[01:15.560 --> 01:17.040]  Do you have audio for us?
[01:18.320 --> 01:22.320]  It has to pre-render before I can really go through it, but I can try.
[01:23.720 --> 01:25.820]  Oh, but I'm showing everyone all my slides.
[01:26.740 --> 01:27.620]  Here, hold on.
[01:28.560 --> 01:29.760]  I'll not have it pre-render.
[01:34.240 --> 01:34.680]  Audio.
[01:34.940 --> 01:36.060]  How about just some music?
[01:36.420 --> 01:36.660]  Okay.
[02:04.370 --> 02:05.490]  One minute, you can start.
[02:11.150 --> 02:12.450]  Just make sure.
[02:12.710 --> 02:16.070]  Just hold a t-shirt up over the head.
[02:16.870 --> 02:18.430]  We did it with a t-shirt.
[02:18.670 --> 02:19.950]  But that's a big whole t-shirt.
[02:23.590 --> 02:25.770]  Yeah, like build it out.
[02:26.130 --> 02:27.250]  It's a small light source.
[02:27.470 --> 02:28.030]  It's a big video.
[02:57.490 --> 02:58.530]  I mean...
[02:58.530 --> 02:59.310]  Should I start?
[03:01.050 --> 03:01.930]  Hi, everyone.
[03:03.110 --> 03:04.110]  I'm Nadya Peek.
[03:05.770 --> 03:07.450]  I'm going to speak for the next hour.
[03:07.630 --> 03:09.690]  So if you didn't want to listen to me speak, you're in the wrong room.
[03:10.750 --> 03:14.010]  I'm a PhD student at MIT in the Center for Bits and Atoms.
[03:14.150 --> 03:17.270]  I work on machine control systems and machine design.
[03:18.330 --> 03:23.490]  And I'm going to be talking about machine design as a form of descent.
[03:24.490 --> 03:25.270]  I don't know...
[03:25.270 --> 03:26.510]  I'm going to start with like a short story.
[03:26.650 --> 03:28.290]  I don't know if any of you guys remember...
[03:28.830 --> 03:29.230]  Hello.
[03:31.330 --> 03:34.690]  I don't know if any of you guys remember this red balloon challenge.
[03:35.310 --> 03:37.590]  DARPA released it in 2009.
[03:38.010 --> 03:48.970]  There were 10 red weather balloons that they kind of put out throughout the United States and said, whatever team is able to find all the locations of all of the balloons is going to win $40,000.
[03:49.430 --> 03:58.910]  And I thought it was a really thinly veiled attempt to try to gather information about how public might use social media for dissent.
[03:58.910 --> 04:03.030]  And I thought it was stupid that taxpayer money was paying for it.
[04:03.210 --> 04:04.250]  And I thought...
[04:04.250 --> 04:06.410]  A lot of the things about it I thought were really problematic.
[04:06.670 --> 04:07.610]  Red balloon searching.
[04:07.930 --> 04:17.530]  And so we decided that we're going to hack the competition by putting up fake red weather balloons.
[04:20.450 --> 04:22.210]  I happen to actually...
[04:22.210 --> 04:25.890]  Some of the research at MIT is funded by DARPA, including mine.
[04:25.890 --> 04:28.470]  And so I happen to have a lot of DARPA paraphernalia.
[04:28.590 --> 04:32.130]  So I can pretty justifiably pull off DARPA impersonator.
[04:32.650 --> 04:36.950]  And on the right is Jeff Warren from Public Lab.
[04:37.570 --> 04:40.450]  And they do balloon mapping with red weather balloons.
[04:40.590 --> 04:45.370]  So we actually also had a big supply of big red weather balloons that we could put up all over the place.
[04:45.590 --> 04:50.890]  And I work in this really quite well-equipped laboratory where we have input...
[04:51.670 --> 04:54.130]  We call them input and output devices across length scales.
[04:54.130 --> 04:57.290]  So we have laser scanners, water jet cutters.
[04:57.470 --> 05:02.750]  We have CT scanners, AFM lasers, and a whole bunch of fabrication equipment.
[05:03.010 --> 05:08.810]  And so making the little flag balloon things that we could put up to impersonate...
[05:08.810 --> 05:12.750]  To impersonate the red balloon challenge was really relatively trivial.
[05:12.950 --> 05:16.050]  It took us about five minutes of fabrication and a little bit of helium.
[05:17.990 --> 05:25.810]  And the kind of tools that we have access to because we're a research organization that works on fabrication is really quite astounding.
[05:25.930 --> 05:29.630]  We have all this stuff that we can do DNA folding with.
[05:29.750 --> 05:36.530]  You know, some of the stuff that came out of our labs includes e-ink or that annoying sensor that beeps at you if you don't buckle your seatbelt.
[05:38.470 --> 05:47.870]  And I kind of feel like us being able to make something to be able to subvert a narrative about DARPA is something that everyone should be able to do.
[05:48.430 --> 05:50.790]  Because we really had a lot of fun doing it.
[05:50.790 --> 05:57.490]  And so using the tools in our lab, we teach a class called How to Make Almost Anything.
[05:57.790 --> 06:01.230]  I don't know if any of you guys have heard of my advisor, Neil Gershenfeld.
[06:01.530 --> 06:18.050]  But How to Make Almost Anything takes you through making everything from electronics to large format furniture or houses using these tools that we have access to.
[06:18.050 --> 06:23.670]  And we don't actually stipulate that anyone has to do whatever they... they can do whatever they want with the tools.
[06:23.870 --> 06:30.670]  And so people end up making all these very useful things like web browsers for their parrots or alarm clocks that you have to wrestle with.
[06:30.890 --> 06:41.270]  Or, you know, really personal projects that have nothing to do with business models or economics or products that are aimed for mass manufacturing.
[06:41.270 --> 06:47.650]  And so digital fabrication as defined as, you know, using a computer to control a fabrication device.
[06:47.930 --> 06:51.930]  Making it easy for you to just alter some code and then remake something in a different way.
[06:52.150 --> 06:53.930]  It's really quite a powerful technology.
[06:54.910 --> 07:00.050]  But I'm going to go back in history a little bit to think about where this maybe all came from.
[07:00.050 --> 07:06.870]  And how history might inform how we might use it going forward.
[07:07.770 --> 07:10.770]  So I'm pretty sure none of you remember the Industrial Revolution.
[07:12.710 --> 07:19.930]  But when it started, one of the things that was really exciting about it was people were able to get products because products were becoming cheaper.
[07:20.070 --> 07:21.410]  Because manufacturing was becoming easier.
[07:21.550 --> 07:29.770]  And we had all of these distribution networks where people could buy things because factories could ship goods through the railroad system or through the road system to different places.
[07:29.770 --> 07:34.530]  And we got automobiles, which also enabled personal mobility.
[07:35.050 --> 07:44.130]  And a lot of the impetus for developing better fabrication tools kind of came from the war.
[07:44.810 --> 07:51.410]  There was a big need for precision manufacturing of things so that you could make things like missile nose cones.
[07:51.610 --> 07:53.950]  Really nice stuff that everyone loves working on.
[07:53.950 --> 08:03.510]  And that kind of confluence of scientific and engineering efforts was extremely strong during the war.
[08:03.630 --> 08:11.150]  And as soon as the war was over, the question kind of became what is it that we're going to focus on now as scientists and technologists.
[08:11.150 --> 08:13.170]  And I don't think...
[08:13.170 --> 08:13.410]  Good afternoon.
[08:14.330 --> 08:15.990]  I don't think that any of...
[08:15.990 --> 08:16.130]  I don't think that any of...
[08:16.130 --> 08:16.730]  You're waiting for it?
[08:17.110 --> 08:17.250]  No.
[08:18.710 --> 08:20.110]  If anyone wants to go over there.
[08:21.070 --> 08:22.030]  So the...
[08:23.290 --> 08:26.150]  So after the war was over, the...
[08:27.750 --> 08:28.510]  Just kind of distracted.
[08:30.230 --> 08:37.550]  It was kind of difficult for different kinds of people to work on, or to figure out what we're collectively going to be working on next.
[08:37.610 --> 08:44.350]  If there isn't imminent threat of global destruction, then how are you going to work together on the next thing?
[08:44.610 --> 08:55.350]  But, continued making tools for fabrication, and a lot of the tools that were being developed were still very strongly grounded in kind of a military application.
