Haischel Dabian & Dan Acristinii – Kryha – Blockchain on Satellites


So, Hi, it’s Haischel Dabian (laughs)
(applause) Alright, so first before I start, I’d like
to introduce our company, our startup, and what we do, and then go into what we did at
the Hackathon. So, our company is Kryha.io, and what we do is we help organisations better
understand, and basically adopt blockchain technology. So, we help them form the ‘ideation’
phase, the validate use cases. We have a platform-agnostic view, so we choose the right platform for
a given use case. And then we actually build prototypes in-house.
So, this is half of our team. This is mostly the developers – this picture was taken
at the Hackathon. So, as you can see, we look quite bad, we didn’t have much sleep that
weekend. But totally worth it, because we won.
So, I’ll start at the Hackathon, this is basically a timeline to keep track of what’s
been going on since the Hackathon. So, the Hackathon’s Blockchaingers Hackathon, and
we participated in the machine2machine economy track. And what this means is it’s the crazy
shit track, as Jan-Pieter always says, and it’s about envisaging the world of tomorrow.
So, the future world where us humans are liberated from performing tedious, hard, and especially
dangerous tasks. A world where we have autonomous machines basically collaborating and figuring
out stuff by themselves and doing all this works so we can lay back and enjoy our lives,
and do whatever we want. So, this was a tough challenge, we had a lot
of debates with Jan-Pieter, he basically was pushing us to break through the thoughts of
today, and basically think about the world of tomorrow.
So, what did we come up with? So, we came up with Grex World State Consensus. So this
is, we built a small piece of the whole concept for the whole idea, just to prove basically
that a very small piece is possible. And what is it? So, it’s autonomous agents – so
think about drones, autonomous vehicles, or whatever. Autonomous machines that basically
agree on a world state. So, they all roam around, they sense, they gather data, and
then they agree on what they see. So, they agree that maybe it’s raining in some continent
or it’s, whatever they agree on, yeah, so they agree on that. Next, once they agree on a certain state of
the world, they are able to collaboratively perform actions. And it’s all built on distributed
ledger technology – for the Hackathon, we specifically built on top of BigchainDB 2.0,
and it was quite a hassle back then because it was still in the alpha phase, we had to
collaborate really a lot with the Bigchain team to basically pull it off. So, once you have all of these autonomous
machines, we specifically chose drones for the Hackathon, they can start doing all kinds
of stuff. So think about planetary observations, even building buildings – so they deliver
resources, they stack ‘em up, put ‘em all together, search and rescue. So think
about a new species of drones that are flying around, roaming, constantly looking for people
that are in trouble, and trigger some action, so that other drones come and help them, or
save them if they are in desperate need. Or wildlife protection – so some drones hunting
poachers basically, the hunters being hunted. But yeah, so once you have these machines
collaborating with one another, we can do all kinds of crazy cool shit. Sorry for cursing
there. So, what did we do at the Hackathon? We actually
brought a few hardware, we brought rolling drones, we brought AR Parrots, we weren’t
allowed to fly them by (inaudible), but we used the small one (inaudible) exactly. And
what we did was we retrofitted these drones with Raspberry Pis, and on top of that we
ran a whole BigchainDB node. So each drone has its own node, they share information with
one another, whenever one drone sends an update to the ledger, every other drone automatically
knows what’s going on. And what we built was, we built a grid. So
every drone has the same view of the grid, and they are just flying around. We simulated
search and rescue specifically. So they are flying around, they identify a point of interest,
so someone in trouble for example. And they send a transaction to the ledger. So now every
other machine that’s participating in this network knows that there’s someone in trouble
at a specific point. So they send, once it’s updated to the ledger, another drone that
is able to carry this person can go, go to the spot, pick up the person, and save them. So that’s just one small piece that we built
at the Hackathon, but we’re here for Nature 2.0. And the whole concept was you have drones,
but in order for these drones to fly, they need energy, right? So, you have other species
of machines that also participate in this network. You have solar energy, basically
plants, so these drones can just come, fly themselves in, and then be on their way, continue,
always continue moving – doing their mission. And then, 3D printing factories, so whenever
a drone is damaged, they go there, if they can still fly, they get repaired by the 3D
printer, or if they are totally dead, some other drone can pick them up, deliver them,
get them repaired, or just you print new drones – new drones to be added to the network. And lastly we have beacons as well, so beacons
we place in areas that are more popular, so if there’s an area people always go, instead
of having drones flying around you, you can just put a beacon there, and it’s always
checking if someone is in trouble. And once you connect all of these machines together,
you have a whole ecosystem of machines that are self-sustaining. And they can always keep
