Most wall-climbing robots that we’re familiar with use one of just a few different techniques to stick to vertical surfaces. Generally, they’re either using magnets, vacuums, or the recently popular gecko foot adhesive pads. There are other more exotic systems as well, like microspines, cloth grabbing, elecrostatics, and hot glue.
No matter what technique you use, you’re always taking a risk with a climbing robot that doesn’t depend on external infrastructure: if your climbing system of choice ever fails, you robot will very shortly find itself transformed into a sad little pile of brokenness, thanks to gravity. To be safe, your climbing robot needs to be able to fly, too.
Researchers at KAIST’s Urban Robotics Lab, in South Korea, who developed this robot, say that other climbing robots have trouble making the transition from lab to commercial deployment because nobody wants to risk their delicate and expensive sensor payloads on a robot designed to climb very high up with a single point of potentially catatrophic failure. KAIST’s robot, on the other hand, can just fly safely to the ground if it ever becomes unstuck.
You might be wondering what the point of doing this is: If the robot can fly, and if its wall-climbing capability takes just as much energy (if not more energy) than flight, why even bother with the wheels and stuff? Why not just fly around the surface that you’d otherwise be climbing? My guess would be that some types of infrastructure inspections require direct contact with structures, whether it’s for a specific type of sensor or just a level of detail and stability that can’t be achieved without very close physical proximity.
In any case, multimodal robots like this offer substantial flexibility at the cost of increased complexity in both construction and control, but for applications where (say) falling to your death at any moment is a serious concern, it’s probably worth all the hassle.