6 ms·

Wow. I imagine there was quite a bit of drift.

eckza 4y ago
Probably not, as it was only optioned to the Accord and Vigor, which are FWD vehicles. http://instantrimshot.com/ http://instantrimshot.com/
- leokeba 4y ago
  Even with data from all four wheels the system as it is described cannot prevent drift altogether. Wheel diameter is variable depending on many factors, tires can lose grip on the road, and also the earth is not perfectly flat. There would be other sources of inaccuracies, but basically in any real-world situation you need some kind of absolute position source to get drift-free location data. The trick is that it can be very noisy and low bandwidth if you integrate it with other sources of data like and IMU.
  - kube-system 4y ago
    The comment you replied to was an attempt at a joke
    - gtirloni 4y ago
      I know explaining a joke ruins it but I failed to understand it so far.
      - kube-system 4y ago
        They are intentionally misinterpreting "drift" as the driving technique (often associated with RWD vehicles) rather than the intended meaning.
      - aidenn0 4y ago
        "Drift" is another term for understeer which is when the rear-end of the car loses traction in a turn. It tends to be easier to induce (accidentally or on purpose) in a rear-wheel-drive car. I should note that my ~2005 MY Ford Focus would absolutely understeer if you enter a turn too fast without applying power.
        eckza 4y ago
        Are you lifting right before you start to turn in? That'll do it, because lifting off the throttle abruptly will shift weight to the front, which will unweight the rear; and sometimes that little weight transfer is all it takes to break the back end out a bit. (Used to do this regularly in a Taurus that I had the pleasure of driving for a few months.)
        kube-system 4y ago
        Yeah, the first gen focus is pretty easy to induce lift-off oversteer in. The rear end weighs nothing and the suspension sits strangely high.
        eckza 4y ago
        > Lift-off oversteer was on the tip of my tongue. Thank you!
        kube-system 4y ago
        Oversteer is when the rear end loses traction. Understeer is when the front end loses traction.
        aidenn0 4y ago
        holy shit I wrote it exactly backwards. I didn't have my coffee this morning.
    - forgotmypw17 4y ago
      This thread just keeps drifting further and further away from the original topic.
  - Xylakant 4y ago
    The device is described as using a gyroscope to measure movement, so it would be independent of wheel diameter and grip. Gyroscopes still drift, so occasional calibration against an absolute position source is still required.
    - kube-system 4y ago
      The device uses a gyroscope to determine direction. But a gyroscope is unable to measure distance travelled. So this device also attaches to the transmission to determine distance. > A special servo gear was also attached to the transmission housing to feed information to the Gyro-Cator to help maintain position, map speed and distance traveled.
  - amelius 4y ago
    You could compensate for drift if you use a map as a constraint on the position.
- Gordonjcp 4y ago
  Well, I thought it was funny.
- kojeovo 4y ago
  Handbrake!!!
  - aidenn0 4y ago
    Don't change the subject; we are talking about cars, not video encoding!
- dfox 4y ago
  You can drift with anything that has wheels, or even, tracks, somebody did that. And drifting with FWD german family car is one of my favorite activities.
elil17 4y ago
Yep - 3.5% of distance traveled (i.e. drive 1 km and you'd have 35 meters of drift)
- pantulis 4y ago
  That's not that bad, specially for solo driving where you cannot use a regular road map (remember those?).
  - dfox 4y ago
    Let me guess: you didn't ever solo drive for work. You can certainly use a paper map while driving. When it gets too distracting to safely drive then you are lost and should stop and use the paper map more thoroughly to find out where you are. Source: I intentionally tried to be a parcel delivery driver without a GPS-based navigation and only with regular road map and printed out rough plan of the route. Not to mention that for most of my long driving trips I used to simply print out the itinerary from RAC RoutePlanner and ignore the navigation (it is a shame that this service apparently does not exist anymore).
- hammock 4y ago
  Not bad. When you consider a city block is about 100m, you could drive 3km before being off by a street so as to miss your turn. On the highway it’s even better- if exits are 5km apart, you can drive 142km before being so far off you might miss your exit
  - dfox 4y ago
    That is surprisingly close to what I used as a good-enough error in fallback route estimator in our TSP/VRP solver. I came up with that essentially by curve-fitting great-circle WGS84 distances and really traveled distances between my favorite pubs and it seems to work (bear in mind that it is the last fallback and there is A* on map data and ML-ish thing before that)
    - hammock 4y ago
krisoft 4y ago
Yes. I found this paper which measured the errors of the system: https://apps.dtic.mil/sti/pdfs/ADA138283.pdf https://apps.dtic.mil/sti/pdfs/ADA138283.pdf They performed two kind of measurements. On a bench they measured the drift of the helium-flow gyroscope in itself, and then they took the whole system to drive test loops and measured the errors in eastings and northings. Quite a thorough job in fact, but I guess that is what you get if you ask a rocket scientist to evaluate a new-fangled vehicle navigation tool.
- analognoise 4y ago
  ...this place is so weird, I love it so much. Like if there was going to be one place where a random comment links to an actual Defense Technical Information Center technical report evaluating the system, it's here.
  - rvnx 4y ago
    Then you may find even more happiness on Reddit where you can find specialists about everything
    - analognoise 4y ago
      I'll bet HNers have a collection of fascinating subreddits. HN is better - it's broader and there's usually amazing takes in the comments (amazing good, not YouTube "amazingly WTF" style comments)
- machinehum 4y ago
  Thanks for the link to the paper, this was an insane amount of work.
olyjohn 4y ago
It doesn't say in the article, but I'm pretty sure you drop in a slide with the map on it, and then you move the pointer to your current location. So you reset the drift every time you use it.
zh3 4y ago
Depends on the qualities of the gyros. My step-dad flew 707's between NY and London, and INS on those used the same kind of system (obviously more pro version) - one of the reasons it was no longer necessary to carry a navigator in the cockpit.
- aidenn0 4y ago
  IIRC some pre-GPS cold-war era spy-planes used an INS that was corrected for error with an analog-computer celestial navigation system.

