highaltitude.log.20090918

[00:02] <SpeedEvil> Yeah - interesting, but not really at a place yet where I can do that.
[00:04] Action: SpeedEvil brb.
[00:04] SpeedEvil (n=user@tor/regular/SpeedEvil) left #highaltitude.
[00:05] SpeedEvil (n=user@tor/regular/SpeedEvil) joined #highaltitude.
[00:28] Matt_APEX (n=chatzill@87.112.25.101.plusnet.ptn-ag1.dyn.plus.net) left irc: Read error: 110 (Connection timed out)
[00:36] GeekShadow (n=Antoine@reactos/tester/GeekShadow) left irc: "The cake is a lie !"
[00:59] jiffe99 (n=jiffe2@209.159.247.189) joined #highaltitude.
[01:02] <jiffe99> so up around 30km, how worried does one have to be about electronics and battery performance with the cold?
[01:09] <SpeedEvil> Moderately.
[01:09] <SpeedEvil> An insulated box helps.
[01:09] <SpeedEvil> Lithium (primary) AA cells work OK.
[01:10] <SpeedEvil> A small heater - or electronics that produce the right amount of heat and a thick polystyrene box can work quite well.
[01:10] kichimi (n=kichimi@86.29.171.185) joined #highaltitude.
[01:10] <kichimi> anyone awake in here?
[01:13] <SpeedEvil> no
[01:13] <kichimi> sleep typing?
[01:14] <jiffe99> the electronics don't produce much heat, may have to insulate it well then
[01:16] <kichimi> i need to either a) learn how to weld, or b) get someone to fix my exaust for me
[01:16] <kichimi> it snapped off last night
[01:17] <kichimi> it took 10 seconds before the raw engine noises started to really hurt me ears
[01:17] <kichimi> pushing a bike 2-3 miles = not fun
[01:23] <kornholi> ;o
[01:25] darknesslord_ (n=darkness@187.146.35.161) left irc: "2nibbles!!"
[01:41] <SpeedEvil> jiffe99: pretty much - also it's easy to test to -20C in your freezer - -55C is unfortuantely harder
[01:43] <kichimi> SpeedEvil: im very lucky, i have access to a commercial freezer that goes to -35
[02:01] Laurenceb (n=laurence@host81-154-154-237.range81-154.btcentralplus.com) left irc: Remote closed the connection
[02:18] KingJ (n=KingJ-FN@oops.i.forgot.to.set.my.hostmask.kingj.net) left irc: "ZNC - http://znc.sourceforge.net"
[02:22] KingJ (n=KingJ-FN@oops.i.forgot.to.set.my.hostmask.kingj.net) joined #highaltitude.
[03:01] KingJ (n=KingJ-FN@oops.i.forgot.to.set.my.hostmask.kingj.net) left irc: Read error: 110 (Connection timed out)
[03:11] KingJ (n=KingJ-FN@oops.i.forgot.to.set.my.hostmask.kingj.net) joined #highaltitude.
[03:18] kornholi (n=kornholi@kornholi.net) left irc: Read error: 110 (Connection timed out)
[04:24] kornholi (n=kornholi@kornholi.net) joined #highaltitude.
[04:33] kornholi (n=kornholi@kornholi.net) left irc: Read error: 104 (Connection reset by peer)
[04:33] kornholi (n=kornholi@kornholi.net) joined #highaltitude.
[06:46] GeekShadow (n=Antoine@reactos/tester/GeekShadow) joined #highaltitude.
[06:54] GeekShadow (n=Antoine@reactos/tester/GeekShadow) left irc: "The cake is a lie !"
[07:38] Simon-MPFH (n=simon@phantom.mpfh.co.uk) joined #highaltitude.
[08:13] junderwood (n=john@adsl.jcu.me.uk) joined #highaltitude.
[08:13] icez (n=icez@unaffiliated/icez) left irc: Remote closed the connection
[08:21] rjharrison (n=rharriso@gateway.hgf.com) joined #highaltitude.
