RunningAHEAD - Topic: Another Garmin Rant (page 3)

Twocat

posted: 2/19/2008 at 10:08 PM

Quote from jEfFgObLuE on 2/19/2008 at 3:04 PM:

Just to be clear, you are calling the direction run as vertical, and side-to-side (with respect to the runner) as horizontal, right? In which case what you says makes sense.

Yep.

Live like you are dying not like you are afraid to die.

Drunken Irish Soda Bread and Irish Brown Bread this way --> http://allrecipes.com/cook/4379041/

Kennyd

posted: 2/19/2008 at 11:42 PM
modified: 2/20/2008 at 12:18 AM

Quote from JakeKnight on 2/19/2008 at 2:56 PM:

Out of curiosity, has anybody checked the accuracy with the "every second" data point recording versus the standard "smart recording?"

I have done comparisons between every second recording and smart recording. I have found the smart recording to be more accurate, let me explain. I have compared multiple runs, on multiple courses, all with three comparison. Me and my Garmin set to every second, my friend and his Garmin set to smart recording, and a a control group, satellite mapping courtesy of RA. In all cases my Garmin set to every second recording measured long. This was not at all what I expected. I anticipated more GPS way points to equate to greater accuracy, Just like Jeffgoblue said. I now reason that I left one variable out of the equation, the variations in GPS tracking accuracy. You can include GPS accuracy as one of the data fields on one of the three screens. You will find that the accuracy fluctuates from about 16 to 32 ft. Now imagine you are running in a straight line. As you run the Garmin is plotting one way point every second. As the signal accuracy fluctuates the way points are recorded with varying degrees of accuracy. Even though these changes in accuracy seem minor, say three or four feet, the frequency of way points being created every second results in a cumulative loss of accuracy. So again picture your running in a straight line while the accuracy is constantly changing. the resulting course will look like a zig-zag. sure the zig-zag deviates from from the actual course by only a few feet, but with the zig-zag potentially occurring every second this does add up. On the other hand, smart recording sets a way point every time directional bearing or speed changes. This allows for more way points to be created on turns, giving greater accuracy, and less way points on straight ways, giving greater accuracy.

Gopher Snail

posted: 2/20/2008 at 12:38 AM

Kudos Kennyd, you make a strong analysis. I would like to see Garmin's software or algorithm to see how they calculate everything. Wonder if they do as you hypothetically describe, with more points on corners and less on straightaways?

Trent

Good Bad & The Monkey

posted: 2/20/2008 at 1:08 AM

The difference between standard recording and smart recording when it comes to running is not great. With standard recording, the device plots a point every second. With smart recording, it is every 5-7 seconds. I don't know about you, but I don't cover that much ground in 5-7 seconds unless I am in an all out sprint.

I'm running somewhere tomorrow. It's going to be beautiful. I can't wait.

Poor baby

jEfFgObLuE

I've got a fever...

posted: 2/20/2008 at 11:05 AM

Quote from Trent on 2/20/2008 at 1:08 AM:

The difference between standard recording and smart recording when it comes to running is not great. With standard recording, the device plots a point every second. With smart recording, it is every 5-7 seconds. I don't know about you, but I don't cover that much ground in 5-7 seconds unless I am in an all out sprint.

I agree. I think the only significant benefit of 1-second sampling that I can think of is that perhaps your "instantaneous" pace (you know, the one you never use because it's too unstable) might be more accurate with more frequent sampling. That's my guess, at least, but I haven't tested this. I find that lap pace is quite sufficient once the lap gets past about 1/4 mile (and the positional error becomes less significant).

On your deathbed, you won't wish that you'd spent more time at the office. But you will wish that you'd spent more time running. Because if you had, you wouldn't be on your deathbed.

kcam

posted: 2/20/2008 at 12:16 PM
modified: 2/20/2008 at 12:37 PM

Quote from Trent on 2/20/2008 at 1:08 AM:

The difference between standard recording and smart recording when it comes to running is not great. With standard recording, the device plots a point every second. With smart recording, it is every 5-7 seconds. I don't know about you, but I don't cover that much ground in 5-7 seconds unless I am in an all out sprint.

Has anyone analyzed the points generated from 'smart recording' tracks? I'd guess that their algorithm makes some kind of determination of when you are running in a straight line and then deletes intermediate points and just uses two endpoints for that segment. When it 'thinks' you are changing direction it most likely retains points in memory at the 1sample/second rate. As you can clearly see in my avatar, I use the GPS during races. I find it let's me know when NOT to go faster. I subconsciously tend to pick up the pace to match that of people that are passing me. I'm more likely to let them go when wearing the Garmin.

