The collective intelligence of ants is used by these ant-robots to escape a prison without a plan or planner.
Individual ants are relatively simple creatures, but an ant colony can perform truly complex tasks, such as building and defending its nests or foraging great distances for food.
Recently, Harvard researchers took inspiration from ants and termites to design a team of relatively simple robots that can work collectively to accomplish complex tasks using just a few basic parameters. O research was published in ELife.
The research team started by studying how carpenter ants work together to burrow and escape a soft enclosure. At first, the ants inside the enclosure moved randomly, communicating via their antennae, before starting to work together to escape.
Ants primarily use their antennae to interact with the environment and other ants, a process called antennae. The researchers observed that the ants spontaneously congregated around the areas where they interacted most frequently. Over time, digging in one of these places was faster than the others, and the ants ended up digging tunnels outside the enclosure.
From these observations, Mahadevan and his team identified two parameters relevant to understanding the ants’ digging task: the strength of collective cooperation and the rate of digging. Numerical simulations of the mathematical models encoding these parameters have shown that ants can only successfully dig when they cooperate with each other strongly enough and at the same time dig effectively.
Based on this knowledge and the models, the researchers built robotic ants, nicknamed RAnts (for Robotic Ants), to see if they could work together to escape a similar enclosure. Instead of chemical pheromones, the RAnts used “photohormones”, fields of light left by roving Rants that mimic pheromone fields or antennae.
The Rants were programmed only by simple local rules: follow the gradient of the photohormone field, avoid more robots where the photohormone density was high, and pick up obstacles where the photohormone density was high and drop them where the photohormone density was low. These three rules allowed RAnts to quickly escape their confinement and, equally important, also allowed researchers to explore regions of behavior that are difficult to detect with real ants.
This approach is very flexible and resistant to detection and control errors. It can be extended and applied to teams of tens or hundreds of robots using different types of communication fields. It’s also more resilient than other collaborative problem-solving methods: even if individual robotic units fail, the rest of the team can complete the task.
REFERENCE
Dynamics of cooperative digging in collectives of ants and robots
