Breaking navigations



  • Continuing the discussion from Hardware 🐛 from our favourite company:

    @loose said:

    I like to induce a nervous breakdown in my sat nav by deliberately going "off route" and then ignoring it's protests. There is, actually, a reason why I do this; but doing it for fun is.... more fun. Anyway, the quicker the sat nav comes up with an alternative route, the better I rate it.The longer it bleats on with "perform a U turn" and it's annoying variants, the quicker it's going to find itself hitch-hiking. Occasional it get its own back when I genuinely have to go off route because the road ahead is blocked: I want an alternative route, it tries to get me back to "where I was". It's almost as if...

    It depends on where you are a lot. If the turn you made is away from the destination, any navigation will be trying to turn you around simply because it still comes up as the “shortest” route to the destination.

    Of course it does depend on how the navigation is implemented. Typically the navigation first tries to return you to a point some distance ahead on the original route, because it is faster than full new calculation. But what happens then depends.

    The navigation I work on automatically starts the full recalculation on background, so it sometimes happens that it tells you to turn somewhere meaning to return you to the original route and couple of seconds later changes its mind, because the full recalculation produced a better route. Our navigation also penalizes the U-turn manoeuvre with rather large cost (something like 10 minutes), so it is really rare you hear that advice. Instead it tries to turn you around the block and it also makes it come with alternate route earlier.

    I did also think about adding a feature that some part of the old route would be automatically blocked, but there is a problem that in some cases you divert because there is a problem and other times you divert because you simply made a mistake and the navigation can't tell. And we never got around to creating a pop-up for it.



  • I currently use my windows phone 8.1 with the Nokia app. I only have two issues with it:

    1. If a road has been blocked for something pedestrianisation, it gets seriously upset with me "continually" going off route. Not in a verbal nagging way - it's quite good about that, but it knows the road is their (after all there are no map updates pending, and this is a Microsoft product after all) and refuses to recalculate.
    2. Sometimes, especially when you need it most in a hurry, it hasn't got a clue where it is.

    I win the game when the display spins round



  • @loose said:

    1. Sometimes, especially when you need it most in a hurry, it hasn't got a clue where it is.

    This would be either hardware problem, connection problem or stupidity of the GPS driver rather than the navigation.

    The thing is, the GPS signal has a whopping rate of one bit per second and the receiver needs 60 bits from each of at least four satellites to be able to calculate position. Add some scanning the frequencies for which satellites are visible and unassisted the receiver can easily take 5-10 minutes before getting a fix.

    However there are two ways to optimize it. The data don't change much, so they can be cached for some time and they can be downloaded from the internet in fraction of the time and with a coarse position (which can be obtained by looking up coordinates of the BTS it is connected over) it also helps selecting which satellites to scan for, so you can get a fix in couple of seconds.

    So if you don't have network and you either didn't use GPS long enough (day or two is long) that the cache is useless or the chip is stupid and didn't cache the data (called almanac), you get a long wait.



  • Yes, I have experienced that - especially with other sat navs. It may well be the case here. I have experienced the "phenomena" with Microsoft products and I normally describe thus:

    • You ask [insert MS App of your choice] for a new... whatever
    • MS look around and finds an instance of it in "cache" and serves it up

    In the case of my sat nav, this can be where I last was. Not a great issue as the difference is great enough to make even MS go :wtf: But mostly, it is "half a mile up the road" - probably because it was too impatient to get a proper fix. And that gives it (and me) serious issues.

    With other MS products, my "support" mantra after any application crash, was / is Reboot!

    .



  • You can also pick up the GLONASS signals, as well as Galileo and BeiDou-2, if they are even finished. Plus IRNSS if you happen to be near India.

    What I don't know is why we aren't using ground-based systems instead. Seems like it would be much more cost-effective. Maybe the atmosphere and curvature of earth are a problem? You could still place a few antennas in densely populated areas (big cities).


  • area_deu

    @anonymous234 said:

    What I don't know is why we aren't using ground-based systems instead.

    Phones do!
    It's apparently less accurate.



  • @anonymous234 said:

    What I don't know is why we aren't using ground-based systems instead.

    Aren't we kinda, with the cell tower and WiFi hotspot assist schemes?

    @anonymous234 said:

    Seems like it would be much more cost-effective.

