Hacker News Reader

>But it is essential. Google (and Apple) maps provide lane specific navigation directions, which are really only possible with assisted GPS. In theory GPS has the accuracy, but remember you are in a big metal box filled with electrical wires. Not to mention other applications like isle specific navigation within stores, which is made possible by bluetooth beacons.

As someone who's degree had a course in sattelite rangefinding(both code and phase), this is bullshit. Code GPS measurement should be enough for that. Especially when you already contain road data to which you can snap, and probably have access to accelerometer too which can easily improve accuracy of 'snapping'.

You don't need to get precision down to a lane on the road for purpose of navigation, you just need to 'snap' to the right road. and warn user beforehand if they need to take a turn. Just like 'dumb' GPS navigation did prior to smartphone era, while having access to less satellite systems.

> As someone who's degree had a course in sattelite rangefinding

So you are well versed in multipath reflection and why it makes unassisted GPS in urban environments with tall buildings close together almost completely unusable?

> You don't need to get precision down to a lane on the road for purpose of navigation

If you want to build an early 2000s Garmin competitor, sure. But people don't buy $1,500 pocket computers every 18 months to keep doing the same thing.

Are you saying every tomtom, garmin, etc used Bluetooth and cell services for navigation all those years?

Not to be argumentative but dedicated GPS devices didn’t always have that option. Also I’m curious: it uses triangulation off cell towers to help. I don’t see how Bluetooth helps improve accuracy. Do they map WiFi spots accurately?

No, they used other technologies, such as DGPS.

https://en.wikipedia.org/wiki/Differential_GPS

> Do they map WiFi spots accurately?

Yes, there are a number of companies that maintain BSSID->Lat/Long Databases. Skyhook is probably the biggest commercial one, and https://www.wigle.net/ the biggest open source one.

>So you are well versed in multipath reflection and why it makes unassisted GPS in urban environments with tall buildings close together almost completely unusable?

Yes, a common problem even for static measurements. Yet it worked with 'dumb' gps navigations. Btw - it's not just a p roblem in urban evirorments, especially for phase measurements - you shouldn't do precise measurements near the trees.

Just to illustrate when doing a phase measurement, under good HDOP constellation, next to the forest we barely could achieve an accuracy of half meter using Leica 1200. We measured the same reference point using total station and neighboring network of reference points.

For a reference, you could easily get sub 2cm accuracy, although below 10cm was a minimum target for that job.

>If you want to build an early 2000s Garmin competitor, sure. But people don't buy $1,500 pocket computers every 18 months to keep doing the same thing.

Sure, but it isn't a CORE REQUIREMENT for the service to function. you can degrade it, and have it work pretty well, without requiring all users to submit other tracking data sources.

> Sure, but it isn't a CORE REQUIREMENT for the service to function.

Again, you don't know Maps product roadmap. Google is taking its "Next Billion Users" initiative very seriously internally. They might be looking at ways to offer navigation using only a Wifi/Bluetooth chip so they can make Android devices as cheaply as possible.

There is also the possibility that GPS may not be functional in areas Google wants to serve. The US Government blacked out GPS over multiple states during a test. China has satellites that can rip other countries satellites out of the sky. Russia intentionally crashed an old weather satellite into an Iridium satellite just as a weird flex.

The Google product roadmap has absolutely nothing to do with whether Google maps — which we concretely know can function without network assistance — should REQUIRE said network assistance for all users at present. Yah, the results may be better with network assurance, but it should still work with out it.

This sentiment also applies to your second point — if Google wants to get some kind of navigation working without using gps in some niche market categories, that’s fine. But that shouldn’t require me, with the gps module I paid for in my expensive smartphone, to share data with Google that I don’t want them to have and they don’t technically need.

And i as an user utterly do not care about their roadmap. I care about my experience.

That still does not excuse them for requiring access to those data sources, as they can still degrade service when user does not give consent. Also there are multiple networks at the moment(USA's GPS, Russia's GLONASS, China's COMPASS/BEIDOU(forgot which was geostationary system, and which one was global),and EU's Galileo), which makes total blackout of a single area unlikely.

