Liama Conquers the Atlantic

From Newfoundland to Scotland in 26 hours

By Rick Allison | Photos by Ron Bennett and Bill Vaglienti

As seen in the April 1999 issue of Model Aviation.

On August 21, 1998, the first unmanned autonomous flight across the Atlantic was completed. This historic event was primarily the work of a small research company called The Insitu Group, led by founder Dr. Tad McGreer.

The vehicle used was one of the group’s model-sized Aerosonde meteorological RPVs (Remotely Piloted Vehicles), dubbed “Liama” after the Latvian goddess of good fortune.

The launch point was Bell Island Airport, just off the southeast tip of Newfoundland, Canada, on August 20 at 7:59 a.m. local time (9:59 UTC [Coordinated Universal Time]). Recovery was made in a grassy field in South Uist, Outer Hebrides, Scotland 26 hours and 45 minutes later, at 1:44 p.m. local time (12:44 UTC).

The transatlantic flight was sponsored by a consortium of Environmental Systems and Services (of Australia), the Australian Bureau of Meteorology, and the US Office of Naval Research, with additional support from the University of Washington, the Boeing Cor., and L-3 Communications.

Tad McGeer with ill-fated Piper. He’s head of Insitu Group and principal designer of the Aerosonde project.

Neither the Aerosonde nor The Insitu Group (Insitu means “on site”) are new to these pages: Model Aviation published an overview of the Aerosonde development project in the May 1996 issue. Readers may recall the many direct links that the Aerosonde has to its aeromodeling ancestry:

The wing is a box-stock 900-square-inch Genesis thermal-duration glider wing with a Selig 7037 airfoil from RnR Products of Milpitas, California. The rest of the airframe, designed by The Insitu Group, is also produced by RnR

The pusher powerplant is a substantially modified Enya 1.20R, running on 100 low-lead avgas (airplane gasoline). The direction of the engine rotation is reversed to allow the use of a regular tractor propeller (APC 20 x 8) in a pusher configuration. A Walbro WT332 carburetor is used, with a custom intake manifold and a custom-made inflight mixture control.

A small 30-volt brushless generator modified from an electric RC car “can” motor is turned from the engine drive washer by a belt-and-pully arrangement. A capacitive discharge ignition system from C&H Electronics is paired with a custom ignition-timing system to provide the spark.

The flight control surfaces are actuated by stock Futaba and JR servos and hookups.

The Aerosonde configuration is a twin-boom pusher with an inverted V-tail. This allows for the correct weight distribution of the relatively heavy fuel and avionics payload, and provides the all-important sensors up front with clear air in which to work. The operational weight range is 16-26 pounds, depending on fuel load and the avionics package carried.

The Aerosonde is launched under the manual control of an RC pilot (from a specially designed launch rack affixed to the roof of a moving car). At altitude, control is passed to the autonomous portion of the flight-control system—basically a PC program.

Ross Hoag and Kip Jackson power-up avionics box and tape nose cone.

An RF modem handles the receiver job, and the ground-control unit—a laptop PC patched to another RF modem and amplifier/antenna unit—is used to input GPS waypoints and altitudes. They create a flightpath for the Aerosonde to track with its onboard GPS receiver and altitude sensors.

As long as the ground-control system is within range of the aircraft, new waypoints may be entered at any time, and the flightpath may be modified. Once the aircraft passes out of range, it becomes truly autonomous.

Recovery is a reverse of launch; the stored GPS waypoints lead the aircraft to within radio range of the ground control system, and new waypoints are entered to bring the Aerosonde within visual range of the ground station. Once the recovery pilot acquires the aircraft visually, control is passed from the program to the buddy box console for the landing.

The Aerosonde’s primary role is to gather meteorological data, which may be collected and stored or transmitted back to base via the downlink telemetry system when the aircraft is within range.

Engineering data includes engine rpm, cylinder head temperature, exhaust gas temperature, the commanded servo positions of all servos, true airspeed, barometric altitude, aircraft altitude and position from the GPS receiver, battery voltage, battery charger status, generator output voltage, and avionics box temperature.

