#robotics Logs

May 23 2020

06:01 AM mrdata_ is now known as mrdata
04:19 PM Samian: [2020:05:22:00:39:29] <rue_bed> Samian, really? <--- yes really
04:19 PM Samian: man I wish this channel was more active!!!
04:19 PM veverak: :D
04:25 PM Samian: I have a degree in CS. I'm an AI engineer. I want to get into robotics so I can use reinforced learning to train robotics
04:27 PM veverak: just another in line :(
04:27 PM Samian: just another in life?
04:27 PM Samian: *line?
04:27 PM Samian: veverak what do you mean?
04:27 PM Samian: I'm sure many people want to be AI robots
04:28 PM veverak: that is what I am saying
04:28 PM veverak: a lot of people are trying to use machine learning in robotics, up to a point that it feels like hype
04:29 PM veverak: -> just another person to allready long line
04:38 PM Samian: veverak I don't think it's hype at all. Many are trying to do this because likely robotics is the future
04:38 PM Samian: veverak 50 years from now, the top people on the forbes 400 list won't be software guys, but founders of robotics companies
04:40 PM veverak: which is not interesting metric :/
04:42 PM Samian: any recommendation for a good software-controllable robotic arm/claw ?
04:43 PM veverak: depends on use case
04:44 PM Samian: like something I can clip to the side of a desk similar to those arms used to hold computer monitors
04:44 PM Samian: a robotic arm/claw about that size
04:45 PM veverak: so self-education thing when you are not as interested in precision?
04:46 PM Samian: oh forget it. I'll need to do much more reading in this subject
04:46 PM Samian: veverak maybe there was a popular robotic arm that made for good educational purposes for beginners
04:46 PM Samian: it appears not
04:47 PM veverak: thinking
04:47 PM veverak: well, I mostly know of DIY stuff
04:48 PM Samian: holy crap this is cool: https://youtu.be/SRupIVknV-A
04:50 PM veverak: and I suspect shitload expensive
04:50 PM veverak: :)
04:51 PM Samian: dude I'm going to start reading all the books on robotics that I can get my hands on
04:51 PM veverak: that won't do it
04:51 PM veverak: just reading books won't work
04:52 PM veverak: just building/using robots won't work either
04:52 PM veverak: one has to balance both :P :)
04:52 PM Samian: [2020:05:23:15:19:54] <veverak> a lot of people are trying to use machine learning in robotics, up to a point that it feels like hype <---- guess what else is hype...
04:52 PM Samian: stock trading bots
04:52 PM Samian: Yet....
04:52 PM veverak: it is, I know
04:53 PM Samian: yet there is Medallion fund, that has outperformed everyone
04:53 PM Samian: https://www.institutionalinvestor.com/article/b1km7z5z7cs1mm/The-Medallion-Fund-Skepticism-and-a-Failure-to-Comprehend
04:56 PM Samian: veverak: https://imgur.com/a/4284xuE
04:56 PM veverak: tooooo long.... :)
04:56 PM Samian: veverak point is... there is such a thing as simply being better than everyone else
04:56 PM Samian: veverak look at the image
04:56 PM veverak: don't care
04:57 PM veverak: but you seem to care about who is better :P
04:57 PM Samian: Renaissance Technologies is a quant trading frim
04:57 PM veverak: (which seems as weird motivation to do stuff to me)
04:57 PM Samian: veverak often the devil is in the details
04:57 PM Samian: and very few can get all the details right
04:58 PM veverak: I very well know that
04:59 PM veverak: but so far you did not said anything interesting :)
05:00 PM Samian: I was countering your claim that AI robotics is overhyped
05:00 PM veverak: how? :D
05:01 PM Samian: by noting that so are trading bots, yet there those few who can do it for real.
