Mapeamento de ambientes utilizando sonares e problemas inversos

Abstract

An autonomous mobile robot must be able to move through an environment without colliding with obstacles. This subject has been investigated by researchers over the years, mainly using sonar sensors to acquire distance data. The traditional approach used in this concept is based on the time of flight technique, in which the distances between the robot and obstacles are computed using the period of time between sending and receiving the ultrasonic wave. This technique aims only at the detection of obstacles, reconstructing the environment in low resolution since it does not consider inherent characteristics of ultrasonic reflections. The present work introduces a new approach, based on recent research on medical image reconstruction. The goal is to achieve environmental mapping in higher resolution, but without the need to acquire multiple ultrasonic bursts. The image reconstructions were made through the acquisition of ultrasonic reflections using sonars in a controlled test environment containing only one obstacle. Experiments were conducted to compare some of the existing image reconstruction methods, focusing on environment mapping. Inverse problems showed promising results, especially when compared to the traditional method of medical image reconstruction, delay and sum beamforming.