A Novel Approach to Environment Mapping Using Sonar Sensors and Inverse Problems

Abstract

The traditional approach for environment mapping using sonar sensors in autonomous mobile robots is generally based on time-of-flight, which results in a sparse representation that unfortunately does not make use of the much richer interaction of ultrasonic waves with the obstacles within the environment. In this work, inspiration is taken from techniques used in ultrasound medical imaging, aiming at higher spatial resolution reconstruction of a cross-section of the environment without the need of sweeping it with multiple ultrasonic bursts. A couple of sonar sensors provide raw analogue data that is used to feed an inverse model of the acquisition system and generate an image reconstruction of what can be interpreted as a top view of the environment. Preliminary experiments in a small controlled environment show promising reconstruction results for inverse problem approaches when compared to beamforming techniques normally used in ultrasound medical imaging.