Reconstructing 2-D maps from multiple sonar scans

The Naval Automation Institute (IAN) of the Italian National Research Council has been carrying out some activities in the field of underwater robotics. The ultimate goal is the development of an autonomous underwater vehicle for research purposes. The paper presents the activity of the IAN in the field of the reconstruction of underwater scenes using sonar. The authors work with real data taken using a high-frequency pencil beam profiling sonar manually positioned in a laboratory tank. The goal is to build a model of the surroundings as a two dimensional map, which could be used by an autonomous vehicle to plan, on-line, the path to reach a target point. At first they describe the basic geometric elements used to model the environment and the synthetic information associated with them. The description of an algorithm to extract basic geometric elements from a single sonar scansion is followed by the presentation of a method to merge the processed data of many scansions to build a more complex 2-D map detecting false echoes