Wind speed on board measurement for autonomous landing?

[Arapajoe]

Hi, guys!
I've been reading this forum for a while and it looks great and has a lot of useful information.

I would like to ask you something, if you don't mind :)
Do you know of any sensor or any way of measuring the wind speed and direction?
I want to achieve an autonomous landing and I would like to take in account that information.
I know there are some ways, like desired direction minus real direction (GPS), but that's not good enough...
I've also read solutions like: GPS, calibrated airspeed, pressure altitude, outside air temperature, and a calibrated electronic compass, which are great, but really hard nonetheless...
Thanks for reading!!

[Felthamg]

Let's see,

how about a pitot tube at the front of your aircraft to measure the dynamic pressure (airspeed) and then use GPS measurements to determine groundspeed.

airspeed-groundspeed = windspeed

thats all I can think of right now, oh and you would need to measure the density of the air to determine the airspeed from the pitot tube measurement.

[dmgoedde]

Well, you must take into account the magnitudes and vectors of aircraft direction and wind direction (speed and heading of each):

GPS Ground Speed = Pitot Airspeed + windspeed*Cos(GPS heading - Wind heading)

Solve for Wind heading part:

Wind heading = GPS heading - ArcCos((GPS Ground Speed - Pitot Airspeed)/windspeed)


How to determine wind speed? It would be 1/2 of the difference between max and min GPS ground speed (180 degrees apart in heading) at constant pitot airspeed. I can't think of a simple equation to simply solve for it, so I would approach the problem iteratively in real time by fitting a model based on a similar equation to above.

[rich smith]

Turn transmitter off. Guaranteed autonomous landing!

Seriously though I have great interest in fully autonomous flight (FAF) and have succeeded to some small degree: takeoff, figure 8, land. Key was super stable model (low Vcg, high dihedral) and zero wind. Just finished arctan routines for AVR so next step longer range w/em406 onboard.

AFAIK GPS can't provide windspeed or airspeed, only groundspeed. I built hotwire sensor couple years ago (airspeed=resistance) for my fully size plane but it was too much fuss so went back to old reliable Hall Airspeed Indicator. There's mini-propellers, bending reed (stall), and pitot (pressure differential) like in Cessnas. I think a few people had limited success with Polaroid modules (sonar) and LIDAR to get AGL for autonomous landing.

Please post progress as few UAV enthusiasts seem interested in truly autonomous flight (incidently, the "A" does not stand for "autonomous" like some think).

Regards,
-Rich


Hi, guys!
I've been reading this forum for a while and it looks great and has a lot of useful information.

I would like to ask you something, if you don't mind :)
Do you know of any sensor or any way of measuring the wind speed and direction?
I want to achieve an autonomous landing and I would like to take in account that information.
I know there are some ways, like desired direction minus real direction (GPS), but that's not good enough...
I've also read solutions like: GPS, calibrated airspeed, pressure altitude, outside air temperature, and a calibrated electronic compass, which are great, but really hard nonetheless...
Thanks for reading!!

[jetblackaircra]

Why do you need to know wind direction and speed? All you need to do to land on a waypoint is to know the direction of the runway, the end of the runway, the vehicle position, and the vehicle ground track. All of this can be done with GPS. In order to align the vehicle on the runway during flare it would be best to know vehicle heading as well as vehicle track. No need for wind speed or direction. You should be able to tell if you're making a downwind landing by using GPS and pitot pressure. If your ground speed is higher than your airspeed you're going with the wind. Land the other direction.

No big deal.

[shanghai_fool]

Now there is a new project. ILS for models. I haven't seen anybody tackle that.

[dmgoedde]

Now there is a new project. ILS for models. I haven't seen anybody tackle that.

I am. "FAF" fully autonomous flight. Automated takeoff to automated landing, 100%.

[rich smith]

dmgoedde,

I've been following your attopilot project for the past year or so but did not see any reference to FAF. Must have missed it. Or was it in one of the behind the scenes beta tester communications or another thread? Do you have any links?

