve the classifier results. The selection of features is an important step of the pattern recognition. In this paper decomposition on wavelet best basis and optimized wavelets are used to extract relevant features from an ultrasonic signal and is compared to other feature extractors.