Scape Technologies compares data from your phone’s camera with a cloud-based, machine-readable 3D map for augmented reality overlays that put existing setups to shame.
Pokemon Go sparked a new wave of augmented reality exploration, but London-based Scape Technologies wants to take AR beyond GPS-driven gaming to provide a more accurate representation of the world around us.
GPS can pinpoint location within a foot or so, but Scape can tell exactly where you’re standing by comparing data from your phone’s camera with a cloud-based, machine-readable 3D map. Its first product is a software development kit that will anchor AR content to specific locations.
Currently running trials in the UK, Scape has created 3D maps of 100 cities, and is live in London and San Francisco. The company also recently dropped a teaser for Holoscape, a massively multiplayer AR game.
We spoke to Scape co-founder and CEO Edward Miller about large-scale AR, what it means to „Scape“ an entire city, and why he believes 3D map data is „the scaffold of the 21st century.“ Below are edited and condensed excerpts of our conversation.
PCMag: You’ve said ‚3D map data is the scaffold of the 21st century.‘ What did you mean by that?
Edward Miller: Over the next few years, we’re going to enter an exciting new era of so-called „spatial computing.“ This refers to a new class of computing devices that will go beyond the boundaries of the 16-by-9 screen and operate within the physical world. Self-driving cars, drones, augmented reality glasses are all examples of spatial computing devices. It’s a field I started working in about nine years ago, first with interactive imagery robotic camera rigs to build early 360-degree video; and it’s [now] about to explode. Essentially it’s a way to connect the physical and digital worlds.
Which is what you’re doing at Scape Technologies. How did the company start?
I joined Entrepreneur First, which is where I met my co-founder and CTO Huub Heijnen. He’d studied his masters in robotics at the world-famous ETH Zurich, and was working as a researcher in Australia before coming to the UK to join the EF program. We got talking in a pub and realized we had the exact skills to collaborate on, and build the infrastructure for the next generation of maps for machines.
How does your AR SDK work?
At Scape, we’ve built a cloud-based Vision Engine that allows camera devices to understand their environment using computer vision. It doesn’t use 3D maps built and then stored locally, but taps into a „shared understanding“ of an environment, which allows content to be anchored to specific locations at an unprecedented scale.
This is city-wide AR, right? How many cities have you ‚Scaped‘ to date?
Collectively, we’ve gathered imagery data for over 100 cities and are currently processing a selective amount of that depending on our rollout schedule, R&D requirements, and financial considerations. We launched in London and San Francisco, and there are many more areas waiting to be rolled out in the coming months, including some geographical regions in Asia, where there will be some particularly exciting events next year.
How is Scape different from Snap’s World Lenses AR overlay?
Several companies have developed the ability to map a small area like a building, road, or landmark, and use that map for augmented reality. However, we need to think bigger than that. We don’t just want to map a building or a street or a block, but entire cities and beyond. This requires a very different approach, which has meant we’ve had to focus on scale from day one.
Explain how what you’re doing is different to Google Street View.
Google built its system for humans to be able to visually identify locations, but we’re creating an entirely different class of map for machines. Using Scape Technologies, devices will be able to understand, via visual features and the information stored in our server-side maps, and triangulate exactly where that device is.
Much more accurate than GPS?
Oh yes, definitely. GPS has the urban canyon problemwithin cities where tall buildings can occlude or reflect signals to the detriment of positions. Also there’s a level of inaccuracy with the phone’s magnetometer (compass) due to influence from steel-structured buildings, which rotate the signal in the wrong direction.
Many people have seen the layering of history on top of today’s world as an exciting use of AR, such as strolling into the Colosseum in Rome, holding up your phone, and ducking to avoid a gladiator’s dagger.
Yes, that’s a great example. We have already mapped out Rome. In fact, the Colosseum is often used as a pillar in research projects by the computer vision community as a benchmark to test internal accuracy localization. There are so many possibilities in this area, in Rome and beyond.
When people think about AR they often say ‚Oh, right, Pokemon Go?‘ Explain how what Scape is doing differently.
That app generated a large proportion of the revenue in AR today so it’s understandable how much attention it received. But do you remember the trailer they ran when the app first got released? The trailer promised full-scale AR and painted a picture of the world and what it could be. Essentially, I believe Scape can fulfil that potential and bring it to life, going much further than they were able due to the constrictions of GPS. We’ll be using not only cloud-based map generation, but deep semantics and networked devices to provide a true city-scale AR future.
Explain how your ScapeKit SDK is able to convert geo-coordinates (longitude and latitude) to Euclidean space coordinates on the fly.
This is how we can provide such precise localization mapping for content persistence. To do this, we work in what’s called Euclidean space, generating a „scene anchor“ within the visual/physical locale and relating to it via our system with transposed coordinates, which our SDK can translate into accurate measurements.
Still at the seed stage, Scape Technologies has raised $8 million to date. What was it about your pitch that struck a chord with investors?
Essentially, I explained that „We’re not building a map, we’re building the map.“ What we’re doing is so much bigger than augmented reality, robotics, or drones alone. We’re building the fundamental infrastructure that will connect all of these industries together, allowing machines and humans to have a shared understanding of the physical environment.
Before Scape, you ran a VR/medtech company called Medical Realities, which sought to democratize access to surgeons in the developing world, and created VR apps for clients including BBC, Jaguar, and the Royal Ballet.
I believe that immersive realities are going to have profound effects in many, if not all, industries, and I’ve done work across sectors exploring this in the past nine years. I stepped away from Medical Realities several years ago now, but it was important for me in learning about early experimentation on integrating new workflows within existing industries.
You’ve recruited computer vision PhDs from all over the world including Greece, South Africa, Switzerland, and Taiwan. Is the UK becoming a hub for this work?
The UK is becoming tremendously important in computer vision, driven by expertise from many of the top universities, including: Imperial and UCL [both part of the University of London], and the universities of Oxford and Cambridge. Many large companies, like Facebook and Snap, are centering their computer vision teams here.