#robotics Logs

May 01 2013

00:23 MrCurious http://www.dailymail.co.uk/sciencetech/article-2313828/Is-really-human-DNA-tests-inch-skeleton-alien-looking-creature-sized-head-prove-actually-human-claim-scientists-new-documentary.html
00:24 GargantuaSauce as you can probably tell from the latest commit i'm working to cut that crap out
00:24 GargantuaSauce poking the registers myself is so much terser/nicer
01:20 MrSnap robots yo!
01:20 MrSnap circuits taken over the world
01:21 MrSnap just you wait
01:21 MrSnap johnny five alive
01:21 MrSnap batteries not included
01:23 MrSnap yes, I am here
01:23 MrSnap your founder has returned
01:24 MrSnap otherwise known as
01:24 wildmage I see things look in good order
01:24 wildmage although they are quite as usual
01:25 wildmage still an idling channel is it?
01:27 GargantuaSauce theres a reasonable amount of robobanter lately
01:32 wildmage oh really?
01:32 wildmage is it shop talk, futurism, or scientific discussion?
01:33 rue_bed wild!
01:33 rue_bed !thislog
01:33 tobbor This one: http://rueshouse.dyndns.org:82/~ircjunk/irclogs/html/%23robotics-2013-05-01.html
01:35 wildmage heya rue
01:35 wildmage how's life?
01:35 RifRaf hello wildmage
01:35 wildmage greetings and salutations rifraf
01:37 wildmage i've been doing software for too long, i haven't had a chance to play with hardware in quite some time
01:37 RifRaf i'll be back in 30 min, about to finish work, good to see ya
01:40 GargantuaSauce might take next week off work so i can make some proper progress on my stuff
01:43 wildmage and what is your stuff?
01:44 GargantuaSauce http://workman-industries.net/?p=27
01:46 wildmage *processing*
01:46 GargantuaSauce no it's all done in c
01:46 wildmage haha... i meant "i'm processing"
01:46 GargantuaSauce i know :V
01:46 wildmage you have a fab?
01:47 GargantuaSauce a fab?
01:47 wildmage a 3d printer?
01:47 wildmage http://workman-industries.net/images/quadrotor/1.jpg
01:47 GargantuaSauce yeah it's a mendel
01:47 GargantuaSauce when you say fab i think semiconductor fabrication
01:48 wildmage yeah... i remembered that
01:48 GargantuaSauce which sadly is not yet approachable on a diy basis :(((
01:48 wildmage RepRap is supposed to be the self-replicating machines correct? or at least in theory capable of constructing another copy
01:48 GargantuaSauce well yeah that's the end goal
01:49 wildmage i started my work in self-configurable robots
01:49 wildmage i was interested in doing a research project on self-replicating robots
01:49 wildmage but... i couldn't come up with a topic to research :P
01:49 wildmage it's so daunting
01:49 GargantuaSauce yeah
01:50 GargantuaSauce have you seen von neumann's cellular automaton?
01:50 wildmage of course
01:51 wildmage I know a guy who sends me anti-obama conspiracy theories and designs for self-replicating turing machines
01:51 GargantuaSauce lol
01:51 wildmage one topic is better research than the other
01:52 wildmage he mostly deals on the grid level like the cellular automota
01:53 wildmage I'm mostly a software guy myself, but I insist on dealing with real physical problems
01:54 GargantuaSauce that is basically where i'm at too
01:54 GargantuaSauce flipping a bunch of bits gets boring
01:54 wildmage I've been mostly working in physics simulations the past 4 years
01:55 wildmage it gives you freedom to explore a lot of computational problems that you would never encounter because you're too busy checking the connections on your cables
01:56 GargantuaSauce what sort of simulations?
01:56 wildmage of course, you have to be careful not to make unrealistic physics assumptions
01:56 GargantuaSauce ha ha
01:56 wildmage well, the engine or the application?
01:56 GargantuaSauce which is a caveat that gets thrown out the window as soon as you ascribe to the models we use ;)
01:56 GargantuaSauce both<
01:57 wildmage I tried a lot of free physics engines. Started at ODE, went to PhysX, and then to Bullet. I'm working with Bullet now.
01:58 wildmage PhysX was good, but it was closed. I had a problem and there was no way to diagnose it or figure out how to workaround it.
01:58 wildmage anyway, my application is a snake robot traveling through confined environments such as pipes and caves
01:59 wildmage the snake is supposed to travel through the pipe and map it with no external sensors.
01:59 GargantuaSauce oh cool
01:59 wildmage this is to study the case of complete sensor failure or entering an environment where your sensors are inapplicable such as muddy water
02:00 GargantuaSauce i have plans to build a similar sort of simulation suite for testing out ideas and controls and so on before implementing them, just havent quite gotten there yet
02:01 wildmage it's much easier than it was 10 years ago
02:01 GargantuaSauce yeah computers are getting stupidly powerful
02:02 wildmage well, the tools are becoming more widely available
02:02 wildmage the computational power was always there
02:02 GargantuaSauce not really
02:02 wildmage well, you should be more focused on what it is you want to accomplish
02:02 wildmage instead of just a vague notion that you want to build a simulator
02:03 GargantuaSauce i am addicted to generalization
02:03 wildmage often times, this is the first thing that grad students say they want to do...
02:03 wildmage usually we let them do it and try to salvage something after they finished
02:03 GargantuaSauce heh
02:07 GargantuaSauce well what i want to accomplish is vision-based SLAM and stabilization, mostly
02:07 GargantuaSauce for the quadcopter at least
02:09 wildmage hmmm
02:09 wildmage big problem
02:09 wildmage what platform
02:09 GargantuaSauce yup
02:11 GargantuaSauce currently i'm working with an stm32f4...CV stuff will be offloaded to an allwinner a10 or a more suitable SoC board if i find or make one
02:11 wildmage what about the robot?
02:12 GargantuaSauce oh, quadcopter
02:12 wildmage do you really need vision to stabilize?
02:13 GargantuaSauce yes
02:13 wildmage a gyro or mems accelerometer can't do the job?
02:14 GargantuaSauce IMUs are good enough to keep em in the air but i want a high degree of stability for indoor navigation
02:15 wildmage ahh yes, the holy grail, indoor navigation
02:15 wildmage I always thought that the best way to do that would be to have a robot that can smack against the walls and still fly
02:15 wildmage like a fly
02:17 GargantuaSauce well i am probably going to need to build the frame that way for the development process :)
02:18 GargantuaSauce and yeah i realize these are some damn lofty goals
02:18 GargantuaSauce taking it one step at a time
02:19 GargantuaSauce also want to build a hexapod, with which i'll approach some more typical locomotion stuff
02:19 wildmage okay, time to sleep
02:19 wildmage good night
02:19 GargantuaSauce later
02:22 Ademan GargantuaSauce: I actually didn't give it much of a read, seeing as my stm32f3 boards just came in though ill give it a look pretty soon
02:22 GargantuaSauce i see
02:23 Ademan I'm actually wondering if there's any value at all (aside from learning) in continuing to work on my MSP430 code...
02:23 GargantuaSauce i think the i2c peripheral is quite different between f3 and f4 so bear that in mind if you do peruse it
02:23 Ademan may as well do it all on the stm
02:23 GargantuaSauce ya i am pretty enamored with the st platform
02:26 GargantuaSauce also i should point out the existence of ##stm32
02:28 Ademan I'll probably be in there a lot