[08:55.350 --> 09:04.870]  So, making more consumer cars, but still with strong analogy to military vehicles, to making flight.
[09:05.670 --> 09:22.750]  And so, automation was something that was really taking the Western world by storm, enabling the mass production of many devices, using economies of scale to be able to give people cheap products.
[09:22.750 --> 09:29.830]  But, those products weren't necessarily the kinds of things that at least someone maybe like me would want to see on the market.
[09:30.230 --> 09:33.610]  So, I have a short video to show you, so that we can think about automation.
[09:34.330 --> 09:35.350]  That's from BuzzFeed.
[11:58.270 --> 12:15.910]  So, I guess, if you think about that video, all the things that you just saw, and all the machines that you just saw doing things, so making glass jars, or making car suspension springs, or making pretzels, those are all custom machines designed specifically for the purpose of making that particular thing.
[12:16.250 --> 12:26.870]  So, if you're the kind of person that makes millions of pretzels a day, then you have enough money to invest in automation to make a machine that's going to make millions of pretzels a day much faster for you.
[12:26.870 --> 12:40.670]  And, back in 1955, or 1954, arguably, MIT kind of stole the idea of an independent inventor and hooked up a milling machine to a computer.
[12:40.670 --> 12:57.750]  And the capability that the computer, therefore, has is you can change the code easily and use a general purpose machine to make different kinds of objects without necessarily needing to reconfigure your machine for different kinds of tooling.
[12:58.310 --> 13:12.930]  And that's pretty exciting, not just because you can make lots of different kinds of stuff, but because you have access to precision manufacturing, which you need to be able to make more precision manufacturing, for example, more precision manufacturing tools,
[13:13.050 --> 13:21.430]  for example, but also any kind of analytical instrumentation or different kinds of machine tools.
[13:21.490 --> 13:27.830]  You're going to have to need a quite high level of tolerance, which is going to take you a long time to figure out how to do by hand.
[13:27.930 --> 13:31.370]  And you might have to put in 10,000 hours to become a master crafts person.
[13:31.490 --> 13:34.670]  Or you can just use a machine that has it programmed into it.
[13:34.670 --> 13:46.170]  And so in case you guys hadn't already seen what a CNC mill would look like, I included this short video of a Moriseki 5-axis milling machine.
[13:46.450 --> 13:47.850]  I can kind of see you guys.
[13:48.010 --> 13:50.510]  How many of you guys have ever used a CNC machine?
[13:50.650 --> 13:50.830]  Anyone?
[13:51.150 --> 13:51.850]  A bunch of you.
[13:51.990 --> 13:52.410]  Okay, awesome.
[13:52.410 --> 13:57.570]  So this one is pretty sexy machine making a turbine blade.
[13:57.850 --> 14:06.410]  And it has XY in the top head, Z in the top head, and then A and B in the rotary aspects.
[14:07.490 --> 14:15.730]  So you see it's using all degrees of freedom that it has to be able to make this complicated part.
[14:17.230 --> 14:20.790]  And so computer control of machine tools is super awesome.
[14:20.950 --> 14:22.490]  We can make all these complicated things.
[14:22.890 --> 14:25.290]  However, it's not changed a whole lot.
[14:25.510 --> 14:29.450]  I was just at NASA Ames two months ago.
[14:29.650 --> 14:38.170]  And this is exactly the same milling machine as it is in that photograph from 1957 in popular science.
[14:38.670 --> 14:42.050]  And the only thing that they changed about it was they updated the computer at some point.
[14:42.050 --> 14:45.290]  But the equipment itself is still exactly the same.
[14:45.490 --> 14:52.850]  And I guess, you know, having fancy milling machines to mill large pieces of titanium at NASA isn't going to change a whole lot.
[14:53.050 --> 14:59.650]  But, you know, it's not necessarily something that people are going to need at a personal level anytime soon.
[14:59.890 --> 15:07.470]  And so without changing the digital fabrication tools, we're also not necessarily going to change what is going to be made with them.
[15:08.330 --> 15:12.790]  So, how many of you guys frequent hackerspace or makerspace?
[15:13.170 --> 15:14.050]  A bunch of you.
[15:14.250 --> 15:15.050]  Yeah, because they're fun.
[15:16.530 --> 15:22.190]  So, you can use digital fabrication tools for personal fabrication in places like Fab Labs.
[15:22.430 --> 15:28.330]  So, I guess our research group is heavily involved in promoting Fab Labs and helping them run.
[15:28.330 --> 15:32.630]  For example, here's one that's in Vesmanair in Iceland.
[15:32.690 --> 15:36.410]  It has a full electronics workbench so you can do hardware hacking.
[15:36.670 --> 15:39.610]  It has a large format milling machine that's only three axes.
[15:39.850 --> 15:41.910]  So, not as fancy as the Moriseki in that video.
[15:42.690 --> 15:46.570]  It has a laser cutter so you can cut plastics and woods.
[15:46.870 --> 15:48.290]  It has a small 3D printer.
[15:48.530 --> 15:51.150]  It has a small milling machine for electronics prototyping.
[15:51.150 --> 15:57.210]  And then it has computers for doing digital design and simulation.
[15:58.170 --> 16:06.890]  And this space is kind of nice because people also come in and are able to interact with other experts that are trying to use these machines.
[16:07.170 --> 16:19.190]  However, we found that using the machines was complicated enough that we had to rewrite some of the design and manufacturing internals so that novices could have access to that.
[16:19.690 --> 16:20.710]  More on that in a second.
[16:21.350 --> 16:30.290]  And so, you know, in Fab Labs, even though the machines are maybe not entirely designed for personal fabrication purposes, people are still making interesting things.
[16:30.630 --> 16:35.590]  So, Sien Stevek, he's working on wooden prosthetic feet prototypes.
[16:36.130 --> 16:40.290]  I don't know if any of you guys saw the Fab 5, a wireless mesh network.
[16:40.470 --> 16:45.170]  There was, in the Fab Lab in Jalalabad, Afghanistan, there was one satellite drop.
[16:45.170 --> 16:57.450]  And so, to be able to share that Internet across all of the NGOs that were spread through the city, they created these parabolic dishes on the three-axis CNC router.
[16:57.990 --> 17:13.290]  And then modified WRX, sorry, not WRX, WRT routers so that they would have a longer wait time before a packet drop so that they could send Internet out to other spaces throughout the city.
[17:13.290 --> 17:19.550]  And that was entirely kind of a, this is our problem, there's not a market that has this problem intervention.
[17:20.430 --> 17:24.930]  There's another student in my lab who wanted a kayak, and so he made a kayak.
[17:25.230 --> 17:32.550]  So, you have all these, you have access to all of this digital fabrication equipment, and you can refactor it to do different things.
[17:32.550 --> 17:41.410]  And obviously, something must be really interesting about it because, for example, in the Fab Lab in Donetsk, I don't know if Kostya is here somewhere, over there.
[17:41.890 --> 17:52.610]  So, Kostya's Fab Lab in Donetsk was occupied because, obviously, you can't have people have access to digital fabrication tools because they might just use them for descent.
[17:53.770 --> 17:56.250]  So, 3D printers may be kind of political after all.
[17:58.730 --> 18:01.130]  But, using all of these tools still kind of sucks.
[18:01.290 --> 18:05.550]  And so, that's kind of what I wanted, the crux of what I wanted to get at to talk about today.
[18:05.550 --> 18:11.070]  If you think about just the CNC milling machine that you saw in that video earlier.
[18:11.310 --> 18:19.170]  To be able to write the toolpath for that milling machine, you have to use horrible, horrible software packages that generally only run on Windows.
[18:20.050 --> 18:24.730]  The controls get exported in G-code, which is not a Turing complete language.
[18:26.950 --> 18:31.770]  I think that somebody's giving a talk on G-code today, if you want to learn more about G-code.
[18:31.910 --> 18:33.830]  But, as far as I'm concerned, we should get rid of G-code.
[18:33.830 --> 18:34.750]  G-code is stupid.
[18:35.650 --> 18:47.710]  And then you have like a really fancy interface where you take a USB stick and you put the G-code that just says go from this coordinate to this coordinate to a machine that has an amazing interface like this.
[18:48.850 --> 19:01.510]  And then you say... there's actually a button that says load tape when you put your USB stick in because nobody bothered to update it since... I don't know.
[19:02.450 --> 19:07.150]  I guess machinists are also recalcitrant trend types that don't necessarily want you to be changing all of their vocabulary.
