#robotics Logs

Mar 27 2014

18:43 phone291 i want to pay some one to make me a simple robot
18:45 phone291 any on there
18:49 Tom_itx make one yourself
18:49 Crenn-NAS http://www.robugtix.com/
18:50 phone291 Tom
18:50 phone291 you say that because you want me to learn robotics?
18:50 phone291 its not something really that im intrested in
18:50 phone291 i need a device made
18:51 phone291 very simple
18:51 phone291 and willing to pay a fair amount
18:51 phone291 to get it made
18:52 Crenn-NAS No stores sell what you want?
19:50 phorce Hello
19:51 GargantuaSauce_ greetings
19:51 phorce How are you? :)
19:52 GargantuaSauce_ kind of sleepy
20:01 phorce urghh tell me about it
20:01 phorce i'm up writing up my thesis
20:01 phorce :/
20:03 Elleo phorce: what's it about?
20:03 phorce spectral segmentation of calls.. signal processing mainly
20:04 phorce hoping to do my PhD in robotics though :D
20:04 phorce Robotics and signal processing, sorry!
20:04 Elleo :)
20:04 phorce I want to do multi-agent robotics, that interact with sonar pulses etc..
20:05 phorce I don't like image processing :(
20:05 Elleo heh
20:05 phorce aha what do you do, btw?
20:06 GargantuaSauce_ i really like the idea of communication and localization using the same medium
20:06 Elleo you can create a robotic systemthat involves image processing without having to actually do much image processing yourself
20:06 Elleo there's plenty of off-the-shelf solutions that you can just build on top of
20:07 phorce I like the idea of localization; particularly using such things as sonar etc.. The other main interest of mine typically is the radio waves, so yeah
20:07 phorce and ha Elleo I know, but, in general it just bores me :(! I like to implement my own things ha
20:07 Elleo now days I'm working on the Ubuntu Touch platform for mobile phones, but I used to work on developmental robotics stuff (trying to get robots to learn in a similar way to children)
20:08 Elleo mostly focusing on learning early motor skills and basic communication
20:08 Elleo (on humanoid robots)
20:09 phorce Ahh sounds pretty good :) Elleo, I got a suggested project.. Basically, it's the idea of how children / humans learn through speech, how they basically learn to recognise words etc.. It seemed pretty interesting - So that kind of things?
20:09 phorce Or was your research a different approach
20:10 Elleo I'd be careful about wanting to implement everything yourself, with a PhD it can be good to be able to focus really specifically on just one thing and not have to worry about every problem under the sun
20:10 Elleo I was looking more at the ways words could be given meaning to the robot, by grounding them in its own sensorimotor experiences
20:10 phorce Yes, aha, through my MSc I've come to realise this :P I've adapted languages such as Matlab and taken on smaller tasks to fully understand a problem
20:10 GargantuaSauce_ i kind of want to get back into academia but am super adverse to the whole specializing thing
20:11 Elleo but with the actual word recognition done by standard speech recognition software
20:11 Elleo there is plenty of interesting work being done on early phoneme acquisition though (just not by me :P)
20:12 phorce Ahh, I see! Sounds good :) I like this idea of learning through words, and sounds and yes, the problem is funding.. There is very little funding available and it's very specific in the subject area
20:12 Elleo I was also incorporating gestural communication into that
20:12 Elleo in the form of point gestures learnt out of the earlier motor learning experiences
20:12 phorce Ahh, so I'm guessing you used such methods as HMM's etc..?
20:13 Elleo I was luckily able to get my work funded through a couple of EU FP7 projects, ROSSI - RObotic Sensorimotor and Social Interaction and IM-CLeVeR (Intrinsically Motiviated, Cumulative Learning, Versatile Robots)
20:14 phorce wow ha impressive!! Was this a masters or full phd? :)
20:14 Elleo nah, this was all based around a schema learning technique
20:14 Elleo PhD and then a bit of time as an RA
20:14 phorce uk or..? :)
20:15 Elleo so the same mechanism used for learning the motor skills also got used for the language, gesture, etc. stuff
20:15 Elleo yep, UK
20:15 Elleo at Aberystwyth University
20:15 phorce nice nice, that's really good that you got such a research funding
20:15 Elleo where abouts are you based?
