I’ve been seeing some articles about using tiny robots in collective action to form a swarm to then accomplish tasks. One big advantage of many small robots would be the manufacturing efficiency as we’ve seen in consumer computers. OTOH, current robot users tend to use the military model, big and expensive and customized robots.
While many small/cheap has often beaten few big/expensive, there is the challenge of how to coordinate many robots to do tasks harder than any one of the small robots can accomplish. So this is an active area of research mostly via simulations (which can be much faster than real-world testing of actual robots, plus cheaper).
So here’s an interesting interactive site you can try a few of these things.
While this site gives you a good feel for how a swarm can do things the more interesting question is how the swarm could do the task WITHOUT the external input (i.e. the human), both in terms of what algorithms (per individual robot) get the task done and even more interesting whether the swarm, as a whole, can “learn” the solution to the task via “evolution” of each individual.
Unfortunately this is probably too hard to program (like a week or so) to just try as a fun little experiment, so it goes on my todo list as yet another way to burn up my time and excess compute cycles.