UC Doctoral Student Develops UAVs with Human Reasoning and Logic

Nick Ernest, UC

College of Engineering and Applied Science 

CEAS

aerospace engineering

PhD student, has been collaborating with the

US Air Force (USAF)

and the

Air Force Research Laboratory (AFRL)

on research into autonomous control of next-generation UCAV’s (unmanned combat aerial vehicles). He presented the novel intelligent system he developed, called LETHA (Learning Enhanced Tactical Handling Algorithm), at the

SAE 2014 Aerospace Systems and Technology Conference (ASTC)

September 23-24, 2014.

Ernest began working with the US Air Force (USAF) at

Wright-Patterson Air Force Base

in 2008, initially as an undergraduate co-op student at the

Landing Gear Test Facility

. Upon presenting his initial co-op research findings in 2011, Ernest was awarded a $220,000

Dayton Area Graduate Studies Institute (DAGSI)

fellowship by the USAF/AFRL to support his efforts over a 3-year span. He then moved to work in the AFRL and through his graduate co-op rotations and DAGSI fellowship, he has been with them ever since.

With his DAGSI fellowship funding, Ernest successfully developed an intelligent system that trains and controls a UCAV (or set of UCAV’s) capable of autonomously completing missions. This all occurs inside an aerial combat simulation environment he has created for his research. Inside the simulation these UCAVs have the ability to fly through a combat zone, utilize self-protection systems, and destroy enemy air defenses and critical targets.

Currently, these simulated UCAV’s are equipped with a limited supply of self-defense missiles—which can shoot down incoming enemy missiles as well as air interceptors—and a laser weapon system that burns through enemy missiles over time but has a maximum charge that quickly depletes. This charge needs to be renewed throughout the mission, and different missiles require different amounts of the laser’s charge to be destroyed. The UCAV’s are also equipped with air-to-ground weapons which can destroy critical enemy targets and surface-to-air missile sites.

Enemy forces are also equipped with electronic warfare stations which can block all communications, and time requirements are enforced on the mission representing how long the squadron has until additional enemy forces arrive. Thus, the problem lies in the mission’s dependence on intelligent routing of the UCAV’s throughout the battle space and the proper usage of the limited and renewable defensive systems.

The artificial intelligence that Ernest has developed is called the Learning Enhanced Tactical Handling Algorithm (LETHA). LETHA is a Genetic Fuzzy Tree which is Ernest’s new type of genetic fuzzy system. These algorithms combine human reasoning and logic with survival of the fittest and evolution. A population of controllers is run through a simulation, each of their performances are measured, and then a new generation of controllers is created based upon these results where the strong have the highest chance of producing offspring. The resulting controller more resembles a doctrine of air combat written by an ace pilot rather than a series of mathematical equations and nodes.

Ernest explains, “This research is the first application of genetic fuzzy systems to UCAV control, and the results obtained thus far are incredibly groundbreaking. The techniques applied to this problem are very advanced; the algorithms have to do a great deal of learning and the resulting controllers go against extreme opposition. This opposition takes the forms of an overpoweringly strong enemy force, imperfections and uncertainties in the friendly weapon systems, and the presence of electronic warfare stations which can temporarily shut down all communications between the friendly forces.

The ingredients to this work have been around for decades and proven in industry, and this combination of them is applicable to quite truly almost any problem. LETHA is specific to this aerial combat problem, but whether the application area be destroying enemy air defenses, automated drafting and design, or large-scale resource management, a Genetic Fuzzy Tree could be created to bring intelligent control to almost any cyber or cyber-physical system.”

Ernest has been working closely with his UC-CEAS advisor, Kelly Cohen, PhD and professor in the CEAS

Department of Aerospace Engineering and Engineering Mechanics

, throughout his fellowship. He reflects, “I’d like to thank both Dr. Kelly Cohen for his guidance and support, and Dr. Corey Schumacher (AFRL and DAGSI program manager) as well as DAGSI for giving me the opportunity to perform this research.

The techniques I utilize are extremely taboo in my field. They are vastly different than methods that solve comparable problems, utilizing much more human reasoning and common sense and rely on much simpler mathematical algorithms. I have gone ‘all-in’ with these methods and it has truly paid off. We have achieved success in every application, often either surpassing desired benchmarks in problems that all others have failed, or when comparable solutions are obtainable, my work is more accurate and many orders of magnitude more computationally efficient.”

Dr. Schumacher applauds Ernest, “I am extremely satisfied with the work that Nick Ernest and Professor Kelly Cohen are doing under this DAGSI fellowship. Nick has shown great energy and ability in developing solutions to this particularly challenging research problem, and has made great progress. His novel approach using a genetic cascading fuzzy strategy to develop autonomous control algorithms for the UAV team is very promising. We look forward to integrating Nick’s approach into our AFRL simulations later this year.”

Ernest has worked with UC and obtained a provisional patent for LETHA. His achievements only continue to accumulate and, this past July, he opened his own company,

Psibernetix Inc (www.psibernetix.com)

. Ernest is greatly interested in continuing work at AFRL, and will also pursue other applications in both the public and private sector upon graduation from his PhD program in spring 2015.

About the SAE 2014 Aerospace Systems and Technology Conference (ASTC)

The theme for the

SAE 2014 Aerospace Systems and Technology Conference (ASTC)

is Propulsion, Power and Avionics - Performance through Integration. The conference combined the strength of the Power Systems Conference (PSC), Aerospace Electronics and Avionics Systems Conference (AEAS), and Counterfeit Parts Avoidance Symposium (CPAS) with new tracks for space and unmanned systems—for which Ernest presented.

The ASTC provided an open technical forum for aerospace systems professionals, especially engineers working in avionics, environmental, human factors, power systems, propulsion, safety, systems, space systems, unmanned systems, to gather valuable technical knowledge, insight, and information on emerging and applied technologies with respect to commercial, military, space, and unmanned aviation. Technical sessions, lively panel discussions, and keynote presentations were highlights of this premier event that is essential for engineers, scientists, designers, program managers, as well as government and research personnel. This event provided an invaluable opportunity for attendees to renew and develop important business relationships within the international aerospace industry.

Unlike other aerospace and defense events, this conference focused on the practical application and implementation of technology, and technology in development in commercial, military and general aviation. Participants included technical leaders such as:

Boeing

,

Airbus

,

Northrop Grumman

,

U.S. Air Force

,

Honeywell

,

Pratt & Whitney

,

Rolls-Royce

, and

GE

.