[08:59] <rjharrison> moring all
[08:59] <rjharrison> UKHAS is back up
[09:26] <gordonjcp> morning
[11:23] Matt_APEX (n=chatzill@87.113.91.113.plusnet.pte-ag2.dyn.plus.net) joined #highaltitude.
[11:26] Matt_APEX (n=chatzill@87.113.91.113.plusnet.pte-ag2.dyn.plus.net) left irc: Client Quit
[11:45] Laurenceb (n=laurence@host81-154-154-237.range81-154.btcentralplus.com) joined #highaltitude.
[12:18] <Laurenceb> http://www.rcgroups.com/forums/showthread.php?t=1004918&page=3
[12:21] <SpeedEvil> neat
[13:19] <Laurenceb> http://www.rcgroups.com/forums/showthread.php?t=1004918&page=3
[13:22] <Laurenceb> oops
[13:27] <SpeedEvil> still neat
[13:39] <Laurenceb> http://www.youtube.com/watch?v=8kl3ofWXBP8&feature=player_embedded
[13:41] <SpeedEvil> Yeah - stupid amounts of power lightweight are a game-changer for small airships
[13:44] <SpeedEvil> http://www.youtube.com/watch?v=RecckjgLyqo&feature=fvw
[13:44] <SpeedEvil> neat
[13:44] <SpeedEvil> pointless - but neat
[13:47] <Laurenceb> if you use v large ducted fan units and run them at low power you could get 30 minutes flight duration
[13:48] <Laurenceb> 2 x 4inch diameter fans or more for a 1Kg GLOW
[13:48] <SpeedEvil> I think big slow props are the way to go over ducted fans
[13:49] <SpeedEvil> So Ve is a couple of times the maximum speed or so - for best N/W
[13:49] <Laurenceb> http://www.hobbycity.com/hobbycity/store/uh_viewItem.asp?idProduct=4235&Product_Name=EDF_Ducted_Fan_Unit_7Blade_5inch_127mm
[13:49] <Laurenceb> yeah
[13:50] <Laurenceb> I like the two vectored thruster design
[13:51] <Laurenceb> its the 1/2mv^2 versus mv relationship
[13:52] <Laurenceb> you want to shift as much air as possible
[14:43] Laurenceb (n=laurence@host81-154-154-237.range81-154.btcentralplus.com) left irc: Remote closed the connection
[15:32] GeekShadow (n=Antoine@reactos/tester/GeekShadow) joined #highaltitude.
[15:43] natrium42 (n=alexei@CPE000625d867e2-CM0014045885be.cpe.net.cable.rogers.com) left irc: Remote closed the connection
[16:22] sbasuita (n=sbasuita@unaffiliated/drebellion) joined #highaltitude.
[16:23] edmoore (i=836f0142@gateway/web/freenode/x-zxbgruhmedxwxbxj) joined #highaltitude.
[16:43] <edmoore> rjharrison: ping a ding
[16:53] DanielRichman (n=daniel@unaffiliated/danielrichman) joined #highaltitude.
[16:56] GeekShadow (n=Antoine@reactos/tester/GeekShadow) left irc: Read error: 104 (Connection reset by peer)
[16:56] GeekShadow (n=Antoine@reactos/tester/GeekShadow) joined #highaltitude.
[17:07] mc- (n=mfcastle@cpc2-glfd1-0-0-cust559.glfd.cable.ntl.com) joined #highaltitude.
[17:07] <mc-> rjharrison, you're welcome to borrow my Art of Electronics - saw you talking about that a few days ago
[17:14] <DanielRichman> oh man.
[17:14] <DanielRichman> <-- is an idiotg
[17:19] <DanielRichman> 45MB of 0x2d just got written to the sd card due to a typo
[17:19] <DanielRichman> infinite loop; nice one!
[17:20] GeekShadow (n=Antoine@reactos/tester/GeekShadow) left irc: "The cake is a lie !"
[17:24] <sbasuita> DanielRichman, you should post that on facebook and see what sort of response you get
[17:24] <kichimi> hey
[17:24] <DanielRichman> erm no
[17:25] <DanielRichman> sbasuita, how many of my facebook friends would care?