    Not sure about that. You need GPS anyway, for folk on open sea / any other place where not even the strongest land signal can reach you. How is "using two systems" cheaper than "using one system"?



  • The satellites are up their, broadcasting their position regardless of what their purpose is. GPS systems just take advantage of that fact, and that the system is "in place". Various land based navigations systems have been used, abandoned and reused. The issue with them is coverage and accuracy (especially) at distance from the transmitter(s).



  • @Maciejasjmj said:

    Aren't we kinda, with the cell tower and WiFi hotspot assist schemes?

    Yeah, but that's kind of a hack on top of an existing system (especially the WiFi one), I wonder if it couldn't be made better.

    @Maciejasjmj said:

    Not sure about that. You need GPS anyway, for folk on open sea / any other place where not even the strongest land signal can reach you. How is "using two systems" cheaper than "using one system"?

    I guess that's true... specially if you have four redundant global systems by competing countries :P . But still, wouldn't a handful of extra land antennas help a lot with precision and fix times? Oh wait we already have that...

    How about indoors positioning? For autonomous robots, centimetre (or even millimetre) precision is crucial, and probably useful on mobile phones for reasons I can't think of (other than tracking people in stores). I predict sooner or later we're going to see those things developed, standardized and installed in pretty much every building.



  • @anonymous234 said:

    How about indoors positioning?

    Microsoft is working on it, and anyway you can probably whip your own with a couple of RF transmitters.



  • @anonymous234 said:

    What I don't know is why we aren't using ground-based systems instead.

    We are. GSM can estimate distance to the BTS (it needs this, because it is time mutliplex and it needs to time the transmission) and triangulation can be done from that or nearby Wi-Fi hotspots can be used (I wonder why they estimate the distance there; it does not have much in a way to provide precise timing).

    However, this is not as accurate. DME has nominal accuracy about 0.1 nmi (~600 ft, ~200 m) and the GSM distance estimation would be similar.

    @anonymous234 said:

    Yeah, but that's kind of a hack on top of an existing system (especially the WiFi one), I wonder if it couldn't be made better.

    The GSM and Wi-Fi systems are only used because they just take advantage of the system built for another purpose. Adding repeaters with more precise timing or very precise time signal would have extra cost and GPS is already there and works.

    GPS has atomic clock on each satellite to provide the very precise time signal, but there is only 31 of them active and only 65 were built and launched so far.

    @anonymous234 said:

    GNSS augmentation

    That just receives the satellite signal at fixed position and reports the measured error.

    @anonymous234 said:

    Seems like it would be much more cost-effective.

    No, it wouldn't. FAA talks about retiring VORs and not installing new ILSs because they are not const-effective. Because GPS is there and it covers every place on Earth while ground-based navigation systems only cover small areas. And short wave transmissions that can be broadcast over long distances don't provide the precision since the length of the signal path differs with atmospheric conditions.



  • @anonymous234 said:

    What I don't know is why we aren't using ground-based systems instead.

    Most phones already do. Cell towers mostly, and some use known wifi MACs as well.



  • @loose said:

    The satellites are up their [...]

    *twitch*



  • What I would really like to know is how the Wi-Fi location provider estimates the distance. In GSM it is easy, because there the response to initial connect has timed delay, since the protocol requires to know the delay. But Wi-Fi is broadcast whenever, cross your fingers and hope there won't be collision and has no responses with deterministic delay.



  • @Bulb said:

    What I would really like to know is how the Wi-Fi location provider estimates the distance.

    Signal strength? Given enough hotspots with known locations, you can probably triangulate pretty fine just based on that.



  • Probably. Given how the signal varies randomly in practice I don't think it would give much precision, but if you can pick a Wi-Fi from 100 m and don't really expect better precision than 100 m it's probably good enough.



  • @Bulb said:

    What I would really like to know is how the Wi-Fi location provider estimates the distance.

    Haven't you always noticed all those trained dogs running back and forth between cellphones and access points, then barking out the distance in dog-code?

    Point is: dogs.



  • Sort of :hanzo: @Maciejasjmj

    Triangulation and fuzzy logic (if I understand fuzzy logic correctly)

    Each tower knows exactly how far away it is from any other. Each tower receives a (specific) "heartbeat" pulse from a phone at a specific time and strength. The exact strength and time is immaterial because the signal originated at a given time and strength, everything else is relative after that. This puts the phone, from the perspective of an individual tower, somewhere within a circle. Where such circles intersect, is where the phone is.

