New Study on Fruit Flies may Lead to Better Autonomous Vehicles

According to a AAA annual survey conducted on autonomous driving, more than 70% of respondents have reported that they are afraid of traveling in a fully autonomous car.

This means manufacturers, like Tesla, may need to start all over again before they can launch their fully autonomous self-driving systems into the market.

However, a new study performed by Northwestern University has demonstrated that it would be better if fruit flies are put behind the wheel in the place of robots.

Drosophila fruit flies have been the focus of science ever since humans have been conducting experiments in laboratories. But considering the size of these insects, it is easy to speculate the kind of things that can be learned by studying them.

A study recently published in the Nature Communications journal has demonstrated that decision-making, memory, and learning are used by fruit flies to carry out basic functions, such as evading heat. Now, this understanding is being used by scientists to defy the way people think about autonomous cars.

The discovery that flexible decision-making, learning and memory are used by flies during such a simple navigational task is both novel and surprising. It may make us rethink what we need to do to program safe and flexible self-driving vehicles.

Marco Gallio, Study Corresponding Author, Northwestern University

Gallio is also an associate professor of neurobiology at the Weinberg College of Arts and Sciences. According to him, the queries behind this research work are analogous to those vexing engineers who are designing cars that travel on their own.

But how does a fruit fly (or a vehicle) handle novelty? How to build a car that can flexibly acclimate to new conditions? This latest finding demonstrates brain functions in the household insect that are generally linked to more complex brains, such as those of humans and mice.

Animal behavior, especially that of insects, is often considered largely fixed and hard-wired —like machines. Most people have a hard time imagining that animals as different from us as a fruit fly may possess complex brain functions, such as the ability to learn, remember or make decisions.

Marco Gallio, Study Corresponding Author, Northwestern University

To find out how fruit flies are able to evade heat, the research team in the Gallio laboratory designed a minute plastic chamber that contains four floor tiles. The temperatures of the floor tiles can be separately regulated, and these tiles also keep the flies inside.

The researchers subsequently applied high resolution video recordings to plot how a fly reacted upon encountering a barrier between a cool tile and a warm tile.

The team observed that the flies were incredibly good at treating heat barriers as invisible boundaries to prevent harm or pain.

The researchers used real measurements and produced a 3D model to predict the accurate temperature of every part of the fly’s small body all through the experiment.

During other experiments, the team opened a window in the fly’s head and captured brain activity in neurons in which external temperature signals are processed.

According to Miguel Simões, the co-first author of the study and a postdoctoral fellow in the Gallio lab, flies can determine with incredible precision whether the optimal route to thermal safety lies to the right or left.

Plotting the direction of escape, flies “‘nearly always’ escape left when they approach from the right, “like a tennis ball bouncing off a wall,” added Simões.

When flies encounter heat, they have to make a rapid decision. Is it safe to continue, or should it turn back? This decision is highly dependent on how dangerous the temperature is on the other side.

Miguel Simões, Study Co-First Author, Northwestern University

Watching the basic reaction reminded the researchers of one of the traditional ideas in early robotics.

“In his famous book, the cyberneticist Valentino Braitenberg imagined simple models made of sensors and motors that could come close to reproducing animal behavior. The vehicles are a combination of simple wires, but the resulting behavior appears complex and even intelligent,” stated Josh Levy, an applied math graduate student, and William Kath, a member of the Gallio lab and an applied math professor.

Braitenberg contended that most of the animal behavior can be elucidated by the same principles. However, does that mean the behavior of the fly is as predictable as that of one of Braitenberg’s fictional robots?

The team from Northwestern University constructed a vehicle by utilizing a computer simulation of fly behavior with the same algorithm and wiring as Braitenberg’s vehicle to observe how closely they could simulate animal behavior.

After performing model race simulations, the researchers ran a sort of natural selection process, selecting the vehicles that did exceedingly well and slightly altering them before reintegrating them with other high-performing vehicles.

Levy subsequently ran 500 generations of evolution in the rugged computing cluster at Northwestern University, constructing cars which the team believed would eventually do as well as flies at evading the virtual heat.

A simulation like this showed that 'hard-wired' vehicles ultimately emerged to perform almost as that of flies. However, while real flies went on to enhance performance over time and learned to implement better approaches to become more efficient, the vehicles continued to remain inflexible and 'dumb.'

The team also found that even as flies executed the basic task of evading the heat, their behavior remains slightly unpredictable, leaving more room for independent decisions.

The researchers finally noted while flies lacking an antenna figure out and adapt to novel strategies to evade heat, cars 'damaged' in the same manner are not able to handle the new scenario and turn toward the direction of the missing portion, ultimately getting stuck in a spin, similar to a dog chasing its tail.

According to Gallio, the concept that basic navigation involves such complexity provides hope for upcoming analysis in this field.

The research work conducted in the Gallio lab was funded by the NIH (Award No. R01NS086859 and R21EY031849), a Pew Scholars Program in the Biomedical Sciences, and a McKnight Technological Innovation in Neuroscience Awards.

Journal Reference:

Simões, J. M., et al. (2021) Robustness and plasticity in Drosophila heat avoidance. Nature Communications. doi.org/10.1038/s41467-021-22322-w.