Navigation of Scalar Fronts With Multirobot Clusters in Simulation and Experiment

摘要

Environmental fronts are transition regions characterized by a steep gradient in the value of a scalar field. Finding and spatially characterizing these fronts is critical for a variety of scientific applications; however, such work is challenging and usually performed manually, limited to localized characterization, and/or based on assumptions regarding the structure of the front. In this article, we present an automated multirobot control technique to find and spatially characterize general scalar field fronts. This is achieved with a multirobot system that adaptively sizes, orients, and translates based on data sampled by each robot in order to span the width of the front, and to move longitudinally along the front, reversing direction when the end of the front is reached. This span and scan functionality allows the cluster to spatially characterize the front's full extent and to enable comprehensive scientific study of its associated phenomenon. To our knowledge, this is the first comprehensive, adaptive multirobot front-following control strategy proposed for general front structures, and initial verification of its capability is provided through both simulation of behavior on a large-scale front and experimentation with a simple, small-scale indoor multirobot testbed.