The main advantage of Arcona technological platform is an opportunity for the mass user to launch the AR projects remotely in any city on the planet. So far such tools are not offered by any of SDK existing in the market, including most popular at the moment Vuforia and ARKit.
- Our solutions, unlike analogs, are focused not on professional developers, but on ordinary people, - says Konstantin Zhukov, the computer vision engineer of Arcona. - We want to give to the user extremely simple tool to create and manage the AR projects without any technological skills. Besides, all popular AR platforms still require a "manual" binding of 3D content. It means that the developer has to arrive on the location of the project to scan space and to make correct positioning for each virtual object. And the main majority of AR systems do not work stable at open spaces.
For example, ARKit framework from Apple is built on very qualitative SLAM algorithms. Indoors it shows excellent results: quickly localizes the mobile device, detects horizontal and vertical surfaces and also detects return to already visited location. This information is necessary for the correct visualization of virtual objects. Nevertheless, on the open air ARKit can give serious failures. Especially under difficult external conditions, such as snow, rain or fog.
We held comparative testing of ARKit and our current algorithm Arcona Core on the snow-covered stage. From the provided images (graphs) we can see that though Arkit correctly carries out return to an initial location, but the trajectory of the movement is not correct and it jumps. It happens because of big visual similarity of images.
Arcona algorithm in such situation is much more reliable. Thanks to introduced by us the system of hybrid positioning on lines. It means that besides comparison of small feature areas on consecutive images as it is accepted in the classical SLAM systems, we detect and we compare more high-level features, for example, the lines belonging to objects. It allows to determine position of the device more precisely as lines belong to real physical objects while in the areas can be just accidental differences of brightness from shadows. The binding to real objects does tracking stabler and exact. Such approach is ideal for work on open spaces with a difficult relief and in unstable weather conditions. Now our team makes a number of experiments for additional increase in accuracy of a trajectory of the movement and correctness of return to a starting point.
And the most important difference of our system that, it allows not just to build separate trajectories for each project, but also coordinates them in the global map correlated to sites of the Digital Land. As a result the Arcona platform will give the chance to our users independently to create unique scenes in augmented reality and to get the most fascinating AR experience.