Towards an O(N) algorithm for online estimation of range, bearing and span of obstacles using a Forward Looking SONAR: An experimental study

Autonomous Underwater Vehicles are autonomous agents carrying specific payload deep underwater to carry out particular tasks like underwater bathymetry, imaging and other investigative and surveillance operations. Since, they generally work on their own, without any human guidance whatsoever, therefore it is essential to incorporate some collision avoidance technique as an effective means of dynamic path planning. However, for the proper execution of all such avoidance algorithms it is required to detect obstacles with consistent estimation of the range and bearing of obstacles lying along the trajectory and determining potential collisions. In this regard AUVs are very often equipped with SONARs for detecting objects laying ahead the vehicles. Consequently, data from the SONARs need to be processed in order to compute the range and bearing of the objects detected by them. As a result, it remains the objective of this paper to introduce a computationally effective algorithm for computing range, bearing and size of an object as detected by the Forward Looking SONAR used for the AUV-150. The discussed approach is different from the conventional image processing methodologies adopted for feature extraction from SONAR images. Results as presented during sheltered water trials for the AUV-150 exhibit considerable accuracy.