A proposed Global Positioning System (GPS) sensor would give information on approximate attitude as well as on position. Unlike other GPS-based attitude sensors, this attitude sensor would not depend on carrier-phase measurements and thus would not be subject to the difficulties and limitations involved in determining attitude from such measurements. Instead, this attitude sensor would be based on a concept that is related to that of an insect's compound eye and that has been implemented in optical attitude sensors of the star-tracker type.

The sensor would be equipped with multiple directional antennas mounted on a convex hemispherical surface. The number of antennas and their locations on the surface would be chosen to obtain a regular or slightly irregular polyhedral (e.g., half-dodecahedral or half-icosahedral) arrangement. Each antenna would thus be aimed to receive GPS signals from a field of view, called a "visualization cone," approximately coincident with the solid angle intercepted by the corresponding face of the polyhedron from the center of the hemisphere (see figure).

This GPS Sensor Would Include Six Antennas aimed outward from faces of a half regular dodecahedron. Their fields of view would be partially overlapping "visualization cones." The purpose of this arrangement is to deduce the orientation of the sensor from the known location of each GPS satellite "visible" to each antenna.

By virtue of the conventional GPS function, the positions of the GPS satellites and of the sensor would be known accurately and thus the direction from the sensor to each GPS satellite would be known accurately. Therefore, the reception of a signal from a given GPS satellite or satellites through a given antenna would provide partial attitude information: It would signify that the sensor is oriented so that the visualization cone of the given antenna contains the known direction(s) to the satellite(s). In a similar manner, the simultaneous reception of GPS signals through the other antennas would make it possible to draw additional conclusions as to how the sensor must be oriented in order to make possible the observed combination of antennas and signals.

The precision of the attitude estimate obtained in this way depends on a number of factors, including notably the number of antennas and the way in which the signal-reception data are processed. Simplistic processing of raw data yields attitude estimates with errors of the order of visualization-cone angles (tens of degrees in the case of a half-dodecahedron). Errors can be reduced by use of optimization techniques in which, for example, greater weights are assigned to signals from directions that lie in overlaps between cones. Errors could be reduced further by increasing the number of antennas to obtain smaller cone angles and overlaps; for example, a first-order calculation has shown that with 16 antennas distributed over the hemispherical surface (on a "buckeyball"), errors would be reduced to <3°. Refinements in processing should make it possible to reduce the errors to the subdegree range.

This work was done by David A. Quinn of Goddard Space Flight Center and John C. Crassidis of Texas A & M University. This invention is owned by NASA, and a patent application has been filed. Inquiries concerning nonexclusive or exclusive license for its commercial development should be addressed to

the Patent Counsel, Goddard Space Flight Center; (301) 286-7351

Refer to GSC-13966.