💾 Archived View for dioskouroi.xyz › thread › 24994709 captured on 2020-11-07 at 00:49:00. Gemini links have been rewritten to link to archived content

View Raw

More Information

-=-=-=-=-=-=-

A Pedometer in the Real World (2015)

Author: test_user_123

Score: 109

Comments: 12

Date: 2020-11-05 01:02:22

Web Link

________________________________________________________________________________

m0zg wrote at 2020-11-05 04:46:56:

I've done this back in iPhone 3 era, and even made a few thousand dollars on the App Store when you could still make money there.

You could improve this by adding a state machine to the algorithm. Basically, only count _repeat_ events, not single events. When you start walking, you start "provisionally" recording the peaks, which can be done fairly easily, but unless it's N peaks or more (IIRC 4 was a good number) within a certain period of time, you don't count those events and don't display them. Once you enter the steady state, count all events. This eliminates a lot of the "steps" that would be recorded due to user fumbling with the device, or other singular, non-step events. This, plus some tuning of the LPF yields an excellent pedometer.

I don't know how it is now, but back in the day the phone would turn off the accelerometer if you lock the screen, and there were also no background tasks of any kind, other than those pre-installed by Apple. So I also had a music player in the app, as well as a fake "lock screen" that would leave the program running, but looked like a real lock screen. The battery would be gone in about 3 hours - this was before the advent of "energy efficient" cores. Total hack, but people paid for it and left good reviews.

Crazyontap wrote at 2020-11-05 04:56:34:

> _when you could still make money there._

What has changed since? Speaking as someone who has never written apps before but have also wanted to in my spare time, what has changed from before?

Polylactic_acid wrote at 2020-11-05 05:04:36:

People don't pay for little utility apps anymore since Apple/someone else will always end up providing a better version for free.

There is still plenty of money to be made in more powerful apps though. I recently spent $15 on Procreate (A drawing app) and it is currently ranked #1 in its category. The other alternative is subscription pricing which is also very popular with apps these days and often bypasses the Apple tax

m0zg wrote at 2020-11-05 07:15:21:

Lots of competition, and you have to put in a _ton_ of work to get noticed. Most devs lose money on Apple's App Store I'm sure.

jefft255 wrote at 2020-11-05 03:28:34:

In French, we say podomètre so let’s just say I thought a pedometer measured something else than steps.

tzs wrote at 2020-11-05 05:34:06:

I once almost told someone I was going to look for pedophile sites when I was trying to find pedometer nerd sites. Fortunately I realized early enough in the sentence that pedophile is not the right word for "lover of pedometers" and avoided an embarrassing situation.

dilippkumar wrote at 2020-11-05 05:59:01:

I didn’t know there were pedometer nerd sites! As someone who has worked on a widely used pedometer, I’d love to see what is being discussed.

Can you please share a link?

tzs wrote at 2020-11-05 06:31:15:

This was a long time ago, maybe 20 years or so. I don’t remember if I actually found any.

zadkey wrote at 2020-11-05 21:37:24:

In English, as long as you pedometer instead of pedo-meter you shouldn't have any misunderstandings.

737maxtw wrote at 2020-11-05 03:51:40:

Bears?

emmanueloga_ wrote at 2020-11-05 07:18:12:

Nice! I recently got a M5Stack Stick controller, which comes with an accelerometer, but the demo pedometer implementation is super basic compared to this! [1]

What other interesting applications have you seen for an acelerometer?

1:

https://github.com/Moddable-OpenSource/moddable/blob/public/...

dariosalvi78 wrote at 2020-11-05 12:37:18:

interesting work. I don't see the need for detecting the G vector and multiplying acceleration with that vector, you can use the module and you'd probably get a very similar waveform in any case. Modern accelerometers have built-in logic for step counting and sometimes they are surprisingly accurate.

I worked in this space a little, all the algos are open sourced:

https://oxford-step-counter.github.io/