We tend to think of animal-inspired robots as the culmination of a research project, but failed robots can highlight what we don't yet know.
You've studied an animal for years, maybe decades. You think you understand it, so you decide to build a robot based on what you know. And it fails. Spectacularly.
The question is: why? Answering that question leads to more research, and - in some cases - major breakthroughs.
Want a real-life example?
Jelle Atema has spent his whole career studying ocean animals' use chemical cues (essentially, smelling) to navigate. But an attempt to create a RoboLobster who did the same was a complete failure. They scrapped the project.
The missing puzzle piece was animals' ability to sense the water flow - the speed and direction water is moving. Without that, RoboLobster didn't know how to follow a chemical trail. Figuring out how animals sense flow and integrate that information with chemical cues has become a new focus for Atema.
Or there's Jack Costello and his jellyfish. Costello's twenty five years of studying how jellyfish move has revealed that they are the most efficient swimmers of any animal. That sparked the interest of the Navy, who funded a team of students at Virginia Tech to make a robotic jellyfish. Their first model had a rigid bell (that's the round body of the jellyfish), and ... wait for it ... it couldn't swim.
Replacing the hard top with a flexible silicone bell worked wonders, though, and sparked a new question: if flexibility is so important for jellyfish, might it be important for other swimmers and flyers? The answer is a resounding "yes." In fact, Costello and his colleagues recently announced universal bending rules that could help design better aircraft.