#robotics Logs
Sep 18 2017
#robotics Calendar
12:42 AM mrdata: rue_house, a parallel line a set distance from another line would use the same equation y = m x + b, with a value of b changed by + or - the length of the hypotenuse of a right triangle fitted to the lines
12:50 AM mrdata: h = D / sin theta , where theta is the angle between the original line and the y axis
12:50 AM mrdata: for the needed distance D
12:58 AM rue_bed: no dude, you can just walk around doingy =mx+b, you need to do A*x+b*y=C or you will get nailed by verticle lines
12:58 AM rue_bed: buuut, I see what you mean
12:58 AM rue_bed: I think
01:03 AM mrdata: i'm sure you can extend this to the form you like for lines
01:07 AM mrdata: so, if you want to choose the x intercept when the slope gets hairy, sure
09:23 AM rue_house: the goal was functions that works nomatter what numbers you put in
09:23 AM rue_house: :)
09:23 AM rue_house: no divides by zero
10:53 AM pokmo: hi
10:54 AM pokmo: i'm planning to hack the good old WiiMote and reading http://www.instructables.com/id/Wii-Remote-IR-Camera-Hack/
10:54 AM pokmo: at the end of the article, there's this advice: "If you are getting too much noise in the system (i.e. the camera is seeing too many extraneous light sources) you can use the IR filter from the Wii remote. This will filter the lights which are disrupting the camera."
10:55 AM pokmo: but since it's an IR camera, wouldn't adding an IR filter on it black it out?
11:02 AM SpeedEvil: The filter is an IR pass filter, it filters out other light than IR, it does not filter IR
11:05 AM pokmo: oh, it's a bandpass filter
01:29 PM mumptai: anyone got some additions: https://github.com/cajt/list_of_robot_electronics ??
01:32 PM deshipu: mumptai: I don't think you can list it all
01:33 PM mumptai: only the worthy ones
01:34 PM deshipu: if you are planning to build anything advanced, I wholeheartedly can recommend this: https://www.edx.org/course/underactuated-robotics-mitx-6-832x-0
01:34 PM deshipu: especially if you want to build anything that walks
01:34 PM mumptai: how it that electronics?
01:34 PM deshipu: or grasps
01:35 PM mumptai: i think i actually saw some of the video lectures, and they are indeed good
01:35 PM polprog: robotics is mechanical engineering + electronics + programming
01:35 PM deshipu: and the border between programming and electronics is very fuzzy
01:35 PM deshipu: you can do a pid loop in electronics
01:36 PM deshipu: but most of the time you do it in sofware
01:36 PM polprog: there's also mechatronics ;)
01:36 PM deshipu: that odrive thing you linked to, this is mostly software
01:36 PM mumptai: yes yes, i know
01:36 PM mumptai: i'm just trying to collect the open-source electronics for robots
01:37 PM deshipu: I'm selling a servo controller on Tindie, does that count?
01:37 PM mumptai: its like the most boring part of the robot that can't be bodged together
01:37 PM mumptai: dunno, is it oshw and nice?
01:38 PM deshipu: it's cc-by-sa and I personally like it
01:38 PM deshipu: https://www.tindie.com/products/deshipu/16-channel-servo-shield-for-d1-mini-version-10/
01:39 PM deshipu: you can stack up to 64 of them, 16 servos each :D
01:39 PM polprog: deshipu: wait, you're the nokia screen terminal guy
01:39 PM polprog: lol
01:40 PM deshipu: I'd prefer "that cheap quadruped spider robot guy", but sure
01:40 PM deshipu: http://tote.rtfd.org <-- might also be a good fit, btw
01:44 PM mumptai: nice project, but it doesn't really fit the "electronics" aspect to well
01:45 PM polprog: i imagine you want to compile a list of modular pieces of electronics that can be put together to build a working device
01:45 PM polprog: right?
01:46 PM mumptai: (and it looks way more "complete" than many other ...)
01:46 PM mumptai: tha would be one way to use it
01:47 PM mumptai: but that would require them to have some degree of interoperability
01:48 PM mumptai: another use case would be, to give it the local hackerspace visitor wanting to build a robot, or some student group
01:50 PM mumptai: and it is implicitly targeted "beyond" designs that are well suited for a arduino motor shield
01:50 PM mumptai: so it is alsmost always a many MCU/ many PCB system
01:51 PM mumptai: with some type of internal communication
01:51 PM mumptai: and so on
01:51 PM deshipu: well, I think the servo driver thing would fit
01:51 PM deshipu: I made it specifically for that purpose
01:51 PM mumptai: deshipu, its on the list, and will be in the next commit
01:52 PM deshipu: by the way, it's based on Adafruit's servo controller with the same chip
01:52 PM deshipu: and pololu also has a whole series of them
01:52 PM deshipu: might be worth to have a whole section for servo and motor controllers
01:52 PM polprog: yeah, but try to include diverse projects
01:53 PM polprog: so it's not 50 boards with an avr and some driver chips :PP
01:54 PM deshipu: that's what I meant by saying that you can't possibly list all of them
01:54 PM mumptai: i'm still looking for something that interfaces well with ROS
01:55 PM deshipu: mumptai: I have one for you I guess
01:55 PM polprog: ROS?
