In an improvement of the visual-targettracking software used aboard a mobile robot (rover) of the type used to explore the Martian surface, an affine-matching algorithm has been replaced by a combination of a normalized-cross- correlation (NCC) algorithm and a template-image- magnification algorithm. Although neither NCC nor template-image magnification is new, the use of both of them to increase the degree of reliability with which features can be matched is new. In operation, a template image of a target is obtained from a previous rover position, then the magnification of the template image is based on the estimated change in the target distance from the previous rover position to the current rover position (see figure). For this purpose, the target distance at the previous rover position is determined by stereoscopy, while the target distance at the current rover position is calculated from an estimate of the current pose of the rover. The template image is then magnified by an amount corresponding to the estimated target distance to obtain a best template image to match with the image acquired at the current rover position.

Turn-in-Place Experiments show beginning image (left) and end image (right) after 80° rover rotation. As the rover turns, the mast camera turns in the opposite direction to point to the target.

This program was written by Won Kim, Adnan Ansar, and Richard Madison of Caltech for NASA's Jet Propulsion Laboratory. For further information, access the Technical Support Package (TSP) free online at under the Software category.

This software is available for commercial licensing. Please contact Karina Edmonds of the California Institute of Technology at (626) 395-2322. Refer to NPO-42682.

This Brief includes a Technical Support Package (TSP).
Improvement in Visual Target Tracking for a Mobile Robot

(reference NPO-42682) is currently available for download from the TSP library.

Don't have an account? Sign up here.