[17:25] <sbasuita> DanielRichman, 0
[17:25] <DanielRichman> there you go
[17:25] <DanielRichman> so no, I won't.
[17:25] Action: DanielRichman retrieves his programmer.
[17:25] <DanielRichman> lezfix diz bug
[17:28] GeekShadow (n=Antoine@reactos/tester/GeekShadow) joined #highaltitude.
[17:29] <rjharrison> mc that would be most excellent
[17:29] <rjharrison> edmoore ping pong
[17:31] <rjharrison> DanielRichman what are you changing in the code?
[17:32] GeekShadow (n=Antoine@reactos/tester/GeekShadow) left irc: Read error: 104 (Connection reset by peer)
[17:32] <DanielRichman> rjharrison, it was a bug in my messages.c
[17:32] <DanielRichman> ie the get_data function
[17:32] <DanielRichman> the log.c code was working fine; just I messed up the message generation
[17:32] <rjharrison> np I don't think that effects the code I have :)
[17:32] <DanielRichman> nope
[17:33] <DanielRichman> now It appears i have to trudge outside and get a fix with this thing
[17:33] <rjharrison> DanielRichman about to send off brd for pcb fabrication
[17:33] <DanielRichman> rjharrison, ooh, nice, what's on it?
[17:33] <rjharrison> we have a lassenIQ gps emulator
[17:33] <DanielRichman> but I cba to hook it up
[17:33] <DanielRichman> ;)
[17:33] <DanielRichman> and it runs on windows if i remember correctly (eugh)
[17:34] <rjharrison> if you can send data into the Rx port via tty (Linux)
[17:34] <rjharrison> rocketboy wrote the code and I modded it for the iQ
[17:34] <rjharrison> http://www.robertharrison.org/svn/listing.php?repname=the-icarus-project&path=%2Ftesting-software%2Flinux%2F#path_testing-software_linux_
[17:34] <DanielRichman> I'd have to get out the arduino, use the ftdi chip, hack a connector or use crocodile clips or summthat.
[17:35] <DanielRichman> less effort to walk outside
[17:35] <DanielRichman> now it has a functioning radio I don't need to take a pc and can set it up and retire here while it gets a fix
[17:35] <rjharrison> True but the emulator is great for testing tracking over East / West boundries
[17:35] <DanielRichman> heh
[17:35] <DanielRichman> i just squashed that bug
[17:35] <DanielRichman> it was the minus sign that was infinite looping ;)
[17:36] <rjharrison> Yep and given the launch site there is a good chance of crossing that boundry
[17:37] <rjharrison> Perhaps not today buy def try the emulator and GPSgen the generator takes a kml file for input
[17:37] natrium42 (n=natrium4@129-97-192-192.uwaterloo.ca) joined #highaltitude.
[17:38] <DanielRichman> ah throw going outside
[17:38] <DanielRichman> let's hope there are enough sats near my window
[17:39] <natrium42> hi
[17:39] <rjharrison> which is great as you can view the theoretical plot against the actual payload transmission
[17:39] <rjharrison> hi natrium42
[17:39] <rjharrison> Hows it going
[17:40] <natrium42> good, lectures started
[17:40] <natrium42> how are you?
[17:41] <rjharrison> Great about to send of PCB to olimex
[17:41] jeev (n=email@unaffiliated/jeev) left irc: Excess Flood
[17:41] <natrium42> cool
[17:41] <rjharrison> well brd file for PCB production
[17:41] <natrium42> i hope the PCBs i ordered on wednesday arrive today
[17:41] <natrium42> it's for a usb motor controller
[17:42] <natrium42> so next icarus is on schedule?
[17:42] <rjharrison> ohh nice another project on
[17:42] <rjharrison> Yep should be launching in October to see what causes float conditions
[17:42] <SpeedEvil> :)
[17:42] <natrium42> awesome
[17:42] <kichimi> i start building mine next week
[17:43] <kichimi> i need a name
[17:43] <natrium42> btw, have you seen the $150 balloon thing that was in the news?
