Modular robots have more flexibility, and several research groups around the world have demonstrated swarms of robots that are not controlled by a central brain, but instead each have computer programs that allow them to organize and work together. These systems are more robust. If one robot breaks down, the others can step in to do the job without needing to be reprogrammed.
But they come with their own challenges. Mainly, each robot has to be programmed so that when it works in a group, the collective executes the desired task.
“This is a very difficult thing to do,” Dorigo noted.
These new robots are essentially a swarm of modular units that communicate by Wi-Fi. When they join together, they cede control to one unit that takes over the entire organism. Any unit can become the brain, and how they cede control can occur randomly or be more directed according to the overall need.
Once connected, the brain unit issues high-level commands that propagate down the collective nervous system to the rest of the units. If a high-level command is meant for one unit specifically — say, unit F needs to pick up a brick — the robot’s own nervous system translates the command locally into the correct action. That makes programming simpler and requires less computational processing.
In this demonstration, the robots work by rolling across a floor in two dimensions, and the connections between them are rigid. Dorigo and his team would like to advance the technology to robotic units that have flexible joints and can move freely in three dimensions.