The brain is one of the most powerful organs of the human being. In this sense, DARPA has started working on cameras that mimic it, working at low power, in order to enhance data processing.

According to Sapo24, the program is intended to improve military applications, such as autonomous vehicles, robotics, among others.

The technology associated with cameras has developed a lot over the years. In addition to scientists being able to collect increasingly accurate images from space and send them back to Earth, the capability of cameras has also seen a strong positive development. After all, very high quality pictures are achieved by cameras that tend to be smaller.

What Defense Advanced Research Projects Agency (DARPA) is trying to do is to surpass the capacity of the human eye. That is, it has teams working to develop cameras that process images in the same way that the human brain does.

As the DARPAhumans have an exceptional ability to process images. This is because brain has optimized the process by detecting and signaling when something changes. In turn, the cameras being used for now are processing more data and consuming more power, while still lacking the intelligence to locate objects if backgrounds change or become cluttered.

Therefore, DARPA intends to replicate the process inherent in the brain, through a program called Fast Event-based Neuromorphic Camera and Electronics (FENCE). This aims to develop neuromorphic cameras - verisimilar to the brain - that capture events by transmitting information only about pixels that have changed. In this way, the amount of data to be processed will be drastically reduced, as will transmission delays and power consumption during the entire process.

"The goal is to develop a smart sensor that can reduce the amount of information that is transmitted from the camera, reducing the data to consider only the most relevant pixels," said program manager Whitney Mason.

The program that will mimic the brain will involve an event-based infrared focal plane array that will capture the signals and transmit them. This will require a new class of digital signal processing and learning algorithms to facilitate the operation of sensors capable of handling dynamic events.