#robotics Logs

Nov 13 2012

16:24 smeding hi
16:24 tobbor smeding! like, totally tell us about the project!
16:25 smeding that seems bot-like
16:25 Tom_itx you think?
16:37 tomaszek I have a question
16:37 tomaszek is it channel which connects electrical programming wich psychology?
16:38 tomaszek in meaning - neural nets etc
16:38 smeding neural nets have nothing to do with psychology really :)
16:38 tomaszek by name it connects
16:40 smeding artificial neural networks are a part of artificial intelligence, and i'd say that's definitely on topic
16:40 smeding but robotics is broader than artificial intelligence
17:21 tomaszek Is here anybody active?
17:25 smeding no
17:25 smeding if you want people to be active, start an interesting conversation :)
17:31 any70466778 ANN are fuzzily programmed state machines
17:31 smeding ANNs are not programmed, they're trained :)
17:32 dumbass same thing
17:33 ace4016 they're programmed, but in a more organic way :P
17:36 dumbass "a more organic way" = they make mistakes
17:37 dumbass "a more organic way" = they can be confused
17:37 ace4016 indeed :D
17:37 dumbass "a more organic way" = they act randomly when exposed to something they weren't trained on
17:37 ace4016 heard the story about the ANN that was trained to identify and shoot tanks? but during training they only showed day time shots, so when they showed night time shots, it couldn't identify anything? :P
17:38 dumbass right
17:38 dumbass and when training on aircraft they had to show it so many pics from all angles it just got too pricey, and the memory requirements got out of sight
17:41 dumbass they even got into lasers and holograms as storage, and that got physically too big to put into a small air-to-air missile
17:43 ace4016 heh
17:44 ace4016 well, it takes humans months after coming out of the womb to make sense of the world
17:44 ace4016 can't expect robots to do it in a day :P
17:45 ace4016 and even after a few months, they still don't quite understand many "important" concepts
17:45 dumbass there's a huge lack of common sense online, what is it they are to learn from?
17:46 dumbass there is nothing to say that dolphins do not walk on your ceiling at nite
17:47 dumbass everything to dis/prove that occurance hasto be dug up from somewhere available to a computer, and the data just isn't there
17:47 dumbass afk
18:20 dumbass no one had anything to say while i was afk?!
18:23 dumbass k, bbl
19:04 rue_mohr I wrote a program to day Hello
19:05 rue_mohr >++>+++>+++[[>+++++++++<-]<]>>>-->[[>++++<-]<]>>.>+.>..+++.
19:05 rue_mohr say hello...
19:13 dumbass i dunno what that is
19:21 dumbass HelLO
19:22 dumbass hello
19:22 dumbass HELLO
19:22 rue_mohr thats source code to output Hello
19:22 dumbass hELio
19:22 rue_mohr I wrote it this morning
19:22 dumbass i use other source code
19:23 rue_mohr ><+-.,[]
19:24 rue_mohr the instructions are incrememnt memory pointer, decrement memory pointer, increment memory value, decrement memory value, output memory value, input to memory value, conditional jump forward and conditional jump backward
19:25 dumbass ^(><)^
19:26 rue_mohr want to make a small language capable of self-modification
19:26 dumbass i wanna be able to look at it and know what i did after a year away from it
19:26 rue_mohr :)
19:27 rue_mohr what are the most basic instructions you would say a processor should have
19:28 dumbass marshmellow
19:29 rue_mohr Nop, yes, what else?
19:29 dumbass marshmellow
19:29 rue_mohr input, yes, ok, what else?
19:30 mbass goes back to playing vect
19:30 rue_mohr output, ok, what else
19:30 rue_mohr I'm thinking 3 registers, instruction pointer, memory pointer, and "register"
19:32 rue_mohr oo memory mapped registers?
20:12 dumbass Human being usually contacts with the external
20:12 dumbass environment with his hand.
20:42 rue_mohr ok I think I designed an instruction set that forces self modifying code
20:45 rue_mohr 16 instructions
20:50 Tom_itx 90v dc motor control
20:50 Tom_itx reversable
20:50 Tom_itx ?
20:50 Tom_itx 115vac input
20:50 rue_mohr ok
20:50 Tom_itx for rigid spindle tapping
20:51 Tom_itx i can put a relay on the current control
20:51 rue_mohr driving a tap?
