New underwater robot from MIT engineers plans its own missions

A new autonomous underwater vehicle (UAV) from researchers at MIT will help researchers on Earth explore the oceans, survey marine habitats and check on the health of species. It may also help scientists at NASA explore the solar system. The UAV uses artificial intelligence to set priorities, adapt to circumstances and execute minor repairs and maintenance.

Currently engineers spend a great deal of time writing scripts and low-level commands that tell a robot, step-by-step, how to carry out a mission. The new system asks humans for high-level goals and the robot then performs high-level decision making, while monitoring circumstances, to decide how best to achieve the goals it is tasked with.

Brian Williams, a professor of aeronautics and astronautics at MIT, and principal developer of the mission-planning system said that the system is loosely based on Star Trek. In the hierarchy of programs running the UAV, there is a “captain” which makes higher level decisions, plans missions and decides where to go and when. There is a “navigator” that selects the best route to meet the goals of the mission and a “doctor” or “engineer” that does diagnostics and repairs.

“We can give the system choices, like, ‘Go to either this or that science location and map it out,’ or ‘Communicate via an acoustic modem, or a satellite link. What the system does is, it makes those choices, but makes sure it satisfies all the timing constraints and doesn’t collide with anything along the way. So it has the ability to adapt to its environment,” said Williams in a statement.

The team tested the system in March on a research cruise off Australia. The MIT group, along with colleagues from the Australian Center for Field Robotics, the University of Rhode Island and Woods Hole Oceanographic Institute tested a number of UAVs. The MIT system demonstrated the ability to operate safely, even with other UAVs operating in the area.

“We wanted to show that these vehicles could plan their own missions, and execute, adapt, and re-plan them alone, without human support. With this system, we were showing we could safely zigzag all the way around the reef, like an obstacle course,” said Williams.

Williams designed a similar system for NASA following the loss of the Mars Observer in 1993.

“There were human operators on Earth who were experts in diagnosis and repair, and were ready to save the spacecraft, but couldn’t communicate with it. Subsequently, NASA realized they needed systems that could reason at the cognitive level like engineers, but that were onboard the spacecraft,” said Williams.

That system was successfully tested when NASA’s Deep Space 1 probe a flyby on an asteroid in 1999.

“That was the first chance to demonstrate goal-directed autonomy in deep space. This was a chance to do the same thing under the sea,” said Williams.

The research, which was partially funded by NASA, could come into play again with some of NASA’s future missions. The space agency has an interest in several moons in the outer solar system which are believed to have liquid oceans.