Thanks,
-Rich



I am. "FAF" fully autonomous flight. Automated takeoff to automated landing, 100%.

[shanghai_fool]

I worked on avionics when I was in USAF about a hundred years ago. Maybe I should brush up on it and see what we could do. I am sure its a lot easier now. We would have to use different freq band though. Wouldn't want some full scale landing on our runway.

[dmgoedde]

dmgoedde,

I've been following your attopilot project for the past year or so but did not see any reference to FAF. Must have missed it. Or was it in one of the behind the scenes beta tester communications or another thread? Do you have any links?

Thanks,
-RichNothing posted here on RCGroups. I did mention recently that intermediate commercial Atto (Version 1.7) will have ultrasonic sensor (0-22 feet) for automated landing and takeoff. With my sensors the takeoff would not be a runway type, but rather a catapult, hand-thrown, or similar launcher meachanism; Atto has no magnetometer.

I don't want to steal this other person's thread, so I will be silent. :)

[shanghai_fool]

Since my project is a VTOL, I had not thought about any ILS except the ultrasonic altimeter but it would be an interesting project. Using freq between 900-2.4G, you could use small PC type antennas and receivers and probably make a Cat IIIC ground station and with todays PLL's, the 90/150 Hz detectors should be easy enough. I have never done any antenna design but looks like a simple patch type would suffice. It may be difficult for me to get the English info in the Chinese library though.

[jetblackaircra]

ILS is on it's way out. Use GPS just like everyone will be in 10 years. It's way easier!!!!

[Arapajoe]

Thanks for the answers!


Why do you need to know wind direction and speed? All you need to do to land on a waypoint is to know the direction of the runway, the end of the runway, the vehicle position, and the vehicle ground track. All of this can be done with GPS. In order to align the vehicle on the runway during flare it would be best to know vehicle heading as well as vehicle track. No need for wind speed or direction. You should be able to tell if you're making a downwind landing by using GPS and pitot pressure. If your ground speed is higher than your airspeed you're going with the wind. Land the other direction.

No big deal.
The problem with GPS is accuracy.
The solution I've thought includes a Laser Range finder, for altitude. I would like to make it real autonomous, so waypoints doesn't look like a good solution at the moment...
I'ven been thinking in something like beacons on earth and triangulation, but that's not clear yet...
And I would like to introduce windspeed measuring so I could avoid dangerous manoeuvres near the ground...

[jetblackaircra]

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

Have a GPS receiver at your ground station near your landing site and you will be very accurate. With a cheap 5Hz GPS receiver in your airplane and an even cheaper 1Hz receiver at your ground station you should be able to do very accurate landings with GPS only. Throw in a very very inexpensive sonic range finder for altitude and you're golden.

The inaccuracies in GPS come from satellite location drift or a drift in the estimated position of the satellite. If you use a ground based receiver which you know the surveyed location of and can come up with a correction factor to send to the aircraft you can get incredible accuracy.

[matttay]

[url]With a cheap 5Hz GPS receiver in your airplane and an even cheaper 1Hz receiver at your ground station you should be able to do very accurate landings with GPS only.

While on a car trip this year and while riding in the navigator's seat, I wrote a lot of code on the laptop to log GPS from two sources. One was a phone using QCOM GPS, another was a Microsoft USB dongle that I think has a SiRF chipset.

