Singapore branches out onto internet of trees Singapore is obsessed with trees. The island nation, population 5.45 million people, is home to around seven million trees – and manages many of them with an enormous Internet of Things monitoring scheme. Which is a very Singaporean thing to do, because another local obsession is tracking everything. The city-state's goal of …
We could use a few of those accident-averse trees around here in near Berlin: the local roads are full of signs warning that the trees will leap out and strike your car; the local nature reserve warns that trees may drop branches on your head.
Downright malevolent, they are!
I'm building the very thing (no pun) albeit on a smaller scale and have cows involved too! It's all so cool linking water, trees, cows etc with ESA and NASA datasets, LoRa, Grafana, MapBox and making the interface flexible enough to be used by blokes in fields to research scientists.
Lovin' it.
Wow, talk about mis-understanding what technology is capable of. Finite element modelling of a structure like a tree is extremely complex, and probably would require considerably more human effort to get meaningful results than sending a tree surgeon around to have a look. Singapore isn't even that big, so sending someone to have a look at a suspect tree isn't a problem, they'll be there less than an hour after you ask them to come.
I've seen enough rubbish churned out from ""press a button and an FE model prediction model pops out" type analysis to know that without some expert human interaction, incorrect results will pop out which could result in death or injury due to the human believing the results without question.
I'm no expert on FE. I hope I'm not talking complete rubbish here.
Automated analysis is usually much worse than expert analysis. However, that would be comparing apples and oranges.
The purpose of automated analysis is not to replace expert analysis. Rather, it is to be applied on a massive scale, typically several orders of magnitude larger than what you could have humans do - e.g. check every single tree every day. At that point, its results are still individually crap, but they become statistically useful.
You can use that data to inform your decision on where to invest your limited expert analysis resources, in such a way that they are statistically more likely to go where there's a problem.
If your subjects number in the millions, such as in this case, then even piss-poor automation can make a very large difference. Prioritize citizen reports, but after that, you send the few experts you have to the trees that the computer say look bad. Even if the computer is wrong nine times out of ten, it's still better than checking trees at random, because a random tree will be fine far more than 90% of times.
Yes, the computer might not flag a sick tree at all, and you might not send an expert there because of this, and people might get hurt because of this. But, well, this can and does happen with no computer involved. Because you'll never have enough experts to have expert monitoring of all of them at all times. So it's all a probabilities game. The automated analysis just weighs the dice a bit more in your favor, that's all.
At this point, I think the main benefit of FIA is just identifying trees. Which are then logged, inspected, and tagged. You can't 'send a tree surgeon around to have a look' if you don't know that you've got a tree to look at, and you don't know that you've tagged them unless you've counted them.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
