highaltitude.log.20081121

[01:48] edmoore (n=edmoore@chu-gw.churchillcambridge.co.uk) left irc:
[02:09] edmoore (n=edmoore@pomegranate.chu.cam.ac.uk) joined #highaltitude.
[02:11] <Laurenceb> if(!(((u08)*address)&0x7F)) //check the address for overflow for the 7 LSB
[02:11] <Laurenceb> ^ will that code do what it says in the comment?
[02:11] <Laurenceb> address is a u32
[02:11] edmoore (n=edmoore@pomegranate.chu.cam.ac.uk) left irc: Client Quit
[02:19] fnoble (n=fnoblef@fn217.quns.cam.ac.uk) joined #highaltitude.
[02:26] <Laurenceb> fnoble: greetings
[02:27] <fnoble> hello
[02:27] <Laurenceb> whats the clock speed on badger?
[02:27] <fnoble> 60mhz
[02:27] <Laurenceb> I'm thinking about IMU stuff
[02:27] <Laurenceb> hmm ok
[02:28] Action: Laurenceb thinks about it
[02:28] <Laurenceb> hmm I think my idea might be possible
[02:28] <Laurenceb> at >50Hz
[02:28] <Laurenceb> are you still going with the lassen iq?
[02:29] <fnoble> yeah, although at some point i'll put the footprint for the ublox 5 on there too, so it can use both
[02:29] <fnoble> sorry, either, not both at the same time :)
[02:29] <Laurenceb> cool, I havent read the ublox5 datasheet very well yet
[02:30] <Laurenceb> but lassen iq you can get out pseudorange and doppler
[02:30] <fnoble> its nice, i played with one henry had
[02:30] <fnoble> very fast aquisition
[02:30] <Laurenceb> yeah i like it
[02:30] <fnoble> yeah, im not sure what you can get out of the ub5
[02:30] <Laurenceb> the sensitivity isnt really good
[02:30] <Laurenceb> but once it gets a signal its amazing
[02:30] <Laurenceb> have you seen my test data?
[02:31] <fnoble> this the lassen or the ublox?
[02:31] <Laurenceb> lassen
[02:31] <fnoble> nope, not seen any data
[02:32] <fnoble> link?
[02:32] <Laurenceb> http://wiki.ukhas.org.uk/code:parafoil_tests
[02:33] <Laurenceb> yeah, my plan was to use the pseudorange and doppler as position and velocity inputs to the IMU filter
[02:34] <Laurenceb> in fact.... I'm still not sure but I think you might be able to extract phase, in which case 4 lassens could be used to get attitude
[02:34] <fnoble> hehe
[02:35] <fnoble> would be interesting
[02:35] <Laurenceb> I think you could only get a rate output from them
[02:35] <Laurenceb> without some mods ...
[02:36] <Laurenceb> probably making them run off the same TCXO
[02:36] <Laurenceb> yeah if they ran off the same TCXO it may work
[02:37] <Laurenceb> that would be a _relatively_ easy soldering job
[02:37] <fnoble> would be a good experiment to try
[02:38] <fnoble> ok, im off
[02:38] <fnoble> bye
[02:38] <Laurenceb> cya
[02:38] Nick change: fnoble -> fnoble|away
[03:04] <Kornholijo> :p
[03:35] fnoble|away (n=fnoblef@fn217.quns.cam.ac.uk) left irc: Read error: 110 (Connection timed out)
[03:58] Laurenceb (i=83e3dd6d@gateway/web/ajax/mibbit.com/x-851ff57c807f6b8b) left irc: "http://www.mibbit.com ajax IRC Client"
[07:14] Simon-MPFH (n=simon@phantom.mpfh.co.uk) joined #highaltitude.
[07:54] edmoore (n=edmoore@pomegranate.chu.cam.ac.uk) joined #highaltitude.
[08:07] edmoore (n=edmoore@pomegranate.chu.cam.ac.uk) left irc:
[09:30] edmoore (n=edmoore@pomegranate.chu.cam.ac.uk) joined #highaltitude.
[09:56] edmoore (n=edmoore@pomegranate.chu.cam.ac.uk) left irc: Read error: 113 (No route to host)
[12:01] borism (n=boris@195-50-204-153-dsl.krw.estpak.ee) left irc: Client Quit
[13:46] edmoore (n=edmoore@chu-gw.churchillcambridge.co.uk) joined #highaltitude.
