Android Apporama: GPS & Navigation

Will Truman

Will Truman is the Editor-in-Chief of Ordinary Times. He is also on Twitter.

Related Post Roulette

14 Responses

  1. Michael Drew says:

    Sort of apropos, sort of O/T (the life-and-limb hook brought it to mind): riding around the torn-up highways of suburban Twin Cities in the landscaping truck this week, a thought occurred to me (might have been related to MH17 as well & thinking about why public transportation, or at least air travel, is safer than individual driving): over the long term, any who stand in the way of the path to full automatization of driving are “killing people” in the Ezra Klein-James Hanley sense of the accusation.

    It seems obvious to me that, though automatic systems are always prone to disasters, over time taking the massive input of human imperfection (with essentially no prior-restraint performance control at all – you can just get in a car and drive it; no one is going to make sure you’ve slept in the last 48 hours or you’re not drunk high off your ass unless you’ve had repeated f*ck-ups in the past), will be the surest path to a major reduction in our largest cause of accidental death and injury. Driverless cars shouldn’t be a curiosity and something to be overly cautious about. They should be seen as a major heath imperative for our country, especially given the transportation infrastructure/land planning approach we’re pretty much locked into.

    America is a car country; people need to be allowed to have their cars. But let’s let the machines do the bulk of the driving as soon as is feasible.Report

    • @michael-drew Though I don’t like the “People who support Policy X kill” formulation (unless it’s something pretty direct) (not that you are advocating that formulation here), I think you are right about the importance of the enormous safety implications here.

      It’s one of the reasons why I doubt those who say that trial lawyers will kill the robocar, or that regulators will or the cab lobby. Any and all of the above may delay deployment, but the gains are just too immense in the most fundamental of ways (lives, quality of life, money) to truly kill it.Report

  2. Mike Schilling says:

    it estimated within five minutes how long the trip would take, and that’s despite uneven traffic, toll booths, and so on.

    “Dammit, I sandbagged by a half-hour, and now I’m going to get him there too soon. Let’s find a nice … Jack-knifed big rig! Perfect!”Report

  3. greginak says:

    2 questions: How well do these compare to a dedicated GPS. I have a 3-4 year old Garmin which has its eccentricities but has generally performed well. It even did well, with an added SD card, driving all around Ireland.

    Do people actually use the voice mode? I’d rather drive blindfolded then listen to the voice prompts.Report

  4. Road Scholar says:

    Definitely tangential, but something I’ve been wondering about for a long time and can’t seem to find anywhere, probably for proprietary reasons. Does anyone know how these routing algorithms work?

    I use GPS extensively as you might imagine and I notice some oddities. Like, it won’t always give me the same route from point “A” to point “B”. So apparently the algorithm isn’t totally determinate.Report

    • veronica d in reply to Road Scholar says:

      @road-scholar — To actually find a decent point-a-point-b path is pretty easy, mostly using some variation of the A* (read “A star”) algorithm (which you can Google up if you’re painfully curious). The difficulty then comes from dealing with the complexity of the road network plus dynamic conditions (road closures, traffic, etc.). To handle the complexity, it seems most of the systems are hierarchal. Local-ish searches look at local roads. Long distance searches include mostly major highways, with local stuff to deal with the endpoints. Plus I suspect they do a fair amount of pre-computation, by which I mean selecting a few likely routes between certain key points. Then if you are searching from (for example) near Boston to somewhere near NYC, it already basically knows the good paths between the cities and only needs to include the variations on each end. From that it maybe generates a few dozen likely paths and scores them, based on traffic and stuff, and shows you the best few.

      Okay, so I’m guessing, but it’s probably something like this.

      So, if you get a route at the start of a trip, and say it finds a few dozen. Then you start down one of those routes for a few hours. By that time if you search again, you still get a few dozen, but now chosen from different basis points, with a different spectrum of possibilities. Plus the road conditions have changed, so the scoring will be different. You can expect to see different routes.Report

    • Brandon Berg in reply to Road Scholar says:

      I suspect that it uses either historical or real-time traffic data to determine the fastest route.Report

      • Waze does an amazing job with real-time-data, which is why I like it so much and why it is the first GPS my wife was sold on. Others had traffic subscription services (Copilot had a free trial), but I assume that they’re mostly working off speed limits, assumed speed limits, and maybe some purchased historical data.Report

      • veronica d in reply to Brandon Berg says:

        @brandon-berg — Right. In fact obviously they have a great source of traffic information coming from the users themselves, since the GPS can tell how fast you are going. Just upload and aggregate. Easy peasy.Report

  5. Mike Dwyer says:

    Will – regarding the offline Google Maps: They used to have an option where you could select a chunk of any given map and save it for offline use. As long as you can get a GPS lock it should work just fine. They put the option in there for people that want to use theirs while hiking or knew in advance they would be in a remote area.Report