[19:07.570 --> 19:09.410]  But it's not great.
[19:09.570 --> 19:13.770]  And then, you know, you bring your end mills and then you show up in the machine shop and you watch the machine.
[19:13.850 --> 19:15.050]  The machine gets to have all the fun.
[19:15.410 --> 19:23.690]  And generally, especially if you're not running the same job over and over again, if something is going to go wrong in your toolpath, you want to be there before the entire lab burns down.
[19:23.690 --> 19:27.650]  So you just have to kind of sit there and watch the machine do all of the work.
[19:28.230 --> 19:44.670]  And so, if we want to decompose machine tools into what their different parts are, I have this pyramid that I like to talk about where it's kind of the pyramid of machine control.
[19:45.850 --> 19:48.290]  Where you start out with... you have some tool head.
[19:48.410 --> 19:50.630]  So imagine we have a laser cutter, the tool head is a laser.
[19:50.830 --> 19:53.150]  Or if we have a milling machine, the tool head is a spindle.
[19:53.730 --> 19:55.830]  And so... and there's a tool head that we want to move around.
[19:55.990 --> 20:01.670]  And to be able to move it around in a constrained fashion, we have some mechanical system that we're placing the tool head on.
[20:01.910 --> 20:08.210]  So maybe it's a... maybe it's an XY frame, or maybe it's something that we suspend from the ceiling that we want to move up and down.
[20:08.210 --> 20:16.470]  And so once you have it on that mechanical constraint frame, you have to move it somehow and so maybe you can move it by hand.
[20:16.650 --> 20:21.410]  You don't need to have automated control or maybe you need sensors and actuators to move those things around.
[20:21.610 --> 20:31.190]  So the sensors and actuators like stepper motors or servos or maybe you're using hydraulics, those things have to be specified as well.
[20:31.190 --> 20:46.110]  And then once you have your tool head, which is on a mechanical system, which is moving around using some kind of actuator, then you're going to have to have some kind of control system that's going to tell the actuators how to move in response to the sensor input.
[20:46.370 --> 20:49.630]  And the control system could be a PLC.
[20:49.870 --> 20:52.950]  PLC is vulnerable for all kinds of exciting security attacks.
[20:53.570 --> 20:58.650]  It could be just a microcontroller or it could be a single purpose computer.
[20:59.810 --> 21:06.110]  And then to be able to talk to that control system, you have to have some way that you're going to come up with the commands that you're going to tell it.
[21:06.230 --> 21:14.470]  And so you have applications and interfaces like Mastercam or the Autodesk 123D suite.
[21:14.470 --> 21:23.270]  And there's kind of on the left, I have the logos of all of the kind of industry, typical industry standard like things.
[21:23.590 --> 21:25.150]  They're all thousands of dollars.
[21:25.490 --> 21:27.970]  Using them requires hundreds of hours of training.
[21:28.310 --> 21:30.330]  And they're all really quite annoying.
[21:30.570 --> 21:43.170]  And then on the right, there's a bunch of open-source alternatives that while they're not going to be used to run a car factory anytime soon, they're slowly improving.
[21:44.110 --> 21:50.390]  But the problem is or one of the problems is that along this pyramid, there's a lot of stratification between the layers.
[21:50.550 --> 21:55.710]  Each layer here is a pretty well established field and you're going to have people that know how to do it really well.
[21:55.910 --> 21:59.970]  And then you're going to have people that know how to really do it really well and maybe also want to do it in their free time.
[22:00.110 --> 22:01.970]  And maybe we'll develop an open-source alternative.
[22:02.230 --> 22:06.910]  But you don't necessarily have people that are like, oh, you know, I like working on mechanical systems.
[22:06.910 --> 22:11.650]  And I also want to make an application interface for it that's really easy to use.
[22:12.110 --> 22:18.910]  And so going from each layer to the next layer of this pyramid is kind of like a bad party trick of telephone.
[22:19.110 --> 22:22.950]  Where you're like, oh, I made this code and now I'm going to have to try to get it to talk to this machine.
[22:23.010 --> 22:24.570]  And my motors speak this weird protocol.
[22:24.890 --> 22:30.030]  And you end up with really stubborn people like me that like working in this space.
[22:30.030 --> 22:33.090]  But otherwise, everyone is kind of like, ah, everything is just kind of a hack.
[22:33.430 --> 22:41.970]  So even if they're incrementally improving, digital fabrication tools are still not that accessible for people who just want to show up and get something done.
[22:42.650 --> 22:44.850]  Because we have finicky bad interfaces.
[22:45.250 --> 22:55.250]  The applications of digital fabrication machines are really quite limited because they're grounded in this military industrial history of what we want CNC equipment for.
[22:55.250 --> 22:57.270]  Or we want to make really precise metal parts.
[22:57.470 --> 22:59.750]  And we want to make things that are approximately this big.
[22:59.990 --> 23:03.270]  And we want to have the most precision possible.
[23:03.470 --> 23:05.650]  Therefore, we're going to use really expensive materials.
[23:07.670 --> 23:15.830]  And most of the fabricators of digital fabrication equipment are kind of, you can kind of cluster them into only a few corporations.
[23:16.330 --> 23:21.930]  Like for example, in the 3D printer space, you have like a wide range of DIY kind of 3D printer people.
[23:21.930 --> 23:26.370]  And then everyone else is pretty much owned by Stratasys or 3D systems.
[23:26.590 --> 23:29.130]  Both of which I would say are pretty evil companies.
[23:29.750 --> 23:35.750]  So, you know, do we really want them to be controlling how we can interface with our fabrication?
[23:37.130 --> 23:38.750]  I kind of disagree with that.
[23:38.970 --> 23:46.930]  So I guess for me, it's then the question is, can we build all of these things ourselves or at least subvert some of the technology that already exists out there?
[23:46.930 --> 23:57.890]  And so, me and a friend of mine, Elan Moyer, spent some time teaching machine design and machine building to people who otherwise don't think that they should be doing machine design and machine building.
[23:58.070 --> 24:02.510]  So this is last week in Spain during the annual Fab Lab conference.
[24:02.770 --> 24:13.510]  We got anyone interested in trying to build a machine in three days to come up, learn about different kinds of motion systems, different kinds of mechanical systems, different kinds of control systems.
[24:14.390 --> 24:17.490]  And so they came up with a lot of pretty silly ideas for machines.
[24:17.670 --> 24:23.390]  Like this one is like a coffee machine called the good morning machine, where you can choose.
[24:23.430 --> 24:24.470]  You can have tea or coffee.
[24:24.610 --> 24:25.270]  It'll give you sugar.
[24:25.370 --> 24:26.070]  It'll give you milk.
[24:26.070 --> 24:26.950]  It'll write your name.
[24:26.950 --> 24:28.530]  And like, I love you on your coffee.
[24:29.870 --> 24:34.670]  And, you know, that's a very useful machine for that person, I'm sure.
[24:36.410 --> 24:45.130]  And so if prototyping something like this was easier than having to make coffee every morning, then I feel like a lot more people would do it.
[24:45.510 --> 24:48.610]  And so at the conference, we also used machines to make machines.
[24:48.750 --> 24:52.790]  So this is, on the right, you can see there's like a folding 3D printer.
[24:54.010 --> 25:00.370]  That's a prototype that was made by Flores, who's from the Fab Lab in Maastricht.
[25:00.670 --> 25:08.510]  And then on the left is a foam core prototype of a machine that is a small scale version of a bricklaying machine.
[25:08.710 --> 25:14.210]  So imagine you wanted to lay bricks not just in a straight line, but you would want to lay them throughout space however you wanted.
[25:14.210 --> 25:19.070]  So this is a system with four pulleys that can extend in different directions.
[25:19.910 --> 25:23.910]  So depending on where within the frame you are, you can lay bricks in different places.
[25:24.190 --> 25:32.510]  Which was a pretty ambitious machine to want to prototype in nine hours, but they got far enough to make a scale model.
[25:32.630 --> 25:35.630]  So here's some more of the machine prototypes that we were building.
[25:35.630 --> 25:45.050]  And I guess one thing that I like to stress about mechanical design is it seems that there's this iteration...
[25:45.050 --> 25:50.810]  If you want to write some software, you write a bash script and it takes you like two minutes or one minute.
[25:51.050 --> 25:54.450]  And if it doesn't work, you change something and then you can try it again.
[25:54.670 --> 25:57.350]  And then eventually your script is going to do what you want it to do.
