The Arc Transversal Median method improves the azimuth resolution of conventional polaroid ultrasonic sensors by intersecting arcs, filtering intersections, and taking the median of intersections. The result in our experiments is a ten-fold improvement in azimuth resolution.

The Arc Carving method improves the azimuth resolution of conventional polaroid ultrasonic sensors by eliminating portions of readings that are contradicted by subsequent readings. Experimental results verify this approach over areas as large as 5000 square meters.

The Kalman filter is a linear state estimator which fuses multiple sources of information to minimize the uncertainty in the estimation. For our feature-based SLAM, we use an Extended Kalman Filter to linearize odometry, and range and bearing measurements of landmarks to both build a map of landmark locations and maintain the configuration of the robot.

Visual Tracking consists of identifying features and their locations in a series of images and estimating the future locations of those features in order to track them in the environment.