[14:32] Kornholijo (i=kornholi@c-f5b9e253.09-99-73746f40.cust.bredbandsbolaget.se) left irc: Remote closed the connection
[14:58] edmoore (n=edmoore@chu-gw.churchillcambridge.co.uk) left irc:
[17:36] natrium42 (n=alexei@72.136.18.113) left irc: Remote closed the connection
[17:39] edmoore (n=edmoore@pomegranate.chu.cam.ac.uk) joined #highaltitude.
[17:46] Hiena (n=Hiena@81.93.195.181.datatrans.hu) joined #highaltitude.
[17:53] <EI5GTB> evening
[18:43] edmoore (n=edmoore@pomegranate.chu.cam.ac.uk) left irc:
[19:30] jcoxon (n=jcoxon@host86-163-198-244.range86-163.btcentralplus.com) joined #highaltitude.
[19:39] jcoxon (n=jcoxon@host86-163-198-244.range86-163.btcentralplus.com) left irc: "This computer has gone to sleep"
[20:41] Hiena (n=Hiena@81.93.195.181.datatrans.hu) left irc: "-=Halt! Hammerzeit!=-"
[21:11] jcoxon (n=jcoxon@host86-163-198-244.range86-163.btcentralplus.com) joined #highaltitude.
[21:14] jcoxon (n=jcoxon@host86-163-198-244.range86-163.btcentralplus.com) left irc: Client Quit
[21:22] Ei5GTB_ (n=Paul@213-202-176-223.bas504.dsl.esat.net) joined #highaltitude.
[21:30] Simon-MPFH (n=simon@phantom.mpfh.co.uk) left irc: "Leaving"
[21:39] Laurenceb (i=83e3dd6d@gateway/web/ajax/mibbit.com/x-af6c97507ed6493f) joined #highaltitude.
[21:40] EI5GTB (n=Paul@78.16.225.144) left irc: Read error: 110 (Connection timed out)
[21:49] Nick change: Ei5GTB_ -> EI5GTB
[22:39] Ebola (n=Ebola@unaffiliated/ebola) joined #highaltitude.
[22:48] borism (n=boris@195-50-204-153-dsl.krw.estpak.ee) joined #highaltitude.
[22:51] natrium42 (n=alexei@CPE000625d867e2-CM0014045885be.cpe.net.cable.rogers.com) joined #highaltitude.
[22:58] edmoore (n=edmoore@pomegranate.chu.cam.ac.uk) joined #highaltitude.
[23:01] <Laurenceb> hi edmoore
[23:01] edmoore (n=edmoore@pomegranate.chu.cam.ac.uk) left irc: Client Quit
[23:02] <EI5GTB> Laurenceb, was messing with matlab, you have the txt file with the raw data?
[23:02] <Laurenceb> yeah, just a sec
[23:02] <Laurenceb> do you have my script?
[23:03] <EI5GTB> yup, the one you pastebinned a while ago
[23:05] <Laurenceb> http://pastebin.com/m53b6f56e
[23:05] <Laurenceb> save that as rpy90bothways.TXT
[23:07] <EI5GTB> ok, copol cool, im getting graphs
[23:07] <EI5GTB> tnx
[23:12] <EI5GTB> is that script 3 state or 6?
[23:13] <Laurenceb> cant remember what your running
[23:13] <Laurenceb> is the graph in 3D?
[23:13] <EI5GTB> nope
[23:14] <Laurenceb> ah 3 state prob
[23:14] <EI5GTB> k
[23:15] <Laurenceb> http://www.mibbit.com/pb/Fv3j7U
[23:16] <Laurenceb> I got good results calling that with beta=0.000004, alpha=0.01, q=0.00003
[23:21] <Laurenceb> you can call the function from the maiin window
[23:22] <EI5GTB> yea, figured that
[23:22] <EI5GTB> after some time :P
[23:22] <EI5GTB> interesting graphm
[23:22] <Laurenceb> I find it easier to follow
[23:23] <Laurenceb> red green blue is the ends of the x y z axes
[23:23] <EI5GTB> so whats that graph outputting? 3 positions in space?
[23:23] <EI5GTB> ah, ok
[23:23] <Laurenceb> yeas
[23:24] <EI5GTB> (this is where the good understanding of the kalman filter comes in) what are the "alpha" "beta" and "q"
[23:25] <Laurenceb> alpha is accel noise, beta gyro noise and q is process noise
[23:25] <EI5GTB> aha
[23:26] <EI5GTB> ill get it eventually..
