An Israeli researcher, who has studied bats for nearly two decades, is trying to apply his findings to improve the way robots can communicate with one another using sound.
Yossi Yovel, who heads the Bat Lab for Neuro-Ecology at Tel Aviv University, said his team was studying the way that “Mexican fish-eating” bats, found in the Gulf of California, interact with one another.
He explained that these bats feed on small fishes that surface at night. But because of the lack of visibility during night time, they have to search huge areas of water, looking for food.
“So these bats solve this problem by working in small groups,” Yovel told CNBC at the YPO Edge conference in Singapore last Friday. The bats spread out in their search for fish and also pick up on the sounds that the other bats emit and receive, Yovel explained.
“Each bat searches as an individual but they eavesdrop on each other,” he said, explaining how they know what the others are doing without actively communicating.
To collect data, his team puts tiny sensors on the bats. The findings would be used to draw up an algorithm that can train robots to work together as a team and ‘talk’ to one another using only sound.
Those robots could potentially “do anything: They could map an environment, search for something, clean something, whatever you can imagine,” he said.
Bats use echolocation to see in the dark or hunt for food. They make calls as they fly and listen to the returning echoes to build a sonic map of their surroundings — this tells them how far away something is by the length of time it takes for the sounds to return, according to the Bat Conservation Trust’s website.
Yovel’s team is currently building a group of small robots — each robot would have a speaker and two microphones to pick up and emit sound.
They would initially work underwater, where the robots would use sonar technology to move around and search for objects individually, but also listen in on what the others in the pack were finding — this would mimic the way the fish-eating bats hunt for food, according to Yovel.
The research is still in its nascent stage and he estimated that perhaps in five years, his team would have potentially trained a group of robots to use only sound to communicate.
“It’s just because we want to prove the point,” Yovel said. “We are not yet building anything that is applicative. We want to show that they really can perform this task efficiently, and then, hopefully, somebody would be interested and then we can think of actual robots. We want to show that the algorithm works.”
