Feature-based probabilistic map building utilizing time and amplitude information of ultrasonic signals

Abstract

For autonomous operation of mobile robots, intelligent sensor systems are need to recognize environments, and map building (recognition of environment and maintenance of information) is a very basic and important process. Various sensors can be used such as vision, ultrasonic sensors, laser or infrared range finders. Among them, ultrasonic transducers are most widely used due to their low cost and simplicity. In majority of applications, ultrasonic transducers are used to measure time-of-flight (TOF) of the first echo to obtain the distance, given the speed of sound in air. It provides simple but accurate range information between sensor and reflector. Despite their simple and accurate range measurement, use of ultrasonic transducers in map building is plagued by a few shortcomings like directional ambiguity due to an acoustical beam width, erroneous measurements by specular or multiple reflections, and limited sensing speed by crosstalk and speed of sound. There have been many researches to solve these. However, most previous works have common limits like the requirement of enough multiple measurements and the indirect use of amplitude-of-signal (AOS) by a simple comparison of its magnitude. In this thesis, we present a feature-based probabilistic map building algorithm which directly utilizes time and amplitude information of sonar in indoor environments. Utilizing additional amplitude-of-signal (AOS) obtained concurrently with time-of-flight (TOF), the amount of inclination of target can be directly calculated from a single echo, and the number of measurements can be greatly reduced with result similar to dense scanning. Consequently, the scanning speed is increased by reduction of scanning. In the first place, to use AOS directly, we propose a AOS model considering both attenuation factors not only dispersion, but also air absorption. Using this model, a reference amplitude is defined for a target, and AOS can be predicted with the variation of range and an...