Behavior-based High Level Control of a VTOL UAV

Abstract

This work addresses the problem of providing the on-board system of an autonomous unmanned aerial vehicle (UAV) with the capabilities to plan and conduct complex missions with reduced off-board assistance. Such capabilities are referred to in this work as Intelligent Behaviors. In this work, three intelligent behaviors are proposed: Fly Home, Search Object and Object Tracking. The Fly Home intelligent behavior is intended to provide the UAV’s on-board system with the capability of autonomously returning to the starting point of a given mission through a safe path. The behavior was designed considering the absence of online path planning and world model modules. The proposed approach is reorganizing, adapting and performing the tasks previously executed by the UAV. The Search Object intelligent behavior is intended to manage a search mission planned offline, recognizing the areas in which the UAV is allowed to search and managing properly possible interruptions. To recognize the search areas, the total area is divided into convex cells and the Monotone Chain algorithm is used to calculate the perimeter of each individual convex cell. The Object Tracking intelligent behavior is intended to provide the capability of pursuing a moving ground object with the assistance of data provided by a visual recognition system. The motion characteristics of the ground object are estimated using a Kalman Filter. This work was developed with the ARTIS project at the German Aerospace Center (DLR), which aims to develop systems to autonomously operate rotorcraft vehicles.