|
| From: | gisela.noci |
| Subject: | RE: Reply to: Re: [Paparazzi-devel] distance measurement for landing.(Electro-Technik) |
| Date: | Sun, 22 Nov 2009 08:31:17 +0200 |
|
-----Original Message----- > "Even if local atmospheric pressure could be measured
perfectly > accurately, with no errors, would this be the best parameter
to use to > facilitate an automatic landing?" From what i have found out so far by reading and thinking
all possible scenarios a barometer will not do the job. What is needed is a definite, accurate and dependable method to measure
the distance from the ground with an accuracy of at least 10cm and a latency of no more than 40ms. ---10cm is a little
excessive - between 20 to 30cm is fine. A 10cm resolution is also fine. Something like 1) someone sitting to the landing gear with a hand held laser distance meter :-) 2) A laser or optical range meter but for all terrains. --- to heavy and requires
a laser of reasonable peak power to cope with surface specularity variation. 3) A UWB range finder ---- This is a very good
option, and is what I am working on. The problem with this technology is that
the process relies on VERY fast pulses being generated for the TX pulse, an the
range delayed RX trigger sample pulse. At 300mm per nanosecond, the TX and RX
pulse width needs to be least than 250picosecond total duration, to be able to ‘measure’
distances of 150mm.(this gives pulse rise and fall times around 90ps –
WHOOSH!) A 1ns pulse will give a range ambiguity of 300mm. These short
pulse are only possible using quite esoteric (read – expensive -)
components, Step recovery diodes, Marks Bank generators, Tunnel Diodes, etc, and
so such solutions are not yet so readily available, but the growth in this
field is rapid, and we will se more of this. The main driver for this
technology right now is the automotive market – reverse parking ranging
and 360 degree ranging for collision avoidance. The second complex part
of this technology are the antenna. The antenna design is quite different from
conventional RF and Radar antenna in that the antenna should not be resonant and
should be very broad band, from around 2 or so GHz up to perhaps 8 to 10GHz. The
pulse rise times of 90 to 100ps gives an antenna response requirement up to 10
or so GHz. The most critical element of the antenna is the requirement of ‘non-resonance’.
If the antenna is resonant, it will ring electrically when the pulse is
applied. The duration of the ringing will directly influence the resolution of
the system; if the ringing persists for 40 to 50ns ( very typical, sometimes
even longer if for example a simple
dipole cut to resonance at 5GHz is used) the your resolution is approx half the
ring time , say 25ns = 7.5meters!!! Also the antenna should
be directional, so that only the area of interest is ranged. Reflection from
the aircraft itself need to be eliminated. This system works very
well and good results can be achieved for the medium and longer ranges where
resolution can be sacrificed, with easily available components. The very short
ranges (where we need very fast pulses to get good resolution) is a little more
difficult. It is an ideal solution for an airborne seek and avoid sensor, both
for other aircraft in the same air space, and for terrain avoidance. There are some ‘commercial’
units starting to appear : http://www.getradar.com/PDFfiles/PER24.PDF 4) A normal RF range finder --- To heavy and power
hungry ( normal RF Doppler rangfinder) 5) A powerful ultrasonic range finder. --- This is what I am
using for autoland. I use 4 Ultrasonic transducers, pulsed with 100v peak to
peak, 40KHz, 15 pulses per transmission. The transducers are all wired in
phase. A single RX transducer is used for the reflected pulse detection. I can
detect up to 22meters of a hard surface, with no TX/RX misalignment. However,
in flight the range is easily reduced to 3 to 5 meters due to : Surface
reflectivity variation. I fly in the Namibian desert areas, so, soft sand and
gravel plains are the order of the day and such surfaces can absorb more than
80% of the transmitted signal. Aircraft
roll angle. The Ultrasound transducer is VERY directional, and an aircraft roll
angle of 10 degrees will result in a loss of received signal. The
receive transducer amplifier gains are critical, to high results in to many
falls rangings, to low reduces range, and the RX tranducer MUST be vibration
isolated from the aircraft, esp if an IC engine is the prime mover ( vibration). This
system does work and very well indeed. We have implemented autoland as follows: Fly
upwind to the descent spiral. Enter the spiral at 100m AGL, and spiral down to
40m AGL. Exit the spiral on a tangent, heading in to the wind, down the landing
glide slope. We control the descent rate to around 2 to 2.5 m/s, aiming for a
pre-touchdown point an estimated 5m AGL. During this descent we control rate of
descent and airspeed, looking for a valid ultrasound range return. As soon as a
ground range is detected, we aim the aircraft at the touchdown point, trying to
keep descent rate less than 1.5 to 2m/s. At 1m above ground we reduce descent
rate to 1m/s, and as soon as that is achieved, we cut throttle, and hold pitch
and roll level, and at 300mm above ground flare to a pith angle of 5 degrees,
and touchdown. Do not try to use the
maxbotics or similar Ultrasound units for this type of application; there is
simply not enough energy in the TX pulse. 6) A vision range finder either on the airplane or at the ground
station. ---On aircraft is can be
a bit bulky and heavy perhaps. System is reliant on lighting conditions, ground
specularity, etc. Try, for example, to let an auto-focus pocket camera focus on
a flat surface that is strewn with uniform shaped gravel, with little colour or
contrast variation. Sometimes it does not manage to focus, and vision ranging systems
can suffer the same effects. Lighting contrast also affect the CCD
considerable, and a repeatable reliable result is not guaranteed. 6) a wire connected to a potentiometer which drags below the aircraft (say 50 cm lower) :-) --Forget the pot; use
thicker wire and and make hook in the end. String up a wash line and catch it…. Joe Chris _______________________________________________ Paparazzi-devel mailing list address@hidden http://lists.nongnu.org/mailman/listinfo/paparazzi-devel |
| [Prev in Thread] | Current Thread | [Next in Thread] |