01:56 PM deshipu: https://hackaday.io/project/3185-dyio-dynamic-input-output-controller
01:57 PM deshipu: never used it myself, though
01:57 PM mumptai: it doesn't mention ROS
01:58 PM deshipu: I think I remember mentioning it somewhere
01:58 PM deshipu: but he mostly uses Bowler Studio -- his own thing
02:00 PM deshipu: mumptai: this one is for ros: https://hackaday.io/project/21589-core2-a-computer-for-robots
02:00 PM mumptai: maybe something like this, http://wiki.ros.org/andruino but with resonable hw
02:01 PM mumptai: andruino is a good example of a design with a glass ceiling
02:02 PM polprog: the bad thing is that you have to crouch
02:02 PM deshipu: well, tbh most ros robots I know are kinda limited
02:02 PM deshipu: they usually only do SLAM
02:02 PM mumptai: which is okay, it is really low-cost, but it is hard to do much more than reactive behaviors and use more decent sensors or manipulators with it
02:03 PM mumptai: s/and/xor/
02:03 PM deshipu: mumptai: how about stuff like sensors? do you also collect those? I mean, generic sensors like distance and motion, not specialized like lidars
02:03 PM deshipu: this VL53L0X for example
02:04 PM mumptai: there was this lidar pcb on HaD?
02:04 PM deshipu: MappyDot?
02:04 PM mumptai: i think so
02:05 PM deshipu: it's on tindie now
02:05 PM mumptai: but i don't see the point of it anyways
02:05 PM deshipu: clearly you never tried to interface to VL53L0X yoruself :)
02:05 PM deshipu: yourself
02:05 PM mumptai: its basically a breakout with a mcu added
02:05 PM deshipu: yes
02:05 PM deshipu: which saves you about two weeks of your life
02:06 PM mumptai: i guess i could use this: http://xpcc.io/api/classxpcc_1_1_vl53l0.html
02:06 PM deshipu: you could try
02:06 PM deshipu: you still would be able to put multiple sensors on a single i2c bus
02:07 PM deshipu: wouldn't, I mean
02:07 PM mumptai: how broken is its i²c protocol?
02:08 PM deshipu: it has about 700 registers that you have to configure to use it
02:08 PM deshipu: and the i2c address is set through i2c
02:08 PM deshipu: and resets to default on power off
02:10 PM deshipu: mumptai: here's a more polished ros robot: https://hackaday.io/project/21885-rosbot-autonomous-robot-platform
02:10 PM deshipu: mumptai: by the same team that did that core2 thing
02:13 PM mumptai: core2 isn't open hardware
02:13 PM mumptai: let me try to find a schematic
02:14 PM deshipu: hmm, you may be right
02:14 PM deshipu: sorry
02:16 PM mumptai: actually can't find a schematic
02:16 PM deshipu: anything with a cpu is going to be closed, because of the NDAs and such
02:17 PM deshipu: there are no really open hardware computers out there
02:17 PM mumptai: thats okay
02:17 PM mumptai: its about building bots, not chips
02:18 PM deshipu: but then it's mechanical design and programming, and not electronics
02:18 PM mumptai: but i still don't get why they don't hand out the schematic
02:18 PM deshipu: possibly afaraid of getting cloned
02:18 PM deshipu: this is a relatively new startup
02:19 PM deshipu: I know it's stupid
02:19 PM deshipu: if anybody wants to clone them, they will just buy it and reverse-engineer it
02:19 PM deshipu: but some people don't realize that
02:21 PM mumptai: nah, give hte pcb and source code for the stm32 it to me on monday and i'll give you a similar schematic on friday
02:22 PM mumptai: and actually i would not recommmend the board, because I can't debug it without an schematic
02:22 PM mumptai: s/an/a
02:23 PM mumptai: I need at least the interface circuitry, which will 80% of that board anyways
02:25 PM mumptai: the reason i would buy it is because it is a well designed piece of hardware and brings along a nice ecosystem
02:28 PM mumptai: and these things are actually were many oshw design fail
02:30 PM deshipu: fortunately you can take them and improve them ;-)
02:30 PM mumptai: all those one-off projects out there just fall short in this aspect
02:30 PM mumptai: yes, theoretically
02:31 PM deshipu: you just have to re-do all the work they did
02:31 PM deshipu: and then some more
02:31 PM mumptai: but you have to spend resources to get to the point where someone else stoped
02:31 PM deshipu: exactly, that's why oshw makes little sense and little progress
02:31 PM mumptai: which is the step that doesn't happen
02:31 PM deshipu: you can't reuse it like you can software
02:32 PM deshipu: sure, it's nice to have the documentation and all, for debugging
02:32 PM mumptai: most because building the hw is f' hardwork
02:32 PM deshipu: but forget about rebuilding them
02:33 PM mumptai: you can also ask the original author about imrpovements for a consecutive design
02:33 PM mumptai: which also rarely happens
02:35 PM mumptai: anyways, electronics is pretty boring in robots
02:36 PM mumptai: its not really a limiting factor
02:37 PM mumptai: but it is not really easily availiable
02:37 PM deshipu: it really depends on what kind of robot you are building
02:38 PM mumptai: besides extreme environments?
