[00:00.000 --> 00:03.240]  Welcome to the Andreas Orthey podcast series on robotics.
[00:03.900 --> 00:06.480]  My guest in this episode is Sean Murray.
[00:07.380 --> 00:11.920]  Sean is co-founder and director of robotics engineering at Realtime Robotics,
[00:12.300 --> 00:16.920]  a company specialized on generating collision-free motions for industrial robots.
[00:18.200 --> 00:22.040]  In this episode, we will talk about his past as a chemical engineer,
[00:22.620 --> 00:24.640]  how to be effective in the robotics industry,
[00:25.420 --> 00:28.220]  what the biggest challenges are for robot manufacturing,
[00:28.220 --> 00:30.880]  and the future of Realtime Robotics.
[00:31.320 --> 00:33.440]  I'm happy that he was able to join me.
[00:33.800 --> 00:35.680]  Please welcome Sean Murray.
[00:37.360 --> 00:38.300]  Hello, Sean.
[00:38.540 --> 00:40.240]  Thanks for doing the podcast.
[00:40.440 --> 00:41.760]  Thanks for coming to Berlin.
[00:41.980 --> 00:42.480]  Of course.
[00:43.300 --> 00:46.920]  So you are currently located in Montana, right?
[00:47.280 --> 00:49.260]  But today you are here in Berlin, Germany.
[00:49.820 --> 00:53.260]  So what made you actually come around at this time?
[00:53.860 --> 00:58.200]  Yeah, so sometimes it feels kind of surreal that we have an office
[00:58.320 --> 01:02.120]  in Berlin, because it seems like just yesterday there was three of us
[01:02.220 --> 01:04.480]  in a different co-working space in Boston.
[01:04.580 --> 01:11.120]  So yeah, here in Berlin for a few days on my way to visit some partners in Finland.
[01:11.320 --> 01:15.440]  And yeah, decided it made sense to stop by and visit our awesome team here.
[01:15.880 --> 01:16.980]  Cool. That sounds good.
[01:17.380 --> 01:20.820]  And so let me start a little bit with your background.
[01:21.240 --> 01:24.080]  So can you maybe tell me a little bit about the place you were born
[01:24.180 --> 01:26.400]  and how it maybe shaped your upbringing?
[01:26.500 --> 01:27.080]  Sure.
[01:27.180 --> 01:27.200]  Sure.
[01:27.300 --> 01:28.180]  So I was born in Thailand.
[01:28.280 --> 01:33.600]  I was born in Texas, but moved around across the states, you know, growing up.
[01:35.600 --> 01:41.160]  I don't know how much it shaped my thinking, except for like being open to kind of whatever happened, you know.
[01:41.260 --> 01:46.860]  So when I was in grad school and, you know, things happen and we thought about starting a company,
[01:46.960 --> 01:48.220]  I was like, all right, let's do it.
[01:48.440 --> 01:49.440]  Let's see what happens.
[01:49.880 --> 01:50.500]  That sounds good.
[01:50.620 --> 01:55.660]  And so when you were still in school, so what kind of topics that you really enjoy
[01:55.660 --> 01:58.140]  and what kind of topics were you really good at?
[01:58.260 --> 02:24.600]  So, you know, I was always really interested in, you know, computer stuff, you know, you know, when I was a, you know, early, early teen, you know, my, my, my dad gave me one of those like programmable, you know, calculator type things, and I really enjoyed like going on a program that and then from computers and so really enjoyed that enjoyed, you know, math stuff, you know, pretty much anything, you know, science related.
[02:24.600 --> 02:25.480]  I really enjoyed.
[02:25.880 --> 02:28.160]  So what was actually your first contact with a computer?
[02:28.160 --> 02:30.740]  Or maybe like first program you wrote also?
[02:31.300 --> 02:37.160]  Some kind of game, I'm sure, you know, like most kids, you like games, you know, say you like playing games and you want to make your own games.
[02:37.800 --> 02:40.400]  And yeah, so some something like that.
[02:40.820 --> 02:45.420]  And did you already like know that you wanted to do something with programming at that time?
[02:45.500 --> 02:49.160]  Or what was it like your vision for your life when you were still growing up?
[02:49.160 --> 02:55.920]  I think I really had a vision, you know, I wasn't the kind of kid, you know, my brother, for example, knew he was really into like space for me.
[02:55.920 --> 02:56.240]  I'm kidding.
[02:56.380 --> 02:58.120]  Super into space stuff, you know.
[02:58.160 --> 02:59.200]  Or he wanted to be an astronaut.
[02:59.200 --> 03:00.080]  I wasn't really like that.
[03:00.080 --> 03:02.340]  I was just kind of like, I was along for the ride.
[03:03.560 --> 03:07.400]  And so is your brother actually doing something in space?
[03:07.620 --> 03:07.940]  No.
[03:09.400 --> 03:09.900]  Okay.
[03:10.180 --> 03:21.580]  And, but before you actually went to the whole computer science half, you first actually stopped and doing some chemical engineering, right?
[03:21.700 --> 03:22.200]  Yeah.
[03:22.540 --> 03:27.640]  So can you maybe tell me what motivated you to pursue a degree in chemical engineering?
[03:27.740 --> 03:28.140]  Yeah.
[03:28.160 --> 03:39.120]  I, um, so, uh, I originally wanted to go into like bio, um, technology, um, you know, there was a, you know, I was going to school in like 2007 to 2011.
[03:39.560 --> 03:42.620]  And so there are a lot of exciting stuff going on there.
[03:43.280 --> 03:49.060]  So that was my, my goal when I, it was a pretty tough time in the, in the job market.
[03:49.060 --> 03:56.280]  So I ended up doing more like stuff like, uh, for the energy industry, doing reservoir simulation type stuff.
[03:56.680 --> 03:58.060]  Um, did it for a few years.
[03:58.160 --> 04:09.120]  Really enjoyed it, but you know, at some point decided I wanted to, you know, try something new and went back to school and, um, yeah, started to study, um, computer architecture.
[04:09.560 --> 04:19.000]  So, but if you had like a background in like programming, how did you actually think about like going into like, for example, compute, uh, in chemical engineering?
[04:19.000 --> 04:21.400]  So I, I think that it's like a big jump, right.
[04:21.960 --> 04:24.400]  Um, so, so, so how did this actually happen?
[04:24.400 --> 04:26.680]  Did you want it to like program life or something?
[04:26.680 --> 04:45.340]  No, I think it was just really interested in, in, in biosciences at the time, you know, you know, there was a lot of, there was a lot of computing going on, you know, and still is with, you know, bioinformatics, you know, like kind of analyzing, you know, genetic data and, um, you know, it was intrigued by that.
[04:45.340 --> 04:52.860]  Um, and, uh, at Rice university, they had a, they had a really kind of strong collaboration between their chemical engineering and bioengineering departments.
[04:53.120 --> 04:55.700]  So that seemed like a cool place to be.
[04:56.000 --> 04:56.660]  And so, so.
[04:56.680 --> 05:01.300]  What kind of teasers did you write about chemical engineering?
[05:01.300 --> 05:08.120]  And so we actually didn't, I'm trying to think, I don't think we really did theses in our, in our undergrad, um, maybe some projects.
[05:08.440 --> 05:08.860]  Yes.
[05:08.900 --> 05:25.560]  So we, yeah, there was, there was, uh, trying to think of, you know, any of it really stand out, but we did fun classes and kind of like modeling the dynamics of, of, of different, you know, uh, uh, you know, genes and species over time.
[05:25.700 --> 05:26.200]  Um,
[05:26.680 --> 05:30.460]  you know, we did, uh, we had some classes in like protein engineering.
[05:30.520 --> 05:34.120]  Um, I'm struggling to remember any of those projects, you know, so long ago.
[05:35.440 --> 05:35.800]  Yeah.
[05:35.920 --> 05:36.180]  Yeah.
[05:37.440 --> 05:54.180]  But something which seems to be really interesting as what you already like mentioned, like reservoir engineering, can you maybe tell me a little bit, or can you maybe describe what reservoir engineering actually entails to someone who may be like outside the field?
[05:54.180 --> 05:56.120]  And what excited you most about it?
[05:56.180 --> 05:56.640]  Sure.
[05:57.220 --> 06:07.060]  So, reservoir engineering is kind of the, um, science of like how to, uh, kind of best manage a, um, system that can be kind of modeled as a closed system.
[06:07.300 --> 06:10.180]  You know, it has temperatures and pressures and different fluids in it.
[06:10.540 --> 06:14.720]  And as you take, you, you know, you're not just taking things out, you're also putting things in.
[06:14.800 --> 06:24.320]  Um, and, uh, so, you know, usually what you do is you, you model the system, you discretize the system, like most modeling, you discretize the system somehow.
[06:24.680 --> 06:26.560]  You apply like the natural laws of governance.
[06:26.560 --> 06:26.660]  Yeah.
