Overwhelming Humans

Bringing a human into a developing situation is a challenge for artificial intelligence (AI) systems.

“Neuromorphic cameras or event-based cameras are types of sensors, which are paying attention to things that change over time. These could be instantaneous changes or from the last image, or it could be very short duration between two sequences image,” said David Aha, director, Navy Center for Applied Research in AI (NCARAI) at the U.S. Navy Research Laboratory. As the system changes, it applies AI-powered computer vision to draw the human’s attention toward an evolving situation.

“There’s interest in developing architectures for creating agents that have, as their foundation, a theory of attention, detecting when they should be attending to what aspects of their sensor suite at a point in time we’re on a mission,” Aha explained.

This streamlines a process and could free humans from various routine jobs, but it also creates a problem, as vast amounts of data from each camera need to be processed and stored for later comparison. According to Aha, storage and computing power are the greatest challenges in this field.

Nevertheless, there is a possible solution, with more capable devices closer to the edge.

“If you put something on an edge device that has the capability to recognize and do well, in terms of recognizing where to attend to at any point in time, you’re in a better position to be able to use it, limiting the amount of power on that machine,” Aha told SIGNAL Media in an interview.