[17:43] <natrium42> they took your name :P
[17:43] <rjharrison> Yep from MIT get used of name
[17:43] <kichimi> natrium42: is that a reference to the greek myth?
[17:43] <rjharrison> I posted on thier blog and suggested they did a google first
[17:43] <natrium42> hi kichimi, haven't seen you before
[17:44] <rjharrison> kichimi yep
[17:44] <kichimi> natrium42: thats because i saw you first
[17:44] <DanielRichman> gah. No fix.
[17:44] Action: DanielRichman going outside; bbiab
[17:44] <natrium42> kichimi: lol, welcome to the chan
[17:45] Action: natrium42 pokes DanielRichman with a SHARPENED stick
[17:45] <rjharrison> lol
[17:45] Action: rjharrison pokes edmoore
[17:45] <kichimi> im not quite sure where im going to get a balloon and parachute from
[17:45] <rjharrison> edmoore ping pong
[17:45] <natrium42> kichimi: where are you based?
[17:45] <DanielRichman> oh ffs
[17:45] <kichimi> im going to a scrappy tmorrow to get some aluminium sheets and start making the payload box
[17:45] <DanielRichman> i go downstairs to get a tray to take it outside on
[17:45] <rjharrison> rocketboy on here sells both
[17:45] <DanielRichman> get back and guess what
[17:45] <DanielRichman> FIX!
[17:45] <kichimi> natrium42: colchester, essex
[17:46] <kichimi> natrium42: you?
[17:46] <DanielRichman> good gawd it's working
[17:46] <rjharrison> DanielRichman I have had that too many times
[17:46] <natrium42> kichimi: ontario
[17:46] <kichimi> nicey nicey
[17:46] <DanielRichman> and like, properly working
[17:46] <natrium42> kichimi: then rocketboy is your man
[17:46] <DanielRichman> it's only 5 meters away but fldigi is getting that rtty and decoding
[17:46] <natrium42> unless you want to import balloons yourself
[17:46] <DanielRichman> and the checksum confirms
[17:46] <DanielRichman> :D
[17:46] <DanielRichman> sbasuita, success
[17:46] <rjharrison> DanielRichman nice one sir
[17:46] <rjharrison> It's getting to the fun stuff now
[17:47] <kichimi> okay, is rocketboy a member of this chan or is it a site, im not sure at all, ive seen some of the most strange irc nicks before
[17:47] <rjharrison> Yep member
[17:47] <DanielRichman> five whole satellites
[17:47] <natrium42> steve randall
[17:47] <rjharrison> One sec and I'll send you the linnk
[17:48] <sbasuita> DanielRichman, should say something on the blog
[17:48] <sbasuita> just generally
[17:48] <natrium42> sbasuita: srsly
[17:48] <DanielRichman> sbasuita, oh man
[17:48] <natrium42> sbasuita: no updates for so long
[17:48] <DanielRichman> not gun writing it
[17:48] <sbasuita> i've got coursework to do
[17:48] <DanielRichman> natrium42, these two are too LAZY
[17:48] <DanielRichman> i write everything and I'm meant to be doing electronics
[17:48] <DanielRichman> it's AlexBreton's job!
[17:48] <sbasuita> well poke him them
[17:48] <DanielRichman> you can't; it bounces off; he's too fat
[17:49] Action: natrium42 gives DanielRichman his SHARPENED poking stick
[17:49] <natrium42> i got another one
[17:49] <DanielRichman> ah ok, cool
[17:51] <sbasuita> time for bed load angularity...
[17:51] <DanielRichman> oh fun stuff
[17:51] Action: DanielRichman points natrium42's SHARPENED poking stick in the general direction of Reading, hopes for the best, sets the end on fire and hurls
[17:51] <DanielRichman> I think I can hear a fat cat scream
[17:52] <rjharrison> kichimi http://randomsolutions.co.uk/top.htm
[17:52] <kichimi> rjharrison: thanks
[17:57] <rjharrison> kichimi if you need stuff for ballooning these link may be helpfull http://www.robertharrison.org/index.php?option=com_content&task=view&id=29&Itemid=78
[17:59] <rjharrison> Off home
[17:59] <rjharrison> bbiab
[17:59] rjharrison (n=rharriso@gateway.hgf.com) left irc:
[18:01] <DanielRichman> time to test the SMS
[18:01] <DanielRichman> okey that worked
[18:01] Action: DanielRichman is on a rol
[18:01] <DanielRichman> *roll -.-
[18:06] <natrium42> heh
[18:06] <natrium42> all well ed?
