October 2, 2025 • 31 mins
Boris Sofman is the co-founder and CEO of Bedrock Robotics.
Boris' problem is this: How do you teach machines not just to drive, but also to work: to grade roads, move heavy objects and dig big holes at construction sites.
On today’s show, Boris talks about how his work at Waymo led him to found Bedrock, and he explains how autonomous construction equipment could help unleash an American building boom.
See omnystudio.com/listener for privacy information.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:15):
Pushkin.
Speaker 2 (00:20):
When we started this show in twenty twenty two, the
standard line about driverless cars was driverless cars have been
five years away for the past fifteen years, because it
seemed like they were just always around the corner, always
just a few years away, but they never quite arrived.
Nobody says that anymore. Today, in several cities around the country,
(00:43):
getting a ride from a driverless car is just a
normal thing people do, and it'll become normal in more
and more and more cities over the next few years.
Driverless cars are here now, so now we can ask
what's next. I'm Jacob Goldstein and this is What's Your Problem,
(01:05):
the show where I talk to people who are trying
to make technological progress. My guest today is Boris Soffman.
He's the co founder and CEO of Bedrock Robotics, a
company that's figuring out how to retrofit heavy equipment to
make it work autonomously. Boris's problem is this, how do
you teach machines not just to drive, but to do
(01:25):
things like grade roads and move heavy things around construction sites.
Boris's company is starting with excavators, and they plan to
have their first commercial excavators autonomously digging holes on construction
projects next year. Later in the interview, Boris goes big.
He argues that if Bedrock succeeds, the company could help
(01:47):
push forward a broad wave of new building in America.
But first we talked about his time at Weimo and
how the wild evolution of the autonomous vehicles that he
worked on there led him to start Bedrock.
Speaker 1 (02:01):
The time at Weimo was this incredible period where I
was there for about five years from mid twenty nineteen
till through spring twenty four, and that was this really
big period where it was going through this like fifteen
years of R and D, and then it finally transitioned
into this hockey stick of like growth that is helping today.
And so today waymos at over one hundred million miles
(02:22):
fully driver lists. It's at five times safer than a human.
It's millions of miles every single week, and so it's
kind of scaling itsponentially and it's like a genuinely fantastic product.
But I was there when we were like stressing over
the first hundred miles, and it felt like the most
incredible achievement to just go like ten fifty hundred miles
to completely drop less, say now that happens at like
hundreds of thousands of miles every single day. And so
(02:44):
one of the things that really broke through it made
that possible is this shift to machine learning and data
driven approaches as a core of the autonomy stack.
Speaker 2 (02:52):
And just to be clear, like that's as opposed to
a more like euristics or rule based kind of model,
like the old school twentieth century eyes, like, well, just
tell the car all the rules of how to drive
and then it'll drive. Like that's what you're comparing machine
learning to.
Speaker 1 (03:08):
Yeah, like drive, I'm going to like embed the cost
functions and all this. So yeah, so like large scale
search and heuristics and rules and you can embed them
now inside those heuristics, and so you can solve almost
any given problem with that sort of approach.
Speaker 2 (03:21):
So you can solve one problem, but you can't solve
like every single possible problem that would ever arise when
you're driving, which is actually what you have to solve
to do full autonomy.
Speaker 1 (03:31):
Right right with activating whackamall, where like you fix one
problem and it becomes harder and harder to kind of
scale the other ones. And so and so there was
this really conscious shift at WEIMA, which was kind of
bold at the time because it feels obvious in hindsight,
It wasn't obvious at all back then. The shift to
like really embracing this as a data driven solution where
you're learning from human driving and human behavior.
Speaker 2 (03:52):
And when you say you're learning, you mean the machine
learning model. The AI basically is learning.
Speaker 1 (03:58):
The machine learning model, that's right. And so you're basically
taking giant scales of data one hundreds of thousands of
millions of miles, and you're learning the model of how
do you drive and how do you interpret this? Like
infinite compocity and kitchen sink of contacts, like all of
this sensored data, the road structures, the things you see
around you, the way people are moving. How do you
go from a kind of engineered and you know, kind
(04:20):
of trained solution into one that is like dominantly a
warned solution.
Speaker 2 (04:24):
I mean, as you're doing that, you're sort of riding
the historic machine learning AI wave, right, Like that's right.
Presumably you're able to do that at that moment because
of this explosion in AI, which is basically machine learning, right,
you know that.
Speaker 1 (04:38):
And you couldn't have done that five ten years ago.
Speaker 2 (04:40):
That's what you're kind of coming out of at way MO,
that's driving this success at way MOO. How does that
set you up for what you're doing a bedrock?
Speaker 1 (04:48):
The most shocking thing was just how well does generalized
from San Francisco to Los Angeles, Phoenix in Austin and
then becomes almost like a qualification problem eventually, where you're
using data to fill some gaps, but like you need
less and less of it, less and less new things
surprise you, like you're just your competency kind of expands.
And then we unified the technology stack between cars and
trucks where even jumping from a cargo a truck was
(05:09):
needed maybe ten to fifty percent more data, but it
was fundamentally like you're using data to explain why how
you operate a very different potages.
Speaker 2 (05:15):
This is like a big truck, This is like an
eighteen wheel truck. When Google was working on that, yeah,
are way most like.
Speaker 1 (05:19):
Fifty three foot trailer like eighty thousand pounds. And so
that was the big moment where we started thinking about
where else can you apply this? What are the places
where you have all this diversity of challenges and capabilities
that benefit from this type of versatility and also have
the ability to jump between platforms in this really natural way.
And so we looked at a lot of spaces and
(05:41):
really settled on automation of specialized typing machinery, and so
the sipes of machines that you see in construction, like
excavators and wheeloaders and bulldozers, but also frankly the sort
of machines you see in all sorts of industries like
agriculture and mining and lumper and garbage movement. And so
you have these very diverse types of machines that are
interacting with the world around them. They're fairly still moving,
(06:03):
they're in semi controlled environments, and at the end of
the day, there's astronomical scale at which they great at
and a lot of the learnings that we experienced a
weymle actually transfer over incredibly well. But the physics of
the problem is a lot less aperspherial. It's actually like,
really tackle this and get the market.
Speaker 2 (06:19):
Basically, big vehicles operating in semi constrained environments doing things
to the world, interacting with the world in some physical
way in addition to just driving across it.
Speaker 1 (06:30):
That's right, and these are slow moving vehicles where you're
already on a closed site with people who are assumed
to be knowledgeable about the world around you, and you're
moving at like five off an hour, for example, You're
able to slow down and stop. You're able to minimize
your exposure. You always have a minimum safety condition of stopping.
Your complexity is less abound interactions with others on the
road and more about the interactions with the world around you.
(06:52):
And so you can actually tackle safety not through this
sort of statistical methods that drove a lot of the
mileage that we collect, but through a much more direct
measure of your competencies in order to just make sure
that you're like you're actually capable of hurting somebody.
Speaker 2 (07:08):
Yeah, more like like an industrial robot al most right.
Speaker 1 (07:12):
That's right. Yeah, it's like a traditional Susis engineering, Yeah.
Speaker 2 (07:15):
Where it's like, look, even if it can't do the
thing every time, that's fine. Just make sure it's not
gonna like go crazy and kill somebody.
Speaker 1 (07:21):
That's right. You can make it to that. Like your
worst case is your productivity suffers, but safety wise, you're
you're you're solid. And so that's like incredibly enabling because
your long tail is now no longer safety. It's the
versatility what you can do.
Speaker 2 (07:33):
Why did you start with excavators?
Speaker 1 (07:36):
So excavators are the most highly utilized machine, Like they're
usually the highest volume in fleet. So between twenty and
twenty five percent of fleets are excavators.
Speaker 2 (07:45):
Fleets are just like big construction heavy equipment.
Speaker 1 (07:48):
Yeah, like a general contractor that alone, like a thousand
machines two hundred, two hundred and fifty will probably be
excavators on average.