[25:57.670 --> 26:01.270]  But if you do mechanical design, then you have to prototype the whole thing.
[26:01.390 --> 26:05.570]  And then if you made some big mistake, then you have to change a bunch of stuff.
[26:05.630 --> 26:09.890]  And changing that takes a lot longer than just changing code.
[26:10.190 --> 26:19.170]  And so I guess some people criticize our use of foam core and hot glue because real mechanical engineers obviously only use aluminum and metal.
[26:19.550 --> 26:25.450]  But it's a really easy way to kind of experiment with the different form factors that you might have in your machine.
[26:25.650 --> 26:26.590]  And so I like it a lot.
[26:26.730 --> 26:30.250]  You kind of didn't calculate the tolerances entirely correctly.
[26:30.250 --> 26:32.570]  Then you kind of just shove it over and add some glue.
[26:34.290 --> 26:42.390]  Here's a kind of logo turtle-ish machine that these guys built during the conference.
[26:42.590 --> 26:47.150]  And so these are all... this was all just three days worth of machine prototyping.
[26:47.250 --> 26:52.250]  I think we came up with six different machine prototypes each to do a very specific thing.
[26:52.290 --> 26:59.710]  And we did not only the mechanical aspect but also the controls and the software and how people might want to interface with the software.
[27:00.010 --> 27:05.510]  So not necessarily making the assumption that the way that you want to use a machine is by standing in front of a laptop and then using it.
[27:06.070 --> 27:15.050]  And so starting to build your own machine solutions is something that I think felt pretty strongly about and hopefully convinced other people during that workshop to do.
[27:15.610 --> 27:19.710]  And so in terms of some machines that I've built...
[27:19.710 --> 27:26.790]  So Alain and I who taught the workshop, we built this machine at some point which is a suitcase multi-purpose CNC machine.
[27:27.030 --> 27:28.150]  So it has a milling head.
[27:28.310 --> 27:30.010]  In this photograph it's milling a circuit board.
[27:30.170 --> 27:31.790]  It also has a 3D print head.
[27:31.970 --> 27:33.410]  It also has a liquid handling head.
[27:33.510 --> 27:37.110]  So if you wanted to do bio experiments, you could do automated bio experiments.
[27:37.110 --> 27:40.030]  And it has like a little scoring and cutting head.
[27:41.330 --> 27:44.230]  And it folds down into this briefcase so you can take it with you.
[27:44.750 --> 27:48.190]  I've taken it many times as a personal item on a plane.
[27:48.470 --> 27:50.210]  And the TSA never asks me about it.
[27:51.150 --> 27:54.410]  They sometimes ask me about like...
[27:54.410 --> 27:55.750]  I have a space mouse.
[27:55.970 --> 27:56.770]  They ask me about that.
[27:56.930 --> 28:05.110]  But apparently if you make your stuff look like it's very serious, then the TSA is not going to get you in trouble.
[28:05.110 --> 28:06.230]  And so you...
[28:06.230 --> 28:08.190]  Try to put the little pocket knife in there.
[28:08.190 --> 28:09.610]  And put a pocket knife in there.
[28:09.750 --> 28:10.090]  Yeah.
[28:10.390 --> 28:11.970]  Then they'll suddenly recognize it as a...
[28:13.090 --> 28:14.890]  They'll suddenly recognize it as a weapon.
[28:15.170 --> 28:21.470]  So these are all different machines that I've worked on kind of experimenting with this different form factor.
[28:21.590 --> 28:39.790]  Because it's not just that I'm thinking about machine building as something that you want to have for more custom machines, but also how do you make machines more accessible by making them cheaper and more portable and more likely to be encountered in a space like...
[28:39.790 --> 28:46.830]  In a space like a home or a small office rather than a space that has a workshop like MIT.
[28:46.830 --> 28:53.050]  So in these photos there's some small format milling machines for circuit board prototyping.
[28:53.430 --> 28:58.490]  There's like a machine that we made just to draw silly portraits.
[28:59.110 --> 29:03.190]  There's another liquid handling machine for bio purposes.
[29:03.190 --> 29:12.930]  And so we've been thinking more about how do you decompose machine building into different modules.
[29:13.270 --> 29:19.090]  And so right now I'm working a lot with different reconfigurable modules that you can easily use.
[29:19.230 --> 29:23.710]  So say you need a machine that needs to move in one axis and you're moving it back and forth.
[29:23.910 --> 29:28.310]  Or you need something that moves back and forth and then turns something on and then turns something off.
[29:28.310 --> 29:33.970]  We're making all of these kind of modules that snap together so you could easily prototype different kinds of machines.
[29:34.310 --> 29:39.770]  And again, not just the mechanical prototyping is important, but also the control system.
[29:40.090 --> 29:56.290]  And so instead of using one single microcontroller that interprets G-code and sends commands to motors, we're using a networked approach so that every single node on the network addresses a different part of the machine.
[29:56.290 --> 29:58.710]  And you have a virtual machine on your computer.
[29:58.950 --> 30:01.310]  So say it's a milling machine with three axes.
[30:01.690 --> 30:06.370]  The virtual machine says there's a synchronized move between the X, Y, and Z axes.
[30:06.810 --> 30:08.650]  There's the spindle off and the spindle on.
[30:08.810 --> 30:11.790]  And each of those are different nodes on the network that you can call functions on.
[30:11.930 --> 30:19.090]  As opposed to you just streaming a G-code file out of a port and hoping that the G-code interpreter is interpreting it correctly.
[30:19.090 --> 30:28.090]  And if you want to stop and reiterate or change something, you don't have to try to backtrack through your G-code buffer.
[30:28.270 --> 30:29.870]  But you can change things on the fly.
[30:30.370 --> 30:37.270]  And that's maybe not that exciting in the example of a milling machine because milling machines do what they do pretty well.
[30:37.370 --> 30:39.290]  G-code is pretty well developed for milling machines.
[30:39.290 --> 30:45.090]  But if you, for example, wanted to do something else with your milling machine.
[30:45.110 --> 30:51.790]  So say you wanted to mill until you got to the red layer of a material and then switch.
[30:52.010 --> 30:56.230]  You could easily put a camera that would recognize between, say, the red layer and the green layer.
[30:57.230 --> 31:00.130]  But you couldn't necessarily have that interface with your G-code.
[31:00.310 --> 31:10.130]  You would have to write a very complicated garb rule of control systems and software to be able to add that simple functionality to your machine.
[31:10.330 --> 31:12.630]  And so we're trying to get away from that.
[31:13.570 --> 31:17.230]  So here's a, with the modules, just to show you guys, here's a...
[31:23.230 --> 31:34.490]  So this one we made in 22 hours, including the, uh, including the water jetting of all of the, of the, of the side parts.
[31:34.790 --> 31:37.130]  Um, including the soldering of the control system.
[31:37.470 --> 31:39.630]  Including writing the software interface.
[31:40.510 --> 31:43.950]  Um, oh yeah, and it's a, it's a four axis hot wire cutter.
[32:14.590 --> 32:15.550]  Serious researcher.
[32:15.970 --> 32:16.070]  Yeah.
[32:28.190 --> 32:33.370]  Um, and so I guess I should also stress all of the machines that I'm showing you here.
[32:33.370 --> 32:39.590]  That I showed you up until now are, um, that we're developing kind of at MIT are, are all open-source.
[32:39.970 --> 32:45.350]  And all of the, all of the mechanical design files and the control system, everything is available online.
[32:45.610 --> 32:49.130]  Um, if you go to the machines that make project website, um, at MIT.
[32:49.570 --> 32:57.410]  Uh, and so I, I think that we are getting to a point where personal fabrication tools are becoming a thing that companies are thinking about too.
[32:58.170 --> 33:13.630]  Um, instead of having, instead of having this model of the shop where you make things, um, having, uh, having fabrication equipment be a, uh, something that you imagine an individual having more like a power tool and less like a machining center.
[33:13.630 --> 33:22.650]  Um, so this is, uh, Alon left the PhD program and started a company making this handheld digital fabrication tool router.
[33:22.850 --> 33:30.170]  So it has this tape on, this tape that you buy that you put down on the material that will recognize how far apart anything is.
[33:30.170 --> 33:33.930]  And it has this webcam on it with, uh, with a small touch screen interface.
[33:34.250 --> 33:39.250]  So that you can say, oh, I want to be cutting out, I want to be cutting out, uh, a square here.
