Scripps development engineer Andy Nager gives a Remus underwater autonomous vehicle a push so it will dive below the surface and begin a survey of the waters off Palau for lost airplanes. Daniel Terdiman/CNET
PALAU -- "All right, let's go find us an airplane."
As Eric Terrill, director of the Coastal Observing and Research Center at the Scripps Institution of Oceanography, spoke on a perfect afternoon late last month, a bright yellow torpedo-shaped robotic vehicle began diving below the surface of the warm Pacific Ocean waters nearby. Then it was gone.
Known as a remote environmental sampling unit, or Remus, it's a 5-foot-long, 81-pound autonomous underwater vehicle (AUV), a $300,000 package of sensors, cameras, electronics, and other gear designed to survey the ocean.
Scripps and its partners at the University of Delaware School of Marine Science have used Remus for several years, and they make annual pilgrimages, supported by the US Office of Naval Research, to Palau to conduct numerous oceanographic studies on things such as currents, weather data, and the rise in sea level.
Since 2012, the teams have also used their high-tech gear for a mission with much more immediate human consequence. They've been working with a volunteer organization called the BentProp Project to hunt the Palauan waters for lost American planes, shot down by the Japanese during World War II, and the MIAs who were aboard them. "It's the treasure hunt of all treasure hunts," said Scripps Ph.D. student Travis Schramek, "only it's got a real moral story to it."
That Terrill is hoping his Remus will discover a plane is a small bit of hubris. During the previous week, BentProp and its oceanographic partners had found two wrecks: a TBM Avenger, and an F6F Hellcat. BentProp makes its own, self-funded annual expeditions to Palau and had been looking for those specific planes since 2005 and 2004, respectively. Thanks in large part to the Scripps and "UDel" Remus and other oceanographic tools, those planes are no longer lost. The finds were a big win for everyone involved. As BentProp team leader Pat Scannon put it to me, "To find two MIA sites in a single season is unprecedented in our experience."
This was the third year Scripps and Delaware worked with BentProp, and the scientists brought some technology and some new approaches with them.
Everywhere you looked, someone had a GoPro. Both the oceanographic and BentProp teams had used the cameras on their own for some time, given that they're small, easy to use, and can shoot HD video and high-quality photos underwater. Plus, they're relatively inexpensive, so it's not a disaster if they're lost in action. This year, GoPro got on board the Palau expedition to document it as part of a public STEM outreach program to inspire youth in the marine sciences.
The main tool though was still the Remus, and every day, the Scripps and Delaware teams headed out on boats to send the AUVs out on preprogrammed missions surveying new areas of the sea in methodical back-and-forth patterns, employing 900KHz sound to look about 131 feet on either side. The result is a 3D image that shows anything sticking up from the sea floor, and Terrill and his colleagues are experts at spotting the difference between human-made and natural features. When they see something promising, that means someone will soon be scuba diving to look at it up close.
That's how the teams found the TBM Avenger, which was lying in 100 feet of low-visibility water. "Eric [Terrill] was able to penetrate through all that, and find out if a tail section and the main fuselage" were there, Scannon said. "Once you know where it is, you can do a lot more to characterize the debris field."
The decision about where to send Remus is also often scientific. Sometimes research in old military archives turns up a promising spot to investigate, and there are plenty of eyewitness accounts to follow up on as well. Terrill said that Scripps and UDel also now have access to US Navy light detection and ranging (lidar) data that was gathered 10 years ago to create new navigation charts. That data offers the teams very-high-resolution imagery of the sea, which can lead to new mission profiles for the AUVs.
Another new tool at their disposal are Baysian statistical models run by engineers at the University of California, San Diego, that predicted "splash locations of the shot-down aircraft," Terrill explained. They took into account things like flight mechanics, and information from after-action reports such as airspeed, weather, time in the air, bombing locations, and more. And then planned missions to survey some of those sites.
Terrill said BentProp has also found terrific imagery taken during some of the bombing missions in 1944 and fused those pictures with modern day aerial images to try to determine where planes were and where they might have ended up.
Based on all that intelligence, Scripps and UDel would plan daily missions with the AUVs and a portable dive sonar device. But even with all that, Remus could cover every inch of only a square mile of sea per day. Given that Palau's islands range over hundreds of miles, this is clearly slow work. Yet, it's far faster with the technology than BentProp was ever able to do on its own. And there's every reason to believe that these expeditions will continue each year, allowing them to gradually winnow down the search areas.
One last tool at the teams' disposal this year was a set of 3D Robotics aerial drones. Piloted by a Scripps development engineer, they were being used to generate topographic maps of jungle areas BentProp wants to search for MIAs, as well as to take infrared imagery to search for possible metal signatures in what could be hidden grave sites.
Walsh was able to fly a number of missions during the expedition, both over and near water, and in the jungle, collecting valuable data. But one day he was having problems with a drone. As he launched it, the craft barely rose off the ground before crashing. Walsh is a whiz at fixing his machines and soon had it flying again. But not right away, as one of its propellers had snapped off. Asked if that counted as a bent prop, Scannon's eyes lit up and he said it did. It was, he added, "a modern day bent prop."