    The problem is that there could be many candidate locations (this is probably where the fuzzy logic comes in), I think, all but the strongest signals are ignored. Fun happens if there is a great big RF reducing object between phone and tower.

    No, I have never worked with a "telecoms" company, nor do I intend to do so in the near or distant future. However, I have spent many hours on site fixing their systems, and it get lonely and boring after a while and there is nothing left to talk about except work.

    OR. The tower could just say to the phone "WTF are you? Can you see any street names / shops / house numbers / landmarks?" And depending on the smartness of the phone, you get an accurate and meaningful response.



  • Which reminds me:

    I was in a lift with a couple of senior managements types, who were showing me off of the premises after upgrading their switching data system (this was in the early days of Mercury) when one on them pulls out a vodaphone. I, in my normal manner, said " I would have thought it would be a case of instant dismal to use one of those around here". He harrumphed something, and his colleague said "I will be when Phase 1 gets rolled out". We all laughed.



  • One thing I noticed recently in W10M is, once I reset it recently and downloaded maps, and asked it to drive me home, apparently there's a new navigation app built in. I like some things about it better, like how it says things in decimals instead of yards.

    Overall, I've been quite satisfied with it. I usually know how to get onto a main road, and which way to start in, so it has time to figure out where I am, and it always recalculates fairly quickly. I guess YMMV.



  • @loose said:

    Each tower knows exactly how far away it is from any other. Each tower receives a (specific) "heartbeat" pulse from a phone at a specific time and strength.

    I was asking about the Wi-Fi case and the access points definitely don't cooperate in that case.

    AFAIK in good old GSM the system is like this: now and then each tower shouts “does anybody want to talk to me” and waits for responses. The phone responds a specific time after receiving that message and the tower than calls back with “hello, your delay was x μs and your next slot is y” and the phone simply converts x μs to distance. The towers also don't need to know anything about each other. However if they are CDMA, they are using the same “call when you feel like it, cross your fingers and if there is a collision, try again later” as Wi-Fi does (that is what is actually called CDMA) so what you say might be used there.

    Either way, the GSM has resolution somewhere in the range of 0.1–1 μs, giving positional accuracy 30–300 m. I doubt anything based on signal strength can do better. On the other hand GPS is accurate to around 10 ns, which allows the positional accuracy of couple of metres.



  • @loose said:

    OR. The tower could just say to the phone "WTF are you? Can you see any street names / shops / house numbers / landmarks?" And depending on the smartness of the phone, you get an accurate and meaningful response.

    Useful for phone navigation, less useful for tracking the phone though ("yeah, I can totally see my house from here, Mr. Policeman").


  • Discourse touched me in a no-no place

    @Bulb said:

    The thing is, the GPS signal has a whopping rate of one bit per second and the receiver needs 60 bits from each of at least four satellites to be able to calculate position.

    At the risk of :hanzo: I kinda have to :squint: at this and wonder if you're being sloppy in your terminology. I know that typically you'll get a position fix NMEA record every second once position is established, and it's common to see a GPS report position changes with a delta of much less than a minute.


  • Discourse touched me in a no-no place

    @anonymous234 said:

    What I don't know is why we aren't using ground-based systems instead. Seems like it would be much more cost-effective. Maybe the atmosphere and curvature of earth are a problem? You could still place a few antennas in densely populated areas (big cities).

    In the US there ARE ground-based helper systems: WAAS is the one I know of.


  • Discourse touched me in a no-no place

    @Maciejasjmj said:

    Aren't we kinda, with the cell tower and WiFi hotspot assist schemes?

    That just helps the receiver get the initial fix faster. "You're somewhere near 153 NW 17th st, because that's where the cell tower is located that your phone is talking to".



  • @FrostCat said:

    At the risk of I kinda have to :squint: at this and wonder if you're being sloppy in your terminology. I know that typically you'll get a position fix NMEA record every second once position is established, and it's common to see a GPS report position changes with a delta of much less than a minute.

    I admit I was a tiny bit sloppy. But just a tiny bit.

    The thing is that to calculate the position, the receiver needs to know the position of the satellites and for this it needs to download the orbital parameters, ephemerides, that have 60 bits for each satellite (they have much more, but the rest is known constants). From these parameters the position of the satellite can be calculated, so the receiver needs them at initialization and then update them every couple of hours.

