Poster Abstract: Data-Parallel Versus Task-Parallel Swarms for Small Unmanned Aerial Systems

Small Unmanned Aerial Systems (sUAS) are small IoT devices (weighing less than 55 lbs) that fly without human pilots aboard. sUAS use sensors and auto pilot software to hover, fly and land autonomously. Without humans aboard, lightweight sUAS can fly low, only meters above the ground, to capture detailed, high resolution images. These images can inform agricultural crop management by detecting crop health issues, such as nitrogen deficiency [6], [9], overcrowding [5], and soybean defoliation [10]. Other uses for sUAS include search and rescue [3], inspection for buildings and bridges [8], and photography. A mission is the unit of workload for sUAS. It includes aerial maneuvers, sensing, and computation required to complete a task. For example, in digital agriculture [4], a sUAS tasked with generating a report on crop health may conduct a mission that includes flying to a targeted area, capturing images of the crop, and applying machine learning to translate the images into a crop health characterization.