Support Provided ByLearn More
Tech + EngineeringTech & Engineering

Bee-Brained Robot Reveals Nature's Navigation Secrets

ByTimothy James DimacaliNOVA NextNOVA Next

Receive emails about upcoming NOVA programs and related content, as well as featured reporting about current events through a science lens.

How do bees make a beeline for home? European scientists may have found the answer—paving the way for everything from better bee conservation to building smarter robots.

Despite wandering far in search of food, bees head straight for their hive using a process called “path integration”: they remember each time they take a new path and put these memories together to plot a course home.

Support Provided ByLearn More

Researchers from Lund University in Sweden, along with colleagues in the United Kingdom and Australia, discovered exactly how this is done by studying the brains of nocturnal Megalopta genalis bees.

bee-side_2048x1152

“[This study] has practical importance in understanding the challenges bees may face in navigating an anthropogenic landscape.” said John Stotskopf Ascher, a melittologist (or bee researcher) at the National University of Singapore who was not involved with the study. “It shows how bees, like many insects, have great potential as a model for biomimetics.”

By hooking the bees up to a flight simulator and looking at their brains under a microscope, the researchers were able to identify neurons that keep track of speed and direction using information from the insects’ eyes.

“Direction neurons” act like compasses, keeping track of sources of polarized light in the environment—such as from the moon and stars—to remember where the bee is in relation to the light source. Meanwhile, “speed neurons” track movement and serve as a kind of odometer to let the bee know how fast it’s moving.

The researchers devised a computational model of how these neurons interact, which they then plugged into a smartphone-powered robot equipped with a 360-degree camera, which mimicks how bees see.

“We sent [the robot] out on a random route and the model of the bee’s navigation system that we implemented allowed it to find the direct path back to its starting point,” said Lund University biologist Stanley Heinze, one of the researchers.

It was no easy feat to find and identify the neurons responsible for this behavior.

“There are about one million cells in the bee brain, out of which there are exactly four speed neurons. Finding these cells repeatedly in many individuals while blindly probing the brain with an electrode was a huge challenge,” Heinze said.

As for compass neurons, “we recorded from over 160 bees and found [only] ten compass neurons,” he said.

It took several years to accumulate enough raw data for the study, and the translation into a working robot required extensive cooperation between the biologists and roboticists.

The effort, though, could pay dividends in bee conservation and robotics.

Vicki Wojcik, the research director of Pollinator Partnership , said that the study may help scientists understand the subtle effects of pesticide on bees.

Bees, she says, are not necessarily dying from exposure to pesticides. “It’s that they’re encountering pesticides and being subjected to these chronic sub-lethal effects that can impact either their ability to grow and reproduce…or their cognitive ability to forage and find home.”

Heinze is hopeful that his team’s work might lead to more efficient robots in the not-too-distant future.

As it stands, not even the latest state-of-the-art drone technology can hold a candle to the elegant simplicity of bee brains.

“The brain of a bee is way smaller and less complex than ours but still a factor more powerful than drone brains,” said Matthew Cua, the founder and CEO of private drone operator Skyeye Analytics, Inc.

But Heinze’s team is showing just how bees are able to do more with less, and how this makes for hardier robots.

“The advantage of our circuit would be that it only consists of very few neurons that perform a robust computation—it’s very resilient to disturbance—and that it only requires visual input,” Heinze said.

Cua said that nature still has much to teach us about efficient engineering. “We may have the most powerful hardware, but our software can’t maximize it,” he said.