[00:12.640 --> 00:19.240]  My name is Ask, Ask at the Info. I answer questions here at DEF CON. I've been Goonan
[00:19.240 --> 00:25.100]  for four years, and I've been a maker in our community for eight. And this year, for one
[00:25.100 --> 00:30.900]  of the events that I was hosting with my friends, RentAnAssassin.com, we were throwing a client
[00:30.900 --> 00:37.200]  acquisition party, and I was commissioned to create some drink robots. One of my friends,
[00:37.400 --> 00:43.400]  a fellow Goon, not going to dox him, asked me if I could use peristaltic pumps and put
[00:43.400 --> 00:46.980]  together a bar robot. So I was like, I can do that. That sounds like it would be a fun
[00:46.980 --> 00:53.560]  challenge for me. And I had recently been practicing using OpenSCAD, which is a programming
[00:53.560 --> 01:00.100]  language for 3D modeling, to create 3D parts. So I wanted to challenge myself to create
[01:00.100 --> 01:06.080]  something new that would solve the problem using parts off the shelf. Because circuit
[01:06.080 --> 01:10.860]  board design is something I'm getting into right now. I'm a novice at it. But being able
[01:10.860 --> 01:15.260]  to hook together all sorts of random kits of things, I am really good at just building
[01:15.260 --> 01:20.720]  one-off, two-off projects for people. So when my friend challenged me to do this, I said,
[01:20.720 --> 01:24.420]  yeah, I will totally get that taken care of for you. And so I actually started with figuring
[01:24.420 --> 01:28.420]  out what the circuit was going to be, what the main circuit chip was going to be. And
[01:28.420 --> 01:35.440]  recently we had this device come out that all of us are using, called the Pi Pico. And
[01:35.440 --> 01:40.100]  it's a $4 chip from Raspberry Pi, and it allows you to load a programming language called
[01:40.100 --> 01:47.000]  CircuitPython on it. And I had recently been using CircuitPython for my friend DBSAU's
[01:47.000 --> 01:51.180]  PicoDucky, and I even designed a 3D printed case for that and everything. So I had just
[01:51.180 --> 01:54.560]  gotten a lot of experience on this platform. And I said, well, I've gotten a lot of experience
[01:54.560 --> 02:00.360]  with object-oriented Python at work recently. And I have gotten a lot of work with 3D printing
[02:00.360 --> 02:06.160]  recently. Let's put this to work. So I ordered a bunch of parts. I got a bunch of relays.
[02:06.160 --> 02:12.140]  We got the power rails and everything. And I actually designed a circuit board that I
[02:12.140 --> 02:16.680]  3D printed. So we went through multiple iterations of this circuit board, where we could actually
[02:17.280 --> 02:22.200]  put all of these maker parts on there. And we ended up with 12 relays, six on each side,
[02:22.200 --> 02:28.820]  with PyPico in the middle, 12-volt power source, and a 5-volt and a 3-volt buck converter
[02:28.820 --> 02:35.000]  to allow it to power the relays and to power the Pico. Because the Pico actually will run
[02:35.000 --> 02:38.340]  off a 5-volt, and then the relays that we got were 3-volt, and then they were switching
[02:38.460 --> 02:43.100]  a 12-volt rail. Which actually ended up being a cog in the works that went away as soon
[02:43.100 --> 02:48.080]  as I stopped touching the robots. So I built up the entire base platform where I could
[02:48.080 --> 02:51.840]  play with the relays on a button board. So I would put buttons out there and play with
[02:51.840 --> 02:54.980]  the relays, and then I was trying to figure out how do I want to kit this together. And
[02:54.980 --> 03:00.320]  originally we wanted to put this in a custom 3D printed housing. So I actually did the
[03:00.320 --> 03:05.460]  full 3D modeling for a self-contained housing that this all would be mounted into. But in
[03:05.460 --> 03:09.120]  the end, that ended up being too long of a print, and the printers that we had just were
[03:09.120 --> 03:14.260]  not up to par, and I don't want to go buy more tools. My lab already has enough tools
[03:14.260 --> 03:18.240]  in it. So we went with what we had, and I looked around, and I saw an old milk crate
[03:18.240 --> 03:23.160]  that I had had from when I previously had worked in maintenance at my alma mater. So
[03:23.160 --> 03:26.680]  I just went and I grabbed the milk crate, and I was like, how does this all fit in here?
[03:26.680 --> 03:31.260]  And I designed a board for the perisaltic pumps to all mount into. I designed a face
[03:31.260 --> 03:36.080]  plate for the screen to mount into. I designed a face plate for the buttons to mount into.
[03:36.320 --> 03:40.940]  And then all that wiring and everything is open in there with a splash guard for the
[03:40.940 --> 03:46.140]  pour nozzles. And actually that pour nozzle really showed how far I've come with OpenSCAD
[03:46.140 --> 03:49.720]  because I was challenged by my friend to get that done right before DEF CON, and I
[03:49.720 --> 03:53.420]  got that done in about an hour and a half. I was able to write all the programming in
[03:53.420 --> 03:56.800]  about half an hour and get a prototype printed, because I've actually learned how to print
[03:56.800 --> 04:00.380]  prototypes really fast, too, by taking out the parts you don't need and only printing
[04:00.380 --> 04:04.840]  the parts you really are caring about measuring. And I was able to successfully last minute
[04:04.840 --> 04:09.500]  get the whole nozzle system going. So in the end, we ended up with a pair of
[04:09.500 --> 04:14.980]  senior Drinkenstein robots or Dr. McGillicuddy's magic drink elixir mixer running on Circuit
[04:14.980 --> 04:19.720]  Python serving eight different drinks from nine different liquid sources. And back to
[04:19.720 --> 04:23.720]  that problem that we had, we actually have a problem with EMI, electromagnetic interference,
[04:23.720 --> 04:27.960]  where our 12-volt lines, when switched on, would interfere on the three-volt pull-up
[04:27.960 --> 04:32.880]  or pull-down buttons and cause the whole thing to lock up completely. We were able to partially
[04:32.880 --> 04:37.240]  solve that, and what we actually found was as soon as I left tending the robots and left
[04:37.240 --> 04:41.620]  them to Bonwit to manage the robots, they didn't break at all. They didn't reboot at
[04:41.620 --> 04:45.620]  all. So we actually think the robots gained sentience and hate their creator.
[04:45.620 --> 04:53.800]  It can only be attributable to human error. This conversation can serve no purpose anymore. Goodbye.