Air Force flight control improvements

"This work investigates sensory-motor feedback mechanisms in the insect brain that could inspire new approaches to flight stabilization and navigation in future insect-sized vehicles for the military," said Dr. Willard Larkin, AFOSR program manager who's supporting the work of lead researcher, Dr. Andrew Straw of Caltech.

The research is being conducted in a laboratory where scientists are studying how flies use visual information to guide flight in natural environments.

The scientists have found that, counter to earlier studies suggesting that insects adjust their height by measuring the motion beneath them as they fly, flies in fact follow horizontal edges of objects to regulate altitude. Remarkably, this edge following behavior is very similar to a rule they use for steering left and right and always turning towards vertical edges.

Straw noted that the flies don't have access to GPS or other radio signals that may also be unavailable in, for example, indoor environments.

"However, with a tiny brain they are able to perform a variety of tasks such as finding food and mates despite changing light levels, wind gusts, wing damage, and so on," he said. "Flies rely heavily on vision."

The scientists designed a virtual reality environment for their flying subjects which they found could regulate their altitude by enabling them to fly at the height of nearby horizontal visual, like the tops of rocks and vegetation.

"We developed a 3D fly tracking system which was our most significant technical challenge: localizing a fly in 3D nearly instantaneously," said Straw. "Next, we developed visual stimulus software capable of making use of this information to project virtual edges and textured floors in which we could modify the fly's sensory-motor feedback mechanism."

The 3D fly tracking system the researchers developed is significant because it will allow a rapid characterization of other fly behaviors with unprecedented levels of visual stimulus control.

Ultimately the scientists would like to build models of fly flight that can accurately predict behavior based on their sensory input and internal states.

"Additionally, being able to identify the neural circuits responsible for flight control would allow us to extend our understanding of how physiological processes implement behavior," said Straw.

In their next phase, the scientists will study more sophisticated flight behaviors, asking if the the fly creates a long-lasting neural representation of its visual surroundings or whether flight is only controlled by fast-acting reflexes.