OT - ILS and RC UAVs

[caseih]

This is off-topic, but kind of follows from this tangent of talking about ILS and autoland. Basically most UAV platforms including Paparazzi can take off and land our RC planes. However GPS isn't that accurate, so landing isn't precise. Maybe a very low-power ILS-like system may be useful to UAV folks... will have to bring that up on the UAV forum. :)

[brakar]

Not a bad idea. You would need a directional antenna at the ground, yagi or maybe parabolic, and a pre-programmed radio to send a standard message. If the autopilot was powered up at the possition where the antenna was going to be located - the autopilot could store the (approximate) location of the ILS. If the signal was sent at the same frequenz as the autopilots datalink (assumed present) it should be possible to program the autopilot to follow the signal and land the plane.

brakar

[brakar]

OK, a I suppose there are a few difficulties to be owercome. To land a plane correctly you would need the exact location of both the ILS and the plane - which might be difficult to obtain by this method.

Another and simpler approach to the problem might be to connect a second GPS to the ground station/(end of runway). Both the gps in the plane and the gps on ground will provide non-exact coordinats, but the errors should be equal in both cases/(gps's). If I am correct, it should be possible to to calculate exact distances (in x, y, and z) between the two gps's - and it should also be possible to calculate exact parametres for landing. Correct me if I am wrong. (We do not have to mention drift and other difficulties at this point of time).

Come on folks, try to be a bit constructive on this one. I am sure autonomous landing is a goal worthy of some effort!

brakar

[ShadesOfGray]

If you're landing on a runway, then image recognition of the runway and centerline wouldn't be out of the question. Image recognition could be used on grass if you took out a roll of toilet paper. The GPS at the ground station idea is pretty swell too. There a million different ways to tackle the problem...

[HELModels]

http://en.wikipedia.org/wiki/Real_Time_Kinematic

autopilot is even mentioned in the link.

http://www.precision-gps.org/

another link to open source software.

one more link on WAAS and DGPS.

http://en.wikipedia.org/wiki/GPS_WAAS


If you had a 10 -20 cm accuracy of x,y,z you could define a location of the landing slot with a defined slope and width. The trick to a good landing would be to keep it in the slot. The concept is simple enough, like anything the hard part is realizing it.

Different approach?

One final thought is that to keep the processing and equipment load light on the UAV, the ground station makes the correction, assumes navigation, transmits the commands. If you could simulate the onboard receiver behavior on the ground, then you might be able to anticipate the error fast enough to correct. If airborne receiver reports some postion at some moment in time, the error must be such and such, anticipate the next report. were we close enough? send servo command. If that could happen fast enough, faster than the plane reacts, then it might work. The UAV would just send it's own position, altitude, the ground station sends servo commands. That's not quite the kinematic idea, but it would keep the UAV simple and leave the complexity on the ground.

[aperture]

Since non-military gps specs are fairly inaccurate, why not use a bright light for the runway coordinate, and a light sensing unit on-plane for tracking? Night landings would be even more possible with this kind of techno.

to shoot an ils approach using gps, autonomously, might be a stretch.

ken

[aperture]

I expect that insitu uses mil-spec differential gps on their various aircraft to allow them to autonomously catch a rope with their wingtip for the arrested landing. Would be interesting to know for sure...
http://www.youtube.com/watch?v=6LUeiYb1baQ

[Tom Harper]

ILS is a bit weighty and expensive for model aircraft, but you might be able to do it with the system that preceded ILS - AN Range.

AN range uses two transmitters whose output lobes overlap. The right hand transmitter (I think it was the right one) transmits a Morse 'A' dit da. The left hand transmitter output is a Morse 'N' dah dit. Where the two overlap you get a solid tone. Not ILS but it was hot stuff on a WW2 B-25.

A similar system for models could be made using ultrasonics. The range of ultrasonics is short (25 ft) where you are bouncing a signal off of a soft surface like ground vegetation, but for line of sight communication 300 feet should be reasonable.

'AN' does not provide distance to touch down info. You can get that by transmitting an RF pulse at the beginning of each 'AN' pair. The distance to touch down is proportional to the time between the arrival of the RF and Acoustic pulses.

So how would one start? Something like the following.

Assumptions:

1. The small, inexpensive xmtr/rcvr pair has a range of 300 ft. The pair shown in photo 1 has that range as an optimistic spec (SparkFun).

2. Pulsed Acoustic transducers can accept sufficient drive for 300 ft range. Those shown are also from SparkFun.

3. The on board navigation system can align the model with the end of the runway at a suitable approach altitude.

Layout:

The layout shows 5 acoustic transmitters - 2 for the 'AN' pulses and 3 placed at the end of the runway for glide path determination. The aircraft has the RF rcvr and two acoustic transducers - one facing forward and one facing down. The 'AN' transducers and the altitude group (a1,a2,a3 ) are all fired at the same time.

A 25 pulse per second rate should be sufficient.

Comments -