Meteorological data (in the standard configuration) includes wind speed and direction calculated by differential GPS, inside and outside temperature, barometric pressure, and relative humidity.

Postflight inspection: Greg Lipski in middle, Steve Huffman on right, trying to remove Liama’s nose cone.

The stunning accomplishment of this robot Lindbergh should be doubly gratifying to modelers. Besides Liama’s model aircraft heritage and extensive “off-the-shelf” modeling technology and products used in construction, two key members of the six-man Insitu Group transatlantic crew are accomplished RC modelers and AMA members!

Senior test pilot and mechanic Kip Jackson is a well-known Seattle-area RC modeler and longtime FAI RC Pattern competitor. He was not only the launch pilot on the records flight, but he helped develop many of the Aerosonde mechanical systems.

Aeronautical engineer and backup pilot Bill Vaglienti is a former RC Pattern competitor and won an AMA scholarship in 1991. He graduated second in his class at the University of Washington, and joined the Insitu Group in January 1997. Bill was the recovery pilot who brough Liama home safely.

I can’t think of a better way to let you share the trials, troubles, and the ultimate feeling of exultation than to include Kip’s and Bill’s accounts of the flight. As told by Kip:

Dr. Tad McGreer, Insitu Group electrical engineer Ross Hoag, and I arrived at Bell Island, Newfoundland on Friday, August 14th. News of the attempt had preceded us and the reception was very warm; the fine hospitality continued through the week. (Special thanks should go to Gordon and Marilyn Shea, who run the Island Hideaway Bed and Breakfast, our base of operation.)

Early Saturday morning, Ross and I unpacked and assembled the first two aircraft, called Thumper and Piper. We had them ready to go for the following morning, but Sunday’s weather was not good enough for an attempt, so we put Millionaire—our test-bed aircraft—together and headed for the runway at Bell Island Airport for some radio testing, and also to just fly around and get used to the place.

One of the four Aerosondes in cartop rack, awaiting launch.

The wind was heavy (as it would be all week), but straight down the runway. We launched and I flew Millionaire around for a while to look for “nulls” in the radio link. We found one almost immediately.

Off of the east end of the runway, as I turned from downwind to base, I temporarily lost control of the aircraft and had an unpleasant moment or two before Millionaire popped out the other side of the blind spot, and regained control.

I landed by cutting the approach in closer and shutting off the downwind leg short of the blind spot. We made a mental note to stay out of that area and headed back to wait out the weather.

At 4:30 a.m. Monday, the weather looked better, so we headed up to the “shop” (an old trade school building located on the hill overlooking the runway) and loaded Thumper into the car-top launcher before driving to the runway, where we were greeted by press and spectators.

After the usual last-minute checks, we launched Thumper and turned on the autopilot. Tad “flew” the aircraft from the laptop ground station, doing some standard speed sweeps to check the autopilot gains. Everything checked out, so he activated the program and sent Thumper on her way out to sea.

A few hours later we brought Piper down to the runway for launch. The idea behind the multiple launch was to double our chances of making it “across the pond” and maximize (or at least not waste) the weather opportunity.

Unfortunately, just after takeoff, Piper experienced a software bug we had never seen before and crashed into the sea right in front of us, the assembled press, and all the onlookers! Very embarrassing, but such is life in the experimental RPV business.

Tuesday and Wednesday, the weather was bad over the Atlantic, so Ross and I got bullet number three, Liama, ready to go. We also prepared our test-bed aircraft, Millionaire, for an Atlantic attempt by changing the engine and moving the wing position aft for weight-and-balance purposes.

On Thursday, the weather was good, so we repeated Monday’s double-launch scenario, with the happy exception of crashing.

With Liama and Millionaire headed out to sea, my work was done. So I headed for the ferry and ultimately home, to get started on my vacation.

About 3:25 a.m., Ross called with the news that Liama had made it! I couldn’t go back to sleep, despite having been up for 27 hours straight!

Tad McGeer in the van rented for a mobile launch-control center.

