Dragonflies' Strange Ability Could Help Build Robots With Sophisticated Artificial Vision

Researchers hope to use a "visual circuit" recently identified in dragonflies' brains to improve artificial vision systems in robots.

A team of researchers found that ON and OFF pathways in the insects' brains combine when looking at a "simple dark object," a University of Adelaide press release reported.

"To perceive the edges of objects and changes in light or darkness, the brains of many animals, including insects, frogs, and even humans, use two independent pathways, known as ON and OFF channels," lead author Dr. Steven Wiederman, said.

"Most animals will use a combination of ON switches with other ON switches in the brain, or OFF and OFF, depending on the circumstances. But what we show occurring in the dragonfly's brain is the combination of both OFF and ON switches. This happens in response to simple dark objects, likely to represent potential prey to this aerial predator," he said.

The researchers kept a record of the activity in the dragonflies' "target-selective" brain neurons.

The team studied neuron activity while presenting the insects with different intensities of light, and some dark targets. They found the dark objects elicited more of a response in the dragonflies than the team had expected. Response to the dark objects comes from "the correlation of opposite contrast pathways: OFF with ON."

"The exact mechanisms that occur in the brain for this to happen are of great interest in visual neurosciences generally, as well as for solving engineering applications in target detection and tracking. Understanding how visual systems work can have a range of outcomes, such as in the development of neural prosthetics and improvements in robot vision," Dr. Wiederman said.

"A project is now underway at the University of Adelaide to translate much of the research we've conducted into a robot, to see if it can emulate the dragonfly's vision and movement. This project is well underway and once complete, watching our autonomous dragonfly robot will be very exciting," he said.

Real Time Analytics