Re: [Flightgear-devel] Real flight playback with FG, based on recorded GPS data
I added the "GPS data recorder"under ftp://ftp.uni-duisburg.de/FlightGear/Misc_ols/src_03 and the "filtered"data playbackvideoftp://ftp.uni-duisburg.de/FlightGear/Misc_ols/LFBR_Landing_FG_03.AVI (to be compared to "rought" data playback ftp://ftp.uni-duisburg.de/FlightGear/Misc_ols/LFBR_Landing_FG_01.AVI and real flight video ftp://ftp.uni-duisburg.de/FlightGear/Misc_ols/LFBR_Landing_Real.AVI) I've filght tested this new "filtered" algorithm this afternoon and it's still "un-flyable" with 1 Hz GPS data stream: 3 seconds latency is way too much for attitude display. I'm eager to finish my home made MEMS IMU to definitively get rid of this latency. To flit from one subject to another, does anyone know if it would be easy/possible to displaythe flight plan in a "high way in the sky / tunnel" maner? Olivier - Original Message - From: Olivier Soussiel To: FlightGear Sent: Friday, November 12, 2004 1:33 AM Subject: [Flightgear-devel] Real flight playback with FG,based on recorded GPS data I finallely ended up using a Kalman filter to enhancePitch and Roll restitutionforreal flight playback, based on GPS data. I used the open source Bayesian Filtering Classes : Bayes++, in conjunction with Boost for the linear algebra part. Both are excellent libraries, for more details see http://bayesclasses.sourceforge.net/Bayes++.html. I gathered my source code on Martin's FTP site : ftp://ftp.uni-duisburg.de/FlightGear/Misc_ols/src_03 I'll re-videotaped the synthetic Flight Gear playback of the Real LFBR landing so that we can compare "rought" data playback with the "filtered" one. Olivier ___Flightgear-devel mailing list[EMAIL PROTECTED]http://mail.flightgear.org/mailman/listinfo/flightgear-devel2f585eeea02e2c79d7b1d8c4963bae2d ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
Re: [Flightgear-devel] Real Flight PLayback ?
Curtis L. Olson wrote: Martin Spott wrote: If you spend $6k at X-Bow you already get a FOG-based IMU. This is the sort of things I was looking at for my partially autonomous heli (see below), but this sort of equipment out of the price range I can pay. Yeah, that's pretty tough for most people's home budget. I imagine you've looked at the autopilot.sf.net project. They claim to have hovered a heli, but they don't seem to have had much activity in the last year or two. I assume I'm the guy who made you aware of the 'autopilot' project ;-) Yes, they've hovered a heli. Even though this is already a difficult task not only for a 'manual' pilot but also for a computer, this is by far not sufficient for my needs. I need the computer to stabilize the heli in situations where the demands go beyond the pilots skills As long as the heli hovers more or less parallel to the surface, things are quite 'easy' because you move within well defined limits. When you start employing significant tilt angles and accelerations and go over tens of minutes of flight, then you get to the point where the 'repeat accuracy' of affordable solid-state gyros and accelerometers makes the whole thing unusuable. I must admit that I don't have a real proof for this theory, but this is the conclusion after several different mathematical approaches to this subject over the past ten years or so The FMA Direct Co-Pilot Norman mentioned has been used with some success for stabalizing R/C helicopters. When you start to get in trouble, just center the controls and give it throttle. To be honest: I don't trust this device enough to make the 'life' of my experimental heli project depend on it - but probably I simply should give it a try ;-) Cheers, Martin. -- Unix _IS_ user friendly - it's just selective about who its friends are ! -- ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] Real Flight PLayback ?
