Well, lets see. We started about 5 years ago doing some electronics design work for a researcher at NASA Langley. We built a board that fit into a PC-104 stack that could generate PWM signals for up to 32 servos to control a "morphing wing" areodynamic test vehicle. The researcher we were doing this for suggested that my students get involved in the AUVSI Student UAV Competition (http://uav.navair.navy.mil/seafarers/default.htm) - I'm a faculty member in computer engineering at Virginia Commonwealth University in Richmond, VA.

We went to the competition for the first time in 2003. The competition allows the use of commercial autopilots (most teams use Micropilot units which Micropilot will actually donate to the univeristy team), but since we're computer engineers, I decided to have the students build their own autopilot - this also counts as their senior project in computer engineering. We used the modified prop-driven Mig "target drones" (FQM-117B for those who are familiar with them) you see in one of the pictures as our aircraft. Our first autopilot was based on the FMA Copilot unit and a single-board computer with a GPS unit that steared it around. It was kind of a "bang-bang" state-based algorithm that made the plane fly around looking like it had an old escapement radio system in it, but it worked! It had a cheap 2.4 GHz video camera on-board to see the targets. That year, we competed against 7 other teams using commercial autopilots and we won.

The following two years we improved the system. We went to a Linux-based single board computer with programmable hardware, went to full PID-based control algorithms, and a completely functional ground station with video and picture analysis capabilities written in C#. At the 2005 competition, the Admiral who gave the key-note speach at the awards ceremony said "the right vehicle for this competition is a helicopter" so we started moving in that direction. It took two years for us to perfect the helicopter system, but we used it in the 2007 competition. Its a MAH Fury Ion with 16 Amp hours of battery on-board, 810mm blades, and a pan/tilt video system with high res. pictures. All up weight is around 20 lbs. The flight control system is all our own, with the exception of the IMU. We use the MIDG II by Microbotics, as, in our experience, it is much more accurate and tolerant of vibration and manuvering than any other IMU we've tested - and its TINY.

At the 2007 competition, we were the only group using a helicopter, the only group that built their own autopilot that actually flew it in the competition, and the only group that was able to re-task the vehicle in the air to find the "pop-up" target that was given in the middle of the flight. We also had the highest resolution pictures of all of the targets. We got second place - primarily because we finished about 5 minutes after the 40-minute time limit, but the team was delayed in taking off because someone in the pits was stepping on our 900 MHz telemetry link. However, the competition organizers didn't give us the time back. If we would have been able to down-load the pictures in real time, we would have done better on that part too, but we didn't have time to get the planned 2.4 GHz Wi-Fi link for picture downlink working.

In addition to the competition work - done by senior CPE students, I have several graduate students working on sponsored research in the UAV area. One project, again with NASA Langley, involves the development of an autopilot system for their "Generic Transport Model (GTM)" - the model with the NASA "meatball" on it that looks just like a B757. This is an aerodynamic/controls system research bird that is flown from the ground, but needs an autopilot system on-board to return it to the takeoff point if the RF links are lost. No small commercial autopilot system meets the stringent weight, power consumption, and integration/interface needs of the GTM aircraft so we've been building one for them. The jet with the twin-tails (a PCM Models DV8R) is our test aircraft for that system. It too uses the MIDG II IMU and at this point, our autopilot can fly the aircraft at airspeeds from 60 to 120 kts and can also roll the aircraft on command and fly the waypoints inverted (to demonstrate robust control and recovery). We are in the process of installing that system into one of the L1011 models (like the one with the "meatball" in the other picture) that the GTM project uses for pilot training and electronics checkout...

Bob

ps Oh, the grey and orange delta-wing aircraft is an old "ExDrone" we were given by one of our contacts in the Army at Ft. Eustis. We added landing gear to it (normally it was catapult launched) and flew it several times, but we've never had a real application for it so we've never put an autopilot into it...