This paper presents a method for navigating relative to the sea floor using sonar imagery. The aspect dependent terrain recognition problem is addressed by generating acoustic holograms of the sea floor containing images of all aspects of interest. Grazing angle compensation is applied to reduce the required spatial sampling from planar to linear. Sonar images are coherently correlated with centimeter resolution to provide navigational updates. The technique is demonstrated using five short duration missions in St. Andrew Bay and three long duration missions, with a combined extent of 69.7 km, in the Gulf of Mexico.

This paper investigates how to design a robot that uses sonar. Governing equations describing the evolution of geometric measurements are derived as a function of 6-DOF robot dynamics. A non-dimensional analysis of the governing equations is presented, to determine which terms are relevant to specific robots. The contributions of waves and currents are considered. The equations are then expanded to evaluate the effect of transducer location on the robot. Finally, the computational complexity of detecting and tracking arbitrarily curved objects is derived as a function of dynamics and sonar properties.

This paper describes a new data association technique for the interpretation of wide-beam sonar measurements. The goal is to group sets of returns that originate from the same surface of an object. We consider the case of a moving observer that obtains range and bearing measurements of curved and/or faceted objects with binaural sonar operating in the specular wavelength regime. Rather than projecting sensor data into a Cartesian space before processing, we operate on a "raw" representation in terms of range, bearing, amplitude, and time. A binary geometric constraint is applied to pairs of consecutive measurements to rule out impossible assignments. A measurement flow model is proposed for validating triples of measurement. The performance of these techniques is illustrated via a set of experiments using a wide-beam 500 kHz binaural sonar system, demonstrating effective perceptual grouping and mapping with sonar echoes originating from a set of objects, despite the presence of navigation error.

The paper describes a system for structure-from-motion using vanishing points and three-dimensional lines extracted from omnidirectional video sequences. Two novel aspects of this work are its deferred initialization of features using noisy observations from multiple, uncertain vantage points, and its use of dynamic programming for efficient 3D line tracking. We show preliminary results from the system for both indoor and outdoor sequences.

Stochastic mapping is an approach to the concurrent mapping and localization problem. The approach is powerful because the feature and robot states are explicitly correlated. Improving the estimate of any state automatically improves the estimates of correlated states. This paper describes a number of extensions to the stochastic mapping framework, which are made possible by the incorporation of past vehicle states into the state vector to explicitly represent the robot's trajectory. Having access to past robot states simplifies the mapping, navigation, and cooperation. Experimental results using sonar data are presented.