NTB >> NEWS >> WHO'S WHO AT NASA June 2005
	Dr. John Melton Co-Principal Investigator, Networked UAV Teaming Experiment, NASA Ames Research Center Moffett Field, CA In a joint project between NASA’s Ames Research Center and Dryden Flight Research Center (Edwards, CA), investigators completed flight tests over a "virtual" forest fire to evaluate new flight-control software to give uninhabited aerial vehicles (UAVs) the ability to autonomously react to obstacles as they fly pre-programmed missions. The tests were conducted over a remote area of California’s Edwards Air Force Base to explore cooperative flight strategies for airborne monitoring and surveillance of natural disasters, and for atmospheric sampling. Dr. Melton was one of the principal investigators involved in developing and testing the flight control software.

NASA Tech Briefs: How did the UAV project develop at NASA?

Dr. Melton: The NASA deputy associate administrator for Aeronautics Technology, Mr. Terry Hertz, provided funding for systems studies of several advanced concepts that could become new aeronautics initiatives. In late 2003, he asked the systems analysis groups at Ames, Dryden, Langley, and Glenn to put in concepts for one-year, skunk-works-like projects. One of the proposals that we submitted, which was an Ames-Dryden combination, was for studying the control aspects of multiple, small-coordinated UAVs. They liked our proposal, and we had one year to do some software development and some flight-testing, which culminated in the tests that were recently performed.

NTB: What was the purpose of the virtual experiments?

Dr. Melton: There were three different aspects of the Dryden flight experiments. The first was to flight-demonstrate the use of Boids algorithms – motion rules that, for example, you see in movies when they need to simulate a flock of birds or a herd of animals. You take a few very simple rules about spacing with your neighbors, staying aligned with your neighbors, and having the speed of your neighbors, so you don’t run into each other. We used those rules and applied them to our aircraft to help them transit from one corner of our test region over to another, opposite corner. A lot of people have used these rules in simulations of UAVs, but it was always our goal to do as much as we could in actual flight because there is often a large difference between simulation and practice. I think this is especially true when you’re dealing with aircraft because they have a lot of their own idiosyncrasies.

The second aspect was to do what we call dynamic replanning. This was done to simulate, in some sense, a forest service wildfire support mission. The idea is that you would have some futuristic “scout” airplanes that could be sent out after a satellite or ground-based sensor detects a lightning strike. These small airplanes could go out and look for potential places where fires might be starting. With our two airplanes, we established an initial grid of points that we wanted them to search. Our software intelligently assigned the waypoints between the two airplanes and started them off flying and traversing over their specific waypoints. We then simulated one of the aircraft actually finding something – the idea being that the aircraft would have some downward-looking infrared camera. We simulated it seeing a hotspot, so we immediately put both aircraft into an orbit about their current positions and then completely replanned the second airplane’s flight path with the addition of the untraversed waypoints from the first airplane. We kept one airplane orbiting over this virtual fire while the second airplane completed the rest of the search mission.

The third aspect simply was to test our ability to command these aircraft to not only get to their assigned waypoints, but to arrive at those waypoints at specific times. Our goal was always to be within ±2 seconds. Many scientists are looking at using teams of similar small aircraft for Earth observation and environmental monitoring tasks. For example, you could use a coordinated team to sample the air and take measurements in a vertical column. If you can fly the aircraft stacked vertically, you then have a really fabulous way for getting simultaneous data that can’t currently be easily obtained. To accomplish this, the airplanes have to be positioned precisely in both time and space. Although we performed this test using only a single aircraft, the ability to control an airplane and have it arrive at specific waypoints and altitudes within a second or two is a key technology that we wanted to demonstrate.

NTB: Is the new flight-control software an improvement over previous software?

Dr. Melton: The software we wrote was designed to provide several extensions to the very capable commercial autopilot system we purchased from CloudCap Technologies. We used their API to develop an enhanced situational awareness and mission planning GUI that sent commands to and retrieved data from the autopilot in each aircraft.

NTB: Could the software be used for other commercial applications?

Dr. Melton: There are so many missions for these small autonomous airplanes to perform, encompassing a huge range of possibilities. Some important examples are Earth observation, wildlife management, or monitoring fisheries. There is a current problem with ghost drift nets where illegal fishing trawlers, for whatever reason, cut their nets and leave them in the ocean. You have these huge nets floating in the ocean and collecting wildlife, depositing invasive species in places that they shouldn’t be, damaging sensitive reef areas, and generally endangering other animals.

Additionally, there are many homeland security aspects to this work as well. If a dirty bomb exploded or a biological agent was released, these aircraft could sample the air downwind in order to not only inform the populous, but also to help first responders make appropriate decisions for their own safety.

Other possible applications include the monitoring of pipelines and power lines, and all kinds of precision agricultural surveys. Disaster relief is also a possibility. There were several small airplanes that were used over the tsunami regions, some of which had the task of looking for bodies in inaccessible locations.
There are just tons of great missions for small, low-altitude UAVs. All of them are going to require some aspects of the controls that we are developing.

NTB: What is the next step for the UAVs and the new software?

Dr. Melton: Our project was a one-year project. Mr. Hertz recently visited General Atomics to see a NOAA/NASA demonstration flight using a NASA-specialized version of the Predator UAV called the Altair. NASA is also going to be flying the Altair next year in conjunction with the Forest Service in the first actual UAV-wildfire support missions.

I also think there is a groundswell building in the aviation and science communities for smaller aircraft. For us specifically, because we had only one year to go out and prove something, spur some thoughts, and give some feedback, we’re already off on our next mission.

Since the tests, we have had some inquiries from a variety of people. The one that we most likely will study is the flight testing of automatic collision avoidance software that is being developed at Dryden. You can afford to test collision avoidance software on these nearly expendable aircraft in ways that you probably wouldn’t on manned aircraft. You can actually run them close to each other and put them on direct collision courses, and test the software to make sure that everybody moves the right way in order to get out of each other’s path.

NTB: Did you partner with any outside companies in the development and testing of the UAVs ?

Dr. Melton: The two autopilot-equipped, 12-foot wingspan APV-3 UAVs were built by RnR Products of Milpitas, CA. CloudCap Technologies of Hood River, OR, developed the autopilots and ground station that we used. We also learned a lot by doing some cooperative work with the Idaho National Lab located in Idaho Falls, ID, and we received some good feedback from the U.S. Forest Service who gave us guidance on our missions and scenarios. Lastly, we also were assisted by the UAV Applications Center, which is a research institute located in the NASA Ames Research Park.

Contact Dr. John Melton at John.E.Melton@nasa.gov. Visit NASA's Aeronautics Research Mission Directorate at www.aeronautics.nasa.gov.