Shark-inspired target approach strategy for foraging with visual clues

Abstract

Searching is one of the key activities to fulfill any foraging task. When working with robots, most searching strategies depend on image processing, which is one of the most time-consuming processes. In this work, white shark hunting behaviors inspired the proposed strategy for a team of foraging robots to search and approach the objects. Based on current perceptions, a robot can speed up the foraging process by switching between its sensors to approach its target, that is, by depending less on image processing. On the other hand, the visual clues or targets in the environment have different shapes and colors to indicate which task is available. Such targets are provided by a set of landmarks that change their color according to their availability. In particular, we can manipulate the delays to show a new available task in these landmarks. Each robot makes decisions about which type of target to search based on its experiences. Robots can double-check when they identify unclear images, which are also included in the image database. The proposed strategy for search and approach surpassed the hardware limitations allowing robot navigation with little visual information. A small improvement in the foraging mechanism allowed robots to achieve faster adaptations.