Object Detection with sector scanning sonar

Abstract

An object detection subsystem serves to detect obstacles that are in the vicinity of an Autonomous Underwater Vehicle (AUV). Along with the obstacle avoidance, command and control subsystems, it ensures that the AUV can safely execute and complete its mission. The sector scanning sonar was considered over other acoustic alternatives such as echosounders and multibeam as a means for object detection for STARFISH AUVs. The reasons are because of its compact size, lower power consumption and lower data rates.

Experiments were planned to analyze the measurements from the sonar. Two experiments were conducted where the sonar was deployed statically at Nanyang Technological University (NTU)'s diving pool and Republic of Singapore Yacht Club (RSYC). In both of these experiments, datasets were collected from the ensonification of static objects. An experiment with STARFISH AUV integrated with Micron DST sonar was conducted at Pandan Reservoir. In this experiment, dataset was collected from the ensonification of the embankments and static buoys. Another experimental dataset at Fluvia Nautic marina was made available by University of Girona. In this setup, a Tritech Miniking sonar attached to a moving AUV was deployed to ensonify the marina.

The scanline measurements from the sector scanning sonar were analyzed to understand how each element in the scanline measurement corresponds to the intensity return for a given bearing and range bin i. Then, detection and representation methods were explored to determine suitable approaches to represent both the operating environment and detection decisions. The detection methodologies considered are Otsu thresholding and static thresholding. The formulation of the static thresholding was based on an adaptive threshold methodology with constant false alarm rate (CFAR). A mean statistic of the binary detections was used to represent the result from the Otsu threshold. Occupancy grid was used together with the static threshold to represent the probabilistic result of object detection.

Both Otsu thresholding and static thresholding are employed for the datasets. The Otsu threshold works well for the NTU, RSYC and Girona datasets but failed for the Pandan Reservoir dataset. The static threshold works well across all the datasets. The static threshold is more effective as the assignment of the probability of a target given the sonar measurement is based on the decision statistic. Thus, measurement that marginally exceeds the threshold does not yield high probability of an object. The probabilistic detection decision was incorporated into the occupancy grid to attain the probability of occupancy. The probability of occupancy for each grid cells can be updated as and when more measurements are attained. The occupancy grid proves to be an effective representation of the environment and was also effective in localizing the AUV along with the objects.