A View from Emerging Technology from the arXiv
Virtual Reality for Fruit Flies: Plans Unveiled
The ability to track fruit flies in real time is the foundation of a virtual reality system that could revolutionize the study of animal flying behavior.
In 1939, John Kennedy at Imperial College, London, discovered a curious fact about mosquitoes. In flight, they tend to turn toward vertical landmarks such as fixed posts. A similar behavior is also observed in fruit flies.
This is a straightforward repeatable observation suggesting that it is hard-wired into a fly’s neurobiology. Changing the neurobiology, either genetically or chemically, should generate other flying behaviors. And studying these flying activities should give neurobiologists a way of understanding exactly how neurons and genes determine behavior. All you have to do is watch the flies.
That’s easier said than done. High-resolution cameras generally have a small field of view and the data they produce is so great that it cannot be processed in real time. That severely limits the kind of experiments scientists can do.
But all that looks set to change with the development of a system called Flydra by Andrew Straw and buddies from the California Institute of Technology in Pasadena. Flydra is formed from two words: fly, the object of the team’s study, and hydra, the many-headed creature of Greek mythology.
The system consists of many relatively low-resolution cameras each with a wide field of view, recording the same, relatively large volume of space. The images are processed in real time by a number of computers that track the 3-D motion of many flying animals at once. This tracking is entirely automated.
That’s an impressive system because it opens up the possibility of the systematic study of the flying behavior of many creatures over long periods of time. That’s never been done before. Potentially, it could revolutionize our understanding of flying behavior.
The team has already studied the flying behaviour of 20 female fruit flies in a box over a period of 12 hours. During this time, the team varied the contrast of a pattern beamed onto the walls of the box. (To give you an idea of how difficult these experiments must be without automatic tracking, the teams says the flies spent most of their time walking on the walls, floor and ceiling and the entire 12-hour run yielded a total of only 1,760 seconds of flight.)
Animal behaviorists have always assumed that flying animals use these patterns to determine their flying speed.
Sure enough, as the contrast of this pattern is reduced, the flies’ flying speed becomes more erratic, suggesting they’re having more difficulty controlling it.
That’s already a useful result (and confirms the results of other experiments)..
But the real power of Flydra is its ability to track movement in real time (or at least with a lag of only 40ms). This allows the pattern to be changed in real time in response to a fly’s motion–a kind of virtual reality for drosophila.
That opens up an entirely new way to study flying behavior. Imagine, for example, a fly veering towards a vertical post which then continually moves. Fruit fly aerobatics, anyone?
And of course, it’s not just fruit flies that can be studied. Straw and co are already looking at hummingbird behavior. “We are also studying maneuvering in solitary and competing hummingbirds and the role of maneuvering in establishing dominance,” they say.
It looks as if a golden age is nigh for the study of flying animal behavior. John Kennedy, who died in 1993, would have been amazed.
Ref: arxiv.org/abs/1001.4297: Multi-camera Realtime 3D Tracking of Multiple Flying Animals .
AI is here.
Own what happens next at EmTech Digital 2019.