The University of Leeds has been awarded £4.2 million to lead part of a national infrastructure research project in the U.K. with the vision of using small robots to create “self-repairing cities.” The general idea is to create swarms of small robots that will be able to zip around cities, keeping out of the way of people while proactively identifying weak infrastructure and making repairs before anything actually goes wrong.
This award is part of a larger national initiative to explore “how new ways of using robotics and autonomous systems can restore the balance between engineered and natural systems in the cities of the future.” This sounds awesome, and technological optimism is great, but it’s also important to temper expectations with reality, and avoid getting swept up in the hype of a press release. Like, that “of the future” phrase should immediately make you suspicious, because of a.) its rampant overuse in headlines by lazy tech bloggers b.) its inherently nonspecific nature. In other words, if something sounds implausibly good, well, it probably is, especially if there’s not a lot of detail to go along with it.
We do have some details about what the robotics part of this project is going to focus on, and it’s some pretty crazy stuff that we’re guessing is never actually going to happen.
According to the press release, “the researchers will initially develop new robot designs and technologies in three areas”:
• “Perch and Repair” – research to develop drones that can perch, like birds, on structures at height and perform repair tasks, such as repairing street lights.
Perching drones are realistic and do exist, using microspines or adhesive or bird-ish legs. It’s the “perform repair tasks” that really seems nuts. To stick with the street light example, let’s think about what it would take to replace the bulb, a relatively simple repair. The drone would have to have sensors and manipulators with enough intelligence, dexterity, and strength to remove the cover, pull out the old bulb, put in the new bulb, and get the cover back on. At best, this seems impractical.
Honestly, I don’t know why a drone is the right way to tackle something like this. Is it because these things are high up? When humans repair high up things, we don’t do it from helicopters, because that would be silly. We use cranes. It seems like it might be more plausible to develop a mobile robotic boom lift: you wouldn’t have to worry about perching or stability or payload, and you could stick a pair of nice robot arms at the top along with a bunch of sensors and whatever else you might need to do repairs. You wouldn’t be able to reach everything that a drone would be able to, but since most of our infrastructure was created with cranes, most if it can be fixed with cranes as well.
• “Perceive and Patch” – research to develop drones able to autonomously inspect, diagnose, repair and prevent potholes in roads.
As with the perch and repair thing, I have no idea why you’d conceivably want to use a drone for this. You’d either be asking it to navigate at low altitude through a highly complex and dangerous urban environment, or you’d have to invent some kind of sensor system that could locate and analyze a dark hole in a dark road from an improbably long distance. You’d also need the drone to be able to carry whatever repair hardware is necessary, which (if you’re repairing potholes) includes a bunch of rocks. Good luck with that. And how does a drone “prevent” potholes, anyway?
It’s far less crazy, even borderline practical, to do this with a robotic ground vehicle. Like, you know, a truck. If you want to make it fully autonomous, that’s great, but the important part is having some pothole sensors at the front and a pothole repair system at the back. Whenever the vehicle drives over a pothole, the sensors inspect and diagnose it, sending commands to the repair system, which fills the pothole as the vehicle passes slowly over. You’d need to drive every lane of every street for this method to be effective, but with an autonomous vehicle (or a small fleet of autonomous vehicles), that’s not a huge burden.
In the interim, it wouldn’t be a terrible idea to do it with a manned ground vehicle, since we have those already, and they work. For example, the Georgia Tech Research Institute has been developing an “Automated Pavement Crack Detection and Sealing System.” It’s a robotic trailer that gets towed by a (manned) truck, and it’s able to detect and repair cracks in asphalt continuously at 3 mph. Scale this up to be able to deal with potholes, enable autonomous driving, and you have something very useful, and slightly realistic.
• “Fire and forget” – research to develop robots which will operate indefinitely within live utility pipes performing inspection, repair, metering and reporting tasks.
This seems realistic and achievable. Hooray! Of the five things that we’re promised here, “inspection” and “metering” and “reporting” are already done by deployable systems. That leaves indefinite operation and repair to really tackle. Harvesting energy from operating utility pipes seems like a solvable problem, either through microbial fuel cells or (more likely) turbine generators powered by moving fluid.
I’m less sure about how the repair part would work. Cleaning seems more realistic, or perhaps breaking clogs, stuff like that. Fixing a crack would be a lot more difficult, although it’s not out of the realm of near-term possibility if the robot was carrying some kind of glue or epoxy that it could apply in one shot.
Even if the things that this press release suggests might happen don’t actually happen, our hope is that a lot of good incremental research is going to happen here. What we don’t really understand is why such an improbable proposal is being made in the first place: is it not cool enough to suggest a robotic vehicle that can autonomously detect and fix potholes? Are drones really so buzzworthy that everything needs to be turned into a drone project regardless of practicality? I really, really hope not, because I’d much rather see research that embraces achievable results that will make my life better, as opposed to research that can’t possibly live up to the expectations that it generates.
Evan Ackerman is a senior editor at IEEE Spectrum. Since 2007, he has written over 6,000 articles on robotics and technology. He has a degree in Martian geology and is excellent at playing bagpipes.