The ArconaCore framework comprises several intermediate-level units: recognition, navigation, and SLAM. These units share a common property: scalability. They function effectively on both modest mobile devices and powerful platforms such as remote servers. 

Recognition: Our developed recognition approach boasts capabilities comparable to neural-network-based solutions. However, unlike the latter, all processing steps are implemented algorithmically. Consequently, the total number of operations performed is significantly reduced. As a result, recognition achieves satisfactory performance using only the mobile CPU, freeing up the mobile device’s GPU for high-resolution AR content rendering and related tasks

Navigation: The navigation unit geopositions a mobile device by performing bundle adjustment using information from various sources: GPS, Compass, Recognized spatial anchors, Odometry (SLAM) unit. The precision of positioning increases as more information is received from these sources.

SLAM (Simultaneous Localization and Mapping): Our SLAM unit combines the extended Kalman filter with a bundle adjustment approach. Importantly, this unit ensures our solutions remain independent of specific platforms like ArKit or ArCore.

​

In summary, ArconaCore’s modular design allows it to adapt seamlessly across different hardware configurations, making it a versatile choice for augmented reality applications.