[06:26.680 --> 06:36.620]  So fluid dynamics, um, uh, you know, pressure gradients, temperature gradients, uh, you model and see, okay, if we, if we drill these holes in these locations, it's going to drop the pressure in these ways.
[06:36.980 --> 06:40.580]  Some gases are going to come out of solution, uh, which may or may not be desirable.
[06:41.540 --> 06:43.920]  You have rocks with different porosities.
[06:43.980 --> 06:45.320]  And so you're, you're also taking measurements.
[06:45.320 --> 06:51.940]  And so you'll, you might have a reservoir that's, you know, 300 feet thick, but the 300 feet is not homogeneous.
[06:51.940 --> 06:56.520]  It might vary from, you know, uh, very, almost no porosity and no permeability.
[06:56.680 --> 07:04.680]  And so you, you use the data you have to create, uh, models, which are, you know, are always imperfect.
[07:04.860 --> 07:08.340]  And then you try to predict what will happen if you use different things.
[07:08.340 --> 07:19.260]  And so you might be participating or you might, you know, have a goal to, you know, get out 80% of, of fossil fuels in a, in a, in a reservoir.
[07:19.260 --> 07:26.260]  And you can kind of test different, different, uh, strategies and, and, and simulate how that might, uh, what, how different, um, development.
[07:26.680 --> 07:29.260]  Strategies might, uh, maybe work or may not work.
[07:29.260 --> 07:30.640]  Um, okay.
[07:30.640 --> 07:32.800]  And, and so, so what is the biggest challenge there?
[07:32.800 --> 07:38.680]  So, so what would actually happen to the reservoir in the, in the worst case, if you, if you do something wrong?
[07:39.180 --> 07:41.800]  So, I mean, there's a lot of, there's a lot of challenges.
[07:42.020 --> 07:55.620]  I mean, this is a whole field, like there, there's like entire, um, you know, uh, certain sort of whole programs on this, but so like, just as an example, um, so you'll, you'll often, you know, we're, this is going to be a huge tangent, but that's okay.
[07:55.620 --> 08:01.000]  So the way that most of these things are going to start is you might start with seismic data.
[08:01.080 --> 08:02.820]  You might start, you haven't drilled anything.
[08:02.960 --> 08:04.480]  You've just taken measurements from the surface.
[08:05.380 --> 08:08.340]  And, and from that, you have a very, very coarse model.
[08:08.580 --> 08:12.140]  It's not going to be very accurate, but it's going to have been pretty cheap to get that data.
[08:12.900 --> 08:15.080]  Um, then you might drill a test well.
[08:15.460 --> 08:20.620]  And, um, and so you, you drill a hole and you, uh, you log, it's called logging the hole.
[08:20.720 --> 08:25.240]  So you stick sensors down the hole and you take measurements and you see what's in there.
[08:25.620 --> 08:27.200]  Um, what kind of rock is in there?
[08:27.240 --> 08:28.100]  What kind of fluid is in there?
[08:28.460 --> 08:29.860]  And that gives you one data point.
[08:30.000 --> 08:33.460]  And so you might try to extrapolate, you combine that with this other data.
[08:33.460 --> 08:46.340]  And so one thing that might happen if you didn't plan well, um, is, uh, the more you decrease the pressure in these reservoirs, the more you might have things come out of solution.
[08:46.340 --> 08:55.460]  So like if, so in the, in kind of very old school, let's say before reservoir simulation was a thing back when people were just like drilling, drilling and pumping, all they were doing is.
[08:55.620 --> 09:03.740]  You'd only get about, you know, 15, call it maybe max 30% of, of the actual fossil fuels out of reservoir.
[09:04.060 --> 09:10.200]  And then that there'll be it, like, which is obviously pretty, you know, inefficient and kind of wasteful.
[09:10.440 --> 09:12.060]  What happens with the rest of it?
[09:13.080 --> 09:14.760]  You can't really, you can't reach it anymore.
[09:14.760 --> 09:16.260]  So, well, there's a few things that'll happen.
[09:16.260 --> 09:25.500]  So one is if you're, um, if you're just trying to pump stuff out, you reach a point in which, uh, the pressure has dropped so much that, you know, uh,
[09:25.500 --> 09:25.520]  uh,
[09:25.520 --> 09:34.940]  that gas is separated from liquid oil and, um, uh, the pumps are usually designed to move liquids and not designed to move gases.
[09:34.940 --> 09:38.340]  And this, that technology just doesn't really work very well.
[09:38.340 --> 09:50.900]  And so as the industry advanced, you know, different techniques came into where you start, uh, pumping water into the reservoir to keep the pressure up and also push, um, uh, push fluids around, um, which works.
[09:51.440 --> 09:55.340]  You can also do things like mix surfactants into the reservoir.
[09:55.340 --> 09:55.460]  So.
[09:55.520 --> 09:56.600]  So what are surfactants?
[09:56.720 --> 10:04.180]  Surfactants are, are chemicals that kind of, you can think of them like soaps, basically, like, you know, obviously if you've tried to clean, you know, oil with water, it doesn't work very well.
[10:04.180 --> 10:18.600]  You use some kind of, uh, something to break the surface tension, you know, and, uh, as, and you can also, you know, pump, uh, carbon dioxide down in there, which is a, is a great, uh, solvent and, and makes, uh, oil more mobile.
[10:18.940 --> 10:23.460]  I feel like I'm getting into a whole different topic here.
[10:24.640 --> 10:25.340]  That's true.
[10:25.520 --> 10:26.920]  I don't know how useful it's going to be.
[10:27.120 --> 10:29.000]  So maybe let's wrap this topic up.
[10:29.140 --> 10:29.260]  Sure.
[10:29.460 --> 10:38.420]  And so, so what would you say were maybe like some bigger lessons, which you learned from, from your time there, which you can maybe still apply today?
[10:38.520 --> 10:38.740]  Yeah.
[10:38.800 --> 10:41.700]  I mean, it was just an interesting, to me, it was an interesting industry.
[10:41.860 --> 10:46.180]  It had a lot of, it had a lot of inertia behind it.
[10:46.220 --> 10:48.180]  It had a lot of science behind it.
[10:48.220 --> 10:49.580]  It had a lot of R and D behind it.
[10:49.640 --> 10:52.260]  Um, it also used computation in an interesting way.
[10:52.260 --> 11:11.420]  Like that was kind of what appealed to me about it is that, um, you, you, you, you would take this kind of big system that was kind of opaque and you would use, you know, certain measurements and build up a model of how it behaved and try and use that model to predict how different strategies might work out in the long run.
[11:12.160 --> 11:12.640]  Okay.
[11:12.760 --> 11:22.220]  And then after three years as being a reservoir engineer, roughly, you actually made a very drastic change and you started a PhD.
[11:22.220 --> 11:23.100]  In computer science.
[11:23.220 --> 11:32.600]  So was there maybe a moment where you realized that you wanted to go down this road and not the road, which you are currently on?
[11:33.120 --> 11:38.700]  Yeah, I, um, I think I was just, you know, realized that I didn't want to do that for the rest of my life.
[11:38.760 --> 11:42.780]  You know, it was, it was interesting, but you know, nothing is interesting forever.
[11:42.780 --> 11:44.660]  You know, at some point you want to try something new.
[11:44.780 --> 11:51.480]  I was at an age where grad school still made sense and yeah, I was, I was excited about computing where, where, where.
[11:52.220 --> 11:53.780]  Where it was going and, uh, yeah.
[11:54.220 --> 11:58.500]  And what kind of options did you consider before committing to, to do a PhD?
[11:59.140 --> 12:04.040]  Um, I, so I was, I considered a few, I mean, I was pretty certain I wanted to go back to school.
[12:04.040 --> 12:09.100]  I was considering a few different, a few different, you know, topics.
[12:09.440 --> 12:16.160]  I was pretty interested in distributed systems, um, high-performance computing, specialized computer architectures.
[12:16.320 --> 12:18.960]  So I was looking for something in one of, in one of those fields.
[12:19.960 --> 12:22.040]  Um, and then I could improve.
[12:22.220 --> 12:37.200]  And there was a, uh, the Kinda going back and not being able to, to, to, not to, to be a
[12:37.200 --> 12:38.140]  unit at Harvard.
[12:38.360 --> 12:44.280]  For me, that thing was only going to, to get there rather than just going forward.
[12:44.800 --> 12:46.100]  Um, so.
[12:46.720 --> 12:50.020]  It's not, it's not –
[12:50.340 --> 12:51.100]  Yesterday?
[12:51.140 --> 12:51.420]  No.
[12:51.420 --> 12:51.760]  No.
[12:51.820 --> 12:51.840]  I was just like, okay.
[12:51.840 --> 12:51.940]  So one, let's talk about Utah.
[12:51.940 --> 12:51.960]  Present.
[12:51.960 --> 12:51.980]  Schools.