Probably not, as it was only optioned to the Accord and Vigor, which are FWD vehicles. http://instantrimshot.com/ http://instantrimshot.com/

leokeba 4y ago
Even with data from all four wheels the system as it is described cannot prevent drift altogether. Wheel diameter is variable depending on many factors, tires can lose grip on the road, and also the earth is not perfectly flat. There would be other sources of inaccuracies, but basically in any real-world situation you need some kind of absolute position source to get drift-free location data. The trick is that it can be very noisy and low bandwidth if you integrate it with other sources of data like and IMU.
- kube-system 4y ago
  The comment you replied to was an attempt at a joke
  - gtirloni 4y ago
    I know explaining a joke ruins it but I failed to understand it so far.
    - kube-system 4y ago
      They are intentionally misinterpreting "drift" as the driving technique (often associated with RWD vehicles) rather than the intended meaning.
    - aidenn0 4y ago
      "Drift" is another term for understeer which is when the rear-end of the car loses traction in a turn. It tends to be easier to induce (accidentally or on purpose) in a rear-wheel-drive car. I should note that my ~2005 MY Ford Focus would absolutely understeer if you enter a turn too fast without applying power.
      - eckza 4y ago
        Are you lifting right before you start to turn in? That'll do it, because lifting off the throttle abruptly will shift weight to the front, which will unweight the rear; and sometimes that little weight transfer is all it takes to break the back end out a bit. (Used to do this regularly in a Taurus that I had the pleasure of driving for a few months.)
        kube-system 4y ago
        Yeah, the first gen focus is pretty easy to induce lift-off oversteer in. The rear end weighs nothing and the suspension sits strangely high.
        eckza 4y ago
        > Lift-off oversteer was on the tip of my tongue. Thank you!
      - kube-system 4y ago
        Oversteer is when the rear end loses traction. Understeer is when the front end loses traction.
        aidenn0 4y ago
        holy shit I wrote it exactly backwards. I didn't have my coffee this morning.
  - forgotmypw17 4y ago
    This thread just keeps drifting further and further away from the original topic.
- Xylakant 4y ago
  The device is described as using a gyroscope to measure movement, so it would be independent of wheel diameter and grip. Gyroscopes still drift, so occasional calibration against an absolute position source is still required.
  - kube-system 4y ago
    The device uses a gyroscope to determine direction. But a gyroscope is unable to measure distance travelled. So this device also attaches to the transmission to determine distance. > A special servo gear was also attached to the transmission housing to feed information to the Gyro-Cator to help maintain position, map speed and distance traveled.
- amelius 4y ago
  You could compensate for drift if you use a map as a constraint on the position.
Gordonjcp 4y ago
Well, I thought it was funny.
kojeovo 4y ago
Handbrake!!!
- aidenn0 4y ago
  Don't change the subject; we are talking about cars, not video encoding!
dfox 4y ago
You can drift with anything that has wheels, or even, tracks, somebody did that. And drifting with FWD german family car is one of my favorite activities.

