It's quite an interesting idea, but the question is the funding. How will the researchers at MIT be able to get enough dough to produce them commercially? Another problem is interaction between robotic fish and real fish. If a shark ate a robotic fish by mistake, would it kill him? I think that it would be interesting if we studied what parts of the visible light spectrum fish actually see, and then make the regular fish semi-invisible to the fish, so that it would look something like a rock. That way, the fish wouldn't eat them, but they wouldn't run straight into them.