20:15 phorce and aha at a good uni too :P
20:15 phorce well, for robotics and comp sci so i hear
20:15 Elleo yeah, Aber's surprisingly well equipped for robotics considering it's in the middle of nowhere :P
20:16 phorce haha yes, i hear that the train tracks run through the city and only a few trains leave there :P
20:16 Elleo they've got everything from space robots, sailing robots, amphibious robots, flying robots all the way to humanoid robots
20:17 phorce Yeah haha it's an interesting place, I've never been.. I heard that a third year, Computer science is doing their year in industry at CERN - impressive huh?
20:17 Elleo yeah, there's only 1 train every 2 hours, and it takes about 3-3.5 hours to get to a bigger town
20:17 Elleo phorce: yeah, we get a surprising number of students go off to CERN
20:17 phorce aha may i ask a question? :P
20:17 Elleo 3 of the past ones are pretty good friends
20:18 Elleo sure
20:18 phorce Do you have any regards on doing your PhD?
20:18 Elleo regards? regrets?
20:18 phorce sorry
20:18 phorce regrets :P
20:19 Elleo not really, I enjoy most of it; it was pretty stressful towards the end as I was frantically trying to write everything up really close to the deadline
20:19 Elleo I massively extended the scope of what I was doing when I got invited on to the IM-CLeVeR project
20:19 Elleo which made the final thesis much better, but meant I had a lot less time to finish writing it
20:20 Elleo I think it depends a bit where you do your PhD though
20:20 phorce Yeah, I hear a lot of people in robotics who are doing their PhD and have completely gone off scope and now have ran out of funding
20:20 phorce anyway, aha, brb - If you're not here when I get back, good chatting :)
20:20 Elleo some places seem to have a pretty stressful and competitive atmosphere
20:20 Elleo but Aber just takes a fairly small number of PhDs and is generally a pretty relaxed place to be
20:21 Elleo sure :)
20:32 phorce hey, back
20:37 MrCurious robotustra: http://www.pidjin.net/2014/03/27/early-adopters/
20:38 Elleo :)
20:38 phorce hahaaha
20:40 phorce :Elleo
20:40 phorce Want to know what I find most fustrating? You might have had experience of the same problem
20:44 GargantuaSauce_ what
20:44 MrCurious i am awefully curious now...
20:44 GargantuaSauce_ i think the most frustrating thing for me is discovering i've been debugging a hardware problem in software for 3 hours
20:45 phorce There is someone, in our research group who has been given a fully funded PhD and sits and does nothing but play games all day aha! It's annoying because he's got given such a good chance, but he's kind of just blown it
20:46 MrCurious ever throw a rock strait up into the air?
20:46 MrCurious looks like it will escape gravity for a while
20:46 phorce ..? :P
20:49 GargantuaSauce_ i wouldn't be too quick to judge...i sometimes piss around all day and can't get my head in the game until everyone goes home, and then get in the zone for 10 hours
20:50 GargantuaSauce_ but it's certainly far from inconceivable that effort and rewards are not necessarily proportional
20:50 MrCurious i think you need to observe for a longer time period, and collect more data
20:51 MrCurious and possibly try your hand at that game, ,in private at first, but later....
20:51 phorce yahh i totally get that
20:51 phorce :P but, this guy literally does nothing and doesn't stay behind
20:51 phorce i mean, it's now 01:34 - i've been in the lab since 8am this morning
20:52 GargantuaSauce_ yeah ok
20:52 phorce see/ :P
20:52 GargantuaSauce_ robotustra: how much current does your head draw when the eyes are slamming into the stops?
20:54 MrCurious GargantuaSauce_: that read rather funny first parse pass
20:56 Elleo phorce: yeah, that's a shame
20:56 MrCurious made me think of the elecctric eye stimulation shutter glasses blink simulator hoax video :)
20:56 Elleo you definitely shouldn't work too hard though
20:56 phorce I feel bad for him, BUT he did write hello world in C# today ;)
20:56 phorce why ha?
20:57 MrCurious http://hackaday.com/2011/01/16/electrodes-turn-your-eyelids-into-3d-shutter-glasses/
20:57 phorce I disagree :(
20:57 MrCurious hilarious was a hoax
20:57 Elleo spending ridiculous hours in the lab consistently will typically result in worse productivity in the long run
20:57 Elleo not too bad for the occassional crunch time, but if you're doing it regularly you'll just wear yourself out
20:57 phorce I haven't found that, I mean, I do take breaks etc.. I have a day off at least once a week, it's just because I have my thesis to hand in, in a few weeks!!