[18:07] Laurenceb (n=laurence@host81-154-154-237.range81-154.btcentralplus.com) joined #highaltitude.
[18:07] <Laurenceb> hi folks
[18:07] <natrium42> yo laurence
[18:09] <DanielRichman> what's going on?
[18:12] <edmoore> natrium42: yep thanks
[18:19] <Laurenceb> are there any launches soon?
[18:19] <edmoore> us on sunday probs
[18:26] ChezaWho (n=chezawho@adsl-99-38-181-246.dsl.pltn13.sbcglobal.net) joined #highaltitude.
[18:27] <Laurenceb> cool, whats flying?
[18:28] <DanielRichman> sbasuita, quashed the empty sms bug
[18:28] <Laurenceb> SpeedEvil: http://www.hobbycity.com/hobbycity/store/uh_viewItem.asp?idProduct=5582&Product_Name=Polyquest_4550mAh_3S_30C_-_50C_Lipoly_(Version_2) and http://www.hobbycity.com/hobbycity/store/uh_viewItem.asp?idProduct=668&Product_Name=TowerPro_BL__Outrunner_TP2410-12T(Y)__(90W)#
[18:28] <Laurenceb> would give you maybe 30minutes flight time
[18:28] <sbasuita> DanielRichman, what was it
[18:28] <DanielRichman> sbasuita, a simple coding-whilst-tired thing
[18:29] <DanielRichman> messages.c wasn't copying the latest_data struct to sms_data
[18:29] <DanielRichman> simnples
[18:29] <Laurenceb> gtg
[18:30] Laurenceb (n=laurence@host81-154-154-237.range81-154.btcentralplus.com) left irc: Remote closed the connection
[18:30] <DanielRichman> bbl
[18:34] <SpeedEvil> http://pdos.csail.mit.edu/6.828/2008/readings/ritchie79evolution.html
[18:37] natrium42 (n=natrium4@129-97-192-192.uwaterloo.ca) left #highaltitude.
[18:41] mc- (n=mfcastle@cpc2-glfd1-0-0-cust559.glfd.cable.ntl.com) left irc: Read error: 110 (Connection timed out)
[18:59] GeekShadow (n=Antoine@reactos/tester/GeekShadow) joined #highaltitude.
[19:06] natrium42 (n=alexei@CPE000625d867e2-CM0014045885be.cpe.net.cable.rogers.com) joined #highaltitude.
[19:09] ChezaWho (n=chezawho@adsl-99-38-181-246.dsl.pltn13.sbcglobal.net) left irc:
[19:10] DanielRichman (n=daniel@unaffiliated/danielrichman) left irc: Remote closed the connection
[19:25] Matt_APEX (n=chatzill@87.114.73.49.plusnet.thn-ag2.dyn.plus.net) joined #highaltitude.
[19:51] sbasuita (n=sbasuita@unaffiliated/drebellion) left irc: Remote closed the connection
[19:53] sbasuita (n=sbasuita@unaffiliated/drebellion) joined #highaltitude.
[19:59] rjharrison_ (n=rharriso@80.176.172.227) joined #highaltitude.
[20:07] <rjharrison_> ping edmoore
[20:20] jatkins (n=jatkins@79-76-58-73.dynamic.dsl.as9105.com) joined #highaltitude.
[20:20] jatkins (n=jatkins@79-76-58-73.dynamic.dsl.as9105.com) left irc: Client Quit
[20:29] SpeedEvil (n=user@tor/regular/SpeedEvil) left irc: "Leaving."
[20:30] SpeedEvil (n=user@tor/regular/SpeedEvil) joined #highaltitude.