[33:39.490 --> 33:40.930]  Or I want to cut out this complicated shape.
[33:41.050 --> 33:44.370]  And I don't know if any of you guys have ever used a handheld router before.
[33:44.710 --> 33:49.490]  But for example, cutting a straight line with a handheld router without a jig is pretty hard.
[33:49.770 --> 33:55.690]  Um, and because this has, you have gross global positioning with your hands.
[33:55.810 --> 33:57.930]  So you're the one dragging the entire contraption around.
[33:57.930 --> 34:04.030]  But then inside there, those two motors form a linkage that can do precision control with, of the router within it.
[34:04.130 --> 34:09.890]  So as long as you stay within a certain boundary area, it'll, it'll do the rest of the control for you.
[34:10.270 --> 34:18.230]  Um, and, uh, and a really powerful thing about this machine then is that this machine can make things that are much, much bigger than it is.
[34:18.490 --> 34:23.090]  Whereas with other, in other cases, you go up to a machine and you make something that's smaller than that machine.
[34:23.090 --> 34:30.870]  But this machine, um, applies the control through just these, like, custom inkjet printed masking tape parts.
[34:31.150 --> 34:36.810]  Um, and so I think that, that's a, a really exciting direction for digital fabrication tools to go in.
[34:36.930 --> 34:41.210]  Also, uh, I don't know if any of you guys ever used ShopBot routers.
[34:41.210 --> 34:41.790]  Anyone?
[34:42.250 --> 34:42.330]  Maybe.
[34:43.010 --> 34:43.090]  Anyone?
[34:43.090 --> 34:43.090]  Maybe.
[34:43.090 --> 34:48.430]  And, uh, so ShopBot is a, is a, is a woodworking router company that makes CNC table routers.
[34:48.690 --> 34:53.270]  And, uh, and then at some point they were like, oh, we also actually want to do five axis milling of things.
[34:53.390 --> 34:57.890]  Because we need more, we need more figurines and stuff in our life.
[34:57.890 --> 35:05.570]  And, uh, uh, and so they actually prototyped this entire machine, the HandyBot, on a ShopBot as well.
[35:05.770 --> 35:22.530]  So they started using, um, they started using their tools to make more tools, uh, so that they could, uh, so that they could prototype different, uh, different kinds of, uh, different kinds of, uh, different kinds of, different kinds of equipment that people might use in different contexts that aren't necessarily table routing.
[35:22.530 --> 35:32.450]  So this HandyBot, I actually used it once where I was at a dump, like a local garbage collection and sorting area, and they wanted more signage.
[35:32.570 --> 35:37.590]  And so you can hold this up against the side of a wall and then cut signs into the wall.
[35:39.010 --> 35:50.990]  It's kind of a more permanence of graffiti or something kind of tool, which I think is an exciting space of digital fabrication to explore.
[35:51.750 --> 36:01.370]  So without the capability to kind of iterate and design new digital fabrication tools, I really feel like the ecosystem can't grow.
[36:01.670 --> 36:08.330]  You don't want to just end up having people print things on the same 3D printers because you're going to end up with a lot of stuff that looks the same.
[36:08.510 --> 36:18.650]  It's kind of the same way as if kids these days just have iPads and they are just downloading apps on iPads and you can't write an iPad application on an iPad.
[36:18.650 --> 36:22.710]  I guess if you deal break it, you can compile an iPad application on an iPad.
[36:22.910 --> 36:27.990]  But there's no way that you can interface with the technology that you're using.
[36:28.410 --> 36:31.150]  And so I think that's a really important thing to think about.
[36:31.330 --> 36:42.650]  And so I realize that not everyone hangs out at MIT in different kinds of fab lab-ish spaces with access to lots of tools where they can prototype.
[36:42.650 --> 36:49.130]  And so if you don't have access to that, what are other ways that you can build is something I've been thinking about.
[36:49.770 --> 36:55.370]  So what are other forms of access that you might have to precision manufacturing?
[36:56.910 --> 37:03.950]  So to answer this question, I went to Shenzhen, which is just outside of Hong Kong.
[37:04.250 --> 37:07.670]  And it's kind of a hotbed of... it's right next to where Foxconn is.
[37:07.770 --> 37:09.850]  It's where people do a lot of electronics prototyping.
[37:10.090 --> 37:15.270]  And it's where people do a lot of... there's a lot of small factories that you can get to do custom jobs for you.
[37:15.550 --> 37:17.250]  It's a really strange place.
[37:17.270 --> 37:18.350]  I thought it was quite odd.
[37:18.950 --> 37:22.270]  As a city, I think it's only maybe two years older than me.
[37:22.970 --> 37:27.630]  And it has a... it has a... I live in Boston.
[37:27.830 --> 37:33.350]  And Boston takes pride in having the oldest subway system in the world.
[37:33.570 --> 37:34.790]  Well, it's not something I should be proud of.
[37:34.930 --> 37:36.050]  It's a really bad subway system.
[37:36.170 --> 37:37.050]  It's way worse than New York.
[37:37.170 --> 37:38.030]  New York's pretty bad too.
[37:38.230 --> 37:48.650]  But in Shenzhen, they have a beautiful brand new subway system that they built in the last eight years that is much more extensive than the Boston Greater Public Transport System.
[37:48.770 --> 37:53.410]  So it's a strange place where people build things really quickly, seemingly out of thin air.
[37:53.610 --> 38:01.050]  And you have these strange tool and dye shops where you come up and someone is eating some kind of pastry.
[38:01.070 --> 38:06.250]  And then they ask you, do you want this kind of potentiometer?
[38:06.830 --> 38:08.190]  Oh, we don't have that one.
[38:08.190 --> 38:11.950]  But you can come back in four minutes and we'll have made it for you.
[38:13.950 --> 38:16.110]  And there are these cell phone markets.
[38:16.390 --> 38:24.690]  So since a lot of cell phones get made there, there's this kind of liminal area around the cell phone markets that involves cell phone repair.
[38:24.850 --> 38:29.910]  And you can buy all these digitizers if you want to buy...
[38:30.490 --> 38:34.410]  I don't really like iPads, so I'm sorry for harping on them.
[38:34.490 --> 38:36.590]  But if you just want... the display itself is really nice, right?
[38:36.610 --> 38:38.570]  It has a really high pixel count.
[38:38.570 --> 38:48.630]  And if you just want to buy the display and have your own little plug that you interface with it, you can buy one at this kind of... not sure if stolen or found or...
[38:48.630 --> 38:53.110]  For $60, you can get really nice screens and other parts.
[38:53.370 --> 38:58.390]  So here are all unsoldered ICs that would go in the iPhone 5.
[38:59.970 --> 39:02.530]  And here are a bunch of iPhone motherboards.
[39:02.810 --> 39:09.190]  And so if you wanted, for I calculated approximately $12, you could make your own iPhone.
[39:10.910 --> 39:15.130]  It's not necessarily an easy thing to do though, right?
[39:15.250 --> 39:21.450]  I mean, how many... I would say that I'm very skilled at surface mount soldering.
[39:21.450 --> 39:26.410]  And the amount of time it would take me to... I think I would rather kill myself than solder an iPhone.
[39:27.910 --> 39:35.910]  And so access to precision tools doesn't necessarily mean anything unless you have the capability to use precision tools as well.
[39:36.150 --> 39:41.630]  And so, again, with like the Mori Seiki that we looked at at the beginning with the 5-axis control...
[39:41.630 --> 39:48.770]  Since programming that is such a pain, even if you say, oh yeah, you're an artist and you want to use this tool, go ahead.
[39:48.770 --> 39:59.570]  And you can't really expect people to invest 200 hours of software training into just being able to use the machine and then still think that creative things are going to come out.
[39:59.810 --> 40:07.150]  But in China, there are a lot of people that are learning about how to use precision fabrication tools.
[40:07.350 --> 40:18.190]  So this is Shenzhen University, which I think is... or my understanding was that it was kind of more like a vocational school where people were learning how to use milling machines.
[40:18.190 --> 40:22.870]  So you can see on the desks next to these students, there are these small milling machines.
[40:23.110 --> 40:30.270]  And every student is learning about G-code, control, how to change their parts correctly, how to keep things clean.
[40:30.430 --> 40:34.730]  And then once they graduate from this class, they go to the next milling machine class.
[40:34.810 --> 40:39.010]  And they start using these larger... they start using these larger milling machines.