Speaker 2 (07:55):
And it's basically what a kid would call it digger, right,
It's like at a digger bucket an arm, and it
like digs stuff up.
Speaker 1 (08:01):
It's like the equivalent of an arm. Like you can
like dig stuff, you can demolish stuff, you can swap
your tools, you can like lift pipes and put them
in a holes. They can do a ton of stuff
with it. It's kind of crazy. It really is a
personal machine. It also makes it very complicated to work.
So there's like seven degrees of freedom sometimes eight, and
so it's one of the hardest machines to learn, and
it takes four to five years to really become an expert.
And there's a huge difference between an expert and novice,
(08:22):
and so you kind of have this situation where it's
a huge volume of work and it's really hard to learn,
and so you have a really deep pull in the
market for it, meaning.
Speaker 2 (08:31):
A lot of demand, like a lot of people want
somebody who can drive one of these or a machine
that could do it.
Speaker 1 (08:37):
Well, let me tell you about the demand. I've never
seen such a divergence of supply to demand like in
my career in any case, where on the demand side,
you have this astronomical construction industry that's already two trillion
dollars a year in the US, that's obviously very heavily
building and having machinery work. And then you have this
like shortage of operators that already existed, but it's going
in the wrong direction where forty percent of construction workers
(08:59):
is retiring in the next ten years. Our partners are
consistently having trouble filling labor. We've met some that have
one hundred percent turnover.
Speaker 2 (09:09):
Everybody leaves every year.
Speaker 1 (09:10):
It means like a third of people might be lifers,
but then like two thirds transition more than once per year,
and you're constantly backfilling the skill sets very a ton
and one of them said that for every one person
entering the workforce of the quality that they look for,
their seven leaving. And so we end up having is
this shortage of ability to meet this demand, and so
(09:30):
prices go up, jobs don't get done. And what's interesting
is it's not that's like isolated industry that's like just
on its own kind of like having these sort of challenges.
It's a horizontal it supports every industry. You can't build
data centers free eye without it, you can't build houses,
you can't build energy facilities.
Speaker 2 (09:46):
Of like a rate limiting skill or rate limiting machine.
Speaker 1 (09:49):
It's a whole country. That's exactly right. And then you
have this need that starts with labor. But then there's safety.
It's huge amounts of safety challenges. You have, you know,
huge predictability challenges. So if you actually could soften out
some of these constraints, we would build more, more work
would get done, and it would like stimulate the whole economy.
And so that's what's actually pretty exciting about this opportunity.
(10:09):
It's not as yours some game at all good.
Speaker 2 (10:12):
So, like you decide to focus on excavators, you you know,
you raise money for your company, Like, do you go
out and buy a whatever, a million dollar excavator. You
go to what is it bucket and shovel dot com
and buy yourself an excavator.
Speaker 1 (10:25):
They're like three hundred thousand to five hundred thousand, so
they're like still expensive. Yeah, they're pretty expensive.
Speaker 2 (10:30):
So I mean like, did you buy one like we did?
Speaker 1 (10:32):
Yeah?
Speaker 2 (10:33):
Did you drive it?
Speaker 1 (10:34):
Of course, like you have to. It's like it's like
a ride of passage. Yeah, they're really fine. I even
took a lessoners. There's a place in Vegas ALESSI drive
excavators and take away and from like a trade operator.
It was pretty fun.
Speaker 2 (10:44):
So, oh, that's genius. You get to break things. I
was actually as I was walking here today to the
train that there's a playground and there was an excavator
just like breaking up the asphalt and then like prying
it up. I was like, that looks awesome.
Speaker 1 (10:59):
This is the funniest thing. Anybody that gets an excavator,
like your inner six year old comes out. Suddenly you
have like the most mature, sophisticated person is trying to
be all professional. You get an excavator and just like
get like a giant glob of mud and then like
bring as hig as a can and then like PLoP
it down and see what happens. It's amazing, and it's like,
without fail, everybody kind of reverts back to this sort
of a world. Yeah, we drove and I think we
(11:20):
bought like a half dozen excavators at this point, and
then we also then use a lot of excavators from
our partners who are general contractors and subcontractors.
Speaker 2 (11:28):
And it's just what a year or so ago that
you started, like not that long in.
Speaker 1 (11:32):
This less than a year and a half, yeah, like
last last minute, seah, it's a pretty been a good run.
Speaker 2 (11:37):
How much is sort of commodified of autonomy, right? How
much is just like, well, we're gonna buy this, this,
and this in terms of sort of hardware and we
know the software, and then how much is like, oh,
here's the things we have to figure out that nobody
knows how to do.
Speaker 1 (11:49):
So it's one of the other enablers that's like way
better than what we would have had to go through
ten years ago. In the space, we can use a
lot of the existing components on lighter on cameras, on
Imus GPS, there's a lot of tailwind from automotive great
cameras and compute.
Speaker 2 (12:05):
That's like accelerating and like presumably they're buying those things
at scale you can like you've got to go get
them cheap. Essentially, you're like, give me one of those,
one of those, one of those.
Speaker 1 (12:14):
That's right, because you're gonna go to millions and millions
of units as like every car against an autopilot equivalent
over the next two three years. Oh, that's starting to
kind of pump in a cost. So that helps. And
then even the platform, we can retrofit existing machinery.
Speaker 2 (12:27):
When you say the platform, what do you mean in
this context.
Speaker 1 (12:30):
Platform is like the car, the truck, the excavator, and
the way in this case the excavator. Yeah, the machine,
so like the excavator itself. Construction machines, particularly this latest generation,
they're really well designed to where they're already effectively drive
by wire, which means that every signal going through the
machine is electric, and so we're able to spice into
it and both read and write to these signals.
Speaker 2 (12:54):
And control them electric meaning like computerized basically.
Speaker 1 (12:58):
Meaning like you have a joystick that operates the excavatory
that's not physically connected to the hydraulics. It's an electric
signal that's connected to the hydraulics, and so the same
for every signal and the machine, like the sense the
pressure gages, and so we're able to both get the
data from the machine as well as control the machine
through a non invasive integration where we can upit a
(13:20):
machine with our sensors and compute suite in like less
than three hours and make it autonomy capable and it's
completely reversible. We could never do that with a car
or a truck because the platforms were just not designed
this way.
Speaker 2 (13:32):
Yeah, so there's a lot that's there already. The excavators
themselves are sort of easy to integrate with lots of
off the shelf technology. What's not there, Like, when you're
coming in, what do you have to sort of build
that nobody has built before?
Speaker 1 (13:44):
Nobody has actually created autonomy that can solve the really
nuanced problems that you need to solve in order to
operate like an excavator in construction tasks.
Speaker 2 (13:55):
So let's talk like specifically, you buy your excavators, you
got your hardware, you know, whatever, what basic AI model
you're going to use, But like, what's a specific thing.
You have to figure out well a.
Speaker 1 (14:07):
Lot of things. So first of all, it's not trivial
to tap into these machines and to build a platform
or kind of an integration around it where you can
read them, you can control them, and so you have
to basically create like a wrapper around the machine and
also design in a way that scales later to new machines. Right,
So that part's hard. It's a big autonomy problem. So
in that respect is no different than what we tackle
(14:27):
with a WEIMO, where you have to train massive scale models.
You have to collect a huge amount of data.
Speaker 2 (14:31):
And in this case it's like about the digging, like
it's about the digging dumb question? Is the digging the
hard part? Like what happens when the thing hits the ground.
There's different kinds of stuff in the ground.
Speaker 1 (14:41):
Like tell me in the category of what's new, suddenly
a car doesn't change the environment around it. You have
a really complicated tough earth and soil, and you got
to figure out the physics of how you dig through it,
how do you deal with clay versus top soil, how
do you deal with rocks inside? Then so, now if
you want to use simulation to solve parts of these problems.
You have a very complicated simulation problem because you have
(15:03):
to solve not just the sensor side, but also the
manipulation side. You have to structure this problem in a
way where you're learning from the data you collect in
order to actually capture the nuances of how do you
actually interact with the environment. You have to then control
and execute it the right way. You have to think
about how do you actually define the goal?
