This post is an excerpt from the ARtillry Intelligence Briefing: ARCore & ARkit: Accelerating Mobile AR. It pulls from the section of the report that examines the value of cloud-delivered mapping data to unlock AR’s potential. You can preview more of the report here, or subscribe to access the entire thing. 


Beyond mobile operating systems (Android or iOS), AR developer kits (ARCore or ARkit), and content (apps), data and shared cloud intelligence will sit in the background as AR’s unsung hero.

For example, geo-relevant data will play an important – though often overlooked – part in AR user experiences. It’s not widely recognized that Pokémon Go was built on the architecture of Niantic’s Ingress game, whose location tags were set over years and made the whole thing work.

As background, rudimentary AR such as Pokémon Go uses geo-tagged data to position graphics. More advanced AR, such as ARkit, conversely uses object recognition to map and register objects before applying relevant and dimensionally accurate graphics or informational overlays.

Though the latter is certainly a more advanced form of AR, it will still benefit from location data such as business/product details or coordinates. This one-two punch will especially be additive in apps for navigation, local discovery, tourism, retail and several other location-relevant use cases.

Fortunately a few startups have spent years building systems that collect, clean and optimize geo-data. Examples include Aisle 411 for store layouts and product data. Foursquare and Yext validate lat-long coordinates for businesses, and things like customer reviews and menu items.

These data will come in handy with local discovery AR apps like Google Lens, which lets users point their phone at a storefront to get useful info. Google currently accomplishes this through a combination of object recognition using Street View imagery, and its local business data.

Google’s visual positioning service, VPS, is probably a better example. Its unveiling included a controlled-environment demo for an AR-assisted screwdriver search at a Lowes store. But in practice, that requires product and blueprint data, as well as 3d scans of hundreds of stores.

Google of course has the deep pockets and computational muscle to pull this off. But the question is if AR developers will have access to such product or area mapping data. Without it, some AR apps could risk the fate of Apple Maps’ famous launch fail: having lots of flash but little function.

The AR Cloud

One solution is what Super Ventures partner Ori Inbar calls the AR Cloud. It’s a shared library of meta data to serve as an informational backbone for AR apps. It will particularly be useful in mapping — a key component in AR apps’ ability to identify, learn and augment their environments.

As background, ARCore and ARkit perform well in individual sessions of mapping a given space, using surface detection and functions explained earlier. But to map large (outdoor) areas, or come back to previously mapped areas, requires more computational muscle than smartphones offer.

An AR cloud can assume that burden, while also assisting AR devices that enter new areas, rather than exhaust computational muscle (and battery life) on chartered territory. It can register devices’ location dynamically, then serve mapping and object recognition data tagged to that location.

Altogether, an AR cloud gives AR apps more functionality, and advances the industry. Think of it like crowdsourcing to build an AR world map. Inbar likens it to Waze, in which drivers get valuable real-time routing information in exchange for traffic and speed data collected by their phones.

Speaking of driving, autonomous vehicles (AVs) will be a key component of the AR cloud. The computer visions that let’s AVs “see” the world relates to AR’s area mapping. So well-funded advancements in AVs will support AR technology, including lots of mapping data for the AR cloud.

Bottom line: The AR cloud gives app developers the capability of a Google (which will develop its own AR cloud), such as mapping data for Lowes Hardware. Developers can focus on UX and business models – the same “democratization” principle behind ARCore and ARkit.

But it will also come with some challenges and considerations. Privacy safety is a must for any location data that people share (a different report). And who owns the AR cloud? How is the data available and platform agnostic? These questions will be answered as mobile AR itself evolves.


For a deeper dive on AR & VR insights, see ARtillry’s new intelligence subscription, and sign up for the free ARtillry Weekly newsletter. 

Disclosure: ARtillry has no financial stake in the companies mentioned in this post, nor received payment for its production. Disclosure and ethics policy can be seen here.