And quite frankly, in case of total blackout we have a bigger problems than civilian GPS navigation.

Not correct. Wifi is not used to boost precision at speed. That is generally done via dead reckoning using the gyroscope and accelerometer. Wifi helps with bootstrapping a coarse location quickly before satellites can be acquired, and with indoor positioning.

And (along with cellular data) with acquiring the almanac data via the Internet instead of waiting for up to 30 minutes for the satellites to broadcast it.

Last time I checked it took 12.5 minutes for full almanac broadcast... Did something change?

Yes, but only under optimal conditions.

Just by downloading recent data without any prior knowledge of your location (aside of maybe determining which continent you're on, which most of the time is already known because of which cellular network you're connected to), you can usually easily reduce that to around a minute with no WiFi, BT or cellular triangulation necessary - and that's on a cold start; hot start will be instantaneous anyway.

> But it is essential. Google (and Apple) maps provide lane specific navigation directions, which are really only possible with assisted GPS.

Assuming that's true (I'm skeptical), there's no reason why it can't gracefully degrade without the added information.

Regardless, I'm not sure what you mean by lane-specific navigation. Yes, I've seen Google put up a little drawing of the multiple lanes, with the correct lanes for the next turn in bright white, and the incorrect lanes in grey, but that doesn't need to know what lane you're currently in, as it just highlights which lanes are correct and incorrect, regardless of where you are.

Lane specific driving directions?

How many Wifi access points and Bluetooth beacons is Google expecting to find on the highway while moving at 70mph?

I would presume they're hoping to find the devices of other nearby users?

But those would be in random cars around you, which are constantly moving. How could those help retrieving your own location on a map?

If there is a car travelling the same direction as you but a bit to the right, then you are not in the rightmost lane.

How would the phone know the wifi signal is coming from your right?

Because we also know about the car going in the other direction, and you came closer to it. Hypothetically.

Or if gps errors are correlated between phones.

Good point. Ok what about this then: Check the relative strength of cars travelling opposite way compared to same way.

I suppose if there are enough cars, and if you have access to the wifi and location data for enough of the phones, it might be possible to determine someone's relative location by triangulating the signal strength from multiple vantage points.

So my phone by itself doesn't know the direction of the signal, but two or more other phones could combine our location data with relative wifi signal strength and potentially get a better idea of where the signal is coming from. (I had forgotten that we were talking about an entity that potentially knows the data from all the devices, not just one endpoint.)

How about in town, going 20mph?

Most of the android tracing that I've seen logged on location history is wifi multilateration based. Since wifi is on most of the time, it's the quickest way of finding your location (since you're most likely on the road), and you can turn off GPS saving about 25mW or more of power.

But it does mean you need WiFi on negating any power saving you make from disabling GPS anyway.

This may be Waze beacons which use bluetooth to help position the user when they're going through tunnels, where there is no GPS coverage.

https://support.google.com/waze/partners/answer/9416071?hl=e...

https://www.theverge.com/2019/1/29/18202359/waze-beacons-nyc...

...positioned to give you a 1-dimensional measurement: how far are you along this specific tunnel? Expecting triangulation is, well, overly optimistic. By an order of magnitude, at least (and would require a far denser mesh of beacons, whereas this only needs one beacon every $measure ; or two, for added redundancy).

In other words, it's still useful for navigation like "you are now 2000 meters/yards/cubits into the tunnel, therefore you should be taking either of the two middle lanes", but the trick is, this doesn't require the app to know your current lane.

Maps routinely tells me to "use the right two lanes to exit", but it doesn't show any awareness of where I am. It says the same thing whether I'm in the left lane or the exit lanes. Are others getting directions like "merge two lanes to the right to exit"?

Similarly, I wish Maps would tell me the best lane to be in right now so that I have the minimum frustration with being in the proper lane for the next step.

Wifi and cell tower location are much less precise than GPS. They are irrelevant for lane detection.