[12:51.980 --> 12:52.080]  Who stay in Utah.
[12:52.080 --> 12:52.160] chnet your background?
[12:52.160 --> 12:52.200]  Again, since you've been here for three years.
[12:52.220 --> 12:59.780]  they probably had better options but luckily um dan soren at duke for whatever reason uh saw
[12:59.780 --> 13:06.280]  something in my resume and interviewed and decided to give me a chance and yeah and do you think your
[13:06.280 --> 13:12.160]  time as a chemical engineer actually gave you maybe like a unique perspective on problems in
[13:12.160 --> 13:17.520]  computer science or gave you somehow like an edge compared to like classical computer science
[13:17.520 --> 13:22.800]  students i don't i don't think the specific industry gave me an edge but i think having
[13:22.800 --> 13:29.080]  worked at all i think definitely gives you perspective that you don't get from just going
[13:29.080 --> 13:34.340]  to school and so i definitely felt like those of us because i wasn't the only one who had a little
[13:34.340 --> 13:38.220]  bit of work experience but i definitely felt like those of us who'd spent some time in industry had
[13:38.220 --> 13:41.680]  a sense for like a little bit more of you know you just know a little bit more about how the
[13:41.680 --> 13:46.280]  world works and any any experiences could be helpful in that you would say more down to earth
[13:46.280 --> 13:47.500]  compared to others
[13:47.520 --> 13:54.380]  students i mean down a little bit of down earth but just kind of uh you know when you're when
[13:54.380 --> 14:00.180]  you're when you're 22 and maybe just finished undergrad you know there's a lot you haven't
[14:00.180 --> 14:04.720]  experienced in the world and um you're just only just having a maybe a little bit broader range of
[14:04.720 --> 14:10.660]  like you know what what goes on out there is helpful yeah i guess so and so how did you
[14:10.660 --> 14:16.500]  actually your phd start and and what were your main goals at the beginning yeah so i was
[14:16.500 --> 14:17.500]  working on a teaching degree i was working on a teaching course and i was working on a book
[14:17.500 --> 14:23.880]  I was working with Dan Soren in the, so it was actually in computer engineering, which
[14:23.880 --> 14:25.780]  is tied in their electrical engineering department.
[14:26.400 --> 14:33.380]  And I was interested in computer architecture, which is what Dan works in.
[14:33.940 --> 14:36.820]  And I didn't start with any specific goals.
[14:36.920 --> 14:39.900]  Like, I wasn't the kind of student that came in and was like, okay, this is the specific
[14:39.900 --> 14:41.420]  topic I wanted to dive into.
[14:41.420 --> 14:48.460]  And so I want to say after, I want to say it was only maybe six months, about six months
[14:48.460 --> 14:56.660]  into it that we met George Connery, who had recently started in the robotics department.
[14:57.820 --> 15:03.440]  Or I guess it was the CS department, kind of a budding robotics, you know, subfield
[15:03.440 --> 15:08.340]  there, and talked with him and him and Dan had kind of been talking about how computer
[15:08.340 --> 15:09.320]  architecture could help.
[15:09.320 --> 15:15.080]  Um, maybe help in some particular computing problems in robotics.
[15:15.160 --> 15:19.480]  And that's when, you know, we started talking about motion planning and in particular, we
[15:19.480 --> 15:23.920]  were, we started off working on, um, specialized computer architectures for collision detection.
[15:24.200 --> 15:27.100]  Uh, so we hadn't even really gotten to the real motion planning parts yet.
[15:27.200 --> 15:31.560]  Just, just collision detection, how to tell if, if, if two things are intersecting, you
[15:31.560 --> 15:32.320]  know, really quickly.
[15:32.320 --> 15:32.940]  Mm-hmm.
[15:33.540 --> 15:39.320]  And then at some point you published a very influential paper, namely Robotics.
[15:39.320 --> 15:42.560]  Motion Planning on a Chip, which is your highest cited paper.
[15:42.620 --> 15:48.040]  Um, so can you tell me a little bit about the ideas which maybe came already, like from
[15:48.040 --> 15:53.140]  this whole collision checking and how did they actually mature into this paper?
[15:54.000 --> 15:54.500]  Yeah.
[15:54.920 --> 16:03.440]  So we, so we were doing really fast collision checking, um, on FPGAs and, um, uh, we thought
[16:03.440 --> 16:08.500]  it would be cool to kind of do a, uh, you know, kind of an end to end demonstration
[16:08.500 --> 16:09.200]  of what you could do with it.
[16:09.200 --> 16:19.040]  So we, uh, we were using the Canova Jayco arms, um, which was, they were at the time, one
[16:19.040 --> 16:22.560]  of the kind of popular ones to use in academic, you know, kind of academic papers because
[16:22.560 --> 16:24.460]  they were, um, they were very safe arms.
[16:24.560 --> 16:26.060]  They were, they didn't move super fast.
[16:26.680 --> 16:30.600]  Um, uh, they had a, they, they came with a built-in gripper, so you didn't have to like
[16:30.600 --> 16:31.680]  buy your own gripper.
[16:32.740 --> 16:38.800]  And, uh, so we, we set up a, I'm trying to remember the demonstration exactly, but, um,
[16:38.800 --> 16:44.520]  so we basically, you know, built a, you know, a roadmap of motions, you know, um, for the,
[16:44.580 --> 16:50.900]  for the arm and, uh, and we were using, um, Kinect cameras, you know, which I'm sure you
[16:50.900 --> 16:58.540]  remember it like back in the, um, 2010s were super, super big thing, you know, this video
[16:58.540 --> 17:03.960]  game, uh, video game, uh, camera, um, cause you could get, uh, and I think this is, you
[17:03.960 --> 17:08.180]  know, the, the proliferation of those 3d cameras is probably, you know, it's, it's, it's, it's,
[17:08.180 --> 17:08.780]  it's, it's, it's, it's, it's, it's, it's, it's, it's, it's, it's, it's, it's, it's, it's, it's,
[17:08.780 --> 17:12.560]  it's probably a big reason of why, um, not only us, but like, I feel like a lot of other
[17:12.560 --> 17:17.440]  companies kind of became founded around the time is that these cheap 3d cameras, which
[17:17.440 --> 17:20.720]  hadn't really been available before, like all of a sudden, you know, it went from the
[17:20.720 --> 17:25.140]  Kinects to the real senses, you know, and like then after that, like a bunch of 3d cameras
[17:25.140 --> 17:25.760]  became available.
[17:25.980 --> 17:29.740]  And, and, um, uh, before then there's, there weren't very many options for that.
[17:29.740 --> 17:34.940]  Um, but when you could get, you know, 30, 60 or 90 frames per second of fairly decent
[17:34.940 --> 17:38.760]  3d data, um, a lot of, you know, researchers started to do things with that data.
[17:38.780 --> 17:44.780]  And so, yeah, back to that, um, that first project, we, um, I want to say that we, we,
[17:45.720 --> 17:49.600]  uh, had like cardboard boxes and we would have the robot pick up, I don't know if it
[17:49.600 --> 17:51.240]  was like a ball or like a, some kind of a toy.
[17:51.900 --> 17:55.740]  And, um, uh, we weren't even like at that point, I don't think we even really moved
[17:55.740 --> 17:57.080]  them around while the robot was moving.
[17:57.140 --> 18:01.420]  We just kind of like moved it around and it would like kind of demonstration was that,
[18:01.460 --> 18:05.820]  you know, you would like hit, I think we put like a big red button to make it really visible.
[18:05.820 --> 18:08.260]  Like when we would send the data to the, to the FPGA.
[18:08.780 --> 18:10.760]  And it was, you know, super fast.
[18:10.860 --> 18:14.080]  They would like just do all the collision checking and then find a plan for the roadmap
[18:14.080 --> 18:14.780]  around the boxes.
[18:14.900 --> 18:19.740]  And, uh, and yeah, that was kind of, it was kind of it, it was, uh, it was pretty primitive
[18:19.740 --> 18:21.720]  compared to what we do now, but, uh, yeah.
[18:21.720 --> 18:22.040]  Yeah.
[18:22.460 --> 18:31.540]  So, so it was basically collision checking on an FPGA class connect censoring and everything
[18:31.540 --> 18:34.340]  combined with like standard motion planning algorithms.
[18:34.340 --> 18:34.700]  Right.
[18:34.780 --> 18:37.500]  So that was basically how you could describe it.
[18:37.500 --> 18:38.700]  It's like in a very simple.
[18:38.780 --> 18:38.940]  Yeah.
[18:39.940 --> 18:44.240]  Um, I don't even know if it was, I wouldn't even, you know, this kind of standard motion
[18:44.240 --> 18:44.740]  planning algorithm.
[18:44.880 --> 18:47.020]  I think it might've been just like a fixed roadmap.
[18:47.140 --> 18:49.260]  I don't even think it was doing any real planning.
