| 01:18.38 | louipc | At one point I used to actually like helicopters. |
| 02:14.09 | Axman6 | Twingy: i'd imagine that's pretty damn hard right? |
| 02:14.44 | Twingy | requires a pretty good understanding of control theory, embedded programming, electronics, and dynamics |
| 02:15.01 | Twingy | but yes, tough problems |
| 02:15.39 | Twingy | building the autopilot is only solving 10% of the problem, the control laws are the other 90% |
| 02:16.00 | Axman6 | the uni i'm at atm has possibly one of the world's most advanced 4 rotor helicopter prototype... things... |
| 02:16.31 | Twingy | which one? |
| 02:16.47 | Axman6 | it's like a meter across, but it's being built so it can fly right up to walls (like dams and stuff), and it can cope if a rotor fails |
| 02:16.51 | Axman6 | ANU |
| 02:16.51 | Twingy | I think CMU is working on one |
| 02:17.23 | Axman6 | i saw it on the tour of the engineering department :P |
| 02:17.42 | Axman6 | sounded a lot more impressive when the tutor was telling us about it |
| 02:18.11 | Twingy | you can make just about anything fly with the right control law |
| 02:18.23 | Axman6 | heh |
| 02:18.25 | Twingy | adaptive, fuzzy logic, model predictive just to name a few |
| 02:18.45 | Axman6 | think they said they'd been working on this for about 10 years now |
| 02:18.51 | Twingy | most students start with a basic PID |
| 02:19.19 | Twingy | most people I've met spend between 5 and 10 years on the autopilot before it flight worthy |
| 02:19.27 | Twingy | (the kind that spend 80 hours a week on them) |
| 02:19.37 | Axman6 | yeah |
| 02:19.40 | Twingy | planes are about half that time |
| 02:19.53 | Axman6 | i can't even imagine where i'd start on something like that |
| 02:20.14 | Twingy | first you decide on a platform |
| 02:20.40 | Axman6 | "everything" |
| 02:20.51 | Axman6 | universal autopilot |
| 02:20.53 | Axman6 | ;) |
| 02:20.54 | Twingy | then you build an autopilot, after that you do testing and integration, then spend a long time testing the control algorithm |
| 02:21.00 | Twingy | no such thing yet |
| 02:21.32 | Twingy | the one I'm working on interfaces with x-plane over a udp connection |
| 02:21.48 | Axman6 | i was about to ask if you'd used Xplane at all |
| 02:21.49 | Twingy | for the simulation part of it |
| 02:22.12 | Axman6 | my dad bought that for me for christmas year before last i think |
| 02:22.15 | Twingy | for the actual testing I've got about $4k in hardware so far for the parts and the platform (E-Raptor 50) |
| 02:22.31 | Twingy | the gps alone was $1500 |
| 02:22.44 | Axman6 | wow |
| 02:22.46 | Twingy | and it's not even one of the fancy L1+L2 |
| 02:22.56 | Axman6 | L1+L2? |
| 02:23.10 | Axman6 | is that using thr military channel thing too? |
| 02:23.12 | Twingy | the frequency bands the gps operates on |
| 02:23.17 | Twingy | L2 used to be military only |
| 02:23.22 | Axman6 | yeah thought so' |
| 02:23.27 | Twingy | L2 gets you under 0.5m accuracy |
| 02:23.43 | Axman6 | that was my next question |
| 02:24.06 | Twingy | the one I have is 20hz position, 20hz heading (dual antenna) with 1dof gyro and 2dof accelerometer for interpolation when gps blinks out |
| 02:24.36 | Axman6 | sounds good |
| 02:24.42 | Twingy | I was designing a USB to serial board for it today |
| 02:25.25 | Twingy | basically an FT232RL + 4 resistors + 4 capacitors and a 11-pin header, 4-pin header (usb), 2-pin header power, and a 3.3V and 5.0V regulator |
| 02:25.41 | Axman6 | brb, i have how cross buns to eat |
| 02:25.46 | Twingy | we have a circuit plotter at work, so I can juts sketch up a circuit and print it |
| 02:26.16 | Axman6 | why are you using USB anyway? |
| 02:26.27 | Twingy | because the pico-itx only has 4 usb and 1 serial |
| 02:26.35 | Axman6 | ah ha! |
| 02:26.40 | Twingy | and the serial goes to the servo board I made |
| 02:26.42 | Axman6 | i want one of those board so badly :( |
| 02:26.54 | Axman6 | pico-itx that is |
| 02:27.13 | Twingy | it's not bad, it doesn't make sense to build an autopilot around a microcontroller anymore |
| 02:27.18 | Axman6 | if it had another ethernet port, it would make an excellent router with OpenBSD |
| 02:27.24 | Twingy | everything is 0402 and all on a single chip now |
| 02:27.41 | Twingy | I can hand solder TQFP 100's and QFN stuff, but you still can't get anywhere near the size |
| 02:28.01 | Twingy | if I were to make an autopilot around an AT91SAM7S64 (32-bit ARM) I still couldn't get anywhere near the size |
| 02:28.12 | Twingy | and next year the mobile-itx will be out, so instant reduction in size and mass |
| 02:33.41 | Axman6 | awesome |
| 02:48.28 | Twingy | it's a lot of fun, but I go home with a headache just about every day |
| 02:48.38 | Twingy | it's really really hard stuff |
| 02:49.10 | Axman6 | i assume you'll be running some OS on this board right? what're you using? |