[20:42] sbasuita (n=sbasuita@unaffiliated/drebellion) left irc: Read error: 104 (Connection reset by peer)
[21:27] <edmoore> rjharrison_: ping
[21:27] <kichimi> i feel like a caged animal
[21:28] <kichimi> i should be out
[21:28] <rjharrison_> hehe welcome to HAB
[21:28] Action: kichimi backs away
[21:30] Simon-MPFH (n=simon@phantom.mpfh.co.uk) left irc: "Leaving"
[21:38] Laurenceb (n=laurence@host81-154-154-237.range81-154.btcentralplus.com) joined #highaltitude.
[21:38] <Laurenceb> hi
[22:05] <Laurenceb> forrestv: ping
[22:08] <forrestv> Laurenceb, hello!
[22:09] <Laurenceb> hi
[22:09] <Laurenceb> theres a nice way to do attitudfe determination
[22:09] <Laurenceb> if you have the gyro biases as the state vector
[22:10] <Laurenceb> then have the actual attitude determination outside the filter
[22:10] <forrestv> what do you mean 'outside the filter'?
[22:10] <forrestv> you mean integrate the gyro measurements outside?
[22:10] <Laurenceb> yes
[22:10] <forrestv> and then how are they corrected with accelerometers?
[22:11] <forrestv> just the biases ..?
[22:11] <Laurenceb> each time you have accel/magno/gps data, you compute the attitude
[22:11] <Laurenceb> then compare to the solution integrated from gyros
[22:12] <Laurenceb> from the discrepancy you can compute an estimate of the biases
[22:12] <Laurenceb> you treat this step with an extended kalman filter
[22:12] <Laurenceb> but it only needs to be a 3DOF filter
[22:13] <Laurenceb> meanwhile you could compute the attitude using the gyros with quaternions - giving good numerical stability
[22:14] <Laurenceb> I havent tried this myself but I have found some papers - unfortunately not looking at this in great detail
[22:14] <Laurenceb> I need to try some octave/matlab code
[22:14] <forrestv> hmm
[22:15] <Laurenceb> the only disadvantage is it takes slightly longer to stabilise when turned on
[22:25] <SpeedEvil> So you run in a coordinate system of the gyros?
[22:25] <SpeedEvil> sort-of?
[22:26] <SpeedEvil> pig Laurenceb
[22:26] Action: SpeedEvil sighs.
[22:26] Action: SpeedEvil cleans sand out of keyboard.
[22:26] <SpeedEvil> Though my connection had died.
[22:27] <Laurenceb> lol
[22:27] <Laurenceb> you run a propogator that integrates the gyros
[22:27] <Laurenceb> and you have an attitude quaternion it integrates into
[22:28] <Laurenceb> this is sperate from the state vector
[22:28] <SpeedEvil> So you have a gyro state - which is a slowly drifting representation of a real vector
[22:28] <Laurenceb> its a quaternion :P
[22:28] <SpeedEvil> Sorry - it's been too long since I did the actual maths.
[22:28] <Laurenceb> then you have a bias vector with is 1x3
[22:28] <SpeedEvil> I need to brush up on it.
[22:29] <Laurenceb> also you have an error matrix , 3x3
[22:29] <SpeedEvil> You're basically cheating.
[22:29] <Laurenceb> I havent worked all this out yet...
[22:29] <Laurenceb> but you propogate integrating the gyros and adjusting the error matrix accordingly
[22:29] <SpeedEvil> And using the fact that the gyros drift slowly to mean that you can not put them in 'properly' to the attitude solution.
[22:29] <Laurenceb> yeah
[22:29] <SpeedEvil> To reduce CPU
[22:29] <Laurenceb> yep
[22:30] <SpeedEvil> Assuming the gyros don't do nasty things with couplings to accelleration.
[22:30] <SpeedEvil> Though you can first order calibrate that out
[22:30] <SpeedEvil> hmm
[22:31] <SpeedEvil> You meant the above lipo and thrusters for a blimp?
[22:31] <Laurenceb> no
[22:31] <Laurenceb> for a two thruster VTOL
[22:31] <SpeedEvil> Ah
[22:32] <Laurenceb> actually with the zippy LIPO cells on that site and 12'' props you could get 40 minutes hover time
[22:32] <SpeedEvil> Hmm.
