An Integrated Risk-Based Strategy for Real-Time Decision Making and Mission Reconfiguration in Unmanned Aerial Vehicles

Abstract

Small unmanned aerial vehicles (UAVs) have been increasingly popular in the last years, being employed in a wide range of applications in diverse areas, including, for instance, military, medicine, and package delivery. These aerial vehicles are commonly energized by rechargeable batteries, and, as a result, their autonomy can be significantly affected by several uncertainty sources, including variable ambient conditions or mechanical mishaps. For this reason, the ability to predict their electric consumption for a determined path is critical to design UAV missions and complete them successfully. Additionally, due to the potential occurrence of unexpected events (e.g., mechanical failures or changes in the weather conditions), airborne implementation of real-time decision-making schemes for mission replanning is of utmost importance. Hence, this paper presents an integrated Risk-based strategy utilizing a State-of-Charge (SOC) prognostics algorithm to quantify the risk associated with a given path in terms of future consumption, intending to make real-time decisions on the UAV operation. More specifically, we consider a UAV mission framework with a discrete set of possible targets, where each one has costs, rewards, and failure risks, which is characterized by calculating the total probability using sample-resampling methods. Then, we seek the optimal choice by employing a Bayesian-Risk inspired approach whose outcome results in a compromise between risks and rewards.