As anyone who’s ever observed an ant farm will attest, ants tend to pick the most efficient route through their maze to a food source, even when confronted with two seemingly identical routes. Scientists weren’t exactly sure how ants are able to do this, especially with limited vision and limited brainpower. So a team of researchers built a robotic ant army to find out.
For their study on swarm intelligence, researchers from the French National Center for Scientific Research (CNRS) and the New Jersey Institute of Technology turned to Alice, a type of tiny, wheeled cube robot developed several years ago by Swiss scientists. The researchers programmed the robots to mimic some basic ant behaviors, such as to avoid obstacles and to follow light trails, which were used to represent pheromone trails that real ants leave for each other. They discovered that relying on just these simple behaviors, the robots were able to select the shortest path through a maze most of the time (over 70 percent of trials), approaching ant-like levels of success.
The high level of performance of robots using only simple programs led the researchers to conclude that ants aren’t using any “complex cognitive process” to pick their route, and instead rely mainly on a loose obstacle-avoidance system and directions laid out by peers that have already successfully navigated the maze. The scientists published their results in an open access journal PloS Computational Biology this week. Their next step is to figure out how ants come up with the structure of their mazes in the first place. Don’t worry about any robotic ants getting loose, though — the researchers returned their robots to a supply company, as one of the scientists leading the group told The Verge.