[18:49.420 --> 18:52.940]  So it was, it was honestly, it was, it was pretty primitive what we were doing.
[18:53.800 --> 19:00.080]  And so it was still primitive, but still it gave you like this idea of maybe there's something
[19:00.080 --> 19:01.140]  bigger behind it.
[19:01.180 --> 19:04.340]  And maybe we can actually go out there and start a company.
[19:04.620 --> 19:08.760]  Um, can you maybe walk me a bit through this whole process of like, you know, like, you
[19:08.780 --> 19:13.380]  know, conceiving this idea, um, and when did it actually, or made sense and when did it
[19:13.380 --> 19:16.020]  actually click for you that we should do a company?
[19:16.340 --> 19:20.300]  It exposed, I think it exposed us to a little bit to like the problems of, and challenges
[19:20.300 --> 19:21.140]  that exist in robotics.
[19:21.140 --> 19:26.220]  And then, um, it's, it connected us with people that were interested who kind of would be
[19:26.220 --> 19:30.080]  each person we talked to explain like a different aspect of the problem, because one of the
[19:30.080 --> 19:33.180]  things in robotics is it's a very, obviously a very broad term.
[19:33.280 --> 19:35.200]  Everyone's using these robots for different things.
[19:35.340 --> 19:36.740]  So they have different challenges.
[19:36.740 --> 19:38.740]  Um, and, um,
[19:39.300 --> 19:42.460]  we were trying to identify like commonalities in those challenges.
[19:42.460 --> 19:45.780]  And so one commonality is that robots are really easy to crash into stuff.
[19:45.800 --> 19:49.400]  You know, um, anyone who's tried to get a robot to do stuff, realize that if you're not
[19:49.400 --> 19:51.080]  careful, it's going to, it's going to crash into stuff.
[19:51.500 --> 19:58.160]  Um, and so our goal was to, uh, make it easier to deploy robots and not have to think that.
[19:58.220 --> 20:02.540]  And if you're trying to do robots to do X, Y, or Z, you can focus on X, Y, or Z and not on
[20:02.540 --> 20:03.620]  how to not crash into stuff.
[20:04.020 --> 20:06.860]  And we wanted to kind of make a product that would do that.
[20:06.860 --> 20:07.760]  And we do it very fast.
[20:07.760 --> 20:14.360]  Um, and I think that it probably, we probably marinated on the, uh, you know,
[20:14.360 --> 20:18.360]  ideas for probably at least a year or so.
[20:18.620 --> 20:22.820]  Um, but we kind of started to get more interest from industrial tech companies.
[20:22.820 --> 20:26.480]  And, um, at a certain point we realized it made sense to, um, give it a go.
[20:26.880 --> 20:29.420]  Um, and, uh, and, and make a company out of it.
[20:30.140 --> 20:34.260]  And how did you envision your first product and maybe which market segment
[20:34.260 --> 20:35.820]  that you wanted to tackle in the beginning?
[20:36.480 --> 20:36.880]  Yeah.
[20:36.880 --> 20:37.380]  Um.
[20:37.760 --> 20:45.460]  So at the beginning we were definitely, um, uh, thought that, um, we'd be operating in
[20:45.460 --> 20:52.240]  a very dynamic space with, uh, lots of like, lots of ops, lots of obstacles that were not
[20:52.240 --> 20:52.640]  predicted.
[20:53.100 --> 20:58.540]  Um, this is around the time when like collaborative robots were becoming a more of a thing, you
[20:58.540 --> 21:04.220]  know, uh, the, the URL robots had, you know, had, you had come out somewhat recently and
[21:04.220 --> 21:07.220]  there was this vision that a lot of people had of like having, um,
[21:07.760 --> 21:10.580]  pretty close human robot collaboration.
[21:11.440 --> 21:15.140]  And, uh, we had envisioned being in kind of that space.
[21:15.980 --> 21:24.300]  And so our first product was, uh, you know, an FPG, an FPGA board, um, a PCI, which you
[21:24.300 --> 21:30.400]  would put it on a PCIe slot and, um, and, and the library, uh, to talk to that board.
[21:30.700 --> 21:37.740]  And, um, we, we, our thought was that we would be a component in a much more complex system where you would feed it.
[21:37.760 --> 21:39.380]  Um, obstacle data.
[21:39.380 --> 21:43.000]  And we kind of had like a standard for how you would communicate obstacle data to this, to this board.
[21:43.140 --> 21:47.320]  And it would, it would tell you back which motions were safe and which motions were not safe.
[21:47.660 --> 21:55.140]  Um, and, and we very quickly realized that that was not, not what, not what the industry wanted.
[21:56.400 --> 21:59.640]  How did you actually find out that this was not what the industry wanted?
[21:59.680 --> 22:07.640]  Well, by, by trying, you know, some of these things, you, some of the things you can learn by just talking to people and guessing, but then some things you just try.
[22:07.640 --> 22:07.740]  Yeah.
[22:07.760 --> 22:16.820]  So we tried and we worked with some great partners and, and, and sent them, you know, and, you know, send them these, these, these prototypes and, and these boards.
[22:16.860 --> 22:22.300]  And, um, uh, what we realized is that, uh, that was, that was a pretty low level interface.
[22:22.660 --> 22:37.460]  Um, and most of the, and that in, you know, kind of in real industry, most, most things are kind of, people want to be kind of plugging play where they don't want, you know, most integrators, you know, outbuilding robotic work cells.
[22:37.760 --> 22:54.740]  Uh, don't want like a, like a C header file in a library to program against, they want something that has like a, um, that they can basically plug in, you know, that has some industrial protocol to talk to it and, and, and just go, they, they weren't really interested in, in, in, in something.
[22:54.940 --> 22:56.060]  So, so low level.
[22:56.840 --> 23:03.500]  And then how did it actually happen that you found your first customer and what was the story about that?
[23:03.500 --> 23:07.760]  Um, I want to say that our, one of our first touch points with, with industry.
[23:07.760 --> 23:16.780]  Was back when we were at Duke, I want to say that, um, Amazon robotics, uh, hosted some kind of like a symposium or something like that.
[23:17.220 --> 23:24.260]  And we're just kind of like, you know, talk with the different robotics, you know, faculty at the university.
[23:24.860 --> 23:28.340]  It might've even been like multiple universities.
[23:28.340 --> 23:37.480]  I can't quite remember, but, uh, yes, we talked with them there and they, they, you know, we, we did some early projects with them that were really beneficial in terms of like stuff that we learned.
[23:37.760 --> 23:38.200]  Nice.
[23:38.260 --> 23:43.680]  And then I guess one of the biggest challenges for any startup founder is of course, to get funding.
[23:44.760 --> 23:47.280]  So, so how did he actually acquire funding in the beginning?
[23:47.280 --> 23:51.540]  And who did you consider actually asking for funding?
[23:52.640 --> 24:01.000]  So I'd say that the, the, the, the best thing we did for getting funding was, was find a great CEO.
[24:01.060 --> 24:06.540]  So we, one of the, one of the first things we did was we, we, um, we found Peter Howard, who was our.
[24:06.540 --> 24:17.820]  It was, I've been our CEO since we, since we started and we looked, uh, high and low, I think we were open to all sources of funding at the beginning.
[24:18.240 --> 24:35.420]  Um, uh, I'm trying to remember now, um, you know, we, so we started when we went full time, um, at mass robotics, which is a robotics focused coworking space in Boston and that.
[24:35.420 --> 24:47.000]  Connected us with a, a lot of different, uh, industry partners and some of those became our, our first funders among them, uh, you know, Mitsubishi has been one of our strongest supporters for the longest time.
[24:47.000 --> 24:49.400]  And so they, they've been involved from, from very early.
[24:49.940 --> 24:50.180]  Cool.
[24:50.180 --> 24:56.780]  And what do you think was maybe the biggest challenge that you faced in the beginning of the company and how did you overcome it?
[25:00.460 --> 25:05.040]  I mean, it seems looking back, it seems like it was just one, we were, we were taking it, you know, week by week.
[25:05.420 --> 25:06.740]  One challenge after the other.
[25:06.800 --> 25:16.280]  Um, I mean, you know, there was, there was, yeah, there was just a lot of stuff early on, you know, getting our first.
[25:17.960 --> 25:23.600]  So initially we had been using, um, off the shelf FPGA development called development boards.
[25:23.600 --> 25:28.340]  So there are FPGAs combined with a bunch of other stuff that people commonly use to use for prototyping.
[25:28.660 --> 25:35.400]  Um, and, but initially, well, we knew that, uh, to be kind of commercially viable, we would need to make.
[25:35.420 --> 25:37.640]  Our own, um, our own circuit boards.
[25:37.640 --> 25:44.540]  So, um, getting the first, you know, and getting, so getting our first custom circuit boards out was, was probably a big, a big first step.