on operating, so the drone delivers resources to the solar energy plant, and so they live
in symbiosis (did I pronounce it right?). Same for the 3D printing factory – it needs
resources. The same drones that got built there, are the ones that bring resources to
the 3D printing factories. So yeah, that’s how you close the whole ecosystem, and it’s
fully self-sustaining. Alright, so same as the Hackathon, we have
been busy basically validating the concept, so have the initial proof-of-concept set.
And we have been busy talking to all kinds of, basically everyone that we know, or that
we find interesting to participate and contribute. So the green tags are areas we’ve been,
and physically visited and met with people, and the yellow ones are people we’re contacting,
and especially the one in Massachusetts is the guy that basically inspired the idea,
his name is Eduardo Castello, and he’s been researching this topic for some time now.
And, yeah, we’re trying to figure out how we can collaborate and basically bring this
closer to reality. So, now Dan will take over. Dan is responsible
for R&D. Thank you very much. I really like how we
are talking today about visions, and about how this could be about future ecosystems
– it’s nice, but this is one moment when I really didn’t click with Jan-Pieter at
the beginning, because I really think – I have a business background, I have been a
hacker for a very long time, but I also have a business background – so I have put everything
in the world of today, I have to understand how to make the business go forward.
(mic issue – lower? okay, fine. Higher, okay) So, after this, we’ve been talking , we’ve
been trying to understand what did we build? I’m trying, sorry, What did we build? What’s
the point of this? How are we moving forward? So, we started doing these things, it’s
called the pillars of Grex, it’s a very weird idea, but just stick with it. So, we’ve
built a world state consensus, that’s great if you have drones, if you have drones with
data. (prompt about microphone technique)
Oh my God – I think you can still hear me – okay sure. So, that’s great of an idea, but you need
data for that, you need some way to manage which drone participates in your network,
you need to afterwards be able to make sure that these drones are autonomous, so that
they can learn, they can take missions if they want. Afterwards, it’s really complicated,
it’s really annoying, and when you’re head of R&D, and you’re just thinking about
this for the entire week, start really going insane. So we split it up: these are the pillars.
The Grex network is just a distributed data network. It’s really IPFS-style (inaudible),
we’re still developing it, (inaudible) with a couple of (POCs?) we have built. But that’s
it, you need a decentralised data storage network in order to enable this. Then what
you want is you want, we are calling it market, by the way, these are all domain names that
we own – I love buying domain names – stick with it. So, market is just the decentralised
user access control that we have. In order to have people access this data, or the drones
to have access to it, you need to make sure that there’s some way of controlling who
has access to it. So you don’t want to have a central server, you don’t really want
to have some third party authority governing this, but what you really want to have is
this decentralised user access control. Right now we’ve thought about Tendermint, we implemented
with Tendermint, but we’re still thinking about it.
And on top of that, you need to have somebody who grants this control on this network. So,
something that is on a permission-less network, but can govern something that permissions.
First idea that came to mind was a DAO, so technically doable – we’re not at the
stage where we’re implementing that because we still need the use-case and the client
for doing this specific one. Then you have the world, this is the world state consensus,
it’s been a bit explained, just everybody agreeing on parse the data. So just make sure
that first you have the umpire’s data, and then once everybody understands what data
is, what’s the current state of the world, that’s when you have the parsed data. Then,
the beautiful part is decentralised federated learning using blockchain, which I’m going
to talk about a bit later. So, why is this so important to Nature 2.0? So we really tried
to come up with one way of explaining it: it’s called a decentralised infrastructure
for autonomous agents and multi-agent systems. Why is this interesting? Because in our words,
Nature 2.0 is a self-governing multi agent system composed by autonomous agents. See
the link? The part, the self-governing part, that’s
where we think is the AI DAO, we don’t have yet the skills or the technology to do that,
so we’re going to work on that later. After that, we were taking so, in order to
have a decentralised system of a world state consensus, you need data. If you have drones
for data: bad idea – regulations are horrible. And having drones flying all over the world
– people get terrified. We scratched that and we went: okay, what happens if combine
three pillars? So we took three of them and we were like:
oh, wait, what about satellite constellations? So currently there is a lot of companies working
on satellite constellations, we have OneWeb, SpaceX – those two are focussed on internet.
There’s Planet – has a beautiful idea of scanning the world once every day – everything.
So, you have this data now. Okay, but the frequency is only one time a day. How do you
incentivize people to have higher frequency of data updates? You incentivize them to collaborate
