John Dabiri, a professor at Caltech, has diverse interests in the environment. One of them is exploring the deep ocean, measuring many conditions – the layout or topography of the floor, the chemistry of the water, the organisms and the ecology of their interactions with each other.
The unexplored depths amount to 90-95% of the ocean. Measurements from surface ships continue, but they go only so far. Drones are an effective way to go, IF we can direct them to take the most informative routes far down there, routes that commonly aren’t obvious except in the moment, on the fly. We can’t drive the drones from the surface 20,000 feet down with any good control; there’s limited potential info about the “roadbed” that can be relayed to the surface. The drones also run into currents that divert them from their courses, and we can’t sense them on the surface.
Dabiri’s group calls in artificial intelligence to the rescue, endowing the drones with that capability. The drones learn in real time, using the flows they sense with electronic accelerometers and gyroscopes. They then make “intelligent” decisions on movements – where’s the next interesting place that I can reach. Good decisions also save power that the drone must reserve to get back to the home ship; it has to know how to coast, like an unpowered aerial glider or a hawk when the time is ripe. I have to say, this is cool, and not just because the deep ocean is cold!
This has been an outreach activity of the Las Cruces Academy, viewable at GreatSchools,org.
Ref.: Caltech news