02:38 PM deshipu: if it's a line photovore, a simple op amp will do
02:38 PM deshipu: s/line//
02:38 PM deshipu: a line follower can run on any mcu on the market
02:39 PM mumptai: also most robots don't need cutting edge electronics
02:39 PM deshipu: they might, but it all depends on the design
02:40 PM deshipu: the sensors will often be a part of the mechanical design
02:40 PM deshipu: not mentioning the actuators
02:41 PM deshipu: and in practical designs, you are not actually building a robot, you are building a XXX machine, where XXX is the function you want it to perform
02:41 PM mumptai: i think there is more need development no sensors, gearboxes, behaviors and models (which is mostly math)
02:41 PM deshipu: like CNCs are technically robots
02:41 PM deshipu: as well as 3d printers
02:41 PM deshipu: but I don't see you collecting oshw designs for those on your page
02:42 PM mumptai: repraps are basically bad robots
02:42 PM mumptai: so little sensors
02:42 PM mumptai: inputs
02:43 PM deshipu: they do the job the were build for
02:43 PM deshipu: built
02:43 PM deshipu: that's the point though — the more a particular use case for a robot matures and becomes usable, the less likely we will call them robots anymore
02:44 PM mumptai: they are algorithmically more similar to a tape recorder than a simple line follower
02:44 PM deshipu: it's almost as if robots are by definition impractical and clumsy
02:44 PM deshipu: and as soon as they become practica and reliable, we call them something else
02:45 PM deshipu: mumptai: how about pick and place machines? enough sensors for you?
02:45 PM mumptai: the new ones: yes
02:45 PM deshipu: there is a bunch of oshw pnp designs out there
02:46 PM mumptai: force-sensing and vision on these is quite nice
02:46 PM mumptai: but the oshw PNPs hw is mostly reprap stuff
02:47 PM mumptai: the smarts is in the PCs software
02:48 PM deshipu: because it doesn't have to be locked up in a proprietary black box
02:49 PM deshipu: welded to the machine you sell
02:50 PM mumptai: i guess i consider 3d-printers a comodity now, and similar there g-code stepper driver cards
02:51 PM mumptai: i still need a nice BMS and power distribution
02:51 PM deshipu: washing machines would have been considered robots 60 years ago
02:51 PM deshipu: or dishwashers
02:51 PM mumptai: todays washing machines actually contain interesting behaviors
02:52 PM deshipu: sure, and have sensors
02:52 PM deshipu: like the scales for weighting your laundry
02:53 PM deshipu: there are microwave ovens with cameras
02:53 PM deshipu: and with vision algorithms
02:58 PM Tom_itx is now known as Tom_L
02:58 PM mumptai: anyways i'd like to bridge the gap across boring electronics divine to get to the interesting robotics stuff
02:58 PM deshipu: sadly, the interesting robotcis stuff is boring programming :(
02:58 PM mumptai: kinda the tool to build the behaviors
02:59 PM deshipu: I wish there was something like ROS, but more for learning
02:59 PM deshipu: and less complex
03:00 PM deshipu: so that you could still use ready blocks and libraries, but without all this java and xml
03:01 PM deshipu: and I also wish I already knew how to use it, so that I didn't have to spend effort learning it
03:02 PM mumptai: the typical framework/lib problem
03:03 PM deshipu: yeah, learning is hard, so most people write their own, so that they don't have to read the docs
03:03 PM weyland|yutani: deshipu, currently i started dabbling in fpga stuff....its like opening a door and mostly everything you learned before is absolutely irrelevent to the task
03:04 PM deshipu: weyland|yutani: I'm steering clear from FPGAs for the very reason :)
03:05 PM mumptai: low level robtics stuff is actually a great fit for fpgas
03:05 PM weyland|yutani: deshipu, its the closest you can get to "talking with the hardware gods"
03:05 PM mumptai: all the timing critical and parallel interfacing stuff
03:10 PM mumptai: actually there is https://github.com/loa-org which is robot stuff on a stm32 + fpga module
03:11 PM deshipu: weyland|yutani: you can build your cpus out of logic gates or transistors too