[25:44.960 --> 25:48.680]  Um, there was a lot of, a lot of stuff that goes into that.
[25:49.340 --> 25:52.160]  And what would you say was maybe like the best decision?
[25:52.520 --> 25:56.540]  So, so I think real time is already like almost 10 years old.
[25:56.720 --> 26:01.040]  So, so what do you think was like the best decision to set up a real time really for a long term success?
[26:02.460 --> 26:05.240]  I mean, I think the best thing that you can do, um, set up.
[26:05.420 --> 26:09.540]  Any company for long-term success is to hire the right people and have the right culture.
[26:09.560 --> 26:22.640]  Like, you know, because what you realize when you're growing a company is that you very quickly, the capacity of the organization is going to really outstrip what, what you as an individual can do.
[26:22.640 --> 26:33.380]  And so it really relies on you hiring great people kind of, uh, you know, getting them bought into the same mission as you and, um, yeah, knowing that.
[26:33.380 --> 26:35.360]  So I think, I think finding the right people is definitely the.
[26:35.420 --> 26:38.000]  Um, you know, the best decision you can make.
[26:38.300 --> 26:39.140]  Yeah, that's a good point.
[26:39.140 --> 26:41.780]  So, so how do we actually find the right qualified people?
[26:41.780 --> 26:46.080]  And what would you say is like your philosophy behind attracting them?
[26:47.120 --> 27:05.400]  I think that, uh, you know, more so than even just qualifications is, you know, uh, their kind of their passion and their, and their kind of fit in your culture, you know, cause you can, you can, if someone is, you know, has the, you know, the motivation and passion, you can, you can, you know, train them to do pretty much anything.
[27:05.420 --> 27:30.420]  Um, uh, because, you know, and it's going to be more of like, do they have, I think what's a lot harder is figuring out, do they have the, you know, the, the motivation and passion that's going to let them kind of continually, you know, especially in a field like robotics, like it changes super fast, you know, like we're doing, we've probably gone through three major kind of like shifts in terms of what we've been focusing on just as we learn more.
[27:30.420 --> 27:35.060]  And so like, we're working on things that are totally different that, you know, we didn't really expect to be working on.
[27:35.580 --> 27:40.160]  And so it's a question of how do you, how do you find other people?
[27:40.740 --> 27:48.100]  You know, if I, if I, if I had a list of three things to do, um, find other people, I would, I would definitely give them to you.
[27:48.100 --> 27:59.580]  But so it's, I think it's a combination of just, you know, you know, being open-minded, talking to, you know, meeting a lot of people and, um, uh, and yeah, I don't know.
[28:00.320 --> 28:00.760]  Yeah.
[28:00.820 --> 28:02.020]  I don't have a great answer to that.
[28:02.220 --> 28:02.320]  Yeah.
[28:02.320 --> 28:07.580]  And a different question also related to this is of course, how do you keep people?
[28:08.580 --> 28:19.320]  Um, so especially when you have like a startup environment where you have a high risk and maybe you have also not already like implemented all the structures, which other established companies already have.
[28:19.480 --> 28:22.040]  So how do we actually make sure that people are staying?
[28:22.740 --> 28:23.180]  Yeah.
[28:23.280 --> 28:25.340]  Retention is definitely, you know, really key.
[28:25.340 --> 28:29.300]  Um, so I think you, you know, the, you don't have to treat people.
[28:29.520 --> 28:30.980]  Well, you know, that's, that's a no brainer.
[28:31.040 --> 28:32.200]  Like if you, if you don't treat people well.
[28:32.320 --> 28:36.260]  They'll find better opportunities, you know, the, the, the best people have lots of options.
[28:36.260 --> 28:45.660]  And so you, um, you definitely want to treat people well, um, make sure that you have reasonable expectations, you know, you provide good feedback, provide good growth opportunities.
[28:46.340 --> 28:50.360]  Um, uh, I think engineers love working on things where they can see the impact.
[28:50.360 --> 28:53.360]  So being, making sure people get to see the impact of what they're working on.
[28:53.940 --> 29:00.360]  Um, uh, but it's definitely true that, um, as a startup, you oftentimes don't have the resources or the.
[29:00.360 --> 29:08.540]  Kind of capacity to, like, you know, recruit people as some other, uh, some other like bigger organizations.
[29:08.540 --> 29:18.440]  So you have to, um, you know, make up for that in other ways, you know, whether that's, you know, having a great culture, really interesting problems, the ability to, um, work on, uh, you know, things that people want to work on.
[29:18.500 --> 29:23.360]  Um, so yeah, I think you have to kind of use what you, what you can offer and, and yeah.
[29:24.120 --> 29:24.300]  Yeah.
[29:24.300 --> 29:25.960]  And I think there's also a third problem.
[29:25.960 --> 29:29.160]  I mean, even if you get like hiring and retention, right.
[29:29.200 --> 29:29.500]  Mm-hmm.
[29:29.560 --> 29:30.340]  Um, then you're still.
[29:30.340 --> 29:33.840]  Like in a very highly fluctuating startup environment.
[29:33.840 --> 29:34.140]  Right.
[29:34.140 --> 29:42.980]  So how do you actually make sure that for example, that you keep technical expertise still inside of the company, even if they are like people come and go.
[29:43.440 --> 29:43.800]  Yeah.
[29:43.800 --> 29:45.840]  I mean, good documentation is crucial.
[29:45.840 --> 29:48.860]  You know, you definitely want to encourage a culture of writing stuff down.
[29:48.980 --> 29:59.360]  Um, so that, uh, because obviously inevitably, inevitably people will, even if you're the best employer, you know, people are gonna leave at some point and you don't want that to create an abyss of, of.
[29:59.660 --> 30:00.340]  Uh, like a lack of.
[30:00.340 --> 30:00.740]  Knowledge.
[30:00.740 --> 30:18.840]  So creating a culture of like, okay, when you create a new feature or a new, a new architecture in the product, like documenting how it works, um, uh, with the idea that someone should be able to step in and, and after spending like a half a day or like a day reading through it, you know, being able to understand it.
[30:18.840 --> 30:20.940]  And that's obviously easier said than done.
[30:21.220 --> 30:25.120]  Mm-hmm the the more fun part is building the stuff, not writing about the stuff.
[30:25.280 --> 30:29.720]  So, um, you definitely have to kind of constantly, uh, encourage people to, to do that.
[30:29.720 --> 30:39.980]  Mm-hmm and so for people who might be interested in, uh, interested in working for real time, what do you think are like essential prerequisites and maybe what kind of topics should they have studied?
[30:40.680 --> 30:41.000]  Yeah.
[30:41.000 --> 30:55.400]  I mean, I think that, uh, you know, if you wanna, you know, be an engineer in real time, you know, uh, we don't really have a ton of formal, um, requirements, you know, you know, having some kind of a degree.
[30:55.400 --> 30:59.360]  I think we, we generally, you know, if you're, you know, a new graduate, you know.
[31:00.320 --> 31:03.320]  Probably something in, uh, you know, a computing type field.
[31:04.640 --> 31:12.500]  I think that, uh, you know, one of the things that really separates, you know, you know, good engineers and great engineers is the ability to communicate well.
[31:12.620 --> 31:16.040]  Mm-hmm and the ability to reach out for help when needed.
[31:16.100 --> 31:29.520]  You know, I think that, um, there's this, there's nothing quite as frustrating as, as, as, as realizing that like, you know, something could have been done, you know, five times faster if, if someone had just like reached out with like some basic questions.
[31:29.520 --> 31:29.620]  Yeah.
[31:29.640 --> 31:29.700]  Early.
[31:29.720 --> 31:34.220]  On, you know, and avoid it, um, uh, you know, going down the wrong path.
[31:34.280 --> 31:40.100]  Um, and I think that, you know, communicate great communication is, is one of the invaluable, uh, skills that you can develop.
[31:40.520 --> 31:51.840]  So do you think like great communication is like the most critical skill, which people need to have to like stand out and, and deliver real value or as our, as our skills, which you think are also as important?
[31:51.840 --> 31:58.840]  I mean, there are definitely, you know, the kind of basic, you know, engineering skills of knowing how to like break a problem down into subproblems.
[31:59.720 --> 32:19.700]  Um, those are almost like table stakes, you know, um, but I think that, uh, you know, some people obviously do them better than others, but I think, um, when I see, you know, things that make certain, you know, you know, individuals, uh, able to, um, really kind of magnify the impact of their work and kind of how, how effective it is.
[32:19.700 --> 32:22.660]  I think communication is definitely a big part of it.
[32:24.520 --> 32:28.680]  The ability to work well with others is something that, you know, certain engineers, you know.
[32:28.680 --> 32:34.480]  Really, really thrive on and, but other people I'll find challenging for sure.
[32:35.480 --> 32:43.580]  And so you have already like visited the academic world about motion planning and now you're in the industrial world of motion planning.