Martin Spott wrote: [ autopilot.sf.net ] Yes, they've hovered a heli. Even though this is already a difficult task not only for a 'manual' pilot but also for a computer, this is by far not sufficient for my needs. I need the computer to stabilize the heli in situations where the demands go beyond the pilots skills As long as the heli hovers more or less parallel to the surface, things are quite 'easy' because you move within well defined limits. When you start employing significant tilt angles and accelerations and go over tens of minutes of flight, then you get to the point where the 'repeat accuracy' of affordable solid-state gyros and accelerometers makes the whole thing unusuable. Yes, I've heard that the lower end IMU stuff (i.e. $100,000) is pretty bad in terms of fine grained accuracy ... probably fine for stabalizing an aircraft and doing navigation, but pretty poor if you need degree-level accuracy. I am [currently] interested in simple level flight of a fixed wing aircraft and simple point to point navigation. I the the FMA direct copilot will work well for me, but it sounds like your interests/needs are in a totally different league. [ co-pilot ] To be honest: I don't trust this device enough to make the 'life' of my experimental heli project depend on it - but probably I simply should give it a try ;-) People are using it to stabalize helis. I've read that it is scary how well it works. I haven't tried it myself. Also, in the case of a helicopter, stabalized flight and hover are two entirely different things. As I understand it, helicopter pilots are using this device as a way to save their butts when they get crossed up on the controls. Center the controls, add power, and the helicopter should level out ... enough to give the pilot time to get reoriented and take back active control. To me, this device doesn't sound like it would be very useful for your needs, but perhaps if you hacked it up and interfaced to the sensors directly, you could get something out of it? It works on the IR temperature differential between sky and ground and (as I understand) the sensor has a fairly wide swath, so for small deviations for horizontal, you may be able to get some accurate data if you look directly at the sensor output ... but then if you are doing this with a helicopter and are near the ground, I have to wonder if trees and buildings or terrain would contribute enough error in the sensor differential to mess you up. But what do I know ... It's 5am here and I haven't had nearly enough sleep. :-) Curt. -- Curtis Olsonhttp://www.flightgear.org/~curt HumanFIRST Program http://www.humanfirst.umn.edu/ FlightGear Project http://www.flightgear.org Unique text:2f585eeea02e2c79d7b1d8c4963bae2d ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] Real Flight PLayback ?
Curtis L. Olson wrote: Martin Spott wrote: Yes, I've heard that the lower end IMU stuff (i.e. $100,000) is pretty bad in terms of fine grained accuracy ... There are two steps between solid-state- and $100k IMU's. _Simple_ FOG's (Fibre Optic Gyros) ($6k at X-Bow) translate a moving picture (convergent or divergent circles, see Michelson Interferometer, http://www.physics.uq.edu.au/people/mcintyre/applets/optics/michelsc.html) into modulated frequency and from threre to an analogue signal, sometimes with a calibrated A/D added to the output. Despite the translation errors the result is pretty usuable, even for manned aircraft. The real thing (TM) employs a digital counter to interpret the interferometer picture which makes this method nearly error-free. I assume these devices are in the $100k pice range. [ co-pilot ] To me, this device doesn't sound like it would be very useful for your needs, but perhaps if you hacked it up and interfaced to the sensors directly, you could get something out of it? It works on the IR temperature differential between sky and ground and (as I understand) Yes, but for some circumstances the co-pilot is simply too error-prone. I need for a system that doesn't lets the heli go over the top when it faces the infraread CO2 spectrum near a tree or a wall ;-) Cheerio, Martin. -- Unix _IS_ user friendly - it's just selective about who its friends are ! -- ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] Real Flight PLayback ?
Curtis L. Olson wrote: Exactly! I'm trying to decide if I want to mount the IR sensor on some sort of gimble and vary the sensor attitude to control the aircraft attitude, or do I just want to hard mount the sensor and try and steer with the rudder while the copilot fights to keep the wings level ... So the idea is that you would steer the rudder, then the copilot would bank the plane just enough for a coordinated turn? I don't think that would work -- you'd probably end up spinning or rolling the plane. All the best, David ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] Real Flight PLayback ?
David Megginson wrote: Curtis L. Olson wrote: Exactly! I'm trying to decide if I want to mount the IR sensor on some sort of gimble and vary the sensor attitude to control the aircraft attitude, or do I just want to hard mount the sensor and try and steer with the rudder while the copilot fights to keep the wings level ... So the idea is that you would steer the rudder, then the copilot would bank the plane just enough for a coordinated turn? I don't think that would work -- you'd probably end up spinning or rolling the plane. Not quite. The co-pilot does one thing (but one thing well from what I read). It keeps itself level with the horizon. That's it. It has no gyros. It fits inline between your receiver and your servos. So as I understand it, it will only kick in when you have your sticks centered. The question is, with an auto-pilot forcing the wings level, how effective would rudder only steering be. I'm guessing not all that effective, but at R/C scales is it effective enough to control heading and self navigate? Curt. -- Curtis Olsonhttp://www.flightgear.org/~curt HumanFIRST Program http://www.humanfirst.umn.edu/ FlightGear Project http://www.flightgear.org Unique text:2f585eeea02e2c79d7b1d8c4963bae2d ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
RE: [Flightgear-devel] Real Flight PLayback ?