At the DERA (Defense Evaluation and Research Agency) missile test range landing site on South Uist in the Hebrides, some 3,200 kilometers from Bell Island, the landing crew—Insitu Group aeronautical engineer Bill Vaglienti, University of Washington acro/astro department electrical engineer Greg Lipski, and University of Washington aeronautical engineering graduate student Steve Huffman—was in place, waiting.

With Thumper long overdue and presumed lost over the ocean, the tension was heavy and growing. Bill’s account:

The long crossing took Liama well out of communication range, leaving us with no way of tracking her over the ocean. We would have to trust the autonomous nature of the Aerosonde to fly her preprogrammed course, which was designed to take full advantage of the prevailing North Atlantic winds.

Our simulations predicted a midday arrival, but vagaries in the weather could easily alter that by several hours. We were prepared and waiting tensely by early morning on the 21st.

The forecast indicated that the landing weather would be dismal at best. 25-knot winds and a low cloud base had me worried that spotting the aircraft would be a problem. Fortunately, the forecast was wrong; the wind was there, but the skies were CAVU (Clear Air Visibility Unlimited).

About 1 p.m. local, our ground station uttered a beep. All heads swiveled to watch the computer, and nobody breathed for the next few minutes while we waited to learn the status of the aircraft. It was soon clear that Liama was not only out there, but in good health!

Hearts were pounding now. Everyone else had done their jobs, and it was all up to us. Greg used the laptop to set Liama up in an offshore holding circuit while Steve headed out to set up the mobile ground station at the landing area. I waited long enough to make sure that Liama was on her way, and then grabbed my gear and followed Steve to the landing zone.

Greg kept us updated by handheld radio while we finished setting up the mobile station. When we were ready and Liama was in position, the radio link was transferred to the mobile station. A few quick computer commands had Liama “feet dry” and over land for the first time in 26 hours. She would be overhead in a few seconds.

The wind kept up a steady roar; it wouldn’t be possible to hear the Enya 1.20. We would have to depend on the telemetry to tell us when she was overhead. When it did, we turned and looked, and with overwhelming pride and relief, visually acquired the aircraft for the first time. I engaged manual control with a flip of a switch on the standard Futaba transmitter.

With so much on the line, I wasn’t even tempted to try anything fancy. The Aerosonde is a complex aircraft, and by model standards, is fast, heavy, and underpowered (26 pounds on 900 squares with a 1.20 for power!).

One short downwind pass, and a smooth turn to set up on final was all it took on the wide-open grass field. I shut the engine down and concentrated on flying smoothly as Steve called out the airspeeds. A few seconds later, Liama slid to a stop on her belly with barely a scratch. We erupted in celebration (and a few tired sighs).

Liama is the first unmanned aircraft and the smallest aircraft of any type to cross the Atlantic. For the overwhelming majority of her trip, she flew alone and unassisted, with only her programming and sensors for guidance. Postflight analysis indicated that 8-10 hours of her journey was in heavy rain.

In the future, when long-range satellite communications become available, such oceanic crossings may become routine. But for now, we have the satisfaction of doing what no one else has done.

The transatlantic flight was conceived as a demonstration of the Aerosonde’s operational capability. It capped an impressive list of recent accomplishments, including the first autonomous flight (including takeoff and landing) on February 25, 1998, and some 350 hours of successful field missions for various groups and meteorological services in the past year.

Concept testing is complete, and the Aerosonde project is in the process of developing a more-flexible and robust aircraft for routine commercial use, with the first aircraft scheduled to begin operation this year. Since only one of the four aircraft launched completed the Atlantic crossing, increased reliability is a major goal.

Since enroute communication and telemetry were not available for the attempts, the reasons why Thumper and Millionaire were lost (Piper fell victim to a software bug during the pass from manual to automatic control) are not known. However, any of the several low-reliability parts included in the design are assumed responsible.

The facts on Liama’s 26-hour, 45-minute journey: 3.9 kilos of fuel (about 1.5 gallons) burned, 30cc of oil used, and 2,031 miles traveled at an average speed of 75 mph, about 1,350 miles per gallon!

Dr. Tad McGreer, the Insitu Group, and all of the individuals and organizations involved—including Kip and Bill—can be proud of their accomplishment. For the Aerosonde, future skies look CAVU indeed.