Two Lincoln Laboratory teams compete to find airplane black boxes

The Technology Office challenged staff to develop an autonomous method to find black boxes from planes that crash into the sea.

by Meg Cichon | Communications and Community Outreach Office

On 8 March 2014, while flying over the Indian Ocean, Malaysian Airlines Flight 370 disappeared from military radar screens. After the plane failed to make landfall, investigators began an extensive search of large swaths of the sea around where the plane had last been detected. When no wreckage was found, search crews focused on finding the plane's black box, which carries critical data, such as recorded conversation between the pilot and copilot and GPS coordinates, that could help explain why a plane crashed. However, Flight 370 and its black box are still missing.

This fruitless investigation triggered the 2016 MIT Lincoln Laboratory Technology Office Challenge (TOC), which each year addresses problems of relevance to Laboratory mission areas and aims to foster Laboratory-wide collaborations. During the 2016 competition, named Coordinated Autonomous Novel Underwater Sensing and Exploitation Ability (CANUSEA), Laboratory researchers sought an effective approach to black box recovery that could include the use of autonomous vehicles, a swarming technique, specialized communications systems, and innovative mechanical engineering. In an eight-month period, two teams, Golden Retriever and Coordinated Autonomous Naval Detection of Objects (CANDOO), designed and built their systems in preparation for a final competition at Mirror Lake in Devens, Massachusetts, in September.

TOC TeamsTeam CANDOO (bottom) and Golden Retriever (top) tested their designs at Mirror Lake in Devens, Massachusetts. Photo credit: Glen Cooper

Search and recovery operations in the ocean are difficult because the underwater environment creates difficulties in communication, remote sensing, and navigation. Because radio waves, which enable the use of Wi-Fi, walkie-talkies, and other mobile devices, travel poorly in water, underwater communication requires the use of acoustic waves that travel much slower than do radio waves and that quickly lose intensity because their propagation is affected by refraction, absorption, and scattering through the water. In addition, the number of unknown objects throughout the water and on the seabed can impede sensing and navigation. Therefore, CANUSEA was designed to apply Lincoln Laboratory expertise, experience, and creative problem solving to familiarize researchers with the unfamiliar undersea domain.

A black box consists of two shoebox-sized orange aluminum cases, one holding a cockpit voice recorder and one holding a flight data recorder. When a black box is immersed in water, an underwater location beacon (ULB) within the box emits an ultrasonic pulse, commonly referred to as a "ping." The ping can be detected in depths of nearly one mile by a submerged hydrophone, which is typically towed through the water by a vessel. If the ping is detected, an operator monitors the signal strength and records the navigation coordinates. This procedure is repeated until the final position is triangulated. The hydrophones can detect a black-box signal at up to a 20,000 foot depth and within one nautical mile laterally. However, the ULB only has enough power to ping for 30 days, and in some cases, like Flight 370, the ULB battery may be expired because of an oversight in routine maintenance or may not work at all. When the ping is silenced, the chance of finding the box plummets because search teams must use sonar to slowly search the seabed for debris, inch by inch, rather than scanning large expanses and then narrowing the search area when a ping is found. To avoid relying on the ping and its 30-day deadline, the TOC teams worked to design systems that can find black boxes both with and without an active ULB.

TOCTeam CANDOO's approach included two autonomous vehicles that use passive and active sonar. A large vehicle images the seabed, and a smaller vehicle dives underwater to identify, tag, and lift the black box to the surface. Illustration: CANDOO team

Team CANDOO decided to use passive and active sonar approaches with two autonomous vehicles. First, researchers deployed a large vehicle at the surface of the lake. Equipped with an active sonar system, the vehicle scanned the lake to create a depth map and to obtain an image of the surveyed area. A software program analyzed the map data to find the box used during the challenge, which had a distinctive football shape and stripes along its length, and geolocated potential boxes on the depth map. The locations were posted on what the researchers called a "bulletin board" that the two vehicles used to transfer and share collected data. After coordinates were posted, a maneuverable autonomous undersea vehicle (AUV) read the post and headed to the location for a closer look. The AUV used optical cameras to identify whether or not the object was a black box and, if identification was successful, used a magnet to lift the box to the surface.

During this identification and retrieval process, an onshore research team monitored the information transmitted via radio frequency (RF) to ensure that the vehicles identified black boxes and to operate the small AUV. "The limited development time caused us to change some of our initial goals," said team CANDOO member Alexander Bockman. "Although we originally planned for an entirely remote operation, the identification and retrieval process became more human in the loop, and users operated the AUV from the shore."

