Nonlinear three-dimensional trajectory following: simulation and application

Abstract

In light of recent military requirements for unmanned and autonomous vehicles, research into methods of designing arbitrary three-dimensional trajectories and controlling aircraft along them has become vital. In this report, we explore two methods of nonlinear control for the purpose of following three-dimensional trajectories and paths. First, prior work on a dynamic feedback linearization exploiting the differential flatness of the ideal airplane is adapted with the intent of implementing it on a physical testbed in MIT’s Realtime indoor Autonomous Vehicle test ENvironment (RAVEN), but poor behavior—both in simulation and in hardware—under moderate levels of joint parameter uncertainty thwarted attempts at implementation. Additionally, the differential flatness technique in its pure form follows trajectories, which are sometimes inferior intuitively and practically to paths. In the context of unmanned air vehicle (UAV) flight in gusty environments, this motivated the extension of prior work on two-dimensional path following to three-dimensions, and simulations are presented in which the fully nonlinear controller derived from differential flatness follows a trajectory that is generated dynamically from a path. The three-dimensional path-following logic is actually implemented in RAVEN, and results are presented that demonstrate good vertical rise time in response to a step input and centimeter accuracy in vertical and lateral tracking. Future directions are proposed.