Besides the professional divers equipped with underwater camera positioning, the U.S. Navy has trained dolphins or sea lions to search for mines around the ship. However, these solutions are costly and not always bring the best performance in all cases. Robot is the answer seems more reasonable and in recent years, the underwater robot has been constantly investing in research. And one of the notable improvements are the team of scientists at MIT have successfully developed new algorithms help improve orientation and identification of aquatic robot mine.
With this important achievement, the fully automated robots can navigate and map the water turbid environment without previous exploration and mine detection is only 10 cm in diameter. The algorithm is an associate professor of the Department of Franz Hover and mechanical engineering graduate student at MIT Englot Brendan developed a robot programmed for Hovering Autonomous Underwater Vehicle named (HAUV).
robot to provide a complete picture of the large structures, such as ships, including small details such as bolts, rivets or small mine is not an easy thing. With mines only 10 cm in diameter, they are still strong enough to cause serious damage to the ship. Therefore, factors detailed in the image is always very important.
"It is not enough to observe objects from a safe distance. Means to access and swimming around the steering wheel and propeller to scan everything. However, the short-range sensors fitted on vehicles is limited in the field of view, "Hover said.
So as to obtain images detailed scan of the hull, the researchers conducted a two-step approach through the period. First, the robot is programmed to swim in a square space around the ship at a safe distance of 10 m and located using underwater cameras to gather information to create a cloud with landmarks. Although the major propeller of a vessel can be identified in low-resolution images but still not detailed enough to detect small mine.
In addition, the cloud consists of multiple markers robot does not know the two ends of the boat. This is a problem to be solved, if not the robot collided with the propeller. So, to create a three dimensional network model of the ship, the researchers translated into a cloud structure fixed by application of computer graphics algorithms in underwater positioning data.
Once the robot activities accompanies the structure, it can move to Phase 2. Robot is programmed to continue to swim closer to the ship to collect all points on the model outlined in the distance is only 10 cm.
While surface sampling scan is usually done each of a - like the path of mowers, researchers have devised a solution that is efficient use of optimization algorithms for robot that can scan along the structure and save complex 3D shapes of them. Thereby, this technique shortens the distance the robot needs to swim to get the overall look of the ship.
Team inspection algorithms by creating models underwater image of the Curtiss - a war ship 183 m long bodyguards in San Diego and Seneca - a 82 m long sailing boat at the port of Boston. Work will continue testing done this month in Boston harbor. Englot student said: "Our Discovery can compete with the divers in speed, efficiency and ability to search on every square inch of the ship. We believe we are closer to the final results. "
So in parallel with automatic cleaning robot hull Hull BUG, U.S. Navy will soon be able to create a series robot to protect as well as maintenance of expensive military equipment, ships that are here.
Here is a video presentation outlining the 3D capabilities of HAUV when mapping the body of the USS Curtiss SS using new algorithm.