Insect-Inspired Navigation Algorithm for an Aerial Agent Using Satellite Imagery

摘要: Humans have long marveled at the ability of animals to navigate swiftly, accurately, and across distances. Many mechanisms been proposed for how acquire, store, retrace learned routes, yet many these hypotheses appear incongruent with behavioral observations animals’ neural constraints. The “Navigation by Scene Familiarity Hypothesis” originally insect navigation offers an elegantly simple solution retracing previously experienced routes without need complex architectures memory retrieval mechanisms. This hypothesis proposes that animal can return a target location simply moving toward most familiar scene any given point. Proof concept simulations used computer-generated ant’s-eye views world, but here we test familiarity algorithms training satellite images extracted from Google Maps. We find are so rich in visual information be even tortuous low-resolution sensors. discuss implications findings not only also potential development augmentation systems robot guidance algorithms.