03:11 PM deshipu: weyland|yutani: so you definitely can get lower
03:12 PM weyland|yutani: deshipu, well only if i could run my design in your fab and produce my own Asic
03:13 PM deshipu: weyland|yutani: you can build it out of mechanical relays in your basement
03:13 PM deshipu: or even out of marbles, if your really like it
03:14 PM weyland|yutani: i dont think its upscaleable to the point converting radar data in realtime
03:14 PM deshipu: I don't like fpgas myself, because they turn a fun hobby (soldering stuff together) into a boring hobby (writing code and debugging)
03:14 PM deshipu: weyland|yutani: it just has to be very long-wave radar ;)
03:16 PM mumptai: its not that bad, and you can do the really crazy stuff with it
03:17 PM weyland|yutani: deshipu, well debbuging a controlunit made from 5 large breadboards with a docen IC and stuff is equally bad
03:18 PM deshipu: weyland|yutani: the whole trick is that you can never make it grow to those sizes :)
03:18 PM weyland|yutani: deshipu, or for starters build a 8 bit computer on breadboards from scratch and then find your error why it wont work
03:19 PM deshipu: like how I started with quadruped walking robots
03:19 PM deshipu: because with wheeled robots you make the robot and then you have to think about the behaviors
03:19 PM deshipu: with a walking robot I could spend a year focusing justr on the walking
03:19 PM deshipu: without even getting to the hard part
03:20 PM weyland|yutani: yea i know getting the point across that killing people is bad i know
03:20 PM weyland|yutani: not that a robot i ever build did that
03:20 PM weyland|yutani: but you get the drift
03:20 PM weyland|yutani: baby steps and euqal problems
03:21 PM deshipu: pick your battles
03:22 PM weyland|yutani: i do i mostly learn fpga stuff for signal analysis
03:24 PM mumptai: sounds like an good application for the HLS stuff
03:26 PM deshipu: signal analysis by itself is a huge topic
03:27 PM weyland|yutani: i know
03:29 PM weyland|yutani: but as someone who installs intercoms and bus systems on a regular basis i like to take stuff apart and want to see whats behind it
03:41 PM mumptai: fpgas also seem to be rather hobbyist unfriendly for getting starting
03:43 PM deshipu: it's changing
03:44 PM veverak: hmm
03:44 PM veverak: http://www.eminebea.com/en/product/rotary/steppingmotor/pm/standard/pm35l-048.shtml
03:44 PM veverak: current/phase -> if I want to know the watage
03:44 PM veverak: I calculate 2*500mA*24V ?
03:46 PM weyland|yutani: mumptai, i dont care i not your run of the mill hobbyist
03:54 PM mumptai: yeah, have fun, and don't get stopped by the demotivators
03:57 PM weyland|yutani: as long a rue is talking to himself like a madman the universe keeps expanding and so our wisdom
03:59 PM deshipu: what do you mean by "like"?
04:00 PM weyland|yutani: curiosity
04:39 PM peepsalot: heehee, i got some pneumatic barb fittings with m3 threading, so tiny
04:43 PM peepsalot: almost thought i was shipped an empty epacket
04:44 PM weyland|yutani: maybe its just a illusion
04:44 PM peepsalot: https://dl2.pushbulletusercontent.com/0Trj2wn1FoW6KqZEYhJhcZpEjTzeFWWD/IMG_20170918_161302.jpg
04:45 PM weyland|yutani: they are really tiny
05:20 PM rue_shop3: ugh, I should give away 7.5 degree motors, trim down the pile a bit
07:46 PM Luminax-Work is now known as LuminaxWk
08:06 PM Guest67243 is now known as jason__
09:02 PM rue_shop3: hate slow computers
09:14 PM Luminax-Work is now known as LuminaxWk
10:29 PM rue_shop3: Tom_L, you still alive?
10:29 PM rue_shop3: have you ever tightened up the jibs on your sherline
10:29 PM rue_shop3: ?
10:44 PM Tom_L: am so
10:44 PM Tom_L: quite a while back i did
10:52 PM Tom_L: replaced the Z axis bronze drive nut once too
10:54 PM rue_house: need to tighten the jibs
10:54 PM rue_house: are they just held in with the one grub screw and the L bar?
10:54 PM Tom_L: sounds about right
10:55 PM rue_house: k
10:55 PM Tom_L: you can only get it so good
11:54 PM rue_house: Tom_L, has chanserv always been here?
11:54 PM rue_house: strikes me as odd suddenly