There’s a James Bond-style aura around the types of exploratory remote machines that software engineer Sandy Takacs helps develop. Whether it’s developing a military satellite system or a smart mine hunter towed through the water by a helicopter, Takacs has been in the forefront of creating smart software for next-generation vehicles. His latest endeavor is testing undersea robots made by Hydroid Inc., a Bourne-based maker of unmanned underwater vehicles. “These vessels are less mystical once you understand their nuts and bolts and know how they work, but just imagine encountering one of these bright yellow vehicles with a fin and strobe light emerging from the deep,’’ said Takacs, 50. “It’s fascinating how far we have come with this technology.’’
Engineer helps robots get the job done
ON THE JOB
Underwater robots, also called autonomous underwater vehicles or AUVs, helped locate the wreckage of an Air France plane that crashed three years ago. AUVs, equipped with sensors, sonar, and navigation systems, are also used for ocean mapping, pipeline inspection, homeland security, and fishery operations, among other tasks. Takacs’s role is to ensure that the mission assigned to the systems can actually be accomplished.
When you joined Hydroid last summer, what was one of your first challenges?
Once we add code modifications to either the vehicle or its user console, we need to verify that those modifications work as expected and that there are no adverse issues with the new software. To do that, we frequently go out onto Buzzards Bay, Cape Cod Bay, or the Atlantic Ocean off of Chatham to test the new software. One of my first projects required testing the ability of a vehicle to berth to a moving underwater dock. This required that the AUV home in on the location, then successfully latch. It was the first time in history this was done autonomously, so it was a pretty fun project.
And did it work?
With the current, waves, and turbulence from the outboard motors of the surface vessels, it was tough to try to hit the small target. These were all fluctuations that needed to be taken into account within the software algorithms. The algorithms needed to be refined within the vehicle software and reloaded with the new updates. But we were finally able to make a successful attempt that worked 80 to 90 percent of the time.
What exactly is an algorithm?