FR955 (FW 15.19) Auto Pause works poorly

I stopped using auto pause for runs when I realized that it already breaks the activity into run/walk/idle in Connect and it shows you elapsed pace and moving pace. 

I'm using instead custom auto pause with lowest speed setting which works great for me (won't obviously work with faster activities but ok for walking/hiking/running).

I have observed the same.

When you slow down/walk "without stopping", it often pauses, and it then waits until you start running at around 8:00/km to resume again. Toggling Auto Pause between "when stopped" and "custom 36:00/km" makes no difference. Strangely, multiband GPS does not seem help as well.

Here is an activity where Auto Pause easily cut off at least 2km from the total distance. Officially, it was listed as a 22km. https://connect.garmin.com/modern/activity/11834788190

Auto Pause also always kicks in when pressing start, whether you are standing still or in motion. Surely there should be a 5 second delay after you press the start button before the algorithm kicks in.

That's the way GPS works. If you stand still, your position jumps and that causes these starts and stops. It can be filtered if you move and it cannot if you stand. 

For biking - speed sensor helps a lot (it is annoying to get auto pause when climbing steep hill in the woods on mtb) - with it auto pause works perfect. I don't know if for pod will help (I think it will if you set it up "pace from footpod always", but then your pace might be slightly incorrect at certain pace)

They should've doesn't mean it is feasible. That is your expectations based on nothing. If you believe everything shall be fixed I have a long list of surprise for you (starting from GPS being totally unprecise data) - unfortunately. If you use a technology better to understand the limitation, it helps to avoid incorrect expectations

They should've turned it off for walking as they did for swimming HR some time ago - that's the right proposal. There might be some reason why they didn't. 

There's two pieces of advice on this thread how to maybe improve things. Youre open to use any of these. 

Or they could fix it, where stopped means when stopped. 

Using your logic, we would have had no improvements to both GPS and wrist based HR tracking, since it's imprecise technology. 

In 2023, with things like ChatGPT, when someone posts a software bug, we cannot just tell them to suck it up because it's imprecise technology.

"That's the way GPS works" it's not something you can accept. GPS it's a sensor, it's streaming raw informations, software is needed to translate this informations in something useful.

It's definetely something fixable on software side. And it's also easy to fix.

Done a lot of hikings and comparing with OruxMaps (android app), Polar watch, all are doing a better job at filtering false movement.

You're just stopped eating panini at lunch and FR955 keep increasing your walking distance, and yet you're just stopped eating... panini and watching your FR955 like this: -_-

Yes, considering how good Garmin is at creating GPS tracks compared to the competition, this should be an easy fix for them. It's just that no one has got down to fixing it.

Gps is a technology. 

It is hardware (both on your wrist and up there on space), software and physical principals all these is built on. And interferences as well (rain, coulds, trees, buildings etc). And when you improve something you do that at cost of something else.

Yep, it might be the case that Garmin doesn't invest any attention to this issue. But it may turn out it is not that easy as it sounds it as you expect. I remember every update in GPS there were loads of complaints from two sides - "hey, the current pace is too laggy and slowed down and not accurate at the moment" or (obviously from a different crowd) "pace is jumping all over the place and not stable". And they struggled thought this issue for at least 3-4 years (accuracy Vs stability ie how much you apply smooth filter). So I believe it is the latter thing - not that easy. 

I also remember first time I went into the wood on my MTB with an edge 520. I have auto pause for bikes, no pause for runs. Every steep climb it paused, around 6-8km/h. 6km/h is 10 min/km - decent walking speed. And GPS device struggled at that pace. The progress might be there but but enough. I solved an issue by simply putting speed sensor - issue solved. And the fact that they now sell speed sensor in MTB buddle (no hr, no cadence) shows there's no easy fix, I assume. 

I cannot comment polar or your phone (although latter has bigger antenna obviously), but it looks there's some trade off that Garmin doesn't want to sacrifice.

They made huge step forward in track accuracy (just check how accurate track now is with no wobbling, lost/jumped signal near hight buildings etc), but it looks that this one is not easy. Software is not magic, it is just math.