Toward Tactical Autonomy in Aerial Robotics

Abstract

This thesis addresses the issue of the generation of tactical trajectories, as is desirable in the case of robot autonomy in hostile environments. The generation of such trajectory currently lies beyond the purview of contemporary path planning research, which focuses on solutions to the shortest path problem and variations thereof. This novel guidance system is developed as a two-stage process. In the first stage, a graph-based search algorithm is used to generate a global trajectory, as is the norm for state-of-the-art path planners. In the second stage, optimal control techniques are utilized to develop local trajectories that are optimal with respect to user-defined cost matrices over a finite time horizon. The guidance system is implemented for use with quadrotor vehicles. It is tested in simulation and then in flight. In both cases, the trajectory generated by the system makes use of cover present in the environment, rather than fly directly to the goal point. Additionally included is a discussion on modern visual navigation techniques, which, if implemented, would enable autonomous operation in initially unknown environments. In combination with the guidance algorithm developed in this thesis, the resulting system would be fully capable of conducting missions in hostile environments