Speaker 2 (15:22):
So from the user point of view with an excavator,
like how does it look like? I just I want
to dig a hole of this size at this spot?
Or like what is it?
Speaker 1 (15:30):
Yeah, and this is a journey like fast forward you
know next year where card customer does this right, So
what they would do is they would give it a
model of what they want the earth to be dug
towards and so that would be like a three D
representation of the depth. So with length depth, here's the edges,
here's the constraints. The pickups for the trucks are going
(15:50):
to be over here and then go to work. And
basically these projects, like these machines will work for many
months at a time, continually just digging earth and working
it towards the right foundation, and the system will respect
that boundary and basically dig down to that level. It'll
have precision to the edges that you've defined, it'll do
it in a sequence that makes sense so that you
don't trap yourself in a hole, for example. And then
(16:12):
you specify where you're going to have dump truck pickups.
Then there's projects where this happens for nine months straight
or twelve months straight with like many machines, and so
what you basically need to specify is what you want
to dig to, and that what you want to dig
to ends up basically being the foundation that you're working
towards for whatever you're going to construct there. And we
(16:33):
want that to be versatile. So sometimes you're just taking
off a layer of top soil. Sometimes you're digging eight
feet deep. Sometimes you're taking an existing stockpile and moving
in somewhere else. So there's a lot of permutations of this.
Sometimes you're taking rubble from a demolition job and loading
it onto trucks, right, and so what's nice is that
you start to see patterns over and over again, and
this sort of work.
Speaker 3 (16:58):
We'll be back in just a minute.
Speaker 2 (17:11):
So the data problem is interesting here, right, Like you
were talking about the physics, just the physics of an
excavator is quite different, right, Like it's pushing down on
the ground, which is pushing back on the excavator, and
like as you say, the ground is changing because of
its work, and there's not Well where do you get
the data?
Speaker 1 (17:29):
Data is actually an interesting mix. We get it both
on our test sites and also with our design partners.
So we were working with general contractors and subcontractors. Today
we have five that we're partnered with across southern states
like Arizona, Texas, and so we're getting data on their size.
We're getting data on our sites.
Speaker 2 (17:46):
It's not that much data, right, Like I mean, I
guess my reference point is always like image net or
you know, the Internet for large language models. It does
seem like a sort of recurring problem in robotics type
AI applications is sparsity of data.
Speaker 1 (18:01):
How do you get it?
Speaker 2 (18:02):
Yeah?
Speaker 1 (18:03):
Yeah, And so this is where there's a few kind
of intersuties. So first of all, like our partners together
have thousand and thousands of machines and so there's a
lot of choices of which she's a partner. On the
other thing that you can do is actually you can
be very clever on a test site, and when you're
on a real project, you're kind of getting a unbiasample.
You're just getting a random distribution of the things you see,
(18:24):
just like driving around on a road. When you're on
a test site, you can actually up sample the things
you actually want and you can go and you can
collect ten hours of data. There's representative of five thousand
hours of random data, but which is particularly useful for
things like safety situations. Right, you can actually create a
much larger equivalent amount of data on a close course
(18:45):
through like kind of structure testing. And so for example,
safety scenarios where you have weird interactions with people doing
things they shouldn't do. You don't wait to see that
on an open site.
Speaker 2 (18:55):
What is one of those, like what is your nightmare
human behavior scenario?
Speaker 1 (18:59):
In the field? Night very human behavior is a human
is curious, they walk up to your machine, Your machine stops,
and then they get really really close and they are
now in a blind spot where you can't see that.
But you still have to be smart enough to track
them where they're in a hole in front of you
while you're like thinking about digging, right, So occlusions from
humans is probably a huge category which is called plex.
Speaker 2 (19:20):
Occlusions, meaning in your blind spot humans in a place
where the machine can't see them.
Speaker 1 (19:25):
Yeah, or usually your number one priority is anything that
touches on human safety. That's sacred. You never take any
risks on that.
Speaker 2 (19:32):
What's something you haven't figured out yet?
Speaker 1 (19:34):
The things people do with excavators, Like we've seen them
do bizarre like clearing of debris. We've seen them load
a wheeloader with dirt. We've seen them bang tools to
change them. We've seen them bang tools to change them.
What's that one? Like a tool gets stuck and they
like bang it on the ground in order to get
(19:56):
it loose. They use their arm as a pivot point
to turn when the wheels are like stuck in mud.
Speaker 2 (20:02):
It's kind of baller, Like it's pretty baller.
Speaker 1 (20:04):
Yeah, it's like I mean, like when you see the
expert operators, they just show off and it's like it's
increa like they're awesome. And so there's like all these
like weird subtleties of how they'll use these tools in
like really subtle ways, which like the dimensions of on
the product side of the use cases that blew was away.
We thought about the obvious ones, but as we started
like really going deeper and like studying this, there's so
(20:28):
much diversity and interesting things that you can do with
these machines. It's quite powerful.
Speaker 2 (20:32):
I mean, presumably you don't have to do all those
to have a product people will pay you for. Right,
you can just have the like competent automaton that can
dig a big hole just the way you want it.
Speaker 1 (20:44):
That's correct, and that there's a huge amount of work
even in those areas. But there is this like tale
that you go in you over time go and add
more and more capability, and then I think it is
just as softwareupdates and you get more. It's just like
a you know, the product ends up being the digital
operator which gets better over time, and so it's similar
to human getting better.
Speaker 2 (21:01):
What's the business model?
Speaker 1 (21:03):
So we will go operator out next year. So that'll
be our first, like fully operator list product that is
in a spirit of like what this is meant to scale.
Speaker 2 (21:11):
As operator out means driver list means autonomous. Is that
what it means?
Speaker 1 (21:16):
Yeah, nobody there, It's just like operator lists. Yeah, and
so our product is the digital operators. So what I
mean by this is that our customer is a general
contractor or subcontractor, so it's the companies that already buy
and manage these machines we are we sell them a
retrofit of these machines, so we install an upfit that
adds sensors and compute.
Speaker 2 (21:36):
So the contractor already owns the excavator.
Speaker 1 (21:39):
That's right, They've already buy a five hundred thousand dollars
machine or three hundred thousand dollars machine, and so this
is an upgrade that enables autonomy. And then our business
model is selling labor, So we're effectively selling the labor
and then a variety of digital services around it that
operates as machine. And for the surface area of types
of tasks that it's approved for, it's completely driverlests. And
(22:01):
then that surface area increases over time with software updates,
and for everything else, it can still be manually operated.
Speaker 2 (22:06):
So is the core product? Are they paying you by
the hour for digging?
Speaker 1 (22:10):
So we're figuring it out, but it'll be something that
is either by the hour, by the project by subscriptions.
So it's effectively the way that today projects are forecasted
and build by shifts, you know, for labor it's a
parallel of that, but it becomes a huge win because
you have complete flexibility. You can work ten hours a day,
(22:30):
you can work twenty four hours a day. So there's
all sorts of benefits that will operationally give you a
lot of leeway on how you use it, what might
go wrong, what might go wrong. I'll tell you the
things that are like really challenging. They like we worry
about it, we think about the nuance and diversity of
things are high. As much as you want to just
(22:51):
boil it down to a very simple dig and load
sort of operation, there's always little corner cases. There's things
you find in the ground, there's weird ways that trucks
can interact with you, there's things people will do. There's
varieties of machines, but there's like fifty kinds of excavator
models and sizes and so forth. So I think the
long tail is still challenging.
Speaker 2 (23:13):
Presumably there's you don't have to figure out every edge case,
but you probably have to figure out a lot for
your thing to be to work in a functional sense, right.
Speaker 1 (23:22):
Yeah, And so those in the meantime, you're not just
digging and loading and got a reposition you got to
organize the earth, You got to think about the sequencing,
you got to deal with you know, daytime, night time rain,
and so you have like these types of like really
challenging diversities you have to think about and deal with.