20:57 any45135487 http://up-ship.com/
20:57 Tom_itx yes
20:58 Tom_itx http://www.kbelectronics.com/data_sheets/kbcc.pdf
20:58 Tom_itx that might do it
21:20 rue_mohr Iwonder how much that matches the circuit board in a powered flat running thing with handlebars to stay put when running
23:14 NotWorthCr4p marshmellow
23:14 rue_mohr I made my self-modifying language interperter, so far it works
23:15 rue_mohr now I can explore self-modifying code
23:15 NotWorthCr4p i did self-modifying code in the 1980's on a C64
23:16 e_mohr nods, I tip my hat to the 6502 and its patheticly small index regis
23:16 NotWorthCr4p it worked, it beat IBM all the heck and back
23:17 rue_mohr what I have right now is a mating of brainfuck and 6502 assembler
23:17 NotWorthCr4p k
23:17 rue_mohr opcodes:
23:17 rue_mohr 0 NOP
23:17 rue_mohr 1 IN IN->*X
23:17 rue_mohr 2 OUT *X->OUT
23:17 rue_mohr 3 v XSET V *X = V
23:17 rue_mohr 4 v YSET V *Y = V
23:17 rue_mohr 5 v XADD V *X = (*X) + V
23:17 rue_mohr 6 v YADD V *Y = (*Y) + V
23:17 rue_mohr 7 NEGX *X = -(*X)
23:17 rue_mohr 8 NEGY *Y = -(*Y)
23:17 rue_mohr 9 v ADDX V X = X + V
23:17 rue_mohr 10 V ADDY V Y = Y + V
23:17 rue_mohr 11 V XJNZ V jump to V if *X != 0
23:17 rue_mohr 12 V XJZ V jump to V is *X == 0
23:17 rue_mohr 13 V YJNZ V jump to V if *Y != 0
23:17 rue_mohr 14 V YJZ V jump to V is *Y == 0
23:17 rue_mohr 15 HALT
23:17 rue_mohr 1 5 1 2 15 << this program inputs a character, ads 1 to it and outputs it
23:18 rue_mohr V is litteral, not a register
23:18 rue_mohr beacuse V is in memory, and memory can be modified
23:18 NotWorthCr4p lots of work for what one adder chip can do 4 times, and prolly 100x faster
23:19 rue_mohr I want to write software the writes new instructions
23:19 rue_mohr ;) like DNA
23:19 NotWorthCr4p k
23:20 rue_mohr http://paste.debian.net/209098/
23:21 rue_mohr TELunus|2, can you prove it turring complete?
23:24 TELunus|2 Can you implement brainfuck in it?
23:25 TELunus|2 If yes it's turring complete.
23:25 rue_mohr hahaha I dont want to! :)
23:25 rue_mohr I think it can implement itself...
23:25 rue_mohr prolly could do brainfuck...
23:25 rue_mohr anyhow
23:26 rue_mohr I need to do simple things to understand it better
23:27 rue_mohr I hit my finger with a drill today :(
23:27 rue_mohr it had a robinson bit in it at the time
23:28 rue_mohr I dont know if the resulting damage is better or worse than a drill bit would have been
23:32 tWorthCr4p grabs a hammer and offers to examin
23:32 e_mohr examines the crystaline structure within its m
23:33 rue_mohr mmm all the complex material density regions....
23:33 rue_mohr the beauty of the magnetic flux lines
23:34 e_mohr passes some current thru it and watches the electron migration
23:34 NotWorthCr4p marshmellow
23:34 e_mohr eats the marshme
23:34 rue_mohr thankyou!
23:34 e_mohr offers kat some lemon
23:35 NotWorthCr4p no
23:35 rue_mohr your choice.
23:35 NotWorthCr4p my choice is to be worth a crap, but my choice isn't valid
23:36 NotWorthCr4p Brooks has coined a concept that reflects this notion: embodiment.
23:36 NotWorthCr4p The embodiment of a creature defines its actions as �part of a dynamic
23:36 NotWorthCr4p with the world, having immediate feedback on the creature�s own sensations
23:36 NotWorthCr4p through direct physical coupling and its consequences� [18].
23:36 NotWorthCr4p As embodied beings, our morphology, sensorimotor system, and neurological
23:36 NotWorthCr4p apparatus, integrate empirical information that shapes our
23:36 NotWorthCr4p behavior
23:36 rue_mohr without the ascii table, 0x20 wouldn't mean a fraction of what it does.
23:37 e_mohr is curious about what table kat is u
23:38 NotWorthCr4p one where everyone else is changing the values
23:38 e_mohr plays "Dean Martin - Memories are made of t
23:39 NotWorthCr4p an alien sentiment
23:40 e_mohr is mildly offe
23:40 NotWorthCr4p appearance of
23:40 NotWorthCr4p higher level �intelligence� can be imputed to automata by furnishing
23:40 NotWorthCr4p them with biologically inspired motivations
23:40 NotWorthCr4p but it's only an apparance
23:41 rue_mohr heh
23:41 rue_mohr dont be surprised if its all just apperance
23:42 rue_mohr is apparence of inteligence is just sympothetic from an alian perspective
23:42 NotWorthCr4p then i am correct: your robot won't be you, it will only appear to be you
23:42 rue_mohr of course
23:42 NotWorthCr4p that sucks for you
23:42 rue_mohr by definition I die
23:43 rue_mohr but my goal is for my process to continue
23:43 NotWorthCr4p you wanted to move into the robot
23:43 rue_mohr and that is what I deem to be me
23:43 rue_mohr the medium that implements the process is insignificant
23:44 NotWorthCr4p it cannot even appear to be you in detail, like what happened on April 15 in 1984, and you know what i mean
23:44 NotWorthCr4p the robot won't know
23:44 rue_mohr but extraction of the process is extrememly complex
23:44 rue_mohr so I need to bootstrap with an AI that can figure it out before I die
23:44 NotWorthCr4p uh huh, it's follows you around asking "and then what?" tilly ou die
23:45 rue_mohr heh
23:46 Jak_o_Shadows1 NotWorthCr4p: How am i meant to alias you back to Katsmeow when you keep changing your name?
