Jim Morris wrote: > I have checked in my very recent attempt to write a GPS based pedometer for > Qtmoko. > It is at http://github.com/wolfmanjm/qtpedometer >
There is a serious accuracy problem with determining the distance traveled during a trip. I did some research, and there does not seem to be a consensus on how to do it, and the GPS device manufacturers obviously have proprietary solutions they do not disclose or discuss. All the open source programs I looked at do it the simple but very inaccurate way I first tried. They basically calculate the distance between successive fixes (segments) and accumulate that as the distance. This has about a 20%-50% error (estimated). The next attempt I discovered on the web (and how I do it now) is non-intuitive, but basically you take the current speed, and multiply it by the delta time since the last update. This is much more accurate, but still accumulates errors even if you are standing still it accumulates a small error. It works really well when driving though, but for walking (and I suspect biking), it is still inaccurate. Next I added a speed threshold, which you must go over before the distance will accumulate, for walking I set that to .4 miles per hour or 0.1788159993648455703 meters per second. This works well if you are standing still, as it does not accumulate any distance, but it increased the error significantly on a driving trip, when I compared the trip distance of my cars GPS with my version. I haven't tried a walking test yet. So if anyone has any ideas how to make this more accurate I would welcome suggestions. Thanks Jim -- Jim Morris, http://blog.wolfman.com _______________________________________________ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