on a constellation because if you want to have total global earth coverage really fast,
you need at least 50 really amazing satellites, but probably more. So we thought that if you
have a decentralised network in space, where everybody can contribute based on their own
satellites, and they own their own satellite, but the data is public because everybody can
just share the amount of data there. Okay, that’s better, but now you also need the
decentralised ground stations. So, ground station network where you can distribute the
load of the network because only having one single point of retrieving data is horrible
because data costs a lot in space. And we’re also not storing anything in space because
it’s expensive and it’s useless if nobody can access it. This is where token economics
– it’s not even token economics, it’s a very stupid token engineering, I have to
be honest here. You cannot buy into the system, you can only get access to the system if you
contribute – either a security audit, a launch of a satellite, or let’s honestly
say maintenance of a ground network. But also, if you have (what?), if you have your own
ground station, you probably have servers, you probably have access to very decent storage.
What happens is we open a relay. Now you actually can sell this access because I’m pretty
sure nobody that produces parts for satellites is interested in storing data about space,
I mean it’s worthless to some of them. So, sell it on a secondary market, now you generate
a revenue form for them. So now you can distribute the load of a network of a ground station
network. Seems to make sense, we’re validating this idea, there’s a very nice company called
ISIS Space in Delft, they do most of the launches that were in Europe here. We’re validating
it – it seems to make sense, we’re writing the specifications right now for it, but okay,
let’s go. Then, we are like: okay, this is a very far-fetched
idea: it’s not going to happen now, it’s going to happen in a while. What can we do
now because if you tell this to a client, because we are an enterprise solution company,
they’re going to be like: ‘yeah, that’s great, no’. So, we are going to go the academic
route for now, we thought that it makes most sense since this is a, everybody who works
at the company basically has a n AI research background except for me – I’m a weird
one, but we really love academia, and we think that such an approach, this is how it all
started, we should continue, and we should contribute back. So, as Haischel mentioned,
we’re in talks with MIT Media Labs to submit a paper there for review because we’ve got,
it seems to be in-line with them regarding this.
So, yeah, it looks weird, it’s not in a whitepaper for an ICO, don’t worry, don’t
get scared, it’s okay, it’s not a scam. So, we’re just going to going to build this,
but fortunately we talked to Dimi today again, and we found out there’s actually a small
part of this that we can build now, and since we’ve built the World State Consensus at
the Hackathon, it’s just the concept that we’ve built and we’ve proved that it works.
We’re going to document that academically, submit it. But then we realised that doing
distributed federated learning on blockchain is not that hard from our perspective of our
knowledge. So today we’re like: ‘oh, we’re going to build this stuff’. So fortunately,
in the next paper, because this is just a description of what the concept is, and what
technology we used. How you can do distributed federated learning
on BigchainDB. So fortunately you are going to see an update from us in the near – three
to four months, fortunately by December, something like that, with an actual proof-of-concept
of this now. Fortunately, yeah, I don’t think it’s going to be that hard from our
side to do it. In order to do this, we really want some support
from people who are interested, so it’s really not important which way you consider
it, but if you know of a use-case, of such a distributed system for multi-agent system,
it’s really close IoT level, so if you also have things regarding that, talk to us. We
really want to validate this idea in as many sectors as we can. It doesn’t mean like
a full on partnership, just tell us if we’re doing something right, or where are we doing
something wrong – this is really important for us. Then, clients, if you actually know
people who are investing for like 5 to 10 years from now, who are willing to start working
on this technology now, again, we’re an enterprise solutions company – we want to
know. We want to really find the clients for this. But now, which is most important, is
we really want academic support, as we are writing a paper – I’ve wrote master papers,
but not the same as this level of really doing it proper with a good review and everything,
so whoever has the support for this, please talk to us. You will find us here: I am head
of R&D, please anything tech-related, coding deployment, servers , IoT. Haischel is the
founder and the CTO, so, again, anything like programming language-wise, protocol level,
great to talk to him. Alexander also is responsible for Kryha, so everything related to enterprise
solutions, just talk to him, anything related to blockchain, great as well.
Thank you so much. (applause)

One thought on “Haischel Dabian & Dan Acristinii – Kryha – Blockchain on Satellites”

  1. Haischel Dabian & Dan Acristinii well done, I love to see young blood on the blockchain, the world belongs to the young fellows such as yourself, keep up the great work.

Leave a Reply

Your email address will not be published. Required fields are marked *