The animal kingdom is fertile ground for roboticists looking to improve on their designs, with everything from insects, fish, seahorses, jellyfish, caterpillars, snakes and birds providing inspiration. Now researchers at Georgia Tech are turning to cats to help soften robot landings. Rather than strapping some felines to a robot's underside, the team is studying the way cats twist in the air when falling to let future robots land safely from a jump or fall.
Cats have an innate ability to orient themselves as they fall so that they land on their feet. This ability, called the cat righting reflex, is thanks to their unusually flexible backbone and lack of a functional collarbone. While it is theoretically possible for us humans to change our body poses in mid air, the researchers say that joint limits and muscle strength constraints prevent us from doing so in practice.
Robots, on the other hand, could in the future be constructed to mimic a cat's righting ability and this is just what Karen Liu, an associate professor in the School of Interactive Computing (IC) at Georgia Tech, and her team are looking to facilitate by studying the physics of not only falling cats, but also the mid-air orientation of divers and astronauts.
To do this they simulated falls and also studied the impact of landings using a small robot mainly consisting of a main body and two symmetric legs with paddles. They found that although a well-designed robot has the processing power necessary to calculate how to achieve a softer landing, current motor and servo technology isn't able to move the hardware fast enough to produce cat-like impacts.
To overcome this, the team created a reduced-gravity environment in the lab by rigging up a tilted surface similar to an air hockey table outfitted with a leaf blower. This allowed the researchers to simulate the elements of a long fall and give the robot enough time to carry out the poses that allow it to right itself and land on its "feet".
Looking towards future search and rescue robots, the team plans to continue with further research into teaching robots the orientation skills required to reduce impact and damage when landing. But until motor and servo technology improves, they might consider strapping a cat to one side of a robot and buttering the other side to prevent any impacts in the first place.
The team's paper can be found here (PDF), and Professor Liu explains the team's experiments in the video below.
Source: Georgia Tech