Challenges of AR Market in early 2018
The right approach for an AR project
The Augmented Reality landscape is full of players. ARCore, ARkit and Hololens are probably the strongest ones when it comes to AR markerless technology, but this field is full of companies with amazing products. It is not clear if there is a solution that can be considered the best one so far; each one has its own strengths and weaknesses.
As far as it concerns Google ARCore, it is for sure one of the most robust AR solutions, it is the evolution of the well known project Tango, but works with modern devices even if they are not Tango enabled (depth sensors).
On the other hand, Apple ARkit is disrupting the market in the latest months, and as Google ARCore, it doesn’t need devices with depth sensors and uses the same technique, Simultaneous Localization And Mapping (SLAM): a computational problem of constructing or updating a map of an unknown environment while simultaneously keeping track of an agent’s location within it.
To be honest, SLAM itself is a broad term, to be precise both ARKit and ARCore are Visual Inertial Odometry (VIO) systems. Tango was something slightly different, and Hololens systems are different too, providing probably the best tracking mechanism but being strongly hardware dependant.
The “battle” is currently taking place around us, and the scenarios are evolving very fast, still there is a good in-depth but understandable analysis of the solutions from a couple of months ago.
Besides all the fascinating technical aspects, there are still some practical open points that are keeping Augmented Reality markerless technology away from the market:
- Device and OS compatibility
- Hardware dependency
- Detection of vertical surfaces
Yes, AR with marker is almost mainstream and runs smoothly on tons of devices from all tiers. But what about a markerless solution, compatible with low-end devices, that can work with vertical surfaces while keeping the development pipeline uniform reaching a AR cross-platform solution? In these terms it seems impossible.
ARCore works on Android devices, it hasn’t got the same hardware limitations of Tango, but still it is compatible with just a couple of devices: Google Pixel and Samsung S8. There is a recent hack called “ARCoreForAll”, but but the name is actually misleading since it enables just three more phones to use ARCore.
ARkit works only on iOS devices of course, but it is compatible with iOS devices starting from iPhone 6s.
ARKit is making giant leaps forward and just few days ago it has been announced that vertical surface are now detectable with ARKit 1.5.
The main problem remains the hardware dependency and device compatibility. In Qbit, while working on Unical Uview project (Virtual and Augmented Reality product catalogue), we had the clear need to develop something that was compatible with as many devices as possible. Taking for granted that the gyroscope sensor is essential, we tried to reach all the devices with said sensor.
In addition, we had the need to maintain the develop pipeline uniform, so we tried our best to work with one solution compatible both with Android and iOS devices. We wanted to focus our efforts on the overall application features, not on the single differences between operating systems and different approaches on AR.
Last but not least, we needed to work with vertical surfaces, as lots of Unical products are designed to be hanged to a wall. All of these had to work with AR markerless technology.
As said, SLAM is a broad term. So we decided to give up some stability and try a less robust but more compatible solution. The tracking solution we utilized tracks arbitrary parts of the camera image. This can be the entire camera image, including any new parts that come into view, or constrained to a narrower area. It doesn’t require different viewpoints and tracks feature points in the environment to know where it is in relation to its immediate surroundings.
This allows the tracker to correctly place and track objects in the real world. What’s missing in this solution is the mapping part (that is the M in SLAM), so basically the device doesn’t actually build and store a three-dimensional map, but first uses the Inertial Measurement Unit (IMU) to simulate the placement of the object in the real space, then it uses feature points of the environment, extracted from the camera image, to track the 3D object correctly. With this approach, we gained another score point: the markerless system starts instantly without the need of mapping the environment.
It was not an easy design decision, but for this business use case we are sure we made the right call. We are already experimenting with other more advanced AR markerless solutions both with high-end smartphone and HMDs, such as DAQRI smart glasses and Aryzon. We look forward to find the right use cases for these technologies.
Meanwhile, feel free to try Unical Uview app. It is not only about AR markerless, there is an AR marker mode, two VR modes per product and a high quality 3D visualization with different interactions with each virtual product.
Download and try the app for Android
Download and try the app for iOS