23:46 rue_mohr was thinking about 3d location of eletrostatic discharge within a medium of variable conductance
23:46 NotWorthCr4p jak, easy: find out how to negate how rue often makes me feel
23:47 NotWorthCr4p rue: easy: parallel fets tween two planes
23:47 e_mohr sh
23:47 rue_mohr I regard her quite highly
23:47 NotWorthCr4p then why the marshemellow incident?
23:48 NotWorthCr4p you begged me to talk about a project, and then made fun of me, i wasted my time, i was the butt of the joke
23:48 rue_mohr 'higher level intelligence can be imputed to automata by furnishing'
23:48 rue_mohr I made no fun of you
23:48 TELunus|2 NotWorthCr4p: you say often, yet you refer to a single incident.
23:49 NotWorthCr4p the pink string incident, i destroyed the coupling AND the pcb driller
23:49 rue_mohr that was not me
23:49 rue_mohr I'm 100% sure
23:49 Jak_o_Shadows1 tobbor logs everything....
23:49 rue_mohr its the only one I WASNT guilty of
23:49 rue_mohr yes, I dont remember what year it was
23:49 NotWorthCr4p the tree trimming = shooting babies (that one got blue-tagged gov trucks parked outside my driveway gates the next morning)
23:50 rue_mohr yes, logging makes me sick
23:50 rue_mohr I know your not clearcutting
23:50 rue_mohr but I told you that was like talking to you about dogs
23:50 NotWorthCr4p then WHY do I make you sick?!?!?!
23:50 rue_mohr just that topic
23:50 rue_mohr I'm sure everyone has a crux
23:51 Jak_o_Shadows1 Anyway, back to the Fet discussion we had yesterday, is this about right? http://jak-o-shadows.users.sourceforge.net/random/Fetdriver.PNG
23:52 NotWorthCr4p WHY do I make you sick?
23:52 rue_mohr you dont, that tree cutting topic does
23:52 NotWorthCr4p no, use an npn for the Q1, and redesign
23:52 rue_mohr Jak_o_Shadows1, I think something is wrong
23:53 rue_mohr your upside transistor is slowing me down
23:53 NotWorthCr4p most people put the pos voltage to the top of the paper and not the bottom
23:53 rue_mohr Jak_o_Shadows1, let me demontrate
23:53 NotWorthCr4p yea,, Q1 being upside down is screwy
23:54 rue_mohr http://eds.dyndns.org/~ircjunk/tutorials/elex/schematic.jpg
23:54 NotWorthCr4p in your case, swap the thing and the do-dad
23:54 rue_mohr kat, I think he needs a common base? amp for that?
23:55 rue_mohr the C and E of the transistor are definitly backwards
23:55 NotWorthCr4p it's all basackwards,, does his fet need to pull up or down to turn on?
23:55 rue_mohr but removing sw1 later, like he wants to do isn't gonna work
23:55 rue_mohr up
23:56 rue_mohr resistor keeps it off
23:56 rue_mohr Jak_o_Shadows1, I dont have the brainpower to design a common base amp for doing the levelshifting you need
23:56 rue_mohr kat might be able to do it, even tho its like 2am over there
23:57 NotWorthCr4p then his transistor must be on to turn the motor off, ground it's emiter, collector to gate, base (thru a resistor) to 3.3v signal
23:57 rue_mohr he has Q1 wrong
23:57 Jak_o_Shadows1 i don't know if i have any npnp transistors
23:57 rue_mohr he wants to use a 3.3v microcontroller to switch the fet
23:57 rue_mohr npnp? thats a triac right kat?
23:57 NotWorthCr4p or scr
23:57 rue_mohr or a SRC
23:58 Jak_o_Shadows1 npn*
23:58 rue_mohr right, thanks
23:58 NotWorthCr4p he can't keep the fet on
23:58 rue_mohr can you help him make it so the 3.3V signal can switch the gate of the fet to 16V?
23:58 rue_mohr I'm going to bed
23:58 rue_mohr I cant think
23:58 NotWorthCr4p no
23:59 rue_mohr I created a self-modifying language interperter, I'v done my part
23:59 NotWorthCr4p he'd need two transistors
23:59 NotWorthCr4p or a p-channel fet
23:59 rue_mohr go slow swithing with the resistor pulldown