| 02:49.22 | Twingy | it doesn't really matter |
| 02:49.41 | Twingy | as long as you don't have any latencies due to buffering |
| 02:49.44 | Twingy | you are fine |
| 02:50.25 | Axman6 | any good examples? |
| 02:50.25 | Twingy | you can alleviate a lot of that by a call to fflush () |
| 02:50.25 | Twingy | or you can modify the kernel driver |
| 02:50.25 | Twingy | either way is straight forward |
| 02:50.44 | Twingy | you should look at the serial code in the linux kernel |
| 02:50.55 | Twingy | it handles data differently above and below 2400 baud |
| 02:51.10 | Twingy | you can modify the kernel so it'll obey no buffering on setserial |
| 02:51.29 | Twingy | at the moment I'm just using ubuntu server off a 2GB flash drive |
| 02:51.35 | Twingy | the drive weighs 3 grams |
| 02:51.48 | Axman6 | heh, very nice |
| 02:51.59 | Twingy | I just made some nylon 5/8-18 TPI nuts on my cnc mill |
| 02:52.05 | Axman6 | well, except that it's ubuntu, but oh well |
| 02:52.12 | Twingy | to replace a pair of 15 gram steel nuts with 3.4 gram nylon nuts |
| 02:52.21 | Twingy | saved me 24 grams |
| 02:52.43 | Axman6 | what's cnc stand for anyway? i saw ANU's cnc mill the other day |
| 02:52.53 | Axman6 | it was pretty massive |
| 02:52.56 | Twingy | I modeled the entire helicopter skid in solid works and put it into GCAM and cut it out on the cnc |
| 02:53.09 | Twingy | computer numerically controlled |
| 02:53.40 | Twingy | everything is 6061 aluminum |
| 02:54.07 | Twingy | it's cut some restitution and it's got a high tensile strength |
| 02:54.21 | Axman6 | excellent |
| 02:54.59 | Twingy | I'm terrible at dynamics though |
| 02:55.06 | Twingy | that's why I've been taking some classes |
| 02:55.18 | Axman6 | i hope you're not working on this alone |
| 02:55.27 | Twingy | the electronics and programming are the easy part |
| 02:55.47 | Twingy | I'm working with 1 other person |
| 02:55.59 | Twingy | he knows more dynamics and control theory than I do |
| 02:56.05 | Twingy | he doesn't do much programming or electronics |
| 02:56.19 | Twingy | so we divide the work accordingly |
| 02:56.36 | Axman6 | good good |
| 02:56.48 | Twingy | so he's learning and I'm learning |
| 02:57.02 | Twingy | it's a lot for one person to do it all |
| 02:57.13 | Twingy | with 2 people you can get it done in about 2 years |
| 02:57.18 | Axman6 | yeah, that's why i asked |
| 02:57.42 | Axman6 | also kinda dangerous too. need someone else to check for mistakes |
| 02:57.53 | Twingy | autopilots have been around for 20 years, but it's still virgin territory as far as I'm concerned |
| 02:58.02 | Twingy | it's very dangerous on the larger vehicles |
| 02:58.11 | Twingy | we're about to integrate one into our $250k RMAX |
| 02:58.25 | Twingy | I'm working with a $1500 e-raptor |
| 02:59.32 | Twingy | I'm still a noob as far as this stuff goes |
| 02:59.43 | Twingy | it'll be a while before I'm any good at this stuff |
| 03:17.03 | Axman6 | i was messing with mathematica last night. lotsa fun once i understoof it a bit |
| 03:17.07 | Axman6 | d* |
| 03:17.40 | Axman6 | using it for physics at uni, taking astronomical photos, using mathematica to average them to reduce the signal to noice ratio |
| 03:17.45 | Axman6 | noise* |
| 03:17.47 | Axman6 | rawr |
| 03:20.07 | Twingy | we use matlab and simulink here |
| 03:22.24 | Axman6 | ANU did a deal with Wolfram so that students can get a copy and install it on their home computers for free, which is pretty awesome |
| 03:24.31 | poolio | ahh. I have to learn Matlab over the next week :( |
| 03:24.58 | Axman6 | http://www.gamasutra.com/view/feature/3563/microsoft_excel_revolutionary_3d_.php?page=3 <-- DirectX at its best, heh |
| 09:02.08 | *** join/#brlcad Z80-Boy (n=clock@77-56-89-57.dclient.hispeed.ch) | |
| 10:50.35 | *** join/#brlcad Elperion (n=Bary@p54876709.dip.t-dialin.net) | |
| 11:20.07 | *** join/#brlcad elite01 (n=elite01@dslb-088-070-018-248.pools.arcor-ip.net) | |
| 15:21.00 | brlcad | :) |
| 15:21.05 | brlcad | howdy yukonbob |
| 15:32.31 | yukonbob | hey brlcad :) |
| 15:33.33 | yukonbob | wiress here is slower than fsck :P |
| 15:33.38 | yukonbob | *wireless |
| 15:38.37 | yukonbob | check ya later, folks :) |
| 16:02.01 | brlcad | cya |
| 16:11.18 | *** join/#brlcad Z80-Boy (n=clock@77-56-80-198.dclient.hispeed.ch) | |
| 21:03.32 | *** join/#brlcad ibot (i=ibot@pdpc/supporter/active/TimRiker/bot/apt) | |
| 21:03.32 | *** topic/#brlcad is BRL-CAD Open Source Solid Modeling || http://brlcad.org || http://sf.net/projects/brlcad || Channel logs at http://ibot.rikers.org/%23brlcad/ || BRL-CAD is on scan.coverity.com but still offline || Release 7.12.0 coming soon to a desktop near you | |
| 23:51.56 | *** join/#brlcad Axman6 (n=Axman6@pdpc/supporter/student/Axman6) | |