Adaptive Variables for Declarative UAV Planning

Abstract

Unmanned Aerial Vehicles (UAVs) are an important subset of autonomous robotics, offering unique opportunities in domains like merchandise delivery, geographical survey, and disaster recovery. The planning layer of UAVs is made up of high-level directives that instruct the system on how to achieve the plan's goals. UAVs execute their plans in the physical environment, and thus the plans must adapt to changes in the dynamic context. In this paper, we present a simple programming abstraction, adaptive variables, to declaratively define adaptation for UAV flight plans in a dynamic context. Building on top of a declarative language for expressing UAV flight plans, adaptive variables can change during a UAV flight based on predicates over physical data. We implement adaptive variable for Paparazzi and demonstrate its usefulness in adaptive UAV planning with the NPS Simulator.