Re: accuracy of automatic celestial navigation
From: Paul Hirose
Date: 2002 Dec 8, 23:07 -0800
From: Paul Hirose
Date: 2002 Dec 8, 23:07 -0800
George Huxtable wrote: > > Not so difficult to define the direction of a star to that precision, > perhaps, but what worries me more are the errors in establishing the > direction of the vertical reference. An aircraft is different from a > spacecraft in that it has some unavoidable degree of buffeting from the > atmosphere that it's flying in, small changes in engine output, tiny > movements of control surfaces, movement of the crew, together with > gravitational anomalies from the ground it's passing over. In the case of the B-2 AINS (astro-inertial navigation system) the star tracker is inside the same case as an inertial nav system. The inertial and stellar hardware are tightly integrated with each other. So the star tracker's short term pitch, roll, and heading stabilization comes from the inertial platform. Earth's lumpy gravitational field is indeed a factor in inertial navigation; it has to be stored in some sort of internal map or table. Through some magic not clear to me (I worked in maintenance, not engineering!) the AINS feeds the measured directions of stars back to the inertial portion to correct its errors. Giving the otherwise blind inertial system a look at the outside world is enormously helpful. As long as it's got a clear view of the sky the AINS will remain within [classified] meters of the correct position, worldwide and practically independent of mission duration. It doesn't drift away over time like a normal INS. In obscured skies the star tracker keeps trying unless you deselect stellar mode. It can shoot through a hole in the clouds. It lingers on a body only long enough to measure its position, then goes looking for others in its field of view (anything within 45 deg of the zenith). Initially the scope is placed on the star's predicted az/el. It searches outward in a "square spiral" pattern until it locks on or concludes the star is obscured. There are 61 stars in its catalog; unfortunately I don't know their identities. The 57 classic navigational stars plus 4? Another thing I never learned is how the AINS knows the offset between UT1 and UTC. An ATTU (airborne time transfer unit) was installed in the plane shortly before each flight to provide accurate time. Between flights we kept the ATTU connected to a rack-mounted rubidium time standard which in turn was synched to radio station WWV. Distance from WWV was set on thumbwheels to correct for propagation time. But as far as I know the whole rack was on UTC. I never saw any DUT1 readout provision or any other hint of UT1. My guess is that the offset was supplied as part of the computer-generated mission plan uploaded before flight. Physically, the AINS is fairly compact. It would fit in the passenger seat of a small car and weighs roughly as much as a man. Using it to navigate the car would take some doing, though. It needs a powerful flow of chilled air into its cooling duct, and gulps a fair amount of 400 Hz 115 VAC 3-phase and 28 VDC power. In operation the system was quite reliable. I remember only a few times when the AINS had to changed out. Good thing, since it was removed with a crane. The location is easily visible in an overhead view of a B-2: the star tracker window looks like a round dark hole about the size of a dinner plate between the cockpit and the left hand engine air intake. I got to run it outdoors a time or two. It was started like a normal INS. You inserted the lat, lon, and elevation, then selected an align mode. There were several available. It depended on how much of a hurry you were in; of course the best nav performance was attained by using the most time consuming align mode. (It's not enough to simply know the airplane is parked at such and such coords when you align the navs. What *part* of the airplane? In the B-2 the reference point is the nose landing gear; our parking spots at Edwards AFB all had little brass survey marks set in the concrete at the spot where the nose tires would go. I bet you won't learn trivia like that from Discovery Channel!) When alignment was done, you saw a NAV READY message on your little screen in the cockpit. At that point the system was frozen to your starting coordinates, waiting for you to select a nav mode. Again there were choices, basically boiling down to what kind of assistance the inertial platform would receive. For example, you had pure inertial, stellar inertial, GPS inertial, etc. Being an enthusiast I liked to select pure inertial and watch the coordinates drift away from the correct values. Then I switched to stellar inertial and watched the star tracker pull it back to the right spot. Worked as advertised - in broad daylight! Although it was fascinating to do celestial at a superhuman level, I quickly found it boring. All it required of me was to push the correct buttons. There was no pride from mastering tables of mysterious figures, no "eye" or "touch" to develop, no satisfaction at seeing LOPs meet nicely. I found the pinnacle of high tech celestial navigation barren of personal satisfaction, and after a couple visits was satisfied that I had seen enough.