AUTONOMOUS NAVIGATION OF USAF SPACECRAFT (FILTERING).

摘要

The U.S. Air Force is developing satellite-borne sensors to enable autonomous navigation of spacecraft in the near future. This study compares the observations from several medium-accuracy space sensors, such as the existing telescopic space sextant, with those of future matrix-type sensors. The large field of view of matrix sensors will allow them to determine the earth horizon to approximately an order of magnitude better than current infrared sensors by observing atmospheric refraction of stellar light. This horizon determination will give the matrix sensors an accuracy of less than 1 km. The limiting factor in earth-horizon determination is the modeling of atmospheric refraction effects. For high-accuracy requirements (100 meters or less), the Global Positioning System (GPS) offers the only near-term solution. A relative navigation technique using range and doppler data is proposed for autonomous navigation of the GPS satellites. The navigation accuracy of this technique is evaluated by consider covariance analysis and by processing corrupted data through a reduced-order onboard Sequentially Partitioned Algorithm. The algorithm is stable and for the GPS system produces in-plane accuracy of 40 meters over twenty days. However, out-of-plane motion is shown to be unobservable in the GPS-to-GPS tracking mode, and errors of up to 1.5 km over 60 days are experienced. For this reason, a supplemental transmitter on the ground or in a different orbit is recommended.