05:02 PM veverak: which is completely not relevant information
05:02 PM Samian: so yes AI robotics is likely hyped up, but still if you're good enough, you can do it successfuly
05:03 PM veverak: but "somebdoy can do it succeessfully" does not counter claim that it is overhyped :P :)
05:04 PM Samian: veverak instead of say overhyped, why not just say "harder to do with success than you think"
05:04 PM Samian: that seems to be what you really mean
05:05 PM veverak: that's just your interpretation :P
05:06 PM veverak: 'X is hyped' -> a lot of people are jumping into doing it (jumping - not actually thinking) because it's what cool guys are doin and because it IS SIMPLY AWESOME\o/ (overstating how important and good it is)
05:06 PM veverak: hmm, not overstating
05:06 PM veverak: damn it I forgot the word
05:07 PM veverak: over-evaluating ?
05:07 PM veverak: over-estimating?
05:12 PM Samian: veverak now you seem to be saying hype means many people without much knowledge of the subject saying "they're doing it. I can do it too!"
05:13 PM veverak: that's part of it
05:13 PM Samian: veverak if so, becoming a doctor is overhyped. And the people who think they have what it takes but don't fail in school.
05:13 PM Samian: that's how it works.
05:14 PM veverak: no
05:15 PM veverak: doctor fits part of the description, not the part where the thing is also expected to be more valuable than it actually is
05:16 PM Samian: aha so hype would be people who want to be doctors and think they can figure out immortality
05:17 PM veverak: not really, hype was cloud some time ago
05:17 PM veverak: "cloud is awesome! let's do everything in cloud and it would be better!"
05:17 PM veverak: (and a lot of slogans how cloud is perfect thing for everything)
05:17 PM veverak: except
05:17 PM veverak: that it is not
05:17 PM veverak: cloud is a good tool for some stuff, bad tool for others
05:18 PM veverak: and yet few years back half of the internet would try to convince you that it is perfect for everything
05:18 PM Samian: veverak all of these errors in thinking are by mediocres and underachievers. Whoever cares what they think
05:19 PM Samian: let the smart people do it for real
05:22 PM veverak: well, here is part of what I consider "smart people thing" - honesty: I trully believe that person that gives a fuck about "forbes 400", how some company as medallion outperfmormed everyone, thinks in terms of "harder to do with success than you think", brings up "mediocres and underachievers! who cares what they think" ...... well I think that person is mediocre, as he evidently cares a lot about success
05:22 PM veverak: and being better than other (as that's evidently what he is thinking about) .... and I do care about him as I find that more reasonable ;)
05:27 PM Samian: bringing up financial success was meant only as evidence of the technological success
05:28 PM veverak: well, it's up to you whenever I am right or wrong about, I can't tell ;)
05:28 PM Samian: My point is that technological success is possible, understandly with a lot of hard work and years of dedication
05:30 PM veverak: a lot of things are possible, does not imply they are worthy or propable
05:32 PM veverak: to be specific, in case of ML and robotics, unless you understand why ML is good idea AND why ML is bad idea, you can't tell whenever it makes sense to use it or no, that is the key information for me
05:32 PM veverak: understanding limits and benefits of specific technology is the key thing to use it properly, which is the key for widespread usage
05:53 PM deshipu: Samian: I have some very bad news for you: there is a very high chance that your own achievements are going to be pretty average in whatever you do -- that's how statistics work
05:55 PM deshipu: I would turn your question around and rather ask who cares about what the freaks who are 0.1% of the population think, they are not going to buy enough of whatever you are making to make it worth your while
05:56 PM deshipu: on average you are going to be stuck with average people most of the time
09:05 PM Jak_o_Shadows: hyperthreading is generally bad for lots of things
09:06 PM rue_mohr: ok, maybe if I eat, assemble the 100k modules
09:06 PM rue_mohr: figure out how to fit all the modules in the case
10:28 PM Jak_o_Shadows: I swear electronics tools seem so much more primitive than mechanicl CAD
10:28 PM Jak_o_Shadows: and like, you have the same issues if you want to interface iwth anything mechanical
10:28 PM Jak_o_Shadows: so I don't understand
11:08 PM mrdata: layout is still an art
11:33 PM Jak_o_Shadows: So why are the tools so bad!
11:33 PM Jak_o_Shadows: If i want two holes/pins to be exactly 5mm apart, why can't I constrain that
11:37 PM Tom_L: you can but not with free crap tools
11:49 PM Jak_o_Shadows: yeah, fair