#robotics Logs

Sep 25 2017

03:25 AM rue_bed: aha
03:28 AM rue_bed: sequencing in the brain is done with timers from a common point to each event
03:28 AM rue_bed: this is why you can make mistakes that have all the right operations, but in the wrong order
03:28 AM rue_bed: take for example, typing the work the
03:28 AM rue_bed: as t e h
03:28 AM rue_bed: which is so common, my irc client just fixes it
03:28 AM rue_bed: one of the timers is out
03:28 AM rue_bed: this is how you do sequential operations in a parallel machine
05:26 AM drekipus: oh nice
05:27 AM drekipus: didn't know there was a robotics channel
05:27 AM deshipu: there isn't
05:28 AM drekipus: has anyone dealt with using xy sensors (computer mice) for a holonomic robot?
05:28 AM drekipus: deshipu, nice ( ≖‿≖)
05:28 AM drekipus: well if this channel is just for shitposting I'm glad to join it
05:29 AM drekipus: basically I'm having a hard time with this: https://i.imgur.com/5i5Ugx7.jpg
05:31 AM drekipus: I actually had a dynamics class but all the issues were set up wayyyy differently and I can't think of how to take it to a form I can work with
05:32 AM deshipu: I guess you have to integrate in very small steps there, as soon as you lose the information about the actual path, there is no way to figure out the angle
05:33 AM deshipu: of course you will still have some error accumulating
05:33 AM drekipus: mm
05:34 AM drekipus: maybe just do a real small integration over the first case? just get the difference and trig it?
05:34 AM deshipu: but I don't really have any experience with this, just thinking about what I would do
05:35 AM drekipus: mm
05:35 AM drekipus: might work for the quick prototype and I can work on it later
05:36 AM drekipus: but there is a method to go from two vectors in a rotating co-ord frame
05:36 AM deshipu: have you thought about adding an extra sensor for the angle?
05:36 AM deshipu: like a compass?
05:36 AM drekipus: actually i don't think I've googled specifically that...
05:37 AM deshipu: assuming this is moving mostly horizontally
05:38 AM drekipus: nah, haven't used a compass, dunno how well/fast it'd work, but regardless the head huncho for this bot is a jerk anyway and I had to fight to actually have *ANY* form of feedback for control
05:38 AM deshipu: yeah, most robots are jerks :)
05:39 AM deshipu: most physical objects just don't care about your feelings
05:39 AM drekipus: ( ≖‿≖) wasn't what I was talking about but I'll go with it
07:36 AM Jak_o_Shadows: drekipus: from your two sensors, you can get the centre of the coordinate frame on the robot that's moving
07:37 AM drekipus: ye
07:37 AM drekipus: I'm thinking of that and treating it like a differential drive robot
07:38 AM Jak_o_Shadows: I feel like it's an intersection of sometihgn problem
07:39 AM Jak_o_Shadows: Why can't you just get the angle of thel ine between the two sensors?
07:39 AM Jak_o_Shadows: with respect to your global frame, which your sensors are in.
07:39 AM Jak_o_Shadows: oh
07:39 AM Jak_o_Shadows: local trackers
07:39 AM Jak_o_Shadows: dunno
07:39 AM Jak_o_Shadows: will keep it open
08:07 AM drekipus: cheers, I'll let you know if I come across anything interesting. I found a few things online that I didn't understand and then I built a pencil contraption to visualise it better
08:07 AM drekipus: but heading off now
08:07 AM drekipus: ~
09:20 AM rue_house: too bad he left
11:08 PM * DagoRed stretches
11:08 PM DagoRed: Mandlebrot sets will have to finish tomorrow.
11:09 PM DagoRed: I'm just looking forward to starting a new project. Debating on an ARM based driving robot or just start hacking the interface on my motorcycle.
11:09 PM DagoRed: Hmm... I should make a new light up tachometer.
11:28 PM rue_house: make one where the needle stays put and the dial moves
11:28 PM rue_house: avoid n time searching for the needle to get your speed