Autonomous Path Planning, Geometric Nonlinear Control and Estimation, and Their Application to Underactuated Vehicles

Date of Award

August 2020

Degree Type


Degree Name

Doctor of Philosophy (PhD)


Mechanical and Aerospace Engineering


Amit K. Sanyal

Subject Categories



Current Federal Aviation Administration (FAA) regulations limit operations of Unmanned Aerial Vehicles (UAVs) to flights within visual line-of-sight of a human observer to ensure safety. However, with the advances in onboard detection, communication and control capabilities, such a limiting regulation may not be necessary for safe operations, because a high level of safety and stability can be achieved and guaranteed even during beyond visual line-of-sight flights of autonomous vehicles.

In this dissertation, we discuss the challenges for introducing autonomy in Guidance and Control subsystems of rotorcraft UAVs with fixed plane of rotors, and present solutions to overcome them. To address the guidance problem in a cluttered environment, we first propose an optimal trajectory generation scheme to create an optimal path between multiple position waypoints. We show how this scheme can work in conjunction with waypoint planning schemes that ensure an obstacle-free path. Further, we propose a discrete trajectory generator that creates an optimal trajectory between given waypoints, while considering flight duration as a free variable, utilizing the discrete transversality condition.

Considering that attitude of rotorcraft UAVs need to change such that the desired position trajectory is achieved, we look at the attitude control problem and propose a nonlinear geometric attitude controller that is robust to bounded disturbance torque. Analytical proof of robustness to bounded disturbance torque is also presented. To complement the guidance and control subsystems, an observer for the position and translational velocity is proposed in the form of a variational estimator, which can be obtained by taking an “action functional” constructed from an energy-like quantity and dissipating this energy. This observer enhances the autonomy and reliability of autonomous vehicles operating in uncertain environments where internal vehicle dynamics and environmental factors (like disturbance forces and torques) are not known. It is shown how this observer can be utilized for target localization applications.

We conclude this dissertation with discussions on a waypoint planner that utilizes reinforcement learning to ensure an obstacle-free path in a cluttered environment, which is the result of a collaborative work with EE & CS Department.


SURFACE provides description only. Full text may be available to ProQuest subscribers. Please ask your Librarian for assistance.

This document is currently not available here.