20:58 MrCurious so he takes ocasional recharge days and works full steam others, while you continue to plug on in burnt-out low productivity mode
20:58 MrCurious he will likely get a raise soon
20:58 phorce As well as, i see my supervisor, who is really, really intelligent and knows so much more than me and it's intimidating haha
20:59 Elleo also you want to get in to the habit of judging yourself by your output rather than the hours you put in to something
20:59 Elleo lots of people seem to get caught up in trying to work the longest hours
20:59 phorce Yes, very true! But, am I the only one who thinks that I'm 80% more productive during the night? :P
20:59 Elleo whereas it's much better to find the sweet spot where you're producing the most output for the minimum time spent
21:00 Elleo phorce: nothing wrong with being a night owl, I tended to work a lot during the night too, but I almost always still got a full 8 hours sleep even if it meant I didn't get up until mid-day
21:01 phorce aha what time are you typically waking up these days? :P
21:01 Elleo except when I had to drag myself out of bed early to do a bit of teaching or something
21:01 Elleo nowdays I tend to get up around 10 and work until 6
21:01 GargantuaSauce_ nocturnal = best urnal
21:01 Elleo I work from home so I don't have to do some horrid commute or anything
21:02 GargantuaSauce_ i am definitely way more productive at night
21:02 Elleo and the rest of my team is spread over lots of different timezones so I match up better with them by getting up a little later
21:02 phorce That's not too bad tbh :) I couldn't work from home I don't think haha
21:02 phorce I'd need people around me
21:02 Elleo yeah, I need the opposite :P
21:02 Elleo I did a large amount of my PhD work from home too, except when I actually needed to run experiments
21:02 Elleo it's good to meet up with people a few times a week to bounce ideas around
21:03 Elleo but I find it hard to focus on something if I'm in a room with a bunch of other people
21:07 MrCurious best inspiration comes 1 hour before bedtime on sunday nights, and extends for 8 hours
21:07 GargantuaSauce_ yeah...
21:07 MrCurious only way to exploit it is working from home on a rolling schedule
21:07 GargantuaSauce_ or being the living dead
21:08 MrCurious but celebrating a friday night at noon at a bar starts out ok, then quickly goes very wrong
21:08 GargantuaSauce_ https://www.youtube.com/watch?v=Hhnfu3DSWy8 yeah, the living dead again
21:08 MrCurious during that phase of my life i learned bars are open at 6am to service the night shift workers for their happy hour
21:08 phorce i guess aha especially when the people are more intelligent than you
21:08 phorce that's a kill
21:08 MrCurious that was a unsettling discovery
21:09 GargantuaSauce_ bars close at 2am here by law
21:09 GargantuaSauce_ it's fucking stupd
21:09 MrCurious yeah
21:09 MrCurious but they open right back up at 6am
21:09 MrCurious perfect for when the after hours partys wind down
21:10 MrCurious but losing a workable social life is not worth the gains from a rolling creative schedule
21:13 robotustra https://www.youtube.com/watch?v=ExeVTvKDxZs&feature=youtu.be
21:19 robotustra in sync
21:19 robotustra random but in sync
21:22 MrCurious i want to see smooth motion trackinga bouncing ball
21:23 robotustra try to do smooth motion with your own eyes first :)
21:23 MrCurious yeah yeah
21:23 robotustra there is NO smoth motion in eyes
21:23 MrCurious i think it can only be done if you defocus and are drunk
21:23 MrCurious makes that cop test make sense suddenly
21:24 robotustra when you are drunk your eyes are focused on one spot
21:24 robotustra the jump is about 0.15 s
21:24 robotustra and I DON'T loose frames because of motion of eye
21:25 robotustra so I loose about 1-2 frames from 25 per one eye per second
21:28 robotustra the next step is to do capture of images and follow the closest object
21:28 robotustra or any other object
21:33 MrCurious that will be a very cool video
21:36 robotustra hope from that points people start to shit with bricks
21:36 robotustra and I'm the first
21:37 robotustra I enter the room and THAT THING is watching me
22:29 MrCurious be someone who moves the world around, don't be someone who the world moves around...