Pose Estimation with Multiple Sources Using Evolutionary Algorithms

An important feature of heterogeneous multi-robots systems is the availability of sensing information coming from multiple sources or points of view. Team members can exchange localization information regarding a target, or provide extra sensor information to team-mates that lack of certain sensing capabilities. In this paper, we present two examples of multi-robot pose estimation in a team of robots composed by unmanned aerial and terrestrial vehicles. Information integration is achieved through an evolutionary-based vision algorithm. Experiments have been performed part on virtual-reality simulations and part in-field.