If you happen to be a robot, then you happen to have one very good way of getting home: GPS. But as every human driver knows, GPS isn’t perfect. Now, a dog-size robot—inspired by the uncanny navigation skills of desert ants—may offer a robust new alternative. The new system doesn’t require complex processing, and it is more accurate than consumer GPS alone—getting the bot within centimeters of its destination, as opposed to meters.
Ants are well-known for their superior navigation skills, finding their way home via chemical trails left by their compatriots. But in the desert, the scorching sun quickly burns such chemicals away, forcing ants there to evolve some tricks that happen to be much easier for humans to engineer.
Without chemical cues used by other ant species, the six-legged desert dwellers use their limited, ultraviolet-detecting eyesight to find rough patterns in the world around them. They also keep track of how far they’ve traveled by counting their footsteps and monitoring how fast the ground streams past. The ants accomplish this seemingly high-minded task with just a few thousand neurons, compared to about 100 billion in humans.
That limited brainpower made it possible for researchers to accomplish the same tasks using relatively simple computer processors. To establish its heading, “AntBot”—which has six insectlike legs and two simple eyes—uses an eye designed to detect the sun’s ultraviolet light and a pair of rotating polarizing filters to determine its relative position. Just like the desert ants, AntBot also counts its steps and monitors the speed of the ground flowing past. In an era where resolution is measured in megapixels, AntBot’s eyes have just 14 pixels between them.
In outdoor tests, AntBot successfully returned home from a distance of 5 meters to 14 meters 100% of the time. What’s more, it was far more accurate than civilian GPS, landing within 6.5 centimeters of its goal, the researchers report today in Science Robotics.
In situations where a GPS signal is unavailable or unreliable, this ant-style navigation could allow autonomous robots to explore unfamiliar or dangerous environments as far away as other planets. It could also help hapless robots clean up or deliver groceries a little bit closer to home.