[32:43.580 --> 32:48.520]  What do you think are like the biggest differences between them?
[32:48.760 --> 32:58.220]  And do you think there are some things which people in the academic world do wrong or maybe, maybe not wrong, but, but maybe differently or maybe not?
[32:58.680 --> 33:01.480]  Uh, something which produces revalue or so.
[33:01.480 --> 33:21.060]  Yeah, the differences are, are really big and this is true, you know, most feel it's not like this is unique to robotics, but like academia is often focused on, um, you know, pushing the boundaries on, on what is kind of theoretically possible and the industry is focused on, you know, things that, things that you can make, you know, economic sense right now.
[33:21.060 --> 33:27.400]  And, and, uh, and I think so, uh, you know, I think that what we've learned.
[33:28.680 --> 33:45.660]  The more we've learned about industrial robotics, the more kind of like, kind of basic and seems in terms of the problems that the problems they're solving are, are, are fairly seem kind of mundane when you look at it from an academic sense.
[33:45.660 --> 33:58.660]  You know, they're not, they're not so concerned about probabilistic completeness or, or any kind of like guarantees on, on, uh, an algorithm might provide, um, uh, they're a lot more focused on.
[33:58.680 --> 33:59.140]  Um, okay.
[33:59.160 --> 34:03.520]  Is this going to take me one day to set up or five days to set up?
[34:03.960 --> 34:09.280]  Um, the fact that you might guarantee to get set up after eventually, this is, this is interesting.
[34:09.680 --> 34:10.040]  Yeah.
[34:10.660 --> 34:27.280]  Uh, and so, you know, in particular, we've, I think what we've, what the kind of niche that we're currently focused on is, is multi, multi robot work cells and figuring out how to have multiple robots working really efficiently with each other.
[34:27.280 --> 34:28.320]  So yeah, not, not even.
[34:28.680 --> 34:43.280]  Taking like the human cow, kind of throwing all the human collaboration stuff out the window and just focusing on, okay, you have, you know, somewhere between, you know, two and 16 robots in a, in a space that is pretty tight because space is at a premium in these really big factories.
[34:43.280 --> 34:45.100]  So you're, you're pretty space constrained.
[34:45.580 --> 34:57.800]  Um, you're trying to move around in a very cluttered environment, you know, lots of machinery, lots of cables, lots of, you know, clamps and, and you have a lot of moving parts, literally moving parts.
[34:58.680 --> 34:59.120]  Um, you're trying to move around in a very cluttered environment, you know, lots of machinery, lots of clamps and, and you have a lot of moving parts, literally moving parts.
[34:59.120 --> 34:59.180]  Um, you're trying to move around in a very cluttered environment, you know, lots of machinery, lots of clamps and, and you have a lot of moving parts.
[34:59.180 --> 35:02.900]  And how do you keep them all collision free?
[35:02.900 --> 35:20.260]  And, um, that's a pretty different problem than we started on, but, uh, I think that's, that's kind of where, at least in the, you know, the automotive, automotive industry, which is one of our biggest, biggest kind of focus areas is right now is, is, you know, multi robot programming, which is just really time consuming to do conventional.
[35:20.540 --> 35:20.780]  Yeah.
[35:21.960 --> 35:24.140]  So let's talk a little bit more about products.
[35:24.460 --> 35:28.520]  So what would you say is like the product you are most proud of?
[35:28.680 --> 35:35.880]  Um, at real time or maybe the product which delivered maybe like the, the biggest value in the longterm.
[35:36.320 --> 35:36.460]  Yeah.
[35:36.600 --> 35:46.020]  So I'd say that the, what we're focusing on right now, which we're calling our resolver is a tool to do multi robot programming, uh, in an automated fashion.
[35:46.340 --> 35:57.520]  So it's a, it's a tool where you, you know, upload, you know, CAD data for your work cell, you know, which includes, you know, the points of interest that you need the robots to get to.
[35:58.040 --> 35:58.440]  Um.
[35:58.680 --> 36:04.840]  And can include, you know, constraints you might have where certain points might need to be visited for other points.
[36:05.520 --> 36:18.900]  Um, and, uh, and you get back, um, a really well optimized, uh, plan for, you know, all the robots, you know, um, interlock signals that kind of keep all the robots in sync.
[36:19.080 --> 36:28.660]  And avoid collisions and really take this really challenging problem of, okay, how do you, you know, cause you know, in these, in these, in these work cells, you know, you're going to have to, you know, you're going to have to, you know, you're going to have to, you know, you're going to have to do multi robot programming.
[36:28.680 --> 36:33.660]  cells, you might not only have, not only do you have multiple robots, but you have many
[36:33.660 --> 36:34.660]  points of interest.
[36:34.660 --> 36:39.920]  So in a spot welding cell, you might have each robot that needs to visit, you know,
[36:39.920 --> 36:47.720]  15, you know, multiply that by 15 robots and, and you kind of have a lot of different orderings
[36:47.720 --> 36:49.640]  that might work out.
[36:49.640 --> 36:53.640]  So the combinatorics work out so that it would just, a human programmer just can't try all
[36:53.640 --> 36:54.640]  of them.
[36:54.640 --> 36:59.240]  And we have a product that will explore that state space for you pretty quickly and give
[36:59.240 --> 37:03.140]  you a result that is, you know, you know, very, very good.
[37:03.140 --> 37:08.420]  Can you maybe walk me through like, so imagine that I'm like a manufacturer, I have like
[37:08.420 --> 37:12.820]  a lot of robots, I want to like, I don't know, assemble something together.
[37:12.820 --> 37:18.220]  And so what do I actually need to provide in order to make it work?
[37:18.220 --> 37:24.620]  Yeah, so at the bare minimum, you need CAD data for your work cell.
[37:24.620 --> 37:30.780]  Which luckily most, most customers building these complicated work cells already have
[37:30.780 --> 37:34.780]  because they're, they're, the people they're using to fabricate the equipment, everything
[37:34.780 --> 37:36.100]  they need, they need this data.
[37:36.100 --> 37:40.620]  So you need, and this usually comes to us in the form of like a step file or, or JT
[37:40.620 --> 37:41.620]  files.
[37:41.620 --> 37:45.420]  So you need CAD data for the work cell, you need kinematic descriptions of the robots.
[37:45.420 --> 37:51.760]  So you know, because there's a lot of different robot manufacturers out there, you know, they
[37:51.760 --> 37:53.120]  have different ways they move around.
[37:53.120 --> 37:54.420]  So you need the kinematic descriptions of the robots.
[37:54.420 --> 37:59.260]  You need to be able to tell us where your points of interest are.
[37:59.260 --> 38:02.080]  So where you want the robots to go.
[38:02.080 --> 38:07.140]  And yeah, and we, you know, accept all this data through kind of very common, you know,
[38:07.140 --> 38:08.140]  REST APIs.
[38:08.140 --> 38:13.420]  And so if you have an account with our, with us, you know, you can send that to us on the
[38:13.420 --> 38:19.820]  cloud and we'll, we'll crunch the numbers and get back some motion plans for you.
[38:19.820 --> 38:23.580]  And what do you think are the most important factors which actually shape a success?
[38:23.580 --> 38:36.660]  Well, I guess you have to listen, listening to the customer, um, uh, what kind of sh so
[38:36.660 --> 38:40.700]  I guess, you know, I guess you have to get early feedback because you, you'd rather you
[38:40.700 --> 38:44.620]  hear early on if you need to shift the product, um, but a successful one is going to be one
[38:44.620 --> 38:48.540]  that the customer is satisfied with and gets value from, you know, not just one that, you
[38:48.540 --> 38:51.420]  know, even if you sell the first two and you make money, but if the customer is not getting
[38:51.420 --> 38:53.380]  value from it, you're not going to sell anything.
[38:53.380 --> 38:54.380]  And that's the first two.
[38:54.380 --> 38:58.740]  And so you want to make sure that the, uh, the product is bringing value to the customer.
[38:58.740 --> 39:03.500]  So I'm understanding what the pain points are, you know, in the industry.
[39:03.500 --> 39:06.660]  And, and that was one of the reasons, one of the reasons why we're focusing on this
[39:06.660 --> 39:10.400]  now is that, you know, from our early talks, we ended up talking with different people
[39:10.400 --> 39:17.060]  and we kind of got guided from our initial thinking of, you know, dynamic obstacles,
[39:17.060 --> 39:20.340]  you know, dealing with that, we, we would pitch this to people and they'd be like, Hmm,
[39:20.340 --> 39:23.380]  that's not really something I'm looking for right now, but I have this big problem.
[39:23.380 --> 39:27.980]  That it takes me, you know, six weeks to program this, um, work cell that six weeks at the
[39:27.980 --> 39:30.020]  factory is not making cars.
[39:30.020 --> 39:34.620]  And if I could turn that six weeks into two weeks, that would help us out a lot, you know?