In the code I time aligned each reading so that I was comparing the same GPS times on each, and then plotted distance between readings. The difference was significant and highly dependent on speed. Thus, I'd argue that to really dial out the errors you should use identical GPSs to ensure the same filtering on the data. Also, I'd imagine the reading could vary quite a bit if one unit saw fewer sats than the other

[dmgoedde]

The inaccuracies in GPS come from satellite location drift or a drift in the estimated position of the satellite. If you use a ground based receiver which you know the surveyed location of and can come up with a correction factor to send to the aircraft you can get incredible accuracy.That is only a slice of the error sources. Bigger yet is distortion of the GPS signals by the ionosphere. Your description of home-made differential GPS only works if both receivers are locked onto the same exact satellites, otherwise it is a hairy mess. If you have the same brand of GPS in plane and on ground then have a better chance of same satellites being locked to both, but given we all know GPS lock can change from 10 to 6 sats in a few heartbeats, it is a long shot. Even worse if you use 5Hz GPS from one manuf in the plane and a 1Hz on the ground.

The best I have seen is with true differential GPS corrections when I can coax the Locosys LS20031 (Atto's GPS) into lock mode "2" which means it is receiving differential corrections from a local ground station. When this lock type is engaged I see no drift in position at all when plane is sitting on the ground not moving, whereas you normally see drift in a type of random walk fashion.

[jetblackaircra]

What kind of error were you seeing? Keep in mind with both units moving you will see twice the error (or near twice). A stationary GPS receiver with a fix on more than 5 satellites should have a very accurate position fix which will become more accurate as time passes.

The faster the GPS receiver is moving, the less accurately you can measure it's position.

And what kind of accuracy are you trying to get? If you're wanting to measure inches then GPS is not the correct tool. But to line up on a runway centerline and glideslope to within a few feet. Very doable.

The only issues I've ever experienced with GPS position error is a fairly constant bias that changes slowly over time depending on what satellites are visible and where they are in the sky. Keep in mind though I'm talking about accuracy +/- a foot, not inch.

[Cort]

Even better...why not go all the way and use RTK GPS. You have to have a GPS that will handle it but there are many that do. UBlox makes a model that is affordable and can do RTK. Well sub-meter accuracy on the fly; if you loose lock you can then default to differential. I'm not quite sure why you would not utilize GPS and a way point for your landing.

1)initialize aircraft (GPS included) at takeoff and landing point. Make sure platform is aligned in the desired heading for take off and landing. AP inserts that point and heading into its landing thread or pattern or whatever you want to call it.
2) fly your flight
3) activate landing thread. Platform moves to a designated landing entry point. I find these are best set in arbitrary coordinates from the take off point. Meaning just specify and X, Y and Z from the initialization point.
4) platform flys a downwind and base leg (set up however you like). I would suggest using comparisons of groundspeed vs airspeed.
5) final leg (same heading as take off) is where it gets more complicated. you can just choose a glide slope but this is where wind can really mess you up. I think a simple fly to waypoint command with other commands for engine and platform configuration work best. Comparison of ground speed and airspeed are essential.

obviously this gets a lot more complicated if you want to land somewhere different than you took off. But it is do able especially with GPS

sorry I guess i repeated some of what others said....but it makes perfect sense and works real well in practice

[billh117]

There are commercial DGPS systems that are used for things like dredging, cranes positioning, farming, etc. that can be accurate to within Millimeters.

From what I have seen these things weigh pounds and cost thousands but the concept holds true.

I saw a demo of a UAV copter that used several earth stations on spikes and a swarm like software to keep exact position of the aircraft in reference to the ground stations. It landed within 3 inches of its target.

[The Touch]

RTK is a form of DGPS and an RTK solution is what would be required in this situation. I have considered developing an RTK solution for my project as WAAS is not available where I live. Having the same receivers for the base and rover would simplify things.

RTK works by having the base over a known position which is the coordinates and RL (height) are known. Both receivers also have to be locked or atleast only using data provided by the same satellites. The the error between the actual position of the base station and the calculated position can be measured and a correction can be applied to the rover or aircraft.

The hard part is being able to control which satellites the receivers use for their position calculations as well as being able to setup over a known point. There are services available where I live to be able to determine a position within cm accuracy if you provide static survey data in RINEX format which unfortunately most common GPS modules do not support.

[shanghai_fool]

I am starting to think the old fashioned ILS/glideslope may be simpler.