To me "Gecko" implies van der Waals forces[1] which this does not use. It just uses suction cups. The only Gecko part is a tail for balance/counter-force.
Right. I was expecting this to use the gecko grabbing system developed in Mark Cutkowsky's lab at Stanford. That's about fifteen years old now. Works, but applications have been few.
Suction-based climbing robots are available as commercial products.[1]
The biggest application of climbing robots is cleaning ships' hulls. That uses magnets for adhesion. It's a huge win, since hull-cleaning robots don't require a drydock. Some can even be used on ships under way.
It's a great feeling to see a cool robotics project on the front page of Hacker News and realize they used your software (http://edgeimpulse.com) to train the machine learning component!
Congratulations to the authors. Now I better stop reading HN and get back to work :D
Cool bot. I'm guessing there most be some advantages to having a system that climbs the wall instead of descends from a line to justify the extra complexity.
Probably more flexibility? A window cleaning type system only really works for a straight segment of wall, plus if you have a human doing it, it’s potentially dangerous if anything happens to the system during for example a sudden high wind event.
you could use one robot for this, but on a building like the empire state building you would need one for each level's 4+ sides of wall.
This would surely have a safety line if used in real life, no one would want it falling from ten storeys up onto a busy street. So it wouldn't have quite as much freedom of movement as it ha in the lab.
a safety line that can potentially move side to side still is quite a lot more flexibility than just going straight up and down.
also, these things are like the size of an arduino board so i imagine they're cheap to deploy, and easy to replace. plus, if you have a fleet of them, you can probably just have some on standby ready to go, as opposed to a fixed window washing style system.
[1] https://www.pnas.org/doi/10.1073/pnas.192252799