So I think all in all, it's still a complicated
(23:42):
product area where there's a huge amount of diversity of
the things that need to be done. But it's one
of those where, like I personally think there's a handful
of these holy grails of autonomy and physical industries that
are like genuinely transformational opportunities for both impact positive impact
to the country and the world, and also just kind
of scale of industries that are like double digit percentages
(24:04):
of GDP. Transportation is, without a doubt one of them,
Construction is one of them, Culture is not far behind.
You know, mining is a very very significant as well.
Manufacturing is one of them. And so I think that
we're going to see a wave over this next ten
years in autonomy, but it's going to be tackling this
like seventy five percent of the world's GDP that's physical
(24:24):
and not digital, and there's a lot of work, like
a lot of positive impact that can happen across these spaces.
Speaker 2 (24:29):
I mean, give me a little more on that one
pick a time in the future, five years, ten years,
not more than ten.
Speaker 1 (24:35):
So my personal belief is that this idea, like there's
a lot of companies getting flooded for this, but the
idea that like this giant brain for all of robotics,
the foundation model for robotics, like, I personally do not
believe that that's viable in the next ten years because
you have such complexity in really understanding these verticals on
(24:56):
the inputs, the hardware, the products, the neat use case,
the customers, everything, that every single one of those have
such complexity to get data that the idea to get
enough data that bulldos is through a generalization problem. That's
a long ways off. But I do think that word
a perfect time for these vertical solutions where if you
have a focused solution where you're trying to do construction,
(25:17):
you're trying to do fulfillment in a warehouse, you're trying
to do a very focused manufacturing solution, I think that's
actually there's a giant stuff function and how powerful and
male technologies are. But if you're trying to do a
humanoid to do everything in a home for a consumer
market too far away.
Speaker 2 (25:33):
Right now you're saying you basically you got to solve
one problem at a time. Yeah, I want to go
back to open road autonomy almost done. When do you
think that most rides most people take in cars or
trucks will be in driverless cars autonomous vehicles?
Speaker 1 (25:52):
Great question. There's a few pre requisites for most rides.
So first, all the goofs is have to disappear and
the thing just works everywhere that's on a trajectory of
getting there because it's getting more and more efficient as
the scales across the country and then the world. It
has to go from ride healing to personal car ownership
that is both cost down and versatility and everything that. Like,
(26:13):
but that's gonna happen.
Speaker 2 (26:14):
In that universe. When do they take the steering wheel out?
I just took a way mover the first time I
was in San Francisco this summer.
Speaker 1 (26:21):
It's awesome, right, it was awesome.
Speaker 2 (26:22):
Yeah, And like the weirdest thing about it to me
was the steering wheel. Yeah, right, Like if it was
just a box that I got in that looked like
a train car, someone I would have been like, oh sure,
it's like the air train or whatever, but that there's
a steering wheel and it's like a ghost is turning
the steeringhell, Like what's the steering wheel doing there?
Speaker 1 (26:39):
So the steering wheels will disappear pretty quickly because ride
sharing autonomous cars, yeah, have no use for it, and
you need to point you know, there's like special cases.
We need to recover them. But it's a wasted space, right,
It's a wasted spot. So that'll happen soon. But the
idea of really deep penetration, I think it's personal cars.
It's beyond luxury, which is another generation to like actually
(27:00):
make it be something that's affordable. Then you got to
go through a buying cycle, which is like five to
seven years, and so I think it starts to get
serious penetrats in the back half of the thirties and
it'll be the forties where like, okay, fifty percent of
driving is kind of like autonomous. I think it's still
twenty years off in my mind. For example, you bought
(27:21):
a car today, it's going to be in circulation for
the next twelve to fifteen years. It just takes a while.
Speaker 2 (27:31):
We'll be back in a minute with the lightning round. Okay,
let's finish with the lightning round. If you could operate
any machine, what would it be?
Speaker 1 (27:53):
Oh gosh, this is a fun one. If I could
operate any machine at all in the world, not even
in contry anything, Oh gosh, Okay, you know what, I
would love to either operate one of the like Boring
Company gigantic drill machines. Was like so astronomical or there's
like my trucks that are so astronomically big that they
just dwarfed any machine in the world, and it costs
(28:14):
like five million dollars and just a skill that would
be really really fun to try at some point.
Speaker 2 (28:19):
Just to be that high, just have that much momentum, right,
that much mass at your disposal.
Speaker 1 (28:25):
Yeah, like literally the tire is like three stories tall.
It's like it's wild, Like it's absurd.
Speaker 2 (28:30):
What's one thing you remember about immigrating from the Soviet
Union to the US when you were six?
Speaker 1 (28:36):
I was six. Yeah, I was born in Moscow. We immigrated.
I was super young. I remember we ended up having
a pet stop in Europe where there's a standard path
of going to be out of Venice in Rome, why
your paperwork gets processed, and then we went to New York.
I remember running around the rooftops of Venice with a
friend of mine, causing a bunch of you know, trouble
(28:57):
and running away and disappearing for long periods of time
and having a boss for for whatever reason, running around
the rooftops of Venice was in my was ingrained in
my memory, which was a kind of very positive memory
while my parents were going through a whole bunch of strusts.
Speaker 2 (29:09):
It sounds very free. Not to project East versus West
language onto it, but it sounds very free.
Speaker 1 (29:15):
It was very freeing. It's It's funny. I just hit
the age my dad was when they immigrated from the
Soviet Union with two kids. My sister was one month old,
zero money, almost no English, having a fresh start.
Speaker 2 (29:29):
So courageous, right, doesn't it seem so brave?
Speaker 1 (29:33):
Yeah? It does. And they were trying to leave for
like ten years and couldn't leave. Yeah, it's kind of fascinating.
So we're very fortunate to not have to go through
something like that.
Speaker 2 (29:41):
Is that when you think about it that way, does
that like put pressure on you? You think, oh my god,
my parents did all this better? I better deliver.
Speaker 1 (29:50):
It's funny, Uh, a little bit. I mean you kind
of have this I don't know, like adventurous spirit. I
guess maybe it gets ingrained. I think of it now
for my kids, where I'm like, okay, Like now they're
in a really nice and comfortable environment growing up. How
do you I convey that edge in a little bit
of that spirit, Like, how.
Speaker 2 (30:08):
Do you keep them from going so.
Speaker 1 (30:09):
Yeah, it's something you want them to go through anything difficult,
because it's not that I was, you know, for me,
it was actually just an adventure my guards. It was difficult,
But part of it is just, yeah, like conveying that
spirit of being able to like be comfortable trying to
tackle something new and being thrown in a completely different environment.
It's hard to force ound or simulate that when you're
just growing up in the Semptusco area. Right.
Speaker 2 (30:32):
I appreciate your time. Thanks for talking to me.
Speaker 1 (30:34):
It's a pleasure. This is a lot of fun. Thanks
for having me.
Speaker 2 (30:43):
Boris Soffman is the co founder and CEO of Bedrock Robotics.
Just a quick note that this is our last episode
before a break of a couple of weeks and then
we'll be back with more episodes. Please email us at
problem at Pushkin dot Fm. We are always looking for
new guests for the show. Today's show was produced by
(31:04):
Trinamanino and Gabriel Hunter Chang, who was edited by Alexander
Garretton and engineered by Sarah Brugueri. I'm Jacob Goldstein and
we'll be back next week with another episode of What's
Your Problem.
Pushkin.
Speaker 2 (00:20):
When we started this show in twenty twenty two, the
standard line about driverless cars was driverless cars have been
five years away for the past fifteen years, because it
seemed like they were just always around the corner, always
just a few years away, but they never quite arrived.
Nobody says that anymore. Today, in several cities around the country,
(00:43):
getting a ride from a driverless car is just a
normal thing people do, and it'll become normal in more
and more and more cities over the next few years.
Driverless cars are here now, so now we can ask
what's next. I'm Jacob Goldstein and this is What's Your Problem,
(01:05):
the show where I talk to people who are trying
to make technological progress. My guest today is Boris Soffman.