[39:34.620 --> 39:38.680]  Um, and so that's kind of how we got to where we are now is learning about the customer
[39:38.680 --> 39:42.700]  pain points, um, and then kind of diving in deep and understanding, you know, why is this
[39:42.700 --> 39:43.700]  a painful process?
[39:43.700 --> 39:45.940]  Can we do something to help resolve that?
[39:45.940 --> 39:50.380]  Um, and then as we start to develop, develop a product, you know, getting early feedback
[39:50.380 --> 39:53.380]  on how the product works, if it's doing what the customer expects.
[39:53.380 --> 39:58.100]  Cause then you'll hear back, you know, oftentimes you'll hear one thing and you'll address that
[39:58.100 --> 40:02.380]  one thing, but then you'll realize that there was like a few other hidden requirements that
[40:02.380 --> 40:03.380]  you didn't, that you didn't know about.
[40:03.380 --> 40:07.740]  And I think that's, that's definitely been a common theme is that it's quite hard to
[40:07.740 --> 40:10.860]  kind of extract all of the requirements for the product.
[40:10.860 --> 40:15.100]  The first go around because the customer maybe thinks that there's one problem and that problem
[40:15.100 --> 40:19.060]  does exist, but there's also kind of like four or five related problems that because
[40:19.060 --> 40:22.600]  they had this first problem, they didn't think bring up those four or five related problems.
[40:22.600 --> 40:26.320]  But, um, but those always come to the surface eventually.
[40:26.320 --> 40:30.660]  And do you think there are other current challenges in the industry, but real time could maybe
[40:30.660 --> 40:33.720]  address in the future or maybe even right now?
[40:33.720 --> 40:37.720]  I mean, there's definitely a lot of challenges in the robotics industry.
[40:37.720 --> 40:43.020]  Um, uh, I think that, um, we'd like to, to get to, uh, you know, there's a lot of ones
[40:43.020 --> 40:46.560]  that we'd like to get to, you know, I think that we, we definitely, um, have planned,
[40:46.560 --> 40:52.480]  you know, have plans in the long term to get to tackle more of the dynamic, um, um, you
[40:52.480 --> 40:56.740]  know, work cells where, where things are kind of changing every cycle.
[40:56.740 --> 41:06.140]  There's a lot of, a lot of the big players in the robotics industry, which traditionally
[41:06.140 --> 41:09.720]  have been automotive companies are fairly slow to adapt.
[41:09.720 --> 41:17.860]  So I think that, um, as they start to use more kind of like cutting edge technology,
[41:17.860 --> 41:22.280]  they'll probably be open to using more, but kind of get having them modernize their processes
[41:22.280 --> 41:22.400]  is.
[41:22.480 --> 41:28.520]  Kind of a slow, slow, slow to happen.
[41:28.520 --> 41:35.620]  And so, so do you think that real time is moving also to what's like a long-term goal
[41:35.620 --> 41:40.300]  or would you say that it makes more sense to actually go product by product and just
[41:40.300 --> 41:45.820]  look, I would want to head, I mean, you want to have a long-term goal, but I think, um,
[41:45.820 --> 41:51.740]  you also have, um, you know, short term.
[41:52.480 --> 41:55.720]  Business goals that you need, they're kind of hit to, you know, keep the company going.
[41:55.720 --> 41:59.720]  And so we definitely have, you know, much broader ambitions for the long term.
[41:59.720 --> 42:04.200]  Um, I think for the, for the, for the near term, we're really focused on, on this product
[42:04.200 --> 42:09.060]  and, and getting wide adoption in the market and continuing to improve it because there's
[42:09.060 --> 42:13.160]  still a lot, a lot of things we want to do on this product and yeah.
[42:13.160 --> 42:17.280]  And so, so I already mentioned that real time is almost like 10 years old.
[42:17.280 --> 42:21.320]  So where do you think real time will be in another 10 years time?
[42:21.320 --> 42:22.320]  Hmm.
[42:22.320 --> 42:23.320]  Good question.
[42:23.320 --> 42:24.320]  It's hard to say.
[42:24.320 --> 42:25.320]  Um, I hope I'm sorry.
[42:25.320 --> 42:30.180]  I think that we, uh, we have great hopes in our, in our current kind of what we're doing
[42:30.180 --> 42:31.180]  right now.
[42:31.180 --> 42:36.540]  So what I, what we'd love to see is in 10 years that, um, uh, everyone programming multi
[42:36.540 --> 42:40.080]  robot work cells is doing so with the help of, uh, with the help of our tool set.
[42:40.080 --> 42:46.500]  Um, I think that, you know, we, we're pretty confident that it's, you know, the best thing
[42:46.500 --> 42:50.160]  on the market right now for, for multi robot programming.
[42:50.160 --> 42:51.560]  And so we'd love to see that, um, kind of achieving real time.
[42:51.560 --> 42:52.080]  Yeah.
[42:52.320 --> 42:58.800]  I think that's a really, you know, good market market reach and, uh, and, and hopefully at
[42:58.800 --> 43:05.200]  that point, we've also kind of really managed to, um, uh, get back into the space of done
[43:05.200 --> 43:10.460]  up work cells kind of more, more kind of semi-structured work cells where, you know, we're seeing a
[43:10.460 --> 43:16.080]  lot more people doing volume, high variability, you know, manufacturing where you might be
[43:16.080 --> 43:21.180]  making, um, a certain part for a couple months and then you want to retool and make a different,
[43:21.180 --> 43:22.180]  um, different part.
[43:22.320 --> 43:23.320]  Yeah.
[43:23.320 --> 43:26.320]  And I think that with our, with our, your products, you can, you can do that a lot,
[43:26.320 --> 43:27.320]  a lot easier.
[43:27.320 --> 43:28.320]  Mm-hmm.
[43:28.320 --> 43:37.320]  And do you think that the advent of, of like AI tools has changed a lot as a, as a, I don't
[43:37.320 --> 43:42.320]  know, there's a landscape maybe of problems or maybe the company itself.
[43:42.320 --> 43:48.640]  Um, I don't mean, I, I, I think there's a lot of companies out there figuring out how
[43:48.640 --> 43:49.320]  to use, you know, all the, all the recent advances.
[43:49.320 --> 43:50.320]  Yeah.
[43:50.320 --> 43:51.320]  Yeah.
[43:51.320 --> 43:52.320]  Yeah.
[43:52.320 --> 43:53.320]  I think there's a lot of advances in it.
[43:53.320 --> 43:56.800]  So I think that where it's affected probably the, the industries that we're working with
[43:56.800 --> 43:59.040]  the most is in the computer vision space.
[43:59.040 --> 44:03.500]  I think that, you know, the advances in computer vision have, have really enabled a lot of
[44:03.500 --> 44:11.560]  more stuff in logistics, you know, doing, um, kind of dynamic, um, you know, palletization,
[44:11.560 --> 44:15.840]  multi-skew palletization, um, really requires, uh, good computer vision.
[44:15.840 --> 44:20.320]  And I think in the past, it's probably only been in the past, you know, you know, it's
[44:20.320 --> 44:21.320]  hard to say.
[44:21.320 --> 44:24.320]  I don't know if it's five years, 10 years or 15 years, but fairly recently has the computer
[44:24.320 --> 44:26.800]  vision become available to kind of do that really well.
[44:26.800 --> 44:31.260]  And I think that's, that's where I've seen the most, um, I, I, I mean, I don't think
[44:31.260 --> 44:36.320]  the kind of more recent, you know, generative AI stuff has, has affected our customers as
[44:36.320 --> 44:41.120]  much yet, you know, probably will eventually, but, um, I'm, I'm still kind of waiting to
[44:41.120 --> 44:43.840]  see what the big impacts will be there.
[44:43.840 --> 44:49.460]  So do you think that robotics will be revolutionized by, by large language models or something
[44:49.460 --> 44:51.040]  similar in robotics or my.
[44:51.040 --> 44:52.040]  Foundational models.
[44:52.040 --> 44:56.040]  I'm more of a skeptic there.
[44:56.040 --> 45:01.540]  I think that, um, maybe at the layer of, uh, communicating intention, you know, so maybe,
[45:01.540 --> 45:06.660]  maybe, um, at like the layer where they interface with, with humans might, they might, might
[45:06.660 --> 45:13.940]  make advantage of, of, of LMS, uh, but, uh, I'm, I'd say I'm definitely more of a skeptic
[45:13.940 --> 45:15.860]  on, on, on what's going to happen there.
[45:15.860 --> 45:16.860]  So we'll see.
[45:16.860 --> 45:20.960]  So you don't think that tools like for motion planning, which are already like a
[45:20.960 --> 45:27.460]  established, they will not be, I don't know, gone out of business business.
[45:27.460 --> 45:28.460]  I don't.
[45:28.460 --> 45:29.460]  Yeah.
[45:29.460 --> 45:30.460]  I don't think so.