[40:39.010 --> 40:44.710]  So there's a vast expertise there of how to make things.
[40:44.930 --> 40:51.510]  And I think there's a growing desire for understanding why you want to be making different things.
[40:51.870 --> 41:06.230]  So there's like an odd remixing that I observed that I wanted to talk to you guys about of manufacturing things and things that you might desire in this space where people have learned how to manufacture things and have been doing them...
[41:06.230 --> 41:07.610]  And I've been doing that in different ways.
[41:07.810 --> 41:09.130]  So here's a phone I bought.
[41:10.350 --> 41:12.250]  The form factor might look kind of familiar.
[41:12.570 --> 41:16.250]  I paid about 17 U.S. dollars for it.
[41:17.130 --> 41:19.850]  And here it is next to a real iPhone.
[41:21.070 --> 41:25.350]  It's like a skinned Android OS that it's running.
[41:25.750 --> 41:32.990]  It has just kind of Chinese specific apps on it, like WeChat and QQ.
[41:34.030 --> 41:36.830]  Unlike the iPhone, you can change its battery.
[41:38.050 --> 41:40.670]  You can... it implements USB host.
[41:41.290 --> 41:45.150]  And I bet if you tried hard enough, you could get its processor opcodes.
[41:45.790 --> 41:57.890]  And so for some reason, someone in Shenzhen decided it would be a good idea to design this kind of mini cheap version of something that was being designed in California.
[41:58.310 --> 42:08.190]  I also x-rayed it to learn more about its origins, which I found that in China they really know a lot about manufacturing things really cheaply.
[42:08.190 --> 42:09.590]  There's no connectors.
[42:10.010 --> 42:11.150]  Connectors are expensive, I guess.
[42:11.310 --> 42:12.310]  But there's no connectors in there.
[42:12.390 --> 42:13.970]  They're just all sandwiched together.
[42:14.410 --> 42:15.710]  Motherboards that have been soldered together.
[42:16.910 --> 42:18.750]  And so I think that...
[42:18.750 --> 42:21.970]  Yeah, this object is so...
[42:21.970 --> 42:22.550]  Whoops, sorry.
[42:23.050 --> 42:25.770]  This object is so hilarious to me.
[42:26.030 --> 42:28.190]  Because it kind of represents...
[42:28.190 --> 42:30.410]  We have the capability to make whatever we want.
[42:30.530 --> 42:34.450]  We have the engineering hours to make whatever we want.
[42:34.450 --> 42:37.650]  And so what we decided that we wanted was like a smaller, cheaper version of the iPhone.
[42:38.350 --> 42:45.150]  So if we could harness that expertise and put it into other things, you know, that could possibly be pretty cool.
[42:46.110 --> 42:46.610]  Here's another...
[42:46.610 --> 42:49.030]  Here's another cell phone we found.
[42:49.410 --> 42:51.250]  This is the...
[42:51.250 --> 42:51.330]  The...
[42:51.330 --> 42:52.330]  The Prash...
[42:53.170 --> 42:53.530]  Prash...
[42:56.710 --> 42:59.670]  I'm pretty sure it's kind of more like the Ferrari logo.
[43:01.710 --> 43:02.430]  But, yeah.
[43:02.930 --> 43:04.490]  It works great as a cell phone.
[43:04.650 --> 43:07.130]  And it came with like a gold chain so you can wear it around your neck.
[43:10.830 --> 43:11.550]  And so...
[43:11.550 --> 43:17.010]  I guess the takeaway that I had from that was like that factories are really changing and personal fabrication is growing.
[43:17.010 --> 43:22.810]  Even at the level of workers in factories who think that they understand more about the market share that they might be able to interface with.
[43:23.110 --> 43:30.770]  You can say, hey, you know, I think my friends might want to buy a gold Prashi cell phone, and so we're going to make it.
[43:30.990 --> 43:35.510]  And I think that that kind of quick turnaround of prototyping is pretty exciting.
[43:35.790 --> 43:40.370]  But it's still not necessarily something that everyone can come along with.
[43:40.890 --> 43:46.170]  I don't know if... did you guys see the open-source laptop that Bunny and Zobs designed?
[43:46.170 --> 43:53.770]  So it's a full... it's like a laptop that's based on an ARM core that they kick-started.
[43:53.850 --> 44:01.450]  This is an x-ray that I did of it so that you could check to see if there were in any of the closed chips if there was any malware.
[44:01.970 --> 44:07.850]  I'm not really that worried about it, but we know for sure now that there isn't.
[44:07.850 --> 44:13.710]  And they kick-started this... and they kick-started this... or crowd supply, sorry.
[44:14.130 --> 44:20.670]  Sorry, this laptop so that they would be able to have a hacker's laptop that was kind of like the thing that they would use.
[44:21.310 --> 44:27.030]  So the specifications for it include things that you would expect in a laptop, like there's a certain amount of memory.
[44:27.030 --> 44:28.650]  You can use it.
[44:28.850 --> 44:29.690]  It boots Debian.
[44:31.110 --> 44:33.110]  But then there are these stretch goals that they added.
[44:33.290 --> 44:35.950]  So for example, it also has an add-on for software-defined radio.
[44:36.230 --> 44:39.270]  So want your own GSM base station?
[44:39.710 --> 44:40.530]  No problem.
[44:42.690 --> 44:48.670]  And I think that the cost of the laptop is not necessarily representative for its capabilities.
[44:49.030 --> 44:55.830]  I think the base laptop was maybe $2,000, which is more than people generally spend on computational devices right now.
[44:55.930 --> 45:06.870]  But I think if you look at the engineering that went into it and the volumes that they're constructing it, I think that the way that I would see this project much more is this is an exploration...
[45:06.870 --> 45:08.490]  It's almost like journalism.
[45:08.750 --> 45:10.730]  This is an exploration where can...
[45:10.730 --> 45:20.110]  Is it possible for people to make laptops that are completely open, that don't necessarily have a mass market, that are designed to be fully hackable?
[45:20.970 --> 45:22.210]  And how do you do that?
[45:22.430 --> 45:35.850]  So Bunny and Zobs have been working with manufacturers in China, the same places around Huaking Bay Market, where they have the kind of cheap cell phone parts that you can buy.
[45:36.470 --> 45:39.550]  to figure out if you can do this kind of small-scale manufacturing.
[45:39.850 --> 45:47.550]  And so can you circumvent large electronics producers to make things that maybe you're more personally interested in?
[45:47.590 --> 45:49.250]  I think it's a pretty cool project.
[45:49.610 --> 45:51.570]  So with this new kind of...
[45:52.310 --> 46:06.510]  Yeah, you have this kind of new possibility of making stuff, where you have access as an individual to material, you have access as an individual to engineering tools, and you have access as an individual to manufacturing equipment.
[46:07.590 --> 46:12.270]  I think that sourcing is one of the big problems that you encounter.
[46:12.710 --> 46:17.490]  And so I guess that's one of the reasons that I was pretty interested in going to China and figuring out where you can...
[46:17.490 --> 46:22.690]  Where can you go somewhere, find a street market, and buy a chip?
[46:22.690 --> 46:24.850]  Not Radio Shack anymore.
[46:25.250 --> 46:26.590]  Or maybe Radio Shack a little bit.
[46:26.830 --> 46:31.750]  But mostly, you go to Radio Shack and you're going to buy an 8-bit microcontroller that is going to be able to run.
[46:31.970 --> 46:32.990]  Not a whole...
[46:32.990 --> 46:33.930]  It's not going to be some...
[46:33.930 --> 46:38.810]  You're not going to replace a device that you're using on a daily basis with an 8-bit microcontroller.
[46:40.230 --> 46:41.930]  And so here's a...
[46:41.930 --> 46:44.390]  I don't know if you guys have heard of this website, Taobao.
[46:45.630 --> 46:51.370]  But it's kind of a portal to start interfacing with small factories in China.
[46:51.510 --> 46:58.750]  You can order things just off of Taobao or you can start communicating with sellers on Taobao who then help you manufacture things.
[46:58.990 --> 47:03.730]  So for example, for that workshop that we taught, one of the things that I wanted...
[47:03.730 --> 47:13.990]  On the right you can see, or maybe you can almost not see, but on the right I wanted a custom lead screw motor assembly with a specific kind of lead screw nut so that I would be able to move things around in a certain way.
[47:14.170 --> 47:19.170]  And so I made a drawing specification and sent it to a manufacturer in China.
