[theamk]

FYI, here is a part which is still relevant today:

"the system continuously ironed out accumulated errors over time by comparing actual distances driven and turns made with road shapes on the map. Honey calls the technique "augmented dead reckoning."

That means driving through a long stretch of straight highway could begin to trip up Etak's system, since there were no turns and no distinct roads for the computer to algorithmically seize upon. If that happened, the driver could manually reposition the car cursor onto a location on the map using controls on the display."

Those things work, but they are barely "good enough". The biggest problem is, it is easy to lose your position if you are in the city -- maybe you had a few sharp turns, went around the parking cars a couple of times. And once this happens, the system is unlikely to correct itself, your only way will be to manually set position, which most people will not do.

[Animats]

No, Etak could recover position after you made a few turns, unless you were in a city with a very regular grid. The recent path would be matched against the map to find the best match. When Honey demoed the system to me, he drove around a vacant lot in Menlo Park to force it to lose position, and then it recovered after a few turns.

[theamk]

I am sure it could recover most of the time, in some places. I am not so sure it would work well in all places -- regular grid is pretty common for example in New York City.

You could actually see it in use if you used older GPS units -- back in SA time, the GPS was low precision (~100 meters), but still provided mostly accurate heading/velocitry information. Thus, many GPS navigators required mapmatching (see other thread for details). It would always fail on me in the most annoying moments, and would take quite a while to recover.

[jmcdiesel]

Airliners used the same basic system for a long time... and it was, in fact, an airline incident that lead the US gov't to declassify GPS so trans-oceanic flights could be tracked better.

To this day though, when I pilot starts up an airliner, they fire up that old system as a backup, its programed down to the gates at the airports, so they set their departure gate and as soon as they start moving, it tracks, through gyros, the whole flight. There can be a lot of drift in the system though, which is what lead to KAL007 being shot down by the soviets, killing almost 300 and prompting Reagan to declassify GPS to prevent that same kind of thing from happening again...

[garaetjjte]

INS mode wasn't even enabled on KAL007, it flew on HEADING mode.

"According to the ICAO, the autopilot was not operating in the INS mode either because the crew did not switch the autopilot to the INS mode (shortly after Cairn Mountain), or they did select the INS mode, but the computer did not transition from INERTIAL NAVIGATION ARMED to INS mode because the aircraft had already deviated off track by more than the 7.5 nautical miles (13.9 km) tolerance permitted by the inertial navigation computer. Whatever the reason, the autopilot remained in the HEADING mode, and the problem was not detected by the crew."

https://en.m.wikipedia.org/wiki/Korean_Air_Lines_Flight_007

[problems]

A surprising amount of stuff still relies on - or quite recently relied on what they call "inertial navigation systems" (INS for short).

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

I work in aviation and it's still used as a fallback and to compare radio-based equipment with in some cases. Full GPS transitions in aviation are still just getting off the ground in most places.