    From then on, the start edge of each bit serves as the very accurate time signal, so at each bit the position can be updated. But for the calculation, sufficiently recent ephemerides are needed and that is the 60 bits per satellite.

    This also fixes the update rate at 1 s, which is hopelessly inadequate for any automatic control, which is why any automatic control device (in aircraft, missile, vehicle, etc.) needs an inertial reference system that is just coupled to GPS for drift correction.

    @FrostCat said:

    In the US there ARE ground-based helper systems: WAAS is the one I know of.

    Yes, they are ground-based, but they don't transmit their own signals. They just measure error of the satellite signal at known locations and these measurements are collected and broadcast, again via satellite, to the mobile receivers so they can apply correction appropriate for their location.


  • Discourse touched me in a no-no place

    @Bulb said:

    Yes, they are ground-based, but they don't transmit their own signals.

    Ah, TIL.



  • @Bulb said:

    The thing is, the GPS signal has a whopping rate of one bit per second and the receiver needs 60 bits from each of at least four satellites to be able to calculate position. Add some scanning the frequencies for which satellites are visible and unassisted the receiver can easily take 5-10 minutes before getting a fix.

    In car nav has been getting worse for a long time now. Back when it was a $2500 option, the units used to supplement GPS data with input from the speed sensor and inertial guidance via onboard gyroscopes. The unit I had ten years ago worked great when I drove around a basement parking garage under a forty story building. My current unit blanks the screen (it shows a tunnel image) when I go under a slightly longer than average bridge.



  • I think the sat nav in my car estimates how far through the tunnel you are.



  • @Bulb said:

    @loose said:
    I like to induce a nervous breakdown in my sat nav by deliberately going "off route" and then ignoring it's protests. ... doing it for fun is.... more fun.

    :+1: Yes, I like playing "Confuse-a-Garmin."

    The biggest problem I've had (not as a result of trying to confuse it; just raisins) is it occasionally deciding I wanted to go somewhere other than the destination I programmed. Specifically, home. Last time this happened, home was two states away.

    Fortunately, so far I've only had this happen in places that I knew my way around well enough to realize fairly quickly that something was wrong. I did wonder why it was taking through downtown San Francisco and across the Bay Bridge, rather than the more logical (to me) route down the Peninsula and across the San Mateo Bridge, but thought maybe it was slightly shorter. But when it told me to take the freeway north when I knew my destination was south, I knew something was wrong. Fortunately, I knew my way well enough get where I was going without it (I only really needed it for getting from the freeway to my motel; as it turned out, the motel was right next to the freeway, so I didn't need it at all), but if it should happen somewhere that I don't know my way around, I could go way out of my way before I realized anything was wrong.



  • my buttumption is that they don't estimate the distance.

    @Bulb said:

    Probably. Given how the signal varies randomly in practice I don't think it would give much precision, but if you can pick a Wi-Fi from 100 m and don't really expect better precision than 100 m it's probably good enough.

    this + lots of Wi-Fi Hotspots with a known location(~100m) and you can narrow it down a little more.



  • I had similar happen a few times. There have also been times when I have selected / programmed a long distance destination / route I am 100% familiar with (for raisins :laughing:) Only to get that horrible feeling that it is taking me the wrong way.

    Perhaps the sat navs are actually messing with us...



  • @loose said:

    route I am 100% familiar with (for raisins :laughing:)

    Most of the little travel I do is in areas I'm already pretty familiar with, so I only need navigation for the "last mile" — how do I get from the freeway to address, not how do I get to city (or the first mile — how do I get from here back to the freeway). However, since it's easier (and much safer) to program the destination before I start driving, I get to listen to the "helpful" directions for the route I already know, too. This might actually be helpful if I had a fancier one that includes current traffic information, but I don't.



  • mine does :p



  • @HardwareGeek said:

    :+1: Yes, I like playing "Confuse-a-Garmin."

    Garmin is a strangely schizophrenic company. They make aircraft instrumentation where extreme quality control is needed (and checked by the authorities) and they make the handheld outdoor/sailing/aviation units which don't need such certifications, but generally work well nevertheless and then they make the PDAs for cars that are utter crap.



  • Actually, if it is a German company, then you might need some "air time" for things to improve. It could also account for why so many sat navs direct their drivers to "non-existent" river crossings...



