Bumblebees are anything but bumbling: The insects quickly figure out the optimal route for visiting five far-flung flowers, a computational task that even human brains find challenging.
That result suggests that an elaborate mental map isn’t necessary to travel efficiently in unknown territory. Finding a way to mimic the bumblebee’s navigation system may allow programmers to develop robots that adeptly maneuver through unfamiliar places.
The new study, published online September 20 in PLOS Biology, pulls together several lines of previous research into one grand experiment. After training bumblebees to associate artificial flowers with a reward, scientists from the University of Sydney, Rothamsted Research in Harpenden, England and Queen Mary University of London arranged five flowers in a pentagon with sides 50 meters long. One at a time, bumblebees outfitted with a little radar antenna were released from the nest. The bees’ movements were tracked by radar, and motion-sensing cameras on the flowers recorded each visiting bee.
There are 120 possible routes that hit each flower once, but some are far more efficient than others. This puzzle is known as the traveling salesman problem: Given a set number of cities, determine the shortest travel route that takes the salesman to every city once. The bees, without even knowing where the flowers were, figured out the optimum path after trying only about 20 of the 120 different routes, reducing their flight distance from the first attempt to the last by about 80 percent, or 1,500 meters.
A computer analysis of the bees’ movements suggested that the insects were doing some quick comparing. If a bee went from flower A to B and later went from flower A to C, it would compare those routes, adding the one that was shorter to its itinerary and abandoning longer paths. The bees also made adjustments when a flower was moved to a different location. These results suggest that bees don’t need a big-picture map to search their surroundings, says team member Mathieu Lihoreau, a behavioral ecologist at the University of Sydney.
“It’s amazing that these little creatures are as flexible as they are and have evolved these solutions that make maximum use of these little brains they are carrying around,” says behavioral biologist Fred Dyer of Michigan State University in East Lansing.
This research could help develop route-planning algorithms for robots exploring unfamiliar terrain or software bots navigating the Internet. “In many contexts they aren’t experiencing the world holistically, but sequentially,” says Dyer. ”When an engineer doesn’t give you a map, you have to experience the world step-by-step.”