Curtis L. Olson writes: Not quite. The co-pilot does one thing (but one thing well from what I read). It keeps itself level with the horizon. That's it. It has no gyros. It fits inline between your receiver and your servos. So as I understand it, it will only kick in when you have your sticks centered. I haven't examined the specs closely but back to the original question my guess is that instead of wiring this into the servos one could record the output of the device and that assuming proper calibration tests had been done one could translate this signal to attitude. I am also assuming that one is recording GPS output to include time into the same data stream. This would make a slick inexpensive package that could potentially have many commercial applications :-) hint of course it would be even better if there was an OEM version that could be coupled with an OEM GPS chip ... kalman filters etc .. /hint Cheers Norman ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] Real Flight PLayback ?
Curtis L. Olson wrote: The question is, with an auto-pilot forcing the wings level, how effective would rudder only steering be. I'm guessing not all that effective, but at R/C scales is it effective enough to control heading and self navigate? It would give you a very rough, slightly slipping turn, I'd think. There'd be a risk of a snap roll (and possible spin) at slower speeds, if you got to the point that one of the wings stalled. All the best, David ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] Real Flight PLayback ?
David Megginson wrote: It would give you a very rough, slightly slipping turn, I'd think. There'd be a risk of a snap roll (and possible spin) at slower speeds, if you got to the point that one of the wings stalled. That should have been a skidding turn. All the best, David -- Majority rule only works if you're also considering individual rights. Because you can't have five wolves and one sheep voting on what to have for supper. --Larry Flynt ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] Real Flight PLayback ?
Norman Vine wrote: Yup, using 3 differential GPS units has become a 'standard' way of doing this http://www.fig.net/pub/proceedings/korea/full-papers/pdf/session17/stathas-karabelas-liapakis-psarianos-kontaratos.pdf http://waas.stanford.edu/~wwu/papers/gps/PDF/DeLorenzoIEEEPLANS04.pdf Cool, the UMN professor I am working with is listed as a reference for the stanford paper. Small world ... however with a little ingenuity one could build a satisfactory system around this neat device for *much* less money https://www.fmadirect.com/site/Detail.htm?item=1489section=20 Exactly! I'm trying to decide if I want to mount the IR sensor on some sort of gimble and vary the sensor attitude to control the aircraft attitude, or do I just want to hard mount the sensor and try and steer with the rudder while the copilot fights to keep the wings level ... (?) But add an isopod ( $100 from newmicros.com) and an OEM GPS ( $100 from garmin) and some elbow grease and you'd be getting pretty close to being able to do 100% autonomous flight. A radio modem, ground station, and onboard sensor suite would also be nice. As well as onboard wireless video camera, directional antennas, etc. etc. etc.) Continuing to be off topic. Take a look at the July 9-10, 2004 entry at my Mariner web site: http://www.flightgear.org/~curt/Models/Mariner/ Curt. -- Curtis Olsonhttp://www.flightgear.org/~curt HumanFIRST Program http://www.humanfirst.umn.edu/ FlightGear Project http://www.flightgear.org Unique text:2f585eeea02e2c79d7b1d8c4963bae2d ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] Real Flight PLayback ?
Martin Spott wrote: If you spend $6k at X-Bow you already get a FOG-based IMU. This is the sort of things I was looking at for my partially autonomous heli (see below), but this sort of equipment out of the price range I can pay. Yeah, that's pretty tough for most people's home budget. I imagine you've looked at the autopilot.sf.net project. They claim to have hovered a heli, but they don't seem to have had much activity in the last year or two. This was the primary reason why _I_ was aiming at a working IMU: I had the desire to have a large, red button placed on the remote to let the heli rest at whatever position it currently had. The downside: If the system doesn't work out and you crash the aircraft then you not only have to repair the craft but also have to go for another $6k to replace the IMU The FMA Direct Co-Pilot Norman mentioned has been used with some success for stabalizing R/C helicopters. When you start to get in trouble, just center the controls and give it throttle. The co-pilot should right the heli and if you have throttle you should then be climbing away from trouble where you have time and space to collect yourself (so to speak, haha ...) and try again. It won't hold a hover in the wind, and it can take some effort to get it calibrated so you don't have a lot of extraneous drift, but it sounds like people have been using it with good success. Seems like a pretty nifty device, I've read mostly good things about it on the net. I've been watching ebay ... :-) Curt. -- Curtis Olsonhttp://www.flightgear.org/~curt HumanFIRST Program http://www.humanfirst.umn.edu/ FlightGear Project http://www.flightgear.org Unique text:2f585eeea02e2c79d7b1d8c4963bae2d ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] Real Flight PLayback ?
