One of the basic problems which have to be solved by Unmanned Automated Vehicles (UAV) involves the computation of a motion plan
that would enable the system to reach a target given a set of initial conditions in presence of uncertainties on the vehicle dynamics and in the environment. Recent research efforts in this
area have relied on deterministic models. To address the problem of inevitable uncertainties, a low-level control layer is typically used to ensure proper robust trajectory tracking. Such
decision-tracking algorithms correct model disturbances a posteriori, while the whole movement planning is done in a purely
deterministic fashion.
We argue that the decision making process that takes place during movement planning, as performed by experienced human pilots, is not
a purely deterministic operation, but is heavily influenced by the presence of uncertainties and reflects a risk-management policy. This research aims at addressing these uncertainties and developing an optimal control strategy that would account for the presence of
system uncertainties.
The underlying description of UAV trajectories will be based on a modeling language, the Maneuver Automaton, that takes into full
account the vehicle dynamics, and hence guarantees flyable and trackable paths and results in a discretized solution space. Two
optimal control problems, a nominal problem omitting uncertainties and a robust problem addressing the presence of uncertainties,
will be defined and compared throughout this work. The incorporation of uncertainties, will ensure that the generated motion planning policies will maximize the probability to meet
mission goals, weighing risks against performance.
Identifer | oai:union.ndltd.org:GATECH/oai:smartech.gatech.edu:1853/6904 |
Date | 18 April 2005 |
Creators | Topsakal, Julide Julie |
Publisher | Georgia Institute of Technology |
Source Sets | Georgia Tech Electronic Thesis and Dissertation Archive |
Language | en_US |
Detected Language | English |
Type | Thesis |
Format | 1968894 bytes, application/pdf |
Page generated in 0.001 seconds