Where they are useful is when you are indoors. Or when a tracking business wants an excuse to send your location to their servers.

Rando: "So what's your company do?"

Google employee: "I work at a tracking business"

Information about nearby devices is not essential for navigation, otherwise the permission wouldn't have been optional until now. It can improve the service in some situations, so the right way would be to explain the tradeoff and let the user control how much data is being shared.

Tinfoil hat: the number of US GPS satellites in operation has been going down, both in % of total satellites and in actual numbers over the years. I’d venture “US being able to rely on GPS” is probably not a great thing.

That being said, wanting more user data is probably the right answer.

> But it is essential. Google (and Apple) maps provide lane specific navigation directions, which are really only possible with assisted GPS. In theory GPS has the accuracy, but remember you are in a big metal box filled with electrical wires. Not to mention other applications like isle specific navigation within stores, which is made possible by bluetooth beacons.

Tell me exactly how this bluetooth beacons and WiFi APs will help you to determine the lanes? I did several research projects in this area and RF mapping by indoor beacons for navigating outdoors at vehicular speeds is very hard and unreliable. You don't want to get wrong bearing just because an WiFi AP has been moved to another spot. Nice in theory, BS in practice.

>But it is essential. Google (and Apple) maps provide lane specific navigation directions, which are really only possible with assisted GPS.

I'm trying very hard to not be sarcastic, but the past 10+ years it was not essential and served me great every time

If we're at the point someone is jamming GPS, I guarantee lane specific directions is going to be about the last thing I care about.

Are you suggesting that they have no choice but to completely stop locating the user as there's absolutely no way they could offer non-lane specific navigation? The vast majority of it would be identical and I'm sure they could easily and gracefully degrade to that when strictly necessary.

How much does a company need to degrade a service - at potentially great cost - to be compliant before they can claim it then would be better not to provide the service at all to customers unwilling to share data?

I am not claiming that this is where Google needs to draw the line in the sand - but there obviously is such a limit to these considerations.

Google doesn't have lane data for every road in the world, yet clearly navigation is already implemented to work without it?

I don't know why people throw up random excuses for companies instead of accepting the simplest answer - they want as much data as they can.

Google was able to provide the service before, now they patched it out. So clearly it is not a question of ability to gracefully degrade but of willingness.

My 10 year old navigation device (WindowsCE-based) has no wifi nor celluar and lane specific navigation works fine. Also having lane specific navigation is not essential.

Could you share more information about your device? What is it?

Watch your step on that slipper slope

I don't have enough experience with the technical challenges of getting a good signal from a phone's tiny GPS antenna inside a vehicle, but that difficulty shouldn't give a company the right to violate privacy. If hardware is limited, make it better or make the limitation clear to the user and let them decide if they want to give extra data to allow a better experience.

We've had good enough navigation in mobile phones for years, why would they suddenly need to make this change now? And why can't I just keep the slightly worse experience if I prefer GPS-only location?

> But it is essential. description of feature non-essential to navigation

Sure. And stand-alone GPS navigation devices also don't work in cars.

> Heck, we are on the verge of a major world conflict where GPS accuracy may be degraded intentionally by the US or jammed by other forces.

During which (brief?) time we will have other pressing concerns beyond degraded mapping services. Why should this temporary condition that on its own is certainly not 'crippling' be an excuse for accepting that Google/Alphabet and like minded corporations want technical approaches that enable them to know everything about us 24/7?

This reason is complete nonsense. Google tells you that you need to be in the right lane for a turn regardless of which lane you are actually in.

Is nobody capable of looking out the window at the street signs that tell you which lane you need to be in to turn right?

Not really. That accuracy is not done with RSSI measurements from far away SSIDs.

WiFi triangulation is absolutely useless when determining which lane are you on. Your whole comment has been derived from false assumptions.

> Google (and Apple) maps provide lane specific navigation directions, which are really only possible with assisted GPS.

lol wtf. Source?

"But it is essential."

My box full of old TomToms and Garmins guffaws at this patently absurd claim.

This is google apologetic bs.