On Fri, 09 Jul 2004 09:34:04 +0100, Mat wrote in message [EMAIL PROTECTED]: Hi All, Hoping to pick up on flying lessons again and was wondering if I can record NMEA flight data of my lessons and play it back in FG ? ..oughtta work. Some gps gear also record static air pressure and temperature, if you can get a remote sensor into a wee cup, total pressure is possible too. To add to this does anyone know of a way of recording bank angle / glide path in a format usable by FG ? Playback of real flights would be a nice feature. ..only hardware ideas; videotaping the panel, or make a gyro black box using model gyros and record data off that, record local weather, static and total pressures or the instrument readings. ..other ideas includes using 2 or more gps units and their relative positions on the airframe, to calculate airframe attitudes. ..local weather will _always_ be different. ;-) -- ..med vennlig hilsen = with Kind Regards from Arnt... ;-) ...with a number of polar bear hunters in his ancestry... Scenarios always come in sets of three: best case, worst case, and just in case. ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] Real Flight PLayback ?
Mat Churchill wrote: Hoping to pick up on flying lessons again and was wondering if I can record NMEA flight data of my lessons and play it back in FG ? To add to this does anyone know of a way of recording bank angle / glide path in a format usable by FG ? Playback of real flights would be a nice feature. Most GPS units will save a track, but only a few will save altitude information -- I know that the Garmin 295 does, and probably the 195, 196, and 295 as well. Assuming coordinated flight, bank angle can be calculated from airspeed and rate of turn (probably using a moving average), but I don't know if there's code you can use for that anywhere in FlightGear yet. You'll probably need to download the track information from the GPS in a proprietary format and convert it; alternatively, you could hook up the GPS to a notebook or other portable computing device and simply save the NMEA output as it comes out. That sounds like quite a bit of fuss, though, when you're already worrying about flying lessons. In fact, the instructor may not want the GPS in a position that you can see it, and if it's down in your flight bag or back in the luggage compartment, it might not be able to receive the satellite signals. Best of luck getting back, David ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] Real Flight PLayback ?
Mat Churchill wrote: Hoping to pick up on flying lessons again and was wondering if I can record NMEA flight data of my lessons and play it back in FG ? To add to this does anyone know of a way of recording bank angle / glide path in a format usable by FG ? Playback of real flights would be a nice feature. Other's have shared some good ideas, but here's one I haven't seen. People have experiented with mounting 3 gps's on an aircraft. They find that the absolute error of all three gps's are subject to traditional limitations, *but* they all see the same view of the sky and they same satellite arrangement so their *relative* errors are very small. People have had really good luck deriving attitude information from the 3 gps receivers. I'm told the antennas can be mounted relatively close to each other (i.e. a couple feet) and still get pretty good accuracy. I even saw a video tape of a real approach with actual video show next to a synthetic view and the two synced amazingly well. I have heard that companies make triple antenna units specifically to do this, and it's my understanding that you can just use 3 separate gps units. If you could get units that output at 10hz (rather than 1 or 0.5 hz) and did a little smoothing, I think you could end up with some very nice results. But you would need to hard mount the gps antennas and do some very careful calculation of their relative positions, and finally do a bit of number crunching. The other solution as others have mentioned is to get an IMU. Crossbow makes a nice unit that can run in a mode that should spit out absolute roll/pitch/yaw. I don't understand it all, but somehow it's able to factor in the gravity vector + fancy math (kahlman filtering?) to self correct and continuously compensate for gyro drift. http://autopilot.sf.net is probably the lowest cost (but most do-it-yourself) IMU solution. If you look around (u-nav.com, micropilot.com, etc.) there are a number of companies out there making IMU solutions that hit several different price ranges (and presumably a large range of quality/accuracy.) This is an area where you get what you pay for. And unfortunately you are probably looking at thousands of as the entry level just to get something that works. The really high end inertial navigation units are more in the hundreds of thousands of dollars range. This is probably a bit off topic for FG (except for the playback part) but it's something I'm currently interested in, so keep talking. :-) Another interesting option would be to put some sort of radio modem on board and you could have people on the ground monitoring your flight in real time. That would be good for a lot of geek points if nothing else. Such a system would also be good for aerobatics competitions and practice (good for a pilot to review his flight and find flaws