Small drones can do big jobs: Firefighters can use them to find hot spots in blazes, environmental monitors can find the source of hazardous chemical leaks. One just delivered a human kidney for transplant surgery.

But it takes lots of power to spin four helicopter blades fast enough to keep a quadcopter-type drone in the air. Most can only stay aloft for about 30 minutes.

So an international team of roboticists is trying to extend the time a drone can stay on the job. For inspiration, they turned to birds.

"Birds usually fly somewhere and they stay at the top of the roof or some tree branches," says Kaiyu Hang a roboticist at Yale University. "Then they look for their prey or they just stay there without flapping their wings all the time and they can still observe what is happening around them."

And observing takes a lot less energy than flying.

Hang and his colleagues developed a claw-like landing system that lets a drone grab on to a branch or pole, turn off its engines and continue to make observations while it's perched. Hang describes his work in the journal Science Robotics.

The team's new system also provides a way of landing even when there isn't a good place to land.

"We have developed a modularized landing gear framework that allows the drone to not only perch on some structures, but also rest on some structures when perching is not possible," Hang says.

By resting, Hang means the quadcopter can essentially lean against a ledge and stay there using just two of its four rotors — saving energy.

The ability to perch can be handy for a number of other reasons, too.

"If you perch, for example, underneath a bridge or underneath the eaves of a building you can ride out storms or bad weather that would make it hard to fly," says Stanford University's Mark Cutkosky.

The new landing system Hang and his colleagues have invented does have some drawbacks. For one thing, it adds weight to the drone. That means it takes more energy to keep the drone aloft.

"So there had better be a good trade-off in terms of really improving mission time in order to pay for that added weight," Cutkosky says.

And for now, a human has to fly the drone onto the perch site.

To make the landing system really useful, the drone would be able to land on its own. So the question is "what sorts of strategies can we use that would allow this system to discover perch-able sites?" Cutkosky says.

The Yale team has already begun integrating the drone's computer with its on-board camera, so it might not be too long before drones are able to decide on their own where to settle down for a rest.

Copyright 2021 NPR. To see more, visit https://www.npr.org.