PROCESS
The main challenge for this project was creating an AR experience with a remote camera feed from a low-quality camera. It required us to write custom computer vision algorithms for recognizing markers, determining orientation, find the surfaces and planes and perform collision detection. Optimization was a major consideration, we had to reduce the latency and deal with the problems inherent with a remote camera feed over a local Wi-Fi connection.
RESULT
Developed a prototype application which builds on top of an existing FPV Drone and adds Augmented Reality layers. This functionality leveraged fiducial markers to allow users to create an AR racetrack in their home and to fly the drone through this race course with polished AR effects and time/lap tracking.
