Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)

Graduate Group

Mechanical Engineering & Applied Mechanics

First Advisor

Vijay Kumar

Second Advisor

Mark Yim


This thesis addresses coordinated path planning for fixed-wing Unmanned Aerial Vehicles

(UAVs) engaged in persistent surveillance missions. While uniquely suited to this mission,

fixed wing vehicles have maneuver constraints that can limit their performance in this role.

Current technology vehicles are capable of long duration flight with a minimal acoustic

footprint while carrying an array of cameras and sensors. Both military tactical and civilian

safety applications can benefit from this technology. We make three main contributions:

C1 A sequential path planner that generates a C2 flight plan to persistently acquire a

covering set of data over a user designated area of interest. The planner features the

following innovations:

• A path length abstraction that embeds kino-dynamic motion constraints to estimate feasible path length

• A Traveling Salesman-type planner to generate a covering set route based on the path length abstraction

• A smooth path generator that provides C2 routes that satisfy user specified curvature constraints

C2 A set of algorithms to coordinate multiple UAVs, including mission commencement

from arbitrary locations to the start of a coordinated mission and de-confliction of

paths to avoid collisions with other vehicles and fixed obstacles


C3 A numerically robust toolbox of spline-based algorithms tailored for vehicle routing

validated through flight test experiments on multiple platforms. A variety of tests

and platforms are discussed.

The algorithms presented are based on a technical approach with approximately equal

emphasis on analysis, computation, dynamic simulation, and flight test experimentation.

Our planner (C1) directly takes into account vehicle maneuverability and agility constraints

that could otherwise render simple solutions infeasible. This is especially important when

surveillance objectives elevate the importance of optimized paths. Researchers have devel

oped a diverse range of solutions for persistent surveillance applications but few directly

address dynamic maneuver constraints.

The key feature of C1 is a two stage sequential solution that discretizes the problem so that

graph search techniques can be combined with parametric polynomial curve generation.

A method to abstract the kino-dynamics of the aerial platforms is then presented so that

a graph search solution can be adapted for this application. An A* Traveling Salesman

Problem (TSP) algorithm is developed to search the discretized space using the abstract

distance metric to acquire more data or avoid obstacles. Results of the graph search are

then transcribed into smooth paths based on vehicle maneuver constraints. A complete

solution for a single vehicle periodic tour of the area is developed using the results of the

graph search algorithm. To execute the mission, we present a simultaneous arrival algorithm

(C2) to coordinate execution by multiple vehicles to satisfy data refresh requirements and

to ensure there are no collisions at any of the path intersections.

We present a toolbox of spline-based algorithms (C3) to streamline the development of C2

continuous paths with numerical stability. These tools are applied to an aerial persistent

surveillance application to illustrate their utility. Comparisons with other parametric poly

nomial approaches are highlighted to underscore the benefits of the B-spline framework.

Performance limits with respect to feasibility constraints are documented.

Files over 3MB may be slow to open. For best results, right-click and select "save as..."