UAV MISSION PLANNING AND EXECUTION VIA NON-DETERMINISTIC AI PLANNING ON ROS

Abstract

Unmanned aerial vehicles (UAVs) have been attracting civilian attention by its increasing number
of applications and decreasing cost. The vast majority of current UAV tasks still have a low degree of autonomy,
which can lead to lack of eficiency, safety, practicality and viability. This paper proposes using Probabilistic
AI Planning to improve autonomy and robustness in UAV applications. Our main contribution is a framework
that integrates Prob-PRP planner on ROS systems and enables the visualization, execution and monitoring of its
plans using Finite State Machine (FSM). On this article, we describe our framework and apply it for high-level
mission control on a simulated autonomous UAV.