[22:32] <SpeedEvil> That sounds rather too good.
[22:33] <Laurenceb> the velocity is low
[22:33] <Laurenceb> so it wont like wind much
[22:33] <Laurenceb> ~7m/s
[22:33] <SpeedEvil> I suppose I have optimised very hard for a short mission, with high disk loading to ignore wind as much as it can
[22:33] <SpeedEvil> ah
[22:34] <SpeedEvil> yeah - my peak climb velocity is about 3-4 times that :)
[22:34] <Laurenceb> that setup is designed for large slow flyers
[22:34] <SpeedEvil> Helicopter, not harrier.
[22:34] <Laurenceb> the props are only ~30% efficient it seems
[22:34] <SpeedEvil> I was using - propcalc.de
[22:34] <SpeedEvil> which looked interesting
[22:35] <Laurenceb> I have some of those motors
[22:35] <Laurenceb> used one on a rogallo wing
[22:35] <SpeedEvil> It's a pity for big helicopters that you can't really do very light low RPM motors
[22:35] <Laurenceb> gearbox?
[22:36] <Laurenceb> or toothed belt even
[22:36] <SpeedEvil> yeah - of course - but light efficient reliable cheap easy to manufacture gearboxes are hard.
[22:36] <SpeedEvil> Or even any two of those
[23:24] junderwood (n=john@adsl.jcu.me.uk) left irc:
[23:43] <Laurenceb> ok : I worked through the simple case of 3 axis magno, gyros and accel
[23:44] <Laurenceb> if you integrate into a quaternion, ie 4x1 vector
[23:45] <Laurenceb> then you need to have a 4x3 matrix to store how the three bias values mix into the quaternion
[23:45] <Laurenceb> this is effectivel the propogator
[23:46] <Laurenceb> the measurement takes the accel and magno data and calculates a quaternion and 4x4 error covariance matrix
[23:47] <Laurenceb> you then compare the calculated quaternion to the "state" quaternion and pass the difference through the inverse of the bias mixing matrix
[23:47] <Laurenceb> this gives you a 3x1 vector
[23:48] <Laurenceb> you also need to do the error treatment...
[23:48] <Laurenceb> I think it needs to be though about in terms of vector spaces
[23:48] <Laurenceb> the kalman filter operates in the bias vector space alone
[23:50] <SpeedEvil> So you've effectively slashed your computational needs?
[23:50] <Laurenceb> yeah
[23:50] <SpeedEvil> As long as no aliasing or other shit crops up.
[23:51] <Laurenceb> well its almost a fully correct treatment
[23:51] <Laurenceb> the only problem is you stick gyro noise in with the bias drift
[23:52] <SpeedEvil> and any aliasing or other systematic errors
[23:52] <Laurenceb> so you make the amplitude of the red noise for the bias drift larger than reality
[23:52] <Laurenceb> to compensate for the fact your lumping the white gyro noise in there
[23:52] <SpeedEvil> yeah
[23:53] <Laurenceb> the trick is to have a 3x4 matrix generated by integrating the bias terms on each gyro into the attitude quaternion
[23:53] <Laurenceb> that step in done in the propogator
[23:54] <Laurenceb> then you have a matrix that maps attitude quaternion space to bias vector space
[23:54] <Laurenceb> thats what you need to construct a filter in 3DOF bias vector space
[23:55] <Laurenceb> kind of hard to explain
[23:56] <Laurenceb> say there 200ms between measurement updates
[23:56] <Laurenceb> that matrix says how the quaternion can be updated with a 3x1 vector representation of the error in gyro bias over that timeperiod
[23:58] <Laurenceb> the only problem with this filter is it takes a while to stabilise - the filter will adjust the biases to spew round its quaternion until it matches with the accel and magno data
[23:59] Matt_APEX (n=chatzill@87.114.73.49.plusnet.thn-ag2.dyn.plus.net) left irc: Read error: 110 (Connection timed out)
[00:00] --- Sat Sep 19 2009