Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Making regular GPS ultra-precise

norwegianscitechnews.com

Post reply on HN
Print view View original

61–68 of 68 posts

Re: Making regular GPS ultra-precise

#61

Post by SpicyUme »

Havoc wrote: post #42
crote wrote: post #20

Earlier quoted context omitted.

GPS beacons would be stupidly expensive, as you'd need tens of thousands of them per city, and each one of them would need a very accurate atomic clock. It would make far more sense (but still unviable) to go for Eurobalise-style RFID tags embedded in the road surface.

>you'd need tens of thousands of them per city I was thinking more find a tall building and throw a single one there. Or maybe tallest 3. As I understand it even a small number of extra satellites with line of sight can improve results a fair bit. So aiming at low hanging fruit rather than blanket city >very accurate atomic clock. Indeed, but I'd think managing that on ground is easier than in space.

What happens when the building sways multiple feet in the wind? We don't normally think about them moving, but for accurate positioning I'd think the additional distance to the beacon could be a problem.

Re: Making regular GPS ultra-precise

#62

Post by eqvinox »

mapt wrote: post #44
eqvinox wrote: post #6

Nothing in this article is new, and the problem with RTK has always been the (unpaid) availability of reference stations. Good on them for trying to make a package of it, but maybe this "news" site could've used a bit less unchecked enthusiasm. Also, RTK is the opposite of "regular" GPS, it's generally considered a "special" usage mode of GPS. And discussing urban canyons with no mention of QZSS?

Can we agree that with close to 10,000 Starlink birds in the air and counting, that whatever succeeds GPS needs to be a much larger constellation than the 30-unit GPS constellation in MEO, flying much lower? It is absolutely insane to me that Japan would be trying to economize using Molniya orbits or geosynchronous orbits in 2025. Some BOTE math: There are ~40k square degrees in a sphere. If I'm in a dead-end alley i…

> Can we agree that with close to 10,000 Starlink birds in the air and counting, that whatever succeeds GPS needs to be a much larger constellation than the 30-unit GPS constellation in MEO, flying much lower?

For the record, block III GPS satellites are 3880kg. A Starlink satellite is 800kg. But I don't know how much of the GPS satellites' weight is a necessity due to the atomic clocks they carry. Either way the constellation would be at least a bit smaller due to launch costs.

Re: Making regular GPS ultra-precise

#63

Post by stlava »

I had some experience with RTK and sensor fusion about 13 years ago on a college project. At the time the only people that were using RTK in real world applications were tractor companies because it kind of matters that you're precise when seed drilling. I'm not sure how far RTK has come but it had pit falls back then like base station drift when the set of satellites it saw changed due to them going over the horizon.

Tractors don't go _that_ fast but I'm not sure I'd rely on it in an actual car for anything but slightly better GPS.

Re: Making regular GPS ultra-precise

#64

Post by adgjlsfhk1 »

_flux wrote: post #55
adgjlsfhk1 wrote: post #51

Earlier quoted context omitted.

yeah Phones have pretty good IMUs. the TLDR is you combine accelerometers and gyros (preferably a bunch of them to minimize error) with GPS which gives you pretty accurate position and can also give you velocity data with some tricks)

But if you don't have GPS, then you don't have a high-confidence low-precision absolute position input to your system? Whereas with wheels you already get a lot of data about that when they are not turning at all. A friend tried to make a phone ruler app (using IMU) and the experience was that it didn't take very long for the distance to shoot to the horizon. So, thought I should try how this kind of solution would w…

This type of solution works better at large scale where you have space and power budget for a lot more IMUs. Phones are very power and space constrained, so you only have room for 1-2 IMUs. In a car, 10W isn't a problem and there's plenty of room for higher power and higher quality sensors.

Re: Making regular GPS ultra-precise

#65

Post by ingsoc79 »

addaon wrote: post #50
sampton wrote: post #39

Earlier quoted context omitted.

Using building geometry to correct gnss signals is new.

I recall Uber looking into this around 8 years ago? Don’t know if it went past a publication.

Yep, they also used 3D building data for correcting position in urban canyons: https://www.uber.com/blog/rethinking-gps/

Re: Making regular GPS ultra-precise

#66

Post by NoiseBert69 »

adgjlsfhk1 wrote: post #49
eqvinox wrote: post #15

Earlier quoted context omitted.

"with enough integration time" — the article is about live navigation, which generally can't afford that.

You can get integration time while driving. You combine accelerometer and gyroscope data to cancel out movement effects.

That's a classic use-case of Kalman

Re: Making regular GPS ultra-precise

#68

Post by llamajams »

bluGill wrote: post #47
brookst wrote: post #37

Earlier quoted context omitted.

I don't get it. I can see how precision GPS is not required for safety, and how it is not sufficient for safety, but can you elaborate on why reducing noise in this signal is not even useful?

Just reread this and realized I probably didn't answer the question as you meant to ask it. If you read close you said GPS is not helpful for safety, but I didn't say it was useless for other purposes. When you are in a car you are only rarely in a situation where it is even possible for you to be on a different road than what the GPS says. Which is more likely, a car is driving through the fence around my yard, or d…

Disagree with your take about how much you can trust the GPS. Maybe on a sparse road network this may be the case, but on a dense road network, stacked roads with adjacent lanes (like on a highway interchange )it can be downright wrong, even after corrections. But like you said in those cases we would have to rely on other inputs to compute precuse lane level location.
Post reply on HN