[47:19.890 --> 47:35.350]  And then was able for the cost of $17 to get these precision integrated lead screw motors with wear compensating lead screw nuts from China, which I was very, very excited about.
[47:35.350 --> 47:43.670]  Because the equivalent cost if you would buy the components off of the shelf would be around $150.
[47:44.150 --> 47:53.830]  So making more machines that are more suitable for whatever the purpose is that you want to make machines for.
[47:54.230 --> 47:56.610]  So that's pretty much all I wanted to talk to you guys about.
[47:57.210 --> 48:01.370]  Thanks a lot for coming to my talk, even though Thomas Drake is talking right over there, guys.
[48:02.170 --> 48:05.230]  And if you have any questions, I actually have some time.
[48:05.450 --> 48:06.730]  So you can ask them.
[48:07.150 --> 48:07.350]  Yes?
[48:07.570 --> 48:09.590]  What happened with the DARPA red balloon?
[48:10.510 --> 48:11.810]  Oh yeah, that was horrible.
[48:12.030 --> 48:13.070]  The DARPA red balloon challenge.
[48:13.250 --> 48:20.010]  So we took a lot of photos of the balloons and we like changed their EXIF data and we posted them on different places.
[48:20.370 --> 48:22.290]  And I definitely think we threw off...
[48:22.290 --> 48:24.130]  We threw off the trail...
[48:24.130 --> 48:25.730]  We threw some people off the trail of finding it.
[48:25.850 --> 48:32.570]  But 10 hours later, there's another group at MIT who are lovely people.
[48:32.570 --> 48:35.230]  And they created this kind of...
[48:35.230 --> 48:42.330]  They kind of created this social platform where you could invite all of your friends and they could invite their friends and they could invite their friends.
[48:42.430 --> 48:51.390]  And if one of your friends found a balloon, they would get $1,000 and you would get $500 for referring them and then that you would get $250 for referring them, et cetera, et cetera.
[48:51.510 --> 48:58.750]  So they created this massive unregulated experiment where they collected huge amounts of information about people's personal networks.
[48:59.050 --> 49:00.750]  And people all just kind of did it.
[49:00.750 --> 49:05.130]  I was really pissed about that because I was like, why are you just giving up all the people that you know?
[49:05.350 --> 49:10.450]  That's not something that you just want to give to a researcher without knowing what their research agenda is.
[49:10.590 --> 49:15.690]  But it took them, using that model, it took them nine hours to win the balloon challenge.
[49:15.890 --> 49:21.610]  And so they won the $40,000, which then they had to pay out into this massive, unregulated human experiment.
[49:22.050 --> 49:28.090]  And then they were invited to speak at the Tonight Show or something.
[49:28.250 --> 49:32.450]  And they came down to the lab and they said, oh, you know, have you heard of us?
[49:32.590 --> 49:35.250]  We won the Red Balloon Challenge.
[49:35.390 --> 49:37.290]  And now we're going to be on the Today Show or the Tonight Show.
[49:37.390 --> 49:37.950]  I forget exactly.
[49:38.090 --> 49:38.490]  And they're like, good.
[49:38.610 --> 49:40.670]  And we want to make sure that everybody really knows that we're scientists.
[49:40.830 --> 49:43.550]  So we're wondering if you could lend us a lab coat.
[49:49.370 --> 49:54.970]  And then I said, well, I really take issue with the way that you've approached this experiment.
[49:55.030 --> 49:58.070]  And I would like you to know that I'm extremely cross with you.
[49:58.210 --> 50:00.550]  And they said, so will you lend us a lab coat?
[50:02.350 --> 50:04.050]  And so then I lent them a lab coat.
[50:04.310 --> 50:07.330]  And I wrote a really snarky blog post about it.
[50:07.490 --> 50:20.270]  And I think I'm now featured in their presentation about this wonderful, like, social connection mining experiment as like the technological Luddite and the person that is opposing change.
[50:20.470 --> 50:23.590]  Because obviously, they're on the path of smartness.
[50:23.590 --> 50:26.630]  And if you're not part of smart connected devices, you must be dumb.
[50:30.670 --> 50:35.950]  Does anyone else have any other questions?
[50:36.310 --> 50:39.970]  I can tell you all about how to make network control systems.
[50:39.970 --> 50:42.430]  What are you using instead of G-code?
[50:43.050 --> 50:43.610]  Okay.
[50:44.650 --> 50:49.730]  Instead of G-code... So there's... Let me go back to this pyramid.
[50:52.390 --> 50:57.430]  We're using something called Gestalt, which we wrote ourselves.
[50:59.470 --> 51:02.550]  And... Got to go back really far.
[51:02.890 --> 51:18.930]  So the nodes individually are running embedded C to interpret... So the nodes... The nodes in the... The nodes that control one sensor or one actuator each are running embedded C.
[51:19.190 --> 51:35.230]  And that embedded C listens to a bus network protocol that we're calling FabBus, which runs both over RS-485 and over other communication... Depending on what communication networks your machine might already have access to, there's different nodes that's written for them.
[51:35.670 --> 51:52.990]  So you... So you can write an application... Say... Say you have, like, a browser... We have a... We have a... We have one browser-based application that will listen for things from the browser and then send them as packets over this FabBus.
[51:53.430 --> 51:56.590]  And then each individual node will interpret those commands.
[51:56.770 --> 51:58.290]  And you can write those commands yourself.
[51:58.570 --> 52:03.670]  So instead of saying, oh, G92, 000, machine goes home.
[52:03.970 --> 52:08.090]  You can just write... You can write a... You can write a function that says home.
[52:08.630 --> 52:12.930]  And then each node on the network knows where its machine home is.
[52:13.050 --> 52:14.690]  And we'll just move back to that location.
[52:16.030 --> 52:24.110]  And if you... But one thing that you don't necessarily... Or that we didn't necessarily want to do was tell everyone that was using G-Code that they couldn't use G-Code anymore.
[52:24.290 --> 52:27.330]  So if you want, you can stream G-Code over the FabBus as well.
[52:27.790 --> 52:35.590]  And it'll still... As long as it's specified what each distance is in G-Code, it'll still move to the right places.
[52:36.270 --> 52:38.870]  But yeah, the system that we're using is called Gestalt.
[52:38.990 --> 52:41.030]  And we have it up on GitHub if you want to take a look at it.
[52:43.010 --> 52:43.350]  Yes.
[52:44.690 --> 52:46.450]  It's... It's wonderful what you do.
[52:46.690 --> 52:47.550]  And we hope it all works.
[52:47.930 --> 52:49.310]  But it's still too complicated.
[52:50.890 --> 52:59.390]  I don't know enough about how to use or even conceive of these personal machines that they can do.
[52:59.650 --> 52:59.870]  Right.
[53:00.010 --> 53:01.670]  So how am I going to get educated about that?
[53:01.750 --> 53:04.770]  How can I bring that to the children in high school?
[53:06.250 --> 53:10.830]  Yeah, I think that's a... I think it's a really good point.
[53:11.030 --> 53:13.370]  And I fully agree with you.
[53:13.670 --> 53:19.210]  One of the things that I I guess I find kind of frustrating about spaces like MIT is that people always act like...
[53:19.210 --> 53:22.410]  People always act like the thing that you're learning is the technology.
[53:22.670 --> 53:23.490]  Or the thing that you're...
[53:23.490 --> 53:27.110]  If you go to a makerspace, the thing that you're there for is the 3D printer.
[53:27.390 --> 53:28.810]  The 3D printer is really quite useless.
[53:28.810 --> 53:32.930]  Unless there's someone there that can help you learn about the 3D printer.
[53:33.130 --> 53:35.750]  Maybe you want to learn about it through the Internet.
[53:35.990 --> 53:37.890]  Or maybe you want to learn about it from a book.
[53:38.110 --> 53:42.170]  But at the end, like there's some... You have to find some way that you can interface with someone else to learn about it.
[53:42.570 --> 53:45.750]  So I guess there's two things that I'm trying to do to address that problem.
[53:45.830 --> 53:47.630]  Number one is you can talk to me, for example.
[53:48.510 --> 53:50.110]  And hopefully people like me.
[53:50.310 --> 53:56.370]  So like seek out other people in your community that are excited about digital fabrication.
[53:56.370 --> 54:05.970]  And then I guess one of the things that we're trying to do with the fab bus and these control system networks is to make it much easier to make machines.
[54:07.210 --> 54:13.270]  And so not just by teaching people how to build machines, but also by changing the tools with which they can build machines.
