Vehicle networks for gradient descent in a sampled environment

Fish in a school efficiently find the densest source of food by individually responding not only to local environmental stimuli but also to the behavior of nearest neighbors. It is of great interest to enable a network of autonomous vehicles to function similarly as an intelligent sensor array capable of climbing or descending gradients of some spatially distributed signal. We formulate and study a coordinated control strategy for a group of autonomous vehicles to descend or climb an environmental gradient using measurements of the environment together with relative position measurements of nearest neighbors. Each vehicle is driven by an estimate of the local environmental gradient together with control forces, derived from artificial potentials, that maintain uniformity in group geometry.