He's the co founder and CEO of Bedrock Robotics, a
company that's figuring out how to retrofit heavy equipment to
make it work autonomously. Boris's problem is this, how do
you teach machines not just to drive, but to do
(01:25):
things like grade roads and move heavy things around construction sites.
Boris's company is starting with excavators, and they plan to
have their first commercial excavators autonomously digging holes on construction
projects next year. Later in the interview, Boris goes big.
He argues that if Bedrock succeeds, the company could help
(01:47):
push forward a broad wave of new building in America.
But first we talked about his time at Weimo and
how the wild evolution of the autonomous vehicles that he
worked on there led him to start Bedrock.
Speaker 1 (02:01):
The time at Weimo was this incredible period where I
was there for about five years from mid twenty nineteen
till through spring twenty four, and that was this really
big period where it was going through this like fifteen
years of R and D, and then it finally transitioned
into this hockey stick of like growth that is helping today.
And so today waymos at over one hundred million miles
(02:22):
fully driver lists. It's at five times safer than a human.
It's millions of miles every single week, and so it's
kind of scaling itsponentially and it's like a genuinely fantastic product.
But I was there when we were like stressing over
the first hundred miles, and it felt like the most
incredible achievement to just go like ten fifty hundred miles
to completely drop less, say now that happens at like
hundreds of thousands of miles every single day. And so
(02:44):
one of the things that really broke through it made
that possible is this shift to machine learning and data
driven approaches as a core of the autonomy stack.
Speaker 2 (02:52):
And just to be clear, like that's as opposed to
a more like euristics or rule based kind of model,
like the old school twentieth century eyes, like, well, just
tell the car all the rules of how to drive
and then it'll drive. Like that's what you're comparing machine
learning to.
Speaker 1 (03:08):
Yeah, like drive, I'm going to like embed the cost
functions and all this. So yeah, so like large scale
search and heuristics and rules and you can embed them
now inside those heuristics, and so you can solve almost
any given problem with that sort of approach.
Speaker 2 (03:21):
So you can solve one problem, but you can't solve
like every single possible problem that would ever arise when
you're driving, which is actually what you have to solve
to do full autonomy.
Speaker 1 (03:31):
Right right with activating whackamall, where like you fix one
problem and it becomes harder and harder to kind of
scale the other ones. And so and so there was
this really conscious shift at WEIMA, which was kind of
bold at the time because it feels obvious in hindsight,
It wasn't obvious at all back then. The shift to
like really embracing this as a data driven solution where
you're learning from human driving and human behavior.
Speaker 2 (03:52):
And when you say you're learning, you mean the machine
learning model. The AI basically is learning.
Speaker 1 (03:58):
The machine learning model, that's right. And so you're basically
taking giant scales of data one hundreds of thousands of
millions of miles, and you're learning the model of how
do you drive and how do you interpret this? Like
infinite compocity and kitchen sink of contacts, like all of
this sensored data, the road structures, the things you see
around you, the way people are moving. How do you
go from a kind of engineered and you know, kind
(04:20):
of trained solution into one that is like dominantly a
warned solution.
Speaker 2 (04:24):
I mean, as you're doing that, you're sort of riding
the historic machine learning AI wave, right, Like that's right.
Presumably you're able to do that at that moment because
of this explosion in AI, which is basically machine learning, right,
you know that.
Speaker 1 (04:38):
And you couldn't have done that five ten years ago.
Speaker 2 (04:40):
That's what you're kind of coming out of at way MO,
that's driving this success at way MOO. How does that
set you up for what you're doing a bedrock?
Speaker 1 (04:48):
The most shocking thing was just how well does generalized
from San Francisco to Los Angeles, Phoenix in Austin and
then becomes almost like a qualification problem eventually, where you're
using data to fill some gaps, but like you need
less and less of it, less and less new things
surprise you, like you're just your competency kind of expands.
And then we unified the technology stack between cars and
trucks where even jumping from a cargo a truck was
(05:09):
needed maybe ten to fifty percent more data, but it
was fundamentally like you're using data to explain why how
you operate a very different potages.
Speaker 2 (05:15):
This is like a big truck, This is like an
eighteen wheel truck. When Google was working on that, yeah,
are way most like.
Speaker 1 (05:19):
Fifty three foot trailer like eighty thousand pounds. And so
that was the big moment where we started thinking about
where else can you apply this? What are the places
where you have all this diversity of challenges and capabilities
that benefit from this type of versatility and also have
the ability to jump between platforms in this really natural way.
And so we looked at a lot of spaces and
(05:41):
really settled on automation of specialized typing machinery, and so
the sipes of machines that you see in construction, like
excavators and wheeloaders and bulldozers, but also frankly the sort
of machines you see in all sorts of industries like
agriculture and mining and lumper and garbage movement. And so
you have these very diverse types of machines that are
interacting with the world around them. They're fairly still moving,
(06:03):
they're in semi controlled environments, and at the end of
the day, there's astronomical scale at which they great at
and a lot of the learnings that we experienced a
weymle actually transfer over incredibly well. But the physics of
the problem is a lot less aperspherial. It's actually like,
really tackle this and get the market.
Speaker 2 (06:19):
Basically, big vehicles operating in semi constrained environments doing things
to the world, interacting with the world in some physical
way in addition to just driving across it.
Speaker 1 (06:30):
That's right, and these are slow moving vehicles where you're
already on a closed site with people who are assumed
to be knowledgeable about the world around you, and you're
moving at like five off an hour, for example, You're
able to slow down and stop. You're able to minimize
your exposure. You always have a minimum safety condition of stopping.
Your complexity is less abound interactions with others on the
road and more about the interactions with the world around you.
(06:52):
And so you can actually tackle safety not through this
sort of statistical methods that drove a lot of the
mileage that we collect, but through a much more direct
measure of your competencies in order to just make sure
that you're like you're actually capable of hurting somebody.
Speaker 2 (07:08):
Yeah, more like like an industrial robot al most right.
Speaker 1 (07:12):
That's right. Yeah, it's like a traditional Susis engineering, Yeah.
Speaker 2 (07:15):
Where it's like, look, even if it can't do the
thing every time, that's fine. Just make sure it's not
gonna like go crazy and kill somebody.
Speaker 1 (07:21):
That's right. You can make it to that. Like your
worst case is your productivity suffers, but safety wise, you're
you're you're solid. And so that's like incredibly enabling because
your long tail is now no longer safety. It's the
versatility what you can do.
Speaker 2 (07:33):
Why did you start with excavators?
Speaker 1 (07:36):
So excavators are the most highly utilized machine, Like they're
usually the highest volume in fleet. So between twenty and
twenty five percent of fleets are excavators.
Speaker 2 (07:45):
Fleets are just like big construction heavy equipment.
Speaker 1 (07:48):
Yeah, like a general contractor that alone, like a thousand
machines two hundred, two hundred and fifty will probably be
excavators on average.
Speaker 2 (07:55):
And it's basically what a kid would call it digger, right,
It's like at a digger bucket an arm, and it
like digs stuff up.
Speaker 1 (08:01):
It's like the equivalent of an arm. Like you can
like dig stuff, you can demolish stuff, you can swap
your tools, you can like lift pipes and put them
in a holes. They can do a ton of stuff
with it. It's kind of crazy. It really is a
personal machine. It also makes it very complicated to work.
So there's like seven degrees of freedom sometimes eight, and
so it's one of the hardest machines to learn, and
it takes four to five years to really become an expert.
And there's a huge difference between an expert and novice,
(08:22):
and so you kind of have this situation where it's
a huge volume of work and it's really hard to learn,
and so you have a really deep pull in the
market for it, meaning.
Speaker 2 (08:31):
A lot of demand, like a lot of people want
somebody who can drive one of these or a machine
that could do it.
