I have always been fascinated by swarms and their behaviour. Bees can cluster in thousands and their overall behaviour seems to be orchestrated by a "super-bee" that in reality does not exist. Same goes with flocks of starlings and schools of fish.
Very simple rules create complex behaviours.
Hence, no wonder that I got hooked by this news of researchers at Harvard that have created a swarm of robots, 1024 of them to be precise.
Cooperative robots have been studied over several years but nowhere near to this number of robots, because of cost in manufacturing so many robots and the complexity of algorithms to manage them.
The approach taken by Harvard researchers has been to simplify as much as possibile the robots and to enforce very simple rules. The simplification carries with it the loss of control and the emergence of errors. A robot may not know exactly its position, may have problem in moving in a given direction and so on.
The beauty of the system is that quantity can overcome these deficits and the overall result looks like each of the robots behaved perfectly.
As you can see in the clip, the robots can self coordinate to make up a variety of complex shapes, with no single robots in charge. What researchers did was to mark 4 robots as reference points that all the others can use to detect their relative position.
You can get all the details by reading the Harvard article.
What is interesting to me is that the researchers consider this as a test bed to experiment simple rules that can generate complex behaviour to be ready for the million of robots (like self driving cars) that will become commonplace in the next decade.
They also want to understand the emergence of intelligence out of simple behaviours, something that happens in our brain all the time.