  • @loose said:

    Actually, if it is a German company, then you might need some "air time" for things to improve.

    Sorry, I don't get this.

    @loose said:

    It could also account for why so many sat navs direct their drivers to "non-existent" river crossings...

    I would rather blame failed government plans and map maker expectations.

    The map is shipped by the map provider (there are two global map makers, Navteq and TomTom), then the navigation makers have to convert it to the format the navigation application needs, which can easily take days for large countries and weeks or even months for the whole world, and then it is distributed.

    Therefore the map provider will add roads that are expected to be opened in near future. The more sensible map makers at least add a date of expected opening, but Navteq simply estimates the release process will take 6 month and includes everything that is expected to be open in 6 month as simply open. This might have been appropriate time back when maps were distributed on CDs, but nowadays it is too much. Yet the policy sticks.

    Now even 6 month is awfully long for any kind of plans, but it will also take some time to process the entry, so the plan they decide this on is a year or more old when you use the navigation. The probability that some problem will come up during that time that will delay the construction approaches certainty.

    In Prague we had a road that was half a year from opening for couple of years. Most of it was built, but there was a short part where administrative obstacles prevented building it. But since it was a short part, it was expected to be built in couple of months once the administrative obstacles were dealt with and there was no estimate on how long the administrative would take.

    This is how you end up with navigation telling you to turn right to the field beside the road.



  • @Bulb said:

    The thing is, the GPS signal has a whopping rate of one bit per second

    Cite? This has it at 50bps. This seems to concur.



  • @Bulb said:

    Sorry, I don't get this.

    Bless :smile:

    @Bulb said:

    They make aircraft instrumentation where extreme quality control is needed

    My understanding is that "air time" (in the context of a car) is when all the wheels are off the ground at the same time. This could be classed as flying, at which time the instrumentation could be said to be avionic.

    @Bulb said:

    they make the handheld outdoor/ sailing /aviation units which don't need such certifications, but generally work well nevertheless

    The UK papers are full (subjective assessment) of stories about people and their sat navs and the sort of "bad" places they end up because of them. Which mostly seem to be stuck in rivers. This would then make the instrumentation, technically, nautical. My implication being "they (the sat nav) were going home or returning to their natural habitat".

    I will leave the reason for my mention of German, as an exercise for the student.

    :smile:



  • This exists near me - the river and road meet at a ford.

    This is what it looks like approached from the south.

    There has been at least one incident since I've lived near it of someone misunderstanding the sat nav instruction to continue along the road, and ending up going straight into the river. :facepalm:



  • @Jaime said:

    My current unit blanks the screen (it shows a tunnel image) when I go under a slightly longer than average bridge.

    The built-in stuff on my Ford Fusion works going through the tunnels under the Baltimore harbor.


  • Discourse touched me in a no-no place

    @loopback0 said:

    ending up going straight into the river.

    Well, what else are you supposed to do? The road goes right across the river, and that flood gauge, if I'm reading it right, says the level's too low to be problematic.



  • @FrostCat said:

    Well, what else are you supposed to do?

    You could use the ford to get to the road on the other side?


  • area_deu

    What if you have a Volkswagen?



  • @FrostCat said:

    Well, what else are you supposed to do?

    Not go straight, and follow the road.



  • @aliceif said:

    What if you have a Volkswagen?

    :giggity:


  • Discourse touched me in a no-no place

    @boomzilla said:

    You could use the ford to get to the road on the other side?

    Well, I wanted to make sure that's what he meant. Based on his reply below I assume people are getting confused about "do I take the left fork or the right one". Stay on the fork with pavement under the water does seem like sage advice.

    Building a fucking bridge, I dunno, maybe that makes too much sense?


  • Discourse touched me in a no-no place

    @loopback0 said:

    Not go straight, and follow the road.

    Maybe it's just me but I would call the river a right turn not straight.



  • @PJH said:

    Cite? This has it at 50bps. This seems to concur.

    Hm, you are right. I remembered it wrong.

    You do have to listen to each satellite for quite long before you can use it though, unless you get the data faster from internet. And the synchronized edge is also once per second.



  • I don't understand why there isn't a bridge there.


Log in to reply
 

Looks like your connection to What the Daily WTF? was lost, please wait while we try to reconnect.