Speaker 1 (08:37):
Well, let me tell you about the demand. I've never
seen such a divergence of supply to demand like in
my career in any case, where on the demand side,
you have this astronomical construction industry that's already two trillion
dollars a year in the US, that's obviously very heavily
building and having machinery work. And then you have this
like shortage of operators that already existed, but it's going
in the wrong direction where forty percent of construction workers
(08:59):
is retiring in the next ten years. Our partners are
consistently having trouble filling labor. We've met some that have
one hundred percent turnover.
Speaker 2 (09:09):
Everybody leaves every year.
Speaker 1 (09:10):
It means like a third of people might be lifers,
but then like two thirds transition more than once per year,
and you're constantly backfilling the skill sets very a ton
and one of them said that for every one person
entering the workforce of the quality that they look for,
their seven leaving. And so we end up having is
this shortage of ability to meet this demand, and so
(09:30):
prices go up, jobs don't get done. And what's interesting
is it's not that's like isolated industry that's like just
on its own kind of like having these sort of challenges.
It's a horizontal it supports every industry. You can't build
data centers free eye without it, you can't build houses,
you can't build energy facilities.
Speaker 2 (09:46):
Of like a rate limiting skill or rate limiting machine.
Speaker 1 (09:49):
It's a whole country. That's exactly right. And then you
have this need that starts with labor. But then there's safety.
It's huge amounts of safety challenges. You have, you know,
huge predictability challenges. So if you actually could soften out
some of these constraints, we would build more, more work
would get done, and it would like stimulate the whole economy.
And so that's what's actually pretty exciting about this opportunity.
(10:09):
It's not as yours some game at all good.
Speaker 2 (10:12):
So, like you decide to focus on excavators, you you know,
you raise money for your company, Like, do you go
out and buy a whatever, a million dollar excavator. You
go to what is it bucket and shovel dot com
and buy yourself an excavator.
Speaker 1 (10:25):
They're like three hundred thousand to five hundred thousand, so
they're like still expensive. Yeah, they're pretty expensive.
Speaker 2 (10:30):
So I mean like, did you buy one like we did?
Speaker 1 (10:32):
Yeah?
Speaker 2 (10:33):
Did you drive it?
Speaker 1 (10:34):
Of course, like you have to. It's like it's like
a ride of passage. Yeah, they're really fine. I even
took a lessoners. There's a place in Vegas ALESSI drive
excavators and take away and from like a trade operator.
It was pretty fun.
Speaker 2 (10:44):
So, oh, that's genius. You get to break things. I
was actually as I was walking here today to the
train that there's a playground and there was an excavator
just like breaking up the asphalt and then like prying
it up. I was like, that looks awesome.
Speaker 1 (10:59):
This is the funniest thing. Anybody that gets an excavator,
like your inner six year old comes out. Suddenly you
have like the most mature, sophisticated person is trying to
be all professional. You get an excavator and just like
get like a giant glob of mud and then like
bring as hig as a can and then like PLoP
it down and see what happens. It's amazing, and it's like,
without fail, everybody kind of reverts back to this sort
of a world. Yeah, we drove and I think we
(11:20):
bought like a half dozen excavators at this point, and
then we also then use a lot of excavators from
our partners who are general contractors and subcontractors.
Speaker 2 (11:28):
And it's just what a year or so ago that
you started, like not that long in.
Speaker 1 (11:32):
This less than a year and a half, yeah, like
last last minute, seah, it's a pretty been a good run.
Speaker 2 (11:37):
How much is sort of commodified of autonomy, right? How
much is just like, well, we're gonna buy this, this,
and this in terms of sort of hardware and we
know the software, and then how much is like, oh,
here's the things we have to figure out that nobody
knows how to do.
Speaker 1 (11:49):
So it's one of the other enablers that's like way
better than what we would have had to go through
ten years ago. In the space, we can use a
lot of the existing components on lighter on cameras, on
Imus GPS, there's a lot of tailwind from automotive great
cameras and compute.
Speaker 2 (12:05):
That's like accelerating and like presumably they're buying those things
at scale you can like you've got to go get
them cheap. Essentially, you're like, give me one of those,
one of those, one of those.
Speaker 1 (12:14):
That's right, because you're gonna go to millions and millions
of units as like every car against an autopilot equivalent
over the next two three years. Oh, that's starting to
kind of pump in a cost. So that helps. And
then even the platform, we can retrofit existing machinery.
Speaker 2 (12:27):
When you say the platform, what do you mean in
this context.
Speaker 1 (12:30):
Platform is like the car, the truck, the excavator, and
the way in this case the excavator. Yeah, the machine,
so like the excavator itself. Construction machines, particularly this latest generation,
they're really well designed to where they're already effectively drive
by wire, which means that every signal going through the
machine is electric, and so we're able to spice into
it and both read and write to these signals.
Speaker 2 (12:54):
And control them electric meaning like computerized basically.
Speaker 1 (12:58):
Meaning like you have a joystick that operates the excavatory
that's not physically connected to the hydraulics. It's an electric
signal that's connected to the hydraulics, and so the same
for every signal and the machine, like the sense the
pressure gages, and so we're able to both get the
data from the machine as well as control the machine
through a non invasive integration where we can upit a
(13:20):
machine with our sensors and compute suite in like less
than three hours and make it autonomy capable and it's
completely reversible. We could never do that with a car
or a truck because the platforms were just not designed
this way.
Speaker 2 (13:32):
Yeah, so there's a lot that's there already. The excavators
themselves are sort of easy to integrate with lots of
off the shelf technology. What's not there, Like, when you're
coming in, what do you have to sort of build
that nobody has built before?
Speaker 1 (13:44):
Nobody has actually created autonomy that can solve the really
nuanced problems that you need to solve in order to
operate like an excavator in construction tasks.
Speaker 2 (13:55):
So let's talk like specifically, you buy your excavators, you
got your hardware, you know, whatever, what basic AI model
you're going to use, But like, what's a specific thing.
You have to figure out well a.
Speaker 1 (14:07):
Lot of things. So first of all, it's not trivial
to tap into these machines and to build a platform
or kind of an integration around it where you can
read them, you can control them, and so you have
to basically create like a wrapper around the machine and
also design in a way that scales later to new machines. Right,
So that part's hard. It's a big autonomy problem. So
in that respect is no different than what we tackle
(14:27):
with a WEIMO, where you have to train massive scale models.
You have to collect a huge amount of data.
Speaker 2 (14:31):
And in this case it's like about the digging, like
it's about the digging dumb question? Is the digging the
hard part? Like what happens when the thing hits the ground.
There's different kinds of stuff in the ground.
Speaker 1 (14:41):
Like tell me in the category of what's new, suddenly
a car doesn't change the environment around it. You have
a really complicated tough earth and soil, and you got
to figure out the physics of how you dig through it,
how do you deal with clay versus top soil, how
do you deal with rocks inside? Then so, now if
you want to use simulation to solve parts of these problems.
You have a very complicated simulation problem because you have
(15:03):
to solve not just the sensor side, but also the
manipulation side. You have to structure this problem in a
way where you're learning from the data you collect in
order to actually capture the nuances of how do you
actually interact with the environment. You have to then control
and execute it the right way. You have to think
about how do you actually define the goal?
Speaker 2 (15:22):
So from the user point of view with an excavator,
like how does it look like? I just I want
to dig a hole of this size at this spot?
Or like what is it?
Speaker 1 (15:30):
Yeah, and this is a journey like fast forward you
know next year where card customer does this right, So
what they would do is they would give it a
model of what they want the earth to be dug
towards and so that would be like a three D
representation of the depth. So with length depth, here's the edges,
here's the constraints. The pickups for the trucks are going
(15:50):
to be over here and then go to work. And
basically these projects, like these machines will work for many
months at a time, continually just digging earth and working
it towards the right foundation, and the system will respect
that boundary and basically dig down to that level. It'll
have precision to the edges that you've defined, it'll do
it in a sequence that makes sense so that you
don't trap yourself in a hole, for example. And then
(16:12):
you specify where you're going to have dump truck pickups.
