Future of Geospatial Intelligence

Future of Geospatial Intelligence : Darafei Praliaskouski Sheds Light

Future of Geospatial Intelligence Darafei Praliaskouski sits at the forefront of geospatial technology, currently working as the Head of Product development at Kontur Inc, a geospatial analytics platform that uses AI and new mapping technologies to provide visual, location-oriented information.

Darafei specializes in cartography and geographic information systems (GIS). He is the organizer of a local GIS community in Minsk, Belarus called byGIS and has also been a member of the OpenStreetMap Foundation (OSMF) since 2010, a non-profit organization aiming to support and enable the development of freely-reusable geospatial data.

Prior to Kontur, Darafei worked on a number of geospatial projects. He was a GIS engineer at Juno, an Israeli-Belarusian company that provided ride sharing services in New York, and he also worked for other, well-known Belarusian technology companies, such as Wargaming and Maps.Me.

RR: Can you tell us a little bit about your background?

So I’ve been working in geospatial since 2008. At the end of school, I started looking into maps and how to work with them, and then tried to use open source solutions to build on top of that.

Then I started a local GIS community called byGIS and we started taking part in various GIS projects. So I worked with all kinds of projects that were being developed in Belarus and Minsk. At the time, I was working with Juno, a taxi service in New York, and helping them develop. One of my responsibilities was developing their geospatial solutions.

Somehow I got deeper and deeper into the code, and now I’m the head of product at Kontur. I decide which way the company and the product goes and what we are doing as a provider of geospatial solutions for disaster management.

RR: What brought you to work so closely with maps, and more specifically, geospatial analytics?

So back then, I was using Linux and at the time, we didn’t have Google maps, online maps, or the Internet for that matter. The maps were usually on the Windows machines as a local application. Then I switched to Linux, but the application didn’t work on it. So what can I do? Is there an open source solution? I found OpenStreetMap and started taking part in that. And you know, it’s addictive.

JP Morgan’s Head of Machine Learning Strategy on Big Data & ML

The software was pretty bad back then and I was aware of a number of Python scripts that were easy to use. Basically, you can just go open the code and see what’s inside and you can actually understand without the levels of abstractions with interfaces, objects, or anything else. It was perfect learning material and it also led me to go and try some of the things that I wanted to use to answer my questions.

RR: What is the first major project that you completed, geospatial or otherwise?

One of the first projects that I completed when I joined my first mapping workplace was to redesign all the manual tools in the mapping software into something that was more workable. I saw people who were actually splitting their shape files by hand, manually selecting thick yellow lines or thin red lines for their maps.

So, we designed a language called MapCSS, which is like CSS, but for maps. With this new language, we could now say that this object is a primary road and that I want all primary roads to be represented by thick red lines. This could now all be done in a style sheet rather than painstakingly by hand.

RR: What firms do you think could greatly benefit from using geospatial data to guide their decisions?

80% of the data in the world is geospatial data.

Bloomberg’s Alt Data Founder Jeremy Baksht Shares His Insights on the World

So most data sets are actually geospatial data sets even if you don’t consider them to be. Whenever you have a patient’s list with street addresses, a phone book with street addresses, a contact list, a Zoom call with time zones, that’s all geospatial. In fact, it’s hard to find non-geospatial data. For example, one of the first satellites, LandSat, which was made specifically for farmers, was launched by the US government.

Future of Geospatial Intelligence

So farmers were pushing everyone to go to space so that we could see what was in the middle of these huge fields.

The easiest way to do that is to use satellites on a national scale. Actual firms use GIS to figure out the best place to open their business, and Kontur actually has a solution for that. We show you the most populated part of the city, where your competitors are, and where your customers are. You want to find a sweet spot where you can have customers but not competitors, all of that is also part of GIS.

RR: As AI continues to improve, do you think there ever comes a point when people no longer need to monitor the maps and AI can do it by themselves?

Well, it depends how distant in the future we’re looking at. In the distant future, there will be no difference between AI and natural intelligence.

Two Sigma, AQR & Interactive Brokers’ Data Head Steve Cannon with our CEO

However, there are still a lot of problems with the current level of AI. Identification is still a major issue. Identification is still more cost efficient when overseen by human beings. The machine cannot say, hey, this changing terrain is a building. What they know is that these certain points in space have this color. They don’t understand that this collection of points is actually a house. And that this house has this number on this street.

RR: What’s the biggest problem with geospatial analytics today?

The main problem with geospatial analytics today is data collection.

With satellite imagery, you can’t see what’s under a tree or house numbers. For that, you need to use cameras that are mounted on something like a car. But then you will be capturing only what’s around the car. You will not capture what’s around pedestrian footways or closed areas.

You can use a drone to solve this. But it’s sometimes not okay to have a drone flying in someone’s backyard taking precise images of their property.

RR: While we’re on this topic of data collection, are there any current issues with geospatial data violating people’s privacy?

Point72’s Head of Data Sameer Gupta Sits Down

Privacy is the top priority in Europe right now (GDPR). I believe it is the same way in the US as well as in California.

Geospatial data is a tool, and like any tool, you can do damage with it or you can do good. The problem is not with the tool. The problem is with the user. So how do you make everyone a responsible user? A lot of people are trying to solve this problem in different ways. There are a lot of blockchain solutions that are based on the argument that we shouldn’t trust anyone. That’s one extreme.

The other extreme is open data. Geospatial data is out there whether you want it or not, but there is certainly some amount of backlash and protest. Germany does not allow Google Street View because it was banned for privacy reasons. But there are still ways that things can happen.

For example, someone can have imagery on their private drone if they’re allowed to fly a drone. Recently, in Belarus, a lengthy registration process was created that gave people permission to fly a drone. People can apply for one permit, and then the next one, and so on. It’s not actually banned, but you have to go through the military and send them all the photos you took so they can review them.

Data Scraping is Revolutionary

RR: How do these machines see the world differently from people?

There’s a way that humans perceive a map, which we’ve been doing for quite a long time, and then there is a way that machines do. The way they perceive what’s happening really doesn’t have to look like a map. You can have some values growing and fading in time like with time series data. People have difficulty perceiving the time series because you need to visualize it in 4 dimensions. But a machine doesn’t care how many dimensions it has to process. It’s just data.

RR: What do you think the future holds for geospatial analytics?

The most interesting stuff that is happening now is that we are getting more and more computing power and processors are getting larger and larger in mobile, handheld devices. With this, you’re going to get a massive amount of data.

Previously, only huge companies had the ability to have a car with six cameras installed to take photos of a city. Now, anyone with a Tesla has a car with six cameras installed. Now, the only thing that they need to do is to record the data and put it into the workflow of cartographers. This type of data collection is growing more and more popular, and because of this, maps can constantly be updated by the 3D models of the world.

Future of Geospatial Intelligence Written by Kevin Ma

Edited by Gihyen Eom

Physics Nobel Prize Winner MIT Prof Frank Wilczek on String Theory, Gravitation, Newton & Big Bang

Future of Geospatial Intelligence