Commercial GPS updates once per second, and without real DGPS or WAAS is
only accurate to about 10m horizontally and 100m vertically.  Most OEM
GPS receivers accept real dgps data, but a dgps sender is big bucks, and
is not trivial to build.  Using several of the same receivers on the
same location and a computer to computer the difference would be less
accurate than DGPS and perhaps WAAS - you'd still likely have a meter
error.  In any case you should probably throw out the vertical
information from the GPS completely.

Since it only updates once a second you are going to have to use
accelerometers or another positioning system to fill in the gaps - think
about how far an airplane goes in one second, and how far it can go in a
wind gust.  Straight interpolation simply won't work here (unless you
can eliminate all the variables, but even large buildings big enough for
your model are going to have odd air currents)

I suspect you'd spend less time in development if you built your own
positioning system.  Radio would work, using lights on the runway and a
camera in the plane would work, you could even have a transmitter on the
plane, tracked on the ground using three radio receivers with all the
heavy computations done by the computer on the ground.  You may want to
go this route first only to make it easier to program the algorithms on
the computer.

The hardest part, I suspect, is not the actual mechanisms for tracking
and autopiloting the plane but the algorithms that control it:

Example:
Up until recently two legged walking robots have all been built by trial
and error and/or by keeping the center of gravity above points touching
the ground.  The problem with doing it this way is that the algorithms
are specific to the hardware/model they are using and are not
generalized for other models.  When they do get the algorithm working on
another piece of hardware they have to change hundreds or thousands of
variables and re-tweak the algorithm.  Researchers have finally made a
simple model, including equations to describe the model, that can walk
and run in a dynamic fashion - with its center of gravity somewhere
beyond its feet.  This model can be described in just a few variables,
and adapting it to a different piece of hardware is a matter of changing
only those few varaibles, not the algorithm itself.

I imagine you are going to go with the trial and error method, and being
able to program a computer on the ground in the situation is going to
make this a lot easier.  I wouldn't try to do it with a PIC just yet.
 Pick some fast high level language, or even create your own program
with a scripting language so you can go through the trials more easily.

-Adam

Tan Chun Chiek wrote:

>Hi,
>    This idea of an autopilot landing system just popped up in my mind, and
>other details came almost immediately. I thought why hasn't anyone done it
>before? The runway is something like 10m X 50m. GPS is used for
>longitudinal/latitudinal positioning and preferably velocity measurement,
>the aircraft would have a GPS receiver onboard, and 2 other GPS receivers
>would sit on the both long ends at center of the runway, it's something akin
>to differential GPS to provide better accuracy. Height measurement is done
>by a ultrasonic transceiver or maybe GPS, are they accurate enough? And
>pitch and roll is sensed by a pizeo gyroscope.
>
>How feasible is this concept?
>
>Regards,
>Tan CC
>
>--
>http://www.piclist.com hint: The PICList is archived three different
>ways.  See http://www.piclist.com/#archives for details.
>
>
>
>
>
>
>

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email listserv@mitvma.mit.edu with SET PICList DIGEST in the body