Then there's projects where this happens for nine months straight
or twelve months straight with like many machines, and so
what you basically need to specify is what you want
to dig to, and that what you want to dig
to ends up basically being the foundation that you're working
towards for whatever you're going to construct there. And we
(16:33):
want that to be versatile. So sometimes you're just taking
off a layer of top soil. Sometimes you're digging eight
feet deep. Sometimes you're taking an existing stockpile and moving
in somewhere else. So there's a lot of permutations of this.
Sometimes you're taking rubble from a demolition job and loading
it onto trucks, right, and so what's nice is that
you start to see patterns over and over again, and
this sort of work.
Speaker 3 (16:58):
We'll be back in just a minute.
Speaker 2 (17:11):
So the data problem is interesting here, right, Like you
were talking about the physics, just the physics of an
excavator is quite different, right, Like it's pushing down on
the ground, which is pushing back on the excavator, and
like as you say, the ground is changing because of
its work, and there's not Well where do you get
the data?
Speaker 1 (17:29):
Data is actually an interesting mix. We get it both
on our test sites and also with our design partners.
So we were working with general contractors and subcontractors. Today
we have five that we're partnered with across southern states
like Arizona, Texas, and so we're getting data on their size.
We're getting data on our sites.
Speaker 2 (17:46):
It's not that much data, right, Like I mean, I
guess my reference point is always like image net or
you know, the Internet for large language models. It does
seem like a sort of recurring problem in robotics type
AI applications is sparsity of data.
Speaker 1 (18:01):
How do you get it?
Speaker 2 (18:02):
Yeah?
Speaker 1 (18:03):
Yeah, And so this is where there's a few kind
of intersuties. So first of all, like our partners together
have thousand and thousands of machines and so there's a
lot of choices of which she's a partner. On the
other thing that you can do is actually you can
be very clever on a test site, and when you're
on a real project, you're kind of getting a unbiasample.
You're just getting a random distribution of the things you see,
(18:24):
just like driving around on a road. When you're on
a test site, you can actually up sample the things
you actually want and you can go and you can
collect ten hours of data. There's representative of five thousand
hours of random data, but which is particularly useful for
things like safety situations. Right, you can actually create a
much larger equivalent amount of data on a close course
(18:45):
through like kind of structure testing. And so for example,
safety scenarios where you have weird interactions with people doing
things they shouldn't do. You don't wait to see that
on an open site.
Speaker 2 (18:55):
What is one of those, like what is your nightmare
human behavior scenario?
Speaker 1 (18:59):
In the field? Night very human behavior is a human
is curious, they walk up to your machine, Your machine stops,
and then they get really really close and they are
now in a blind spot where you can't see that.
But you still have to be smart enough to track
them where they're in a hole in front of you
while you're like thinking about digging, right, So occlusions from
humans is probably a huge category which is called plex.
Speaker 2 (19:20):
Occlusions, meaning in your blind spot humans in a place
where the machine can't see them.
Speaker 1 (19:25):
Yeah, or usually your number one priority is anything that
touches on human safety. That's sacred. You never take any
risks on that.
Speaker 2 (19:32):
What's something you haven't figured out yet?
Speaker 1 (19:34):
The things people do with excavators, Like we've seen them
do bizarre like clearing of debris. We've seen them load
a wheeloader with dirt. We've seen them bang tools to
change them. We've seen them bang tools to change them.
What's that one? Like a tool gets stuck and they
like bang it on the ground in order to get
(19:56):
it loose. They use their arm as a pivot point
to turn when the wheels are like stuck in mud.
Speaker 2 (20:02):
It's kind of baller, Like it's pretty baller.
Speaker 1 (20:04):
Yeah, it's like I mean, like when you see the
expert operators, they just show off and it's like it's
increa like they're awesome. And so there's like all these
like weird subtleties of how they'll use these tools in
like really subtle ways, which like the dimensions of on
the product side of the use cases that blew was away.
We thought about the obvious ones, but as we started
like really going deeper and like studying this, there's so
(20:28):
much diversity and interesting things that you can do with
these machines. It's quite powerful.
Speaker 2 (20:32):
I mean, presumably you don't have to do all those
to have a product people will pay you for. Right,
you can just have the like competent automaton that can
dig a big hole just the way you want it.
Speaker 1 (20:44):
That's correct, and that there's a huge amount of work
even in those areas. But there is this like tale
that you go in you over time go and add
more and more capability, and then I think it is
just as softwareupdates and you get more. It's just like
a you know, the product ends up being the digital
operator which gets better over time, and so it's similar
to human getting better.
Speaker 2 (21:01):
What's the business model?
Speaker 1 (21:03):
So we will go operator out next year. So that'll
be our first, like fully operator list product that is
in a spirit of like what this is meant to scale.
Speaker 2 (21:11):
As operator out means driver list means autonomous. Is that
what it means?
Speaker 1 (21:16):
Yeah, nobody there, It's just like operator lists. Yeah, and
so our product is the digital operators. So what I
mean by this is that our customer is a general
contractor or subcontractor, so it's the companies that already buy
and manage these machines we are we sell them a
retrofit of these machines, so we install an upfit that
adds sensors and compute.
Speaker 2 (21:36):
So the contractor already owns the excavator.
Speaker 1 (21:39):
That's right, They've already buy a five hundred thousand dollars
machine or three hundred thousand dollars machine, and so this
is an upgrade that enables autonomy. And then our business
model is selling labor, So we're effectively selling the labor
and then a variety of digital services around it that
operates as machine. And for the surface area of types
of tasks that it's approved for, it's completely driverlests. And
(22:01):
then that surface area increases over time with software updates,
and for everything else, it can still be manually operated.
Speaker 2 (22:06):
So is the core product? Are they paying you by
the hour for digging?
Speaker 1 (22:10):
So we're figuring it out, but it'll be something that
is either by the hour, by the project by subscriptions.
So it's effectively the way that today projects are forecasted
and build by shifts, you know, for labor it's a
parallel of that, but it becomes a huge win because
you have complete flexibility. You can work ten hours a day,
(22:30):
you can work twenty four hours a day. So there's
all sorts of benefits that will operationally give you a
lot of leeway on how you use it, what might
go wrong, what might go wrong. I'll tell you the
things that are like really challenging. They like we worry
about it, we think about the nuance and diversity of
things are high. As much as you want to just
(22:51):
boil it down to a very simple dig and load
sort of operation, there's always little corner cases. There's things
you find in the ground, there's weird ways that trucks
can interact with you, there's things people will do. There's
varieties of machines, but there's like fifty kinds of excavator
models and sizes and so forth. So I think the
long tail is still challenging.
Speaker 2 (23:13):
Presumably there's you don't have to figure out every edge case,
but you probably have to figure out a lot for
your thing to be to work in a functional sense, right.
Speaker 1 (23:22):
Yeah, And so those in the meantime, you're not just
digging and loading and got a reposition you got to
organize the earth, You got to think about the sequencing,
you got to deal with you know, daytime, night time rain,
and so you have like these types of like really
challenging diversities you have to think about and deal with.
So I think all in all, it's still a complicated
(23:42):
product area where there's a huge amount of diversity of
the things that need to be done. But it's one
of those where, like I personally think there's a handful
of these holy grails of autonomy and physical industries that
are like genuinely transformational opportunities for both impact positive impact
to the country and the world, and also just kind
of scale of industries that are like double digit percentages
(24:04):
of GDP. Transportation is, without a doubt one of them,
Construction is one of them, Culture is not far behind.
You know, mining is a very very significant as well.
Manufacturing is one of them. And so I think that
we're going to see a wave over this next ten
years in autonomy, but it's going to be tackling this
like seventy five percent of the world's GDP that's physical
(24:24):
and not digital, and there's a lot of work, like
a lot of positive impact that can happen across these spaces.
Speaker 2 (24:29):
I mean, give me a little more on that one
pick a time in the future, five years, ten years,
not more than ten.
