newbie question on speed limits

My assumption is the GPS makers use road types. Mapsource has default values for different road types. Depending on the road type and the number of segments traveled an estimated travel time can be calculated.

I believe this goes back to the early model GPS receivers that Garmin made. I still have an original StreetPilot; there was no provision to generate a route within the device, you had to 1st create the route using MapSource then export it to the StreetPilot. MapSource had and still has under Preferences, Routing, Driving Speeds an editable table of speeds for 5 road types. It uses those values to calculate an ETA for the destination.

I have to assume that the modern GPSr models use a similar algorithm to do the calculations that is independent of the displayed speed limit.

A way to demonstrate this to yourself is to put the nüvi into simulation mode and generate a route that includes one of those roads that posts the speed limit on the screen. While it's "driving" on that route, look at the speed on the trip odometer screen. You will see that it is more than likely cruising at a speed above or below the posted limit. It also slows for curves and brakes and accelerates at intersections which shows the level sophistication that went into the program's design.

I've always wondered if it somehow adapts to the driver's habits using data from the track log. Maybe that's in the works.

The concept of the Nuvi "learning" a driver's speed habits was discusseed at length on another forum. A method to test and demonstrate this ability was developed basically using simulation and ETA on known routes. This method proved that "learning" does happen and affect simulaion speed on roads, at least on some models of Nuvi. When you do a hard reset, it loses this "learning" and the simulation speeds and ETA's go back to default values until the real driving patterns are again learned.

The only learning going on is your deviation from expected norms on a given route, while the route is active - I've not seen any indications anywhere that any GPSr is capable of "learning" in the sense that data is somehow 'saved' and then later included in future calculations.

It is probably not appropriate to link to other forums here, but I would be glad to refer you to the discussion privately if you are interested. In that discussion it was stated that Garmin support was contacted and they confirmed this learning ability.

See http://www.gpsreview.net/train-your-nuvi-eta/ training your Garmin.

thanks for all the answers and the link.

Something tells me deep down inside if Garmin's technology were really capable of learning to the extent people seem to think it is, Garmin would be marketing it all over the place as an advantage to other devices.

If they're marketing this... I've never seen it. I think they haven't made the claim because they can't substantiate it.

Ya, that be pretty cool though and maybe a little scary.

I too feel that Garmin would be all over this in terms of marketing if it were a feature that works. The GPS market is so competitive! In terms of more accurate arrival times, it would be helpful if the GPS units recognized intersections along the route with traffic signals and could use some sort of calculation, like 60% of the time you'll hit an intersection with a green light and 40% of the time you reach an intersection it will be red. I generally add 3-5 min. to my local route if the route I'm being led on has numerous traffic signals.

If you don't believe it, try it. It is pretty easy to demonstrate that simulation speeds and ETA are altered by driving habits if you follow the method given in the link provided by Charlesd45.

That said, I see no evidence that routes are changed because of the learned information. They could be if ETA figures in somewhere in Garmin's routing algorithm, but I have seen no evidence of it. It seems to me that it would only be truly useful if the learning applied to individual road segments, not just to road "classes". I am not surprised that Garmin doesn't use the feature in marketing. I don't think it adds sufficient useful function to make it worth advertising. It would just cause users to have expectations that wouldn't be realized in practice.