Ocularone: Exploring Drones-based Assistive Technologies for the Visually Impaired

Abstract

Fleets of Unmanned Aerial Vehicles (UAVs), also called drones, are becoming commonplace to support diverse applications in logistics and urban safety. These rely on advances in computer vision and Artifcial Intelligence (AI) for autonomous navigation and control, facilitated by onboard sensors and accelerated edge computing devices attached to their base stations. This article examines how drones can be used for social good to enhance the lifestyle of Visually-Impaired People (VIP) through navigational assistance and situation awareness, enabled through visual analytics over drone video streams. We propose Ocularone, a platform for coordinating and managing a heterogeneous UAV feet that ofers drones as buddies to guide users. It uses onboard sensors and edge accelerators, and two-way communication using gestures and audio prompts, to enhance the safety and autonomy of the VIP. We validate our early prototype using Tello nano-drones and Jetson Nano edge accelerators, and present preliminary results for several safety scenarios.