Speaker 1 (24:35):
So my personal belief is that this idea, like there's
a lot of companies getting flooded for this, but the
idea that like this giant brain for all of robotics,
the foundation model for robotics, like, I personally do not
believe that that's viable in the next ten years because
you have such complexity in really understanding these verticals on
(24:56):
the inputs, the hardware, the products, the neat use case,
the customers, everything, that every single one of those have
such complexity to get data that the idea to get
enough data that bulldos is through a generalization problem. That's
a long ways off. But I do think that word
a perfect time for these vertical solutions where if you
have a focused solution where you're trying to do construction,
(25:17):
you're trying to do fulfillment in a warehouse, you're trying
to do a very focused manufacturing solution, I think that's
actually there's a giant stuff function and how powerful and
male technologies are. But if you're trying to do a
humanoid to do everything in a home for a consumer
market too far away.
Speaker 2 (25:33):
Right now you're saying you basically you got to solve
one problem at a time. Yeah, I want to go
back to open road autonomy almost done. When do you
think that most rides most people take in cars or
trucks will be in driverless cars autonomous vehicles?
Speaker 1 (25:52):
Great question. There's a few pre requisites for most rides.
So first, all the goofs is have to disappear and
the thing just works everywhere that's on a trajectory of
getting there because it's getting more and more efficient as
the scales across the country and then the world. It
has to go from ride healing to personal car ownership
that is both cost down and versatility and everything that. Like,
(26:13):
but that's gonna happen.
Speaker 2 (26:14):
In that universe. When do they take the steering wheel out?
I just took a way mover the first time I
was in San Francisco this summer.
Speaker 1 (26:21):
It's awesome, right, it was awesome.
Speaker 2 (26:22):
Yeah, And like the weirdest thing about it to me
was the steering wheel. Yeah, right, Like if it was
just a box that I got in that looked like
a train car, someone I would have been like, oh sure,
it's like the air train or whatever, but that there's
a steering wheel and it's like a ghost is turning
the steeringhell, Like what's the steering wheel doing there?
Speaker 1 (26:39):
So the steering wheels will disappear pretty quickly because ride
sharing autonomous cars, yeah, have no use for it, and
you need to point you know, there's like special cases.
We need to recover them. But it's a wasted space, right,
It's a wasted spot. So that'll happen soon. But the
idea of really deep penetration, I think it's personal cars.
It's beyond luxury, which is another generation to like actually
(27:00):
make it be something that's affordable. Then you got to
go through a buying cycle, which is like five to
seven years, and so I think it starts to get
serious penetrats in the back half of the thirties and
it'll be the forties where like, okay, fifty percent of
driving is kind of like autonomous. I think it's still
twenty years off in my mind. For example, you bought
(27:21):
a car today, it's going to be in circulation for
the next twelve to fifteen years. It just takes a while.
Speaker 2 (27:31):
We'll be back in a minute with the lightning round. Okay,
let's finish with the lightning round. If you could operate
any machine, what would it be?
Speaker 1 (27:53):
Oh gosh, this is a fun one. If I could
operate any machine at all in the world, not even
in contry anything, Oh gosh, Okay, you know what, I
would love to either operate one of the like Boring
Company gigantic drill machines. Was like so astronomical or there's
like my trucks that are so astronomically big that they
just dwarfed any machine in the world, and it costs
(28:14):
like five million dollars and just a skill that would
be really really fun to try at some point.
Speaker 2 (28:19):
Just to be that high, just have that much momentum, right,
that much mass at your disposal.
Speaker 1 (28:25):
Yeah, like literally the tire is like three stories tall.
It's like it's wild, Like it's absurd.
Speaker 2 (28:30):
What's one thing you remember about immigrating from the Soviet
Union to the US when you were six?
Speaker 1 (28:36):
I was six. Yeah, I was born in Moscow. We immigrated.
I was super young. I remember we ended up having
a pet stop in Europe where there's a standard path
of going to be out of Venice in Rome, why
your paperwork gets processed, and then we went to New York.
I remember running around the rooftops of Venice with a
friend of mine, causing a bunch of you know, trouble
(28:57):
and running away and disappearing for long periods of time
and having a boss for for whatever reason, running around
the rooftops of Venice was in my was ingrained in
my memory, which was a kind of very positive memory
while my parents were going through a whole bunch of strusts.
Speaker 2 (29:09):
It sounds very free. Not to project East versus West
language onto it, but it sounds very free.
Speaker 1 (29:15):
It was very freeing. It's It's funny. I just hit
the age my dad was when they immigrated from the
Soviet Union with two kids. My sister was one month old,
zero money, almost no English, having a fresh start.
Speaker 2 (29:29):
So courageous, right, doesn't it seem so brave?
Speaker 1 (29:33):
Yeah? It does. And they were trying to leave for
like ten years and couldn't leave. Yeah, it's kind of fascinating.
So we're very fortunate to not have to go through
something like that.
Speaker 2 (29:41):
Is that when you think about it that way, does
that like put pressure on you? You think, oh my god,
my parents did all this better? I better deliver.
Speaker 1 (29:50):
It's funny, Uh, a little bit. I mean you kind
of have this I don't know, like adventurous spirit. I
guess maybe it gets ingrained. I think of it now
for my kids, where I'm like, okay, Like now they're
in a really nice and comfortable environment growing up. How
do you I convey that edge in a little bit
of that spirit, Like, how.
Speaker 2 (30:08):
Do you keep them from going so.
Speaker 1 (30:09):
Yeah, it's something you want them to go through anything difficult,
because it's not that I was, you know, for me,
it was actually just an adventure my guards. It was difficult,
But part of it is just, yeah, like conveying that
spirit of being able to like be comfortable trying to
tackle something new and being thrown in a completely different environment.
It's hard to force ound or simulate that when you're
just growing up in the Semptusco area. Right.
Speaker 2 (30:32):
I appreciate your time. Thanks for talking to me.
Speaker 1 (30:34):
It's a pleasure. This is a lot of fun. Thanks
for having me.
Speaker 2 (30:43):
Boris Soffman is the co founder and CEO of Bedrock Robotics.
Just a quick note that this is our last episode
before a break of a couple of weeks and then
we'll be back with more episodes. Please email us at
problem at Pushkin dot Fm. We are always looking for
new guests for the show. Today's show was produced by
(31:04):
Trinamanino and Gabriel Hunter Chang, who was edited by Alexander
Garretton and engineered by Sarah Brugueri. I'm Jacob Goldstein and
we'll be back next week with another episode of What's
Your Problem.
Popular Podcasts
Las Culturistas with Matt Rogers and Bowen Yang
Ding dong! Join your culture consultants, Matt Rogers and Bowen Yang, on an unforgettable journey into the beating heart of CULTURE. Alongside sizzling special guests, they GET INTO the hottest pop-culture moments of the day and the formative cultural experiences that turned them into Culturistas. Produced by the Big Money Players Network and iHeartRadio.
Crime Junkie
Does hearing about a true crime case always leave you scouring the internet for the truth behind the story? Dive into your next mystery with Crime Junkie. Every Monday, join your host Ashley Flowers as she unravels all the details of infamous and underreported true crime cases with her best friend Brit Prawat. From cold cases to missing persons and heroes in our community who seek justice, Crime Junkie is your destination for theories and stories you won’t hear anywhere else. Whether you're a seasoned true crime enthusiast or new to the genre, you'll find yourself on the edge of your seat awaiting a new episode every Monday. If you can never get enough true crime... Congratulations, you’ve found your people. Follow to join a community of Crime Junkies! Crime Junkie is presented by Audiochuck Media Company.
The Brothers Ortiz
The Brothers Ortiz is the story of two brothers–both successful, but in very different ways. Gabe Ortiz becomes a third-highest ranking officer in all of Texas while his younger brother Larry climbs the ranks in Puro Tango Blast, a notorious Texas Prison gang. Gabe doesn’t know all the details of his brother’s nefarious dealings, and he’s made a point not to ask, to protect their relationship. But when Larry is murdered during a home invasion in a rented beach house, Gabe has no choice but to look into what happened that night. To solve Larry’s murder, Gabe, and the whole Ortiz family, must ask each other tough questions.