Real time obstacle detection in a water tank environment and its experimental study

This paper presents a kind of real time obstacle detection method for underwater vehicles equipped with 2D forward looking sonar. Due to frequent failure or multipath effects, it is very often uneasy to get satisfactory sonar range measurements, especially in the water tank environment. How to extract as accurate as possible environmental information from these sonar raw (noisy) measurements is crucial for the vehicle's obstacle avoidance. In this paper, we propose a sort of occupancy grid mapping method to detect the obstacles in a cuboid tank environment, which is assumed to be unknown to the vehicle except for its dimensions. Some of Experimental studies are carried out to demonstrate the effectiveness of presented algorithm.