Date of Award

Winter 1999

Document Type


Degree Name

Doctor of Engineering (DEng)


Electrical Engineering

First Advisor

David Cowling


The last ten years have shown a marked increase in research into Autonomous Underwater Vehicles. A key component of this research is the ability to “look” ahead of the vehicle's projected path and translate active sonar returns into three-dimensional (3-D) computer data structures used to navigate and avoid obstacles. This need to avoid obstacles is also common to other underwater vehicles including submersibles.

Of special interest to this research is the “Occupancy Grid Framework.” This technique divides the forward looking sonar field into cells. These cells can be maintained either in spherical or cartesian coordinate systems.

This research demonstrates a method of maintaining an array-type data structure based on the cartesian coordinates of returned sonar echoes. A volume of array elements are colored to reflect the probability of potential obstacles. Also, between sweeps of a scanning sonar transducer array, the locations of these volumes of probability are both rotated and translated inside the data structure as the vehicle turns and/or moves forward. ard. This research is different from prior research in two respects. First, rotation and translation of target probability spheres, located in a 3-D array, are accomplished by rotating and translating the sphere centers rather than the actual voxels comprising the spheres. In this way, probability spheres are continuously being removed, and regenerated at new locations inside the data structure. This relocation can be done in real-time as the vehicle moves.

Secondly, this research shows a method of pre-processing real-time data for increased speed using a series of two microcontrollers located between the sonar transceiver and the host computer, where the data is processed.

Programs used in this system consist of both assembly language programs for the microcontrollers, and a C language program for the host computer. These programs demonstrate a software approach which can be used as a basis for future research.