and areas of improvements, but also good for judges?) Lot's of possible fun here for a geek who can solder and design circuits. Unfortunately I am mostly software guy. Oh, and now to really get off topic. I have some ideas and have started assembling some equipment to build a self navigating autopilot for an R/C aircraft. Initially I see it as a safety system and for pilot training ... it could kick in and return the aircraft home if contact with the transmitter is lost. It could prevent an aircraft from exceeding some maximum radius from home to enforce a safe visual and radio range. It could enforce minimum and maximum altitude thresholds ... again for safety. It could fly an infinite holding pattern if you need to go take a whiz. I have all sorts of ideas. :-) With some luck and a few onboard telemetry type sensors you could imagine it detecting a variety of failure scenarios and taking reasonable action to minimize damage or avoid a crash altogether. I'm planning an IMU-less solution right now since I'm on a very limited personal hobby budget. If there is new ground to be broken here, I want to demonstrate a self navigating autonomous vehicle (very stable, reliable, and robust) that can be constructed for a small (or relatively small) amount of money. Let's say the cost of a decent R/C aircraft setup including radio and engine and airframe is about $400. I'm hoping that for another $400 I can make it fully autonomous and self navigating ... finally, I'm hoping to be able to actually scrounge up the money so I can actually do this by spreading the cost over many months and keep the total costs down by watching ebay and other used market places ... buying reasonable quality from the standpoint of safety, but not being afraid of used either. And, hey, I'm not asking for donations since this is for my own fun, but I probably wouldn't turn you away if you wanted to help or participate at some level. :-) Regards, Curt. -- Curtis Olsonhttp://www.flightgear.org/~curt HumanFIRST Program http://www.humanfirst.umn.edu/
Re: [Flightgear-devel] Real Flight PLayback ?
Curtis L. Olson wrote: If you look around (u-nav.com, micropilot.com, etc.) there are a number of companies out there making IMU solutions that hit several different price ranges (and presumably a large range of quality/accuracy.) This is an area where you get what you pay for. And unfortunately you are probably looking at thousands of as the entry level just to get something that works. The really high end inertial navigation units are more in the hundreds of thousands of dollars range. If you spend $6k at X-Bow you already get a FOG-based IMU. This is the sort of things I was looking at for my partially autonomous heli (see below), but this sort of equipment out of the price range I can pay. Oh, and now to really get off topic. I have some ideas and have started assembling some equipment to build a self navigating autopilot for an R/C aircraft. Initially I see it as a safety system and for pilot training ... it could kick in and return the aircraft home if contact with the transmitter is lost. This was the primary reason why _I_ was aiming at a working IMU: I had the desire to have a large, red button placed on the remote to let the heli rest at whatever position it currently had. The downside: If the system doesn't work out and you crash the aircraft then you not only have to repair the craft but also have to go for another $6k to replace the IMU Cheers, Martin. -- Unix _IS_ user friendly - it's just selective about who its friends are ! -- ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
RE: [Flightgear-devel] Real Flight PLayback ?
Curtis L. Olson writes: Mat Churchill wrote: Hoping to pick up on flying lessons again and was wondering if I can record NMEA flight data of my lessons and play it back in FG ? To add to this does anyone know of a way of recording bank angle / glide path in a format usable by FG ? Playback of real flights would be a nice feature. Other's have shared some good ideas, but here's one I haven't seen. People have experiented with mounting 3 gps's on an aircraft. They find that the absolute error of all three gps's are subject to traditional limitations, *but* they all see the same view of the sky and they same satellite arrangement so their *relative* errors are very small. People have had really good luck deriving attitude information from the 3 gps receivers. I'm told the antennas can be mounted relatively close to each other (i.e. a couple feet) and still get pretty good accuracy. I even saw a video tape of a real approach with actual video show next to a synthetic view and the two synced amazingly well. I have heard that companies make triple antenna units specifically to do this, and it's my understanding that you can just use 3 separate gps units. Yup, using 3 differential GPS units has become a 'standard' way of doing this http://www.fig.net/pub/proceedings/korea/full-papers/pdf/session17/stathas-karabelas-liapakis-psarianos-kontaratos.pdf http://waas.stanford.edu/~wwu/papers/gps/PDF/DeLorenzoIEEEPLANS04.pdf however with a little ingenuity one could build a satisfactory system around this neat device for *much* less money https://www.fmadirect.com/site/Detail.htm?item=1489section=20 Cheers Norman ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel