| --- Log opened Sat Mar 02 00:00:49 2019 |
| 00:01 | | Emmy [Emmy@Nightstar-9p7hb1.direct-adsl.nl] has quit [Ping timeout: 121 seconds] |
| 00:44 | | Derakon[AFK] is now known as Derakon |
| 00:58 | | celmin|sleep is now known as celticminstrel |
| 01:15 | <&McMartin> | [R] I *think* it's predominantly C so it should not be too terribly nasty |
| 01:32 | | Kindamoody is now known as Kindamoody[zZz] |
| 04:26 | | catalyst [Jessikat@Nightstar-5dv16h.cable.virginm.net] has quit [Connection closed] |
| 04:36 | | Emmy [Emmy@Nightstar-9p7hb1.direct-adsl.nl] has joined #code |
| 04:42 | | Emmy [Emmy@Nightstar-9p7hb1.direct-adsl.nl] has quit [Ping timeout: 121 seconds] |
| 04:51 | | macdjord|slep is now known as macdjord |
| 04:52 | | * macdjord ponders his adaptive bisection search algorithm |
| 04:54 | <@macdjord> | I use 3 strategies for picking my test index: |
| 04:54 | <@macdjord> | 1) Linear interpolation between the min and max values of the range |
| 04:55 | <@macdjord> | 2) Linear interpolation, but the result is clamped to never fall within the first or last 10% of the range |
| 04:55 | <@macdjord> | 3) Straight bisection |
| 04:56 | <@macdjord> | I start with strategy 1, but if any given round removes < 10% of the range, I fall back to strategy 2 for the next round. |
| 04:58 | <@macdjord> | If the strategy 2 round take off more than 10% of the range, I resume strategy 1 for the next round. If it only takes off 10% (the clamping ensures it cannot do worse) then I fall back to strategy 3 for the next round. |
| 04:59 | <@macdjord> | And after any round of strategy 3 I try strategy 2 again. |
| 05:00 | <@macdjord> | The idea is that strat 1 can potentially find the answer in O(1) time if the data is ~linear. |
| 05:02 | <@macdjord> | Strat 2 handles the case where the target value is near one end of the range, but linear interpolation keeps guessing values on the small end. |
| 05:02 | <@macdjord> | And then start 3 ensures reasonable performance even against pathological inputs. |
| 05:02 | <@macdjord> | Now, what can I do to improve it? |
| 05:12 | | Vorntastic [uid293981@Nightstar-6br85t.irccloud.com] has joined #code |
| 05:12 | | mode/#code [+qo Vorntastic Vorntastic] by ChanServ |
| 05:18 | | mac [macdjord@Nightstar-grpbnp.mc.videotron.ca] has joined #code |
| 05:18 | | mode/#code [+o mac] by ChanServ |
| 05:19 | | macdjord [macdjord@Nightstar-grpbnp.mc.videotron.ca] has quit [Ping timeout: 121 seconds] |
| 05:23 | | Derakon is now known as Derakon[AFK] |
| 06:16 | | mac [macdjord@Nightstar-grpbnp.mc.videotron.ca] has quit [[NS] Quit: Deep inside, every housecat believes themself to be just a temporarily embarrassed tiger.] |
| 06:22 | | macdjord [macdjord@Nightstar-grpbnp.mc.videotron.ca] has joined #code |
| 06:22 | | mode/#code [+o macdjord] by ChanServ |
| 06:49 | | celticminstrel is now known as celmin|sleep |
| 10:16 | | Kindamoody[zZz] is now known as Kindamoody |
| 11:00 | | Kindamoody is now known as Kindamoody|afk |
| 11:08 | | Emmy [Emmy@Nightstar-9p7hb1.direct-adsl.nl] has joined #code |
| 15:50 | | Vornicus [Vorn@ServerAdministrator.Nightstar.Net] has joined #code |
| 15:50 | | mode/#code [+qo Vornicus Vornicus] by ChanServ |
| 17:32 | | Vorntastic [uid293981@Nightstar-6br85t.irccloud.com] has quit [[NS] Quit: Connection closed for inactivity] |
| 17:40 | | * ToxicFrog bonks her head gently against the keyboard |
| 17:40 | <&ToxicFrog> | The reason none of my debugging output is working is because AVR libc doesn't link in support for printf("%f") by default, because it's huge. |
| 17:40 | <&[R]> | Ha |
| 17:41 | <&[R]> | ... and it doesn't give any error/warning about that? |
| 17:42 | <&ToxicFrog> | It outputs ? instead |
| 17:43 | <&ToxicFrog> | I mean, in the documentation for vfprintf() it mentions this, but there's no compile-time warning or anything |
| 17:43 | <&ToxicFrog> | I think because this is a linker setting, and by the time the linker runs it has no idea what arguments you passed to printf |
| 18:05 | <&ToxicFrog> | Oh no wonder the numbers I'm getting from this thing are so weird |
| 18:05 | <&ToxicFrog> | Magnetic field inclination where I am is like +70° |
| 18:10 | <~Vornicus> | so north is ...down? |
| 18:17 | <&ToxicFrog> | Yes. |
| 18:17 | <&ToxicFrog> | Turns out magnetic compasses are usually counterweighted based on their expected country of use to compensate for this! |
| 18:19 | <&ToxicFrog> | Magnetomechanical compasses, I mean. |
| 18:19 | <~Vornicus> | oh shit that's neat |
| 18:19 | <&ToxicFrog> | While what I have is a solid-state magnetoresistive compass, which just gives me a field vector. |
| 18:19 | <&ToxicFrog> | (since it's actually three individual magnetometers, one per axis) |
| 18:23 | <&ToxicFrog> | Now, as long as it's level I can get away with taking the arctangent of the two horizontal axes and throwing out the third, but if you tilt it more than 20° to the south it starts giving you garbage. |
| 18:23 | <&ToxicFrog> | I may have to implement some sort of inclination correction. |
| 18:27 | <~Vornicus> | proj(v, x) + proj(v, z) |
| 18:27 | <&ToxicFrog> | You're going to have to expand on that |
| 18:29 | <&ToxicFrog> | At the moment it gives me x/y/z field strength values and I take atan2(z,y) (or whatever axes are horizontal in its current mounting) and then range-correct it into [0,360) |
| 18:29 | <~Vornicus> | where proj(a, b) = b * dot(a,b) / dot(b,b) |
| 18:30 | <~Vornicus> | where x and z are your horizontal axes |
| 18:30 | <&ToxicFrog> | (basically, taking the projection of the field strength vector into the horizontal plane, then computing its angle) |
| 18:38 | <~Vornicus> | though I guess v - proj(v, y) would work faster. remember you need to consider the orientation of your thingy according to the gyros |
| 18:39 | <&ToxicFrog> | There are no gyros. |
| 18:39 | <&ToxicFrog> | I have three single-axis magnetometers at 90° to each other and that's it. |
| 18:39 | <&ToxicFrog> | Understanding of which axes are "horizontal" and which axis is "vertical" is determined at compile time based on how I've mounted the compass. |
| 18:40 | <~Vornicus> | aha |
| 18:40 | <~Vornicus> | What is it mounted on? |
| 18:41 | <&ToxicFrog> | At the moment, a breadboard |
| 18:41 | <&ToxicFrog> | Eventually, some sort of clothing |
| 18:42 | <&ToxicFrog> | I may have to eventually pair it with gyros and/or accelerometers to get it to work reliably, but knowing the inclination and approximate field strength in this region should be enough for it to figure out roughly how it's oriented. |
| 18:42 | <&ToxicFrog> | The most immediate aggravation is just that if you tilt it ~20° south or so, what happens is that the up/down axis increases slightly, the east/west axis remains the same, and the north/south axis suddenly reverses. |
| 18:46 | <~Vornicus> | Things I saw coming: that |
| 18:49 | <&ToxicFrog> | But knowing the local inclination should be sufficient to correct for that, I think -- basically, get the field vector, then swing it up by 70°, then project and get the bearing. |
| 18:50 | <&ToxicFrog> | ...assuming we can differentiate between "facing north" and "facing south but tilted down 40°", but I think we can do that with the east-west axis |
| 18:50 | <&ToxicFrog> | But I do not have nearly enough brain today to work out the details of that |
| 18:52 | <&ToxicFrog> | Especially not while on Alex duty with no backup ;.; |
| 18:53 | <~Vornicus> | You only get one vector, don't you? |
| 18:56 | <&ToxicFrog> | Yes -- the readings from the three magnetometers. |
| 18:57 | <&ToxicFrog> | Oh, right, hrm, I can't do that with east-west |
| 18:57 | <&ToxicFrog> | For some reason I was briefly modeling Earth's magnetic field as two overlapping fields at right angles to each other, which...it really isn't |
| 18:58 | <&ToxicFrog> | I probably should not be trying to math while this tired. |
| 18:59 | <~Vornicus> | Right, you don't get full orientation, just one of the two vectors, so you can't distinguish between the two, you're gimbal locked |
| 18:59 | | Derakon[AFK] is now known as Derakon |
| 19:01 | <&ToxicFrog> | Right. So if I want this to be inclination-corrected I also need, at minimum, something to tell it what direction gravity is in. |
| 19:15 | <~Vornicus> | Yep. |
| 19:20 | <~Vornicus> | Cross those two you get east (or west, depending on handedness), cross east and down and you get north (or south) |
| 19:58 | | mode/#code [+o PinkFreud] by ChanServ |
| 20:12 | | Kindamoody|afk is now known as Kindamoody |
| 21:34 | | poxifide [poxifide@Nightstar-09mo82.rcfreak0.com] has quit [G-Lined: Spamming proxy, see ya... (ID: 8ZVABLHNEJ)] |
| 21:35 | | Tamber [tamber@furryhelix.co.uk] has quit [G-Lined: Spamming proxy, see ya... (ID: TMB3TADPNX)] |
| 21:40 | | Tamber [tamber@furryhelix.co.uk] has joined #code |
| 21:40 | | mode/#code [+o Tamber] by ChanServ |
| --- Log closed Sat Mar 02 23:00:54 2019 |
| --- Log opened Sat Mar 02 23:01:00 2019 |
| 23:01 | | TheWatcher [chris@GlobalOperator.Nightstar.Net] has joined #code |
| 23:01 | | Irssi: #code: Total of 32 nicks [24 ops, 0 halfops, 0 voices, 8 normal] |
| 23:01 | | mode/#code [+o TheWatcher] by ChanServ |
| 23:01 | | Irssi: Join to #code was synced in 15 secs |
| 23:30 | | Vornicus [Vorn@ServerAdministrator.Nightstar.Net] has quit [Connection closed] |
| 23:36 | | Emmy [Emmy@Nightstar-9p7hb1.direct-adsl.nl] has quit [Ping timeout: 121 seconds] |
| --- Log closed Sun Mar 03 00:00:50 2019 |