An Energy-Aware Airborne Dynamic Data-Driven Application System for Persistent Sampling and Surveillance
Eric W. Frew, Brian Argrow, Adam L. Houston, Chris Weiss, Jack S. Elston · Procedia Computer Science · 2013
This paper describes an energy-aware, airborne, dynamic data-driven application systems for persistent sensing in complex atmospheric conditions. The work combines i.) new onboard and remote real-time, wind sensing capabilities; ii.) online models for planning based on Gaussian processes for onboard data and dynamic atmospheric models that assimilate Doppler radar data; and iii.) a hierarchical guidance and control framework with algorithms that can adapt to environmental, sensing, and computational resources. The novel aspects of this work include real-time synthesis of multiple Doppler radar data into wind field measurements; creation of atmospheric models for online planning that can be run inside guidance loops; guidance algorithms based on stochastic dynamic programming and ordered upwind methods that can adapt planning horizons, cost function approximations, and mesh representations of the environment; and throttling algorithms that manage the adaptation of the models and guidance algorithms in response to computational resources.