Mabel Ramirez 40 Under 40Team Golden Retriever's approach uses passive sonar sensing to create an optical three-dimensional map. An unmanned underwater vehicle (UUV) uses hydrophones to scan the area for pings. Data are relayed to the UUVs and on-shore ground station via a floating "mothership." Illustration: Golden Retriever team

Team Golden Retriever used passive sonar sensing and optical three-dimensional (3D) mapping. To find the pinging box, they employed a custom unmanned underwater vehicle (UUV) equipped with hydrophones to scan the lake for the sound. Data from the UUV were used to draw a map of the direction and intensity of the pings. The team then sent the UUV underwater to image the seabed at the mapped points.

TOCIn this augmented-reality view of Mirror Lake, the user can view the surface of the lake, lakebed depth contours (blue boxes), potential black box location (orange square), and UUV location and path (green circles). Image: Golden Retriever team

The UUV would also passively image the lake for the silent black box. The team identified possible objects by using custom software that scans the video data from the UUV to pinpoint the color orange, the color of the simulated black box. To help navigate the vehicle, researchers acquired a depth chart of Mirror Lake. They converted the chart into an octomap, i.e., a 3D pixel grid, and uploaded it to a head-up display—virtual-reality glasses that project information and imagery onto real environments. While wearing the glasses, researchers were able to look at the surface of the lake, and to see the UUV location and lakebed depth contours. When potential black boxes were identified, the team planned to use a custom-made UUV with a lifting system to retrieve the object.

All data were transmitted by an optical communication system. A "mothership" on the lake's surface served as a communications relay between the UUV and the ground station. The UUV was connected to the mothership via an Ethernet cable, and the ground station, where team members interfaced with UUVs and analyzed data, was connected to the mothership via Wi-Fi.

According to Golden Retriever team member Adam Shabshelowitz, a scaled-up version of this concept would use multiple UUVs, i.e., a swarm, to obtain data and theoretically locate a black box more quickly and efficiently than using just one UUV. Said Shabshelowitz. "Because the vehicles are inexpensive and use the same software, many of them can be built, and they would all use the mothership as a relay station."

In September, the two teams faced off during a final exercise at Mirror Lake. Each team had two hours to find and tag the boxes and relay data from the boxes—one box was sending pings and one box was silent. The boxes were submerged within a 10,000-square-meter area of the lake, with depths as low as about 18 meters. At the onset of the competition, each team faced major challenges: CANDOO's large vehicle failed and Golden Retriever's communications system malfunctioned. Both teams decided to reschedule the competition. On the second attempt, CANDOO located the box, but did not have enough thruster power to lift the box out of the water. "We thought we did a good job finding the box," said Bockman. "Finding it is 90% of the battle." Golden Retriever did not find the black box on the challenge day; however, one week later, the team went back to the lake and successfully located the box. The Technology Office Advisory Panel named CANDOO the official winner of the challenge.

"We think the teams did a superb job given that the underwater domain is a brand new area for the Laboratory," said Beijia Zhang, Associate Technology Officer. "Finding black boxes under water is analogous to finding many different types of targets underwater. The oceans are vast, but we still don't know much about what is down there. We hope this challenge has given the Laboratory a big boost in developing unique capabilities for underwater needs. We can't wait to see how this area of work evolves."

Both teams agreed that the challenge was exceptionally difficult and rewarding. Said Shabshelowitz, "We only had six months to develop a system from scratch, and we made very good progress. Using many inexpensive platforms to cover a wide area is a good approach for solving this kind of problem and has great potential. We are happy with the outcome, and it was a fun experience. The fact that it was so challenging made it even more fun." Bockman agreed that the competition was challenging and rewarding, and looks forward to seeing his team's technologies used in different programs across the Laboratory. "Our equipment is being redeployed to other projects," said Bockman. "It is great to see that this challenge has created a lot of opportunity and value for the Laboratory."

Team CANDOO included Alexander Bockman, Kristen Railey, Paul Ryu, Matthew Bradford, Joshua Smith, Michael Chan, Scott Hamilton, Nicholas Hardy, Dan, Glenn Schrader, Richard Marino, Bryce Remesch, Antonio Rufo, Edwin LeFave, Benjamin Waltuch and Eric Huang.

Team Golden Retreiver included Adam Shabshelowitz, Cheryl Blomberg, Christopher Nutting, Bakari Hassan, Jessica Brooks, Mark Donahue, Navid Shahrestani, Owen Guldner, Peter Klein, and Stephen O'Keefe.

Posted November 2016

top of page