[45:30.460 --> 45:33.720]  I think that they'll, I, I think that they'll prove their usefulness over the longterm,
[45:33.720 --> 45:39.120]  you know, anything can happen, but, uh, yeah, I don't, I don't think motion planning is
[45:39.120 --> 45:40.120]  going anywhere.
[45:40.120 --> 45:41.120]  Yeah.
[45:41.120 --> 45:42.120]  Okay.
[45:42.120 --> 45:46.960]  And so, so yeah, I, I think that was already like a long journey, like almost 10 years.
[45:46.960 --> 45:50.960]  Um, but is there maybe like some piece of advice, which you give, which you give people?
[45:50.960 --> 45:56.960]  Like if you're younger self, if you could go back in time and say, okay, do this before
[45:56.960 --> 45:59.700]  you actually go and start a startup.
[45:59.700 --> 46:05.240]  We definitely, we definitely, if we, with, with knowledge, with all the knowledge we
[46:05.240 --> 46:07.720]  have now, we would have made very different decisions.
[46:07.720 --> 46:14.860]  Um, I think that usually when we've made changes, we've probably on average made them a few
[46:14.860 --> 46:16.660]  months later than we should have.
[46:16.660 --> 46:19.460]  And so I guess one piece of advice would be, don't be afraid to change quickly.
[46:19.460 --> 46:20.460]  Um, you know, it's, it's, it's, it's, it's, it's, it's, it's, it's, it's, it's, it's, it's,
[46:20.460 --> 46:23.920]  you know, it's, it's a balance though, because you want to focus on things and see them through.
[46:23.920 --> 46:29.040]  We also want to be open to feedback, you know, from customers, from the market, from, you
[46:29.040 --> 46:34.340]  know, industry, if you need to, you know, change that, you want to hear that feedback
[46:34.340 --> 46:38.260]  early and change early rather than, than, than, you know, wasting time on a route that
[46:38.260 --> 46:40.000]  might not be the right one to go down.
[46:40.000 --> 46:44.920]  So, so you would say like pivoting towards it and like a new idea is also super critical
[46:44.920 --> 46:49.300]  instead of like trying very hard to like push your products through.
[46:49.300 --> 46:50.300]  Yeah.
[46:50.300 --> 47:08.540]  I think that, you know, our first couple of years, I mean, a couple of years, but like, I think that we definitely were kind of biased in our thinking the first couple, you know, first while, like, you know, with a vision of what we thought we wanted to make and should have probably been, you know, listening a bit more early and kind of pivoting faster.
[47:08.540 --> 47:15.120]  So I think that's probably a generally applicable piece of advice is to don't be afraid to adapt what you're doing to feedback that you get.
[47:15.120 --> 47:34.960]  Cool. And then I also would like to talk a little bit about the social impact of automation. And so people are sometimes a bit anxious, of course, of robotics. They fear that they are losing their jobs. Do you think this is a valid concern? And what are maybe possible solutions to that?
[47:35.180 --> 47:45.100]  I mean, it's definitely a valid concern. I never want to dismiss anyone's concerns. I think that, you know, from the time they, you know, you know,
[47:45.120 --> 48:15.100]  things to make textiles, which is probably the earliest, you know, forms of automation, you know, there's been people displaced by automation. And I think that, you know, it's never great when someone, you know, loses a job. But I think that there's a lot of opportunities. There's a lot of jobs as well. And there's almost no unemployed robot programmers right now. So, you know, it's no consolation to someone who, you know,
[48:15.120 --> 48:26.900]  just wants to do welding and not program robots. But I think that, you know, the world we live in is, you know, is going to change, and it's going to change rapidly. And I think that, you know, yeah.
[48:27.920 --> 48:32.420]  And so you think that people could also then move on to something bigger and better?
[48:32.740 --> 48:44.360]  I mean, I think that, I think that, yeah, I'm not trying to imply that everyone should just like kind of give up what they're doing. But,
[48:45.120 --> 49:04.600]  you know, technology is going to advance. And it's, it's, I think it's, you want to do so in a responsible manner. But as a, as a company, you know, we just try to, you know, abide by all the all the regulations that are set out. And yeah.
[49:05.060 --> 49:12.180]  And so what are maybe some jobs which you already foresee, which could be become obsolete in the future?
[49:12.180 --> 49:14.680]  Huh, that's a good one. I don't know.
[49:15.120 --> 49:31.000]  I think there'll certainly be things that we look back on. It'll be it'll be funny that humans ever did that. But I don't know if I have any unique insight there. But there's a lot of a lot of dangerous and pleasant jobs out there that we could probably automate automate away.
[49:31.000 --> 49:45.020]  I guess some people also believe that once we automate everything away, we need some kind of like basic universal income or so. And do you also believe that basic universal income is a good thing? And maybe a remedy?
[49:45.120 --> 49:48.040]  Some, some of the problems we have within our workforce?
[49:48.060 --> 50:00.880]  Yeah, I'm certainly not opposed to it. I'm not I'm definitely not like a, you know, an economics expert. But I mean, if a country is thriving, and has developed be something like that, I don't see a reason why not to.
[50:01.840 --> 50:05.400]  So you're not like very capitalistic in this point. And
[50:06.500 --> 50:14.820]  I mean, I'm, I would say that I'm not, I don't have any unique
[50:15.640 --> 50:23.620]  I don't think I have any special thoughts on the topic. I think that it's, it's an interesting one I've heard a lot about. But yeah, I think it's a it's very intriguing.
[50:24.220 --> 50:34.560]  And maybe let's talk a little bit about how you structure your day. Are they maybe like some some daily habits which you do or something which I don't know sets you up for success?
[50:34.620 --> 50:45.100]  I try to at least stay organized, I definitely try to, you know, because in a startup, you're going to have an endless array of tasks to get done. And so I, I try to at least
[50:45.220 --> 50:55.920]  take notes on things that I need to get done on a week to week basis, you know, and I have kind of a way of prioritizing, like kind of making sure I highlight which items are kind of important.
[50:57.420 --> 51:14.820]  So I'll usually start each day and like, take a look at, you know, what I need to get done that week. And, and what's kind of risen to the top of the priority list. One thing that I definitely try to focus on is making sure people aren't blocked by stuff that I'm responsible for. And so I try to make sure that if, you know, people are waiting on
[51:15.120 --> 51:34.120]  product guidance, or kind of like technical reviews or something, I definitely try to make sure that people aren't stalled because of stuff that I have to do. So those are usually things that I prioritize, or if someone else's work is waiting on me, like checking in, and somehow whether that's feedback or review or something like that, I try and get that stuff done first.
[51:34.520 --> 51:44.820]  Yeah, I guess we are all a bit in this profession of where we are all a bit more despondent than other people. And so how do you actually combat that? Do you do something against that?
[51:45.120 --> 52:04.320]  Yeah, I mean, yeah, I mean, I definitely focus better if I've, you know, been getting exercise. That's definitely true. I think most people probably benefit from that. And so yeah, after me after work, I try to, you know, get some form of exercise, you know, so it's, you know, some some running some stuff, stuff outside.
[52:04.320 --> 52:08.160]  So you have like some some goals, which you want to achieve?
[52:08.460 --> 52:12.940]  I'm not super goal oriented when it comes to like hobbies, I used to be.
[52:13.760 --> 52:14.920]  But I
[52:14.920 --> 52:15.000]  I
[52:15.000 --> 52:15.020]  I
[52:15.020 --> 52:15.100]  I
[52:15.100 --> 52:15.120]  I
[52:15.120 --> 52:21.340]  found myself getting injured more often. So I try to just do things that I have fun, I can have fun with, you know, kind of relax.
[52:21.620 --> 52:26.060]  And so what, what kind of exercises do you do on a daily basis?
[52:26.300 --> 52:44.400]  On a daily basis, you know, I like go on trail runs, I really enjoy kind of like running around in the hills. I have some friends that I'll, you know, work out with. It's always fun to get, especially, you know, a lot of people work remotely these days. And it's good to get some some human contact, you know.
[52:44.400 --> 52:53.580]  And so, so what else do you do? Like, so I guess Montana, you could also like do some skiing there or climbing.
[52:53.580 --> 53:07.820]  There's some, there's some really good skiing. We definitely, you know, do that. You know, rock climbing is also a lot of stuff, a lot of that stuff around as well. I definitely try to take advantage of all the stuff that's around us.
[53:08.640 --> 53:11.700]  Cool. Well, then thank you for coming by.
[53:12.220 --> 53:12.540]  Of course.
[53:12.800 --> 53:13.920]  It was a pleasure for me.
[53:13.920 --> 53:14.360]  Excellent.
[53:14.400 --> 53:16.560]  Excited, excited to do it, Andreas. Yeah, happy to chat.
[53:16.720 --> 53:17.820]  Cool. Yeah. Thanks.
[53:18.180 --> 53:18.440]  Of course.