Landing a helicopter on a moving deck is a daunting task for even the most skilled pilot. The task
involves controlling the vehicle, tracking the movements of the ship, and estimating the future
positions of the deck. The difficulty of this task is the motivation for this thesis. In this thesis,
a method of estimating the relative deck states necessary for autonomous landing as well as a
method of generating valid trajectories to the deck surface are presented.
An unscented Kalman filter is presented which fuses measurements from a monocular vision
system with an inertial measurement unit, both of which are carried on board the vehicle. Using
the Kalman filter generated estimates, biologically inspired trajectories are generated to the deck
which satisfy the vehicle dynamical constraints and several terminal landing constraints.
The Kalman filter and biological trajectories were used to land a mid-size quadrotor on a static
and moving platform. Single run and aggregate performance of the Kalman filter are presented.