[54:13.270 --> 54:28.090]  So for example, in September, we're going to have some architecture students at University of Kansas play around with building machine systems from the networked machine controllers.
[54:28.710 --> 54:33.990]  But I realize that's maybe not a very satisfactory answer to your question personally.
[54:37.530 --> 54:39.430]  Yeah, that's a great idea.
[54:40.990 --> 54:42.350]  It's a...
[54:42.350 --> 54:43.350]  What is it?
[54:44.290 --> 54:45.750]  Is this stuff available?
[54:46.170 --> 54:46.950]  Which stuff?
[54:47.290 --> 54:48.410]  The network node?
[54:49.730 --> 54:52.110]  So the boards...
[54:53.830 --> 54:57.970]  The whole machine or just the control system nodes?
[54:57.970 --> 55:06.590]  So all of the design files are online, but you can't buy the boards pre-assembled, but the design files for that are online.
[55:06.670 --> 55:10.770]  So you could buy the board and then solder it yourself, which I realize is not a very satisfactory answer either.
[55:13.550 --> 55:13.950]  And...
[55:16.390 --> 55:17.190]  Yeah, the...
[55:17.190 --> 55:18.510]  I guess the...
[55:19.250 --> 55:21.090]  Maybe if you could wait for like a year.
[55:21.510 --> 55:24.190]  Maybe one day I'll graduate and I'm going to work much harder on it.
[55:27.350 --> 55:27.750]  Yeah.
[55:28.290 --> 55:28.490]  Okay.
[55:28.690 --> 55:29.050]  Yeah, I will.
[55:30.090 --> 55:30.490]  Yes.
[55:41.300 --> 55:42.700]  Yes.
[55:44.260 --> 55:46.640]  Is it a good thing?
[55:49.640 --> 55:51.040]  Uh...
[55:53.720 --> 55:54.420]  Right.
[55:58.060 --> 55:59.100]  So why...
[55:59.100 --> 56:01.220]  Why do you think that...
[56:03.540 --> 56:07.220]  Well, I guess if you're worried about...
[56:07.220 --> 56:11.400]  If you're worried about security of your machine and...
[56:11.400 --> 56:16.100]  Then it's nice that G-code isn't a Turing complete language because you can't execute code on it.
[56:16.100 --> 56:17.920]  But I think that if...
[56:18.700 --> 56:19.220]  Uh...
[56:19.220 --> 56:19.580]  I have...
[56:19.580 --> 56:20.400]  I have like this...
[56:20.400 --> 56:26.140]  I have this photograph I took of a friend of mine who's using a seven axis robot arm.
[56:26.500 --> 56:26.940]  Um...
[56:26.940 --> 56:27.620]  And, uh...
[56:27.620 --> 56:29.880]  He's using it to cut Styrofoam surfaces.
[56:30.580 --> 56:31.100]  And...
[56:31.100 --> 56:33.720]  Styrofoam surfaces have very complicated topology.
[56:33.960 --> 56:34.440]  It's like...
[56:34.440 --> 56:39.880]  It's like a full non-zero Gaussian curvature that he's trying to approximate with swoops back and forth.
[56:40.140 --> 56:41.000]  And, uh...
[56:41.000 --> 56:44.080]  The swoop motion that goes back and forth is fine.
[56:44.080 --> 56:47.060]  He can program that in G-code and have, um...
[56:47.060 --> 56:50.720]  The Kuga robot arm interpret the coordinates that it's sending to it.
[56:51.100 --> 56:51.540]  But...
[56:51.540 --> 56:53.440]  He's standing next to the robot.
[56:53.620 --> 56:55.380]  And this is what I think is striking about the photograph.
[56:55.720 --> 56:57.060]  Is he standing next to the robot.
[56:57.240 --> 57:01.500]  And he's controlling the temperature of the hot arm that's cutting the Styrofoam.
[57:01.800 --> 57:02.540]  By hooking it...
[57:02.540 --> 57:04.040]  He hooked it up to a surge protector.
[57:04.300 --> 57:08.540]  And he has a surge protector in his hand, and he's flipping the switch off and on with his thumb.
[57:09.020 --> 57:10.820]  And so, he's standing next to the machine.
[57:10.820 --> 57:21.640]  And the easiest way for him to interface with the machine's system and change something about the machine is to put a hard stop in it and control it by turning it off and on with his thumb.
[57:22.520 --> 57:25.740]  Thumbs are pretty expensive and it's a frustrating...
[57:28.160 --> 57:51.120]  You could get around it with enough... if you wrote an embedded controller that took something that you decided to specify within the G-code as on and off, add a transistor to the... add a transistor that would turn off and on the hot wire for you and recompile your controller box firmware,
[57:51.480 --> 57:58.480]  put that on the controller box and then talk to that from some software that you custom wrote, then yes, you could do that.
[57:58.660 --> 58:00.480]  However, I think it's a waste of time.
[58:12.860 --> 58:16.580]  I would say it's not... it's more ease of use.
[58:16.800 --> 58:26.140]  If I don't want someone to come up and say, I want to build something and I don't have time... I don't have time to, like, hack all of these existing systems.
[58:26.140 --> 58:27.460]  This is going to take me too long.
[58:27.580 --> 58:30.960]  I'm going to go... I'm going to work on something else that's not fabrication engineering.
[58:32.500 --> 58:36.280]  Do you guys know Travis Goodspeed, the guy who organized NeighborCon?
[58:36.400 --> 58:38.480]  So he did that thing with a satellite, right?
[58:38.580 --> 58:41.440]  Where he had a satellite dish that he was pointing around in different places.
[58:41.720 --> 58:53.820]  And so instead of using the control box that he had in the satellite dish, which probably would listen to G-code, he kind of repurposed... he repurposed the internals of an egg bot.
[58:54.100 --> 58:58.440]  So the thing... the little doohickey machine that draws on Easter eggs.
[58:59.060 --> 59:01.680]  Because that was the easiest way for him to interface with that system.
[59:01.960 --> 59:15.560]  And he happens to be a very tenacious kind of guy who's okay with taking out an entire control system, writing... writing firmware for a new control system and hooking it up to a satellite dish and then looking at different kinds of stars.
[59:15.800 --> 59:21.820]  But I think that the world would be a better place if more people could do things like Travis Goodspeed.
[59:23.220 --> 59:23.740]  Yes?
[59:26.680 --> 59:35.400]  So earlier you were talking about... right now you're talking about how it's like both are like... do these kind of machine-making things.
[59:37.060 --> 59:46.840]  In Shenzhen... obviously they have a lot of wealth of people making... approach phones and whatever.
[59:49.560 --> 59:51.720]  But obviously it's probably still quite difficult.
[59:52.340 --> 59:53.720]  And they have like advantage of that.
[59:53.720 --> 59:53.720]  And they have like...
[01:00:24.730 --> 01:00:35.570]  I think that because of portals like Taobao and other kind of Internet at mediary interfaces, it's getting a lot easier to buy things.
[01:00:36.130 --> 01:00:46.070]  So before I bought those motors in Shenzhen, I also buy motors from Hayden Kirk Switch Co., which is in Connecticut.
[01:00:46.970 --> 01:00:54.470]  And the lead times for their motors are actually much longer than the lead times including shipping for the motors that you order from Shenzhen.
[01:00:54.750 --> 01:01:06.470]  So I think that there's like an argument that you can make that we live in a cosmopolitan society where we have rapid access to different places in the world that have different pockets of expertise.
[01:01:08.110 --> 01:01:19.210]  And I realize that it also is maybe not a super... if you're concerned about the American economy, perhaps you don't want things to be manufactured abroad.
[01:01:19.830 --> 01:01:35.810]  But for me, it's also just a question of where can you get tools, materials, and parts that are to the specification that you need at a price that you can afford in the timeline that you have available to you.
[01:01:36.910 --> 01:01:40.290]  So yeah, I think it's complicated.
[01:01:40.450 --> 01:01:52.030]  And I think that moving closer to the... I think moving closer to working together with small-scale manufacturers in China is definitely a direction I think that we're going in.
[01:01:52.190 --> 01:01:53.330]  And I'm pretty excited about it.
[01:01:53.670 --> 01:01:55.290]  Maybe Obama's not so excited about it.
[01:01:59.850 --> 01:02:00.750]  So that was it.
[01:02:00.990 --> 01:02:01.870]  So long, y'all.