elil17 4y ago

Yep - 3.5% of distance traveled (i.e. drive 1 km and you'd have 35 meters of drift)

pantulis 4y ago
That's not that bad, specially for solo driving where you cannot use a regular road map (remember those?).
- dfox 4y ago
  Let me guess: you didn't ever solo drive for work. You can certainly use a paper map while driving. When it gets too distracting to safely drive then you are lost and should stop and use the paper map more thoroughly to find out where you are. Source: I intentionally tried to be a parcel delivery driver without a GPS-based navigation and only with regular road map and printed out rough plan of the route. Not to mention that for most of my long driving trips I used to simply print out the itinerary from RAC RoutePlanner and ignore the navigation (it is a shame that this service apparently does not exist anymore).
hammock 4y ago
Not bad. When you consider a city block is about 100m, you could drive 3km before being off by a street so as to miss your turn. On the highway it’s even better- if exits are 5km apart, you can drive 142km before being so far off you might miss your exit
- dfox 4y ago
  That is surprisingly close to what I used as a good-enough error in fallback route estimator in our TSP/VRP solver. I came up with that essentially by curve-fitting great-circle WGS84 distances and really traveled distances between my favorite pubs and it seems to work (bear in mind that it is the last fallback and there is A* on map data and ML-ish thing before that)
  - hammock 4y ago

krisoft 4y ago

Yes. I found this paper which measured the errors of the system: https://apps.dtic.mil/sti/pdfs/ADA138283.pdf https://apps.dtic.mil/sti/pdfs/ADA138283.pdf They performed two kind of measurements. On a bench they measured the drift of the helium-flow gyroscope in itself, and then they took the whole system to drive test loops and measured the errors in eastings and northings. Quite a thorough job in fact, but I guess that is what you get if you ask a rocket scientist to evaluate a new-fangled vehicle navigation tool.

analognoise 4y ago
...this place is so weird, I love it so much. Like if there was going to be one place where a random comment links to an actual Defense Technical Information Center technical report evaluating the system, it's here.
- rvnx 4y ago
  Then you may find even more happiness on Reddit where you can find specialists about everything
  - analognoise 4y ago
    I'll bet HNers have a collection of fascinating subreddits. HN is better - it's broader and there's usually amazing takes in the comments (amazing good, not YouTube "amazingly WTF" style comments)
machinehum 4y ago
Thanks for the link to the paper, this was an insane amount of work.

olyjohn 4y ago

It doesn't say in the article, but I'm pretty sure you drop in a slide with the map on it, and then you move the pointer to your current location. So you reset the drift every time you use it.

zh3 4y ago

Depends on the qualities of the gyros. My step-dad flew 707's between NY and London, and INS on those used the same kind of system (obviously more pro version) - one of the reasons it was no longer necessary to carry a navigator in the cockpit.

aidenn0 4y ago
IIRC some pre-GPS cold-war era spy-planes used an INS that was corrected for error with an analog-computer celestial navigation system.