This discussion has been locked.

You can no longer post new replies to this discussion. If you have a question you can start a new discussion

Accelerometer to keep track of direction/turns?

Maybe this has already been considered but what if there were a device that was/could be attached, for instance, to the HRM strap, and that kept track of (upper) body direction changes, in order to aid the GPS device to calculate distance (and speed)? My thinking is that this device would give constant feedback to the GPS device. We all know that GPS points have a spread, but if the GPS device were given concrete info that the runner (etc) did (or did not) actually change direction, and when in time, then (my thinking) this would help the GPS to know better when to discard or keep each GPS point. It most probably has to be tuned and optimized, perhaps even like to footpod, individually.

What do you think?

0 DW152 over 11 years ago

Combining inertial-navigation systems (e.g., accelerometer) with another reference method (in this case, GPS) is a well-known technique. Usually it's implemented the reverse that you describe, that is, the GPS is used to initialize and update the inertial system, and to correct for drift.

But I don't think that improving the position accuracy is what you are suggesting. I think you are suggesting using an accelerometer to validate each GPS trackpoint, and record a GPS trackpoint that deviates from the previous trackpoint only if it correlates to an acceleration. Is that what you're suggesting?

If that's the case, then IMHO it's solving a problem that doesn't exist. For the most part, the GPS provides enough accuracy for our fitness activities, at least for the overall totals. Every now and then the device will record some wildly rogue point, but I've always thought that those could be eliminated simply by requiring two successive track points to be within, say, 1km of each other.

0 alaala over 11 years ago

Combining inertial-navigation systems (e.g., accelerometer) with another reference method (in this case, GPS) is a well-known technique. Usually it's implemented the reverse that you describe, that is, the GPS is used to initialize and update the inertial system, and to correct for drift.

But I don't think that improving the position accuracy is what you are suggesting. I think you are suggesting using an accelerometer to validate each GPS trackpoint, and record a GPS trackpoint that deviates from the previous trackpoint only if it correlates to an acceleration. Is that what you're suggesting?

Yes, but actually only acceleration due to change of direction, not in speed.

If that's the case, then IMHO it's solving a problem that doesn't exist. For the most part, the GPS provides enough accuracy for our fitness activities, at least for the overall totals. Every now and then the device will record some wildly rogue point, but I've always thought that those could be eliminated simply by requiring two successive track points to be within, say, 1km of each other.

I think that it is evident that the problem exists. Most dramatically between tall buildings and under heavy canopy. Especially when combined with frequent turning. Also, it would help, I believe, to better the calculations when running on a track, where the big problem is that it seems the GPS devices don't "believe" in the 100 m long constant acceleration to the left twice each lap.