COASTAL NEW ENGLAND — MIT Lincoln Laboratory's Advanced Undersea Systems and Technology Group tested a diver–autonomous underwater vehicle teaming system in coastal New England waters as part of a project to support U.S. military maritime missions. The project, funded through an internally administered R&D portfolio on autonomous systems, aims to optimize operations including critical infrastructure inspection and repair, search and rescue, harbor entry, and countermine operations.

"Divers and AUVs generally don't team at all underwater," principal investigator Madeline Miller said. Underwater missions requiring humans typically involve manipulation tasks that robots cannot perform, such as repairing infrastructure or deactivating mines. Remotely operated vehicles face challenges in those tasks because their manipulators lack sufficient agility, while humans offer superior dexterity and excel at recognizing objects underwater. Humans working underwater cannot perform complex computations or move very quickly when carrying heavy equipment. Robots, in turn, hold advantages in processing power, high-speed mobility, and endurance.

Miller and her team are developing hardware and algorithms for underwater navigation and perception. "Ultimately, we want to devise solutions for navigation and perception in expeditionary environments. For the missions we're thinking about, there is limited or no opportunity to map out the area in advance. For the harbor entry mission, maybe you have a satellite map but no underwater map, for example." Miller said.

The team built upon earlier work by the MIT Marine Robotics Group, led by John Leonard, which ran simulations under optimal conditions and performed field testing in calm waters using human-paddled kayaks as proxies for both divers and AUVs. Miller's team then integrated the diver–AUV teaming algorithms into a mission-relevant AUV and tested them under realistic ocean conditions, first with a support boat acting as a diver surrogate and then with actual divers.

The team is also developing an AI classifier that processes optical and sonar data mid-mission and solicits human input when it is uncertain about object classifications. That feedback loop requires an underwater acoustic modem for diver–AUV communication. State-of-the-art underwater acoustic communications data rates require tens of minutes to send an uncompressed image from an AUV to a diver. The team is investigating methods to compress information within the constraints of low bandwidth, high latency, and limited hardware.

Using mostly commercial off-the-shelf sensors, the team built a sensor payload designed to integrate into AUVs routinely used by the U.S. Navy. Tests took place in the open ocean near Portsmouth, New Hampshire, using the University of New Hampshire's Gulf Surveyor and Gulf Challenger research vessels as diver surrogates, and on the Boston-area Charles River using an MIT Sailing Pavilion skiff.