[23:26] <EI5GTB> ::P
[23:26] <EI5GTB> so how many states is that?
[23:27] <Laurenceb> 8
[23:27] <EI5GTB> cool
[23:27] <EI5GTB> sounds impressive :P
[23:29] <Laurenceb> I think the 3 state is good enough for a quadcopter
[23:30] <EI5GTB> but why have 3 state when you can have 8 ::P
[23:30] <Laurenceb> youd struggle to run it in real time
[23:30] <Laurenceb> even on arm7
[23:31] <EI5GTB> hmmmm
[23:31] <Laurenceb> wheras the 3 state will run on avr
[23:31] <EI5GTB> i seee..
[23:32] <EI5GTB> at the minute, my first goal is to get the thing to hover in one place
[23:32] <EI5GTB> once it can do that.. the battles one, the rest is trivial
[23:32] <Laurenceb> it would take some good piloting
[23:32] <Laurenceb> wit the three state
[23:33] <EI5GTB> yea, wheras i would like it to hovber with no pioloting
[23:33] <EI5GTB> piloting*
[23:34] <EI5GTB> but then, gps resoloution may not be good enough to hold a good hover
[23:34] <Laurenceb> it wont
[23:34] <EI5GTB> and i dont think it will be too possible to extract the accel. data from the thing
[23:34] <Laurenceb> you could use a ultrasonic rangefinder
[23:35] <Laurenceb> to hover
[23:35] <EI5GTB> hmm
[23:35] <EI5GTB> that covers height
[23:36] <Laurenceb> gps can give good velocity
[23:37] <EI5GTB> thats what im thinking, use gps as X,Y hold
[23:37] <EI5GTB> and then ultrasonic as Z hold
[23:37] <Laurenceb> might work to an xtent
[23:38] <EI5GTB> yea...
[23:38] <EI5GTB> im just wondering how much deviation you will get on the gps signal..
[23:38] <Laurenceb> ~5 to 10m
[23:39] <EI5GTB> heh, hmm
[23:39] <EI5GTB> OR, when you go into hover mode, could you filter the gps data
[23:39] <EI5GTB> i.e kalman..
[23:39] <Laurenceb> yes
[23:39] <EI5GTB> so in a few seconds it will stabalise....
[23:39] <EI5GTB> or does the gps give slow deviation
[23:40] <Laurenceb> unless your a good pilot and cn fly it with jut roll pitch yaw stabilisation
[23:40] <Laurenceb> you really need position and velocity estimation from the filter
[23:40] <EI5GTB> yea
[23:41] <EI5GTB> im thinking, when the user preses forward, the thing moves forward, when they let go, it hovers
[23:41] <EI5GTB> that sort of controll.
[23:42] <EI5GTB> but how to estimate velocity....
[23:43] <Laurenceb> use gps and the accelerometers
[23:44] <EI5GTB> accelerometers.. derived by the gyros
[23:44] <Laurenceb> eh?
[23:44] <Laurenceb> you have a three axis accel on there already
[23:45] <EI5GTB> yea
[23:45] <EI5GTB> oh
[23:45] <EI5GTB> but wont you need to take the accelerometer data with the actual attitude in mind
[23:46] <Laurenceb> yes
[23:46] <Laurenceb> but you can do that with a rotation matrix
[23:46] <EI5GTB> aaah, ok
[23:47] <EI5GTB> hmmm, its probably best for now if i use one avr to astimate attitude, another to astimate position/velocity and then feed them to a flight controll chip to controll motors and take control inputs etc.
[23:47] <EI5GTB> so 3 avrs
[23:47] <Laurenceb> the nice thing is you can go up to a volocity, attitude, position, acceleration, roll pitch yaw rate filter without any more sensors
[23:48] <Laurenceb> nope fraid not
[23:48] <Laurenceb> it all needs to run in one filter
[23:48] <Laurenceb> so you need a very fast processor
[23:48] <Laurenceb> with fpu
[23:48] <Laurenceb> its not for beginners
[23:48] <EI5GTB> hmmm, im figuring
[23:49] <Laurenceb> but you might have some luck with a three state filter and an avr
[23:49] <Laurenceb> you'll be able to fly it by hand if your experienced
[23:49] <EI5GTB> hmm, im not tho..
[23:49] <EI5GTB> i was aiming for a complete uav
[23:50] <EI5GTB> but i guesse i was a little ambitious
[23:52] <EI5GTB> grr, i hate not following through!
[00:00] --- Sat Nov 22 2008