“The idea is to get multiple robots to cooperate to do tasks that a single robot could not do alone,” Dorigo told Digital Trends. “Usually multiple robots would be controlled through some kind of centralized control, such as a CPU which knows what all of the robots are doing and makes decisions about their actions. The problem with this approach is that, as the number of robots grows, it becomes more and more complex and therefore computationally difficult. Additionally, if the computer that is controlling the robots breaks down, the whole system fails. What we instead want to do is to focus on self-organization.”
The innovative concept at the heart of this work is to get the robots to work both autonomously and as part of a group. Dorigo likens it to two people operating separately, then coming together and holding hands — at which point one person’s brain is switched off, and control of their body is passed over to the other person. The result is a “mergeable nervous systems,” in which individual autonomous bots with independent brain units form a larger collective. “This is something you don’t find anywhere in nature, but with robots it is possible — and we think it could be a very useful infrastructure for building a robot nervous system,” he said.
At present, the work is still a proof-of-concept. In principle, however, it’s possible to imagine all kinds of potential applications. “In the future, we are going to have more and more scenarios in which robots have to work together,” Dorigo said. “For example, if you were to have robots working together to build a house or some other structure you want them to do so in a robust coordinated way. That’s what we’ve developed with the software that we have written.”