Gopher Snail

posted: 2/20/2008 at 12:41 PM

Quote from Trent on 2/20/2008 at 1:08 AM:

The difference between standard recording and smart With smart recording, it is every 5-7 seconds. I don't know about you, but I don't cover that much ground in 5-7 seconds unless I am in an all out sprint.

I keep getting stuck on peoples language here. Trent, Im sure you meant to say that you don't cover much ground in 1 second (not 5-7 seconds) unless you are in an all out sprint. You can hopefully cover 40+ yards in 5-7 seconds! (football player sprints for the 40 yard dash are usually 4-5 seconds) Where do you get this 5-7 second number from anyway? Did you pull up some raw GPS data to analyze it?

jEfFgObLuE

I've got a fever...

posted: 2/20/2008 at 12:44 PM

Quote from Gopher Snail on 2/20/2008 at 12:41 PM:

I keep getting stuck on peoples language here. Trent, Im sure you meant to say that you don't cover much ground in 1 second (not 5-7 seconds) unless you are in an all out sprint. You can hopefully cover 40+ yards in 5-7 seconds! (football player sprints for the 40 yard dash are usually 4-5 seconds)

What he said makes sense. He's saying that unless he's sprinting, he's not covering enough ground to necessitate 1-second samples. In other words, he's so slow that 5~7 second sampling is plenty fine for him. Tongue

On your deathbed, you won't wish that you'd spent more time at the office. But you will wish that you'd spent more time running. Because if you had, you wouldn't be on your deathbed.

Ennay

A Saucy Wench

posted: 2/20/2008 at 12:49 PM

I take it "smart recording" and 1 second sampling are features in the x05 series of garmin? I still have a 301. I did find it interesting - my friend and I were out for a run together, I had a 301 she has a 205 and the longer we went the farther apart our miles became. She recorded a longer distance for the run than I did. I guess I have a pessimistic garmin....I prefer that for training!

I have become Death, the destroyer of electronic gadgets

"When I got too tired to run anymore I just pretended I wasnt tired and kept running anyway" - dd, age 7

Neil Gunn

Gandalf the Grey

posted: 2/20/2008 at 1:32 PM

A couple of weeks ago a group of us were out for a trail run of about 18 miles. We had 2x201 and a 205 Garmins in the group. All three never totally agreed at any point on the route (sometimes the 305 was ahead, sometimes the 205) but at the end we were all within 0.1 mile of each other which I think is 'good enough' for most purposes. Just my 2p-worth. Neil UK

Running ... just keep running!

AKTrail

posted: 2/20/2008 at 2:17 PM

Quote from Trent on 2/20/2008 at 1:08 AM:

The difference between standard recording and smart recording when it comes to running is not great. With standard recording, the device plots a point every second. With smart recording, it is every 5-7 seconds. I don't know about you, but I don't cover that much ground in 5-7 seconds unless I am in an all out sprint.

Can't remember if I was using my handheld or FR305, but I've got one trail with about half a dozen sharp switchbacks - about 180 deg change in direction within a few steps. Normal recording was about 3-5 sec and it missed the corners. I eventually hesitated on the corners and increased sampling frequency to fastest I could to get a good map of trail. I know one day I used both my handheld (60csx) as well as FR305 for comparison, but don't remember what settings I used. (Objective was to get a good map - and I had to redo my original one because it missed the corners. Maps, approx total climb, and approx total distance are only reasons I use a gps. Maybe once or twice a year for general pacing.) OTOH, I was running a section line trail on Sat - about 6 mi of straight. I could've used a 5- min setting with no significant loss of data. Wink

"So many people get stuck in the routine of life that their dreams waste away. This is about living the dream." - Cave Dog

milktruck

an amazing likeness

posted: 2/20/2008 at 2:20 PM

Quote from jEfFgObLuE on 2/20/2008 at 12:44 PM:

In other words, he's so slow that 5~7 second sampling is plenty fine for him.

I certainly didn't write the Garmin software, design the hardware, develop the firmware or anything of that nature. I don't even play a Garmin insider on television. (g) But....I have used a Forerunner since May 2006, recording over 2,500 miles on three different Forerunners, varied terrain, some on certified courses. I've been a keen student of the Garmin, I worship the Garmin and its stats. Anyway now that I've established my credentials as being exactly "..just some guy who uses the thing..", I just thought I'd jump in a say that Smart Recording doesn't actually sample location at only the longer interval, it still uses 1 second samples for the current trackpoint. Then it looks back over the past X trackpoints and discards those that it feels aren't adding information. Smart Recording is algorithm about which trackpoints to retain, not a change in sampling rate. I've found that I can alter the number of trackpoints recorded by how carefully I go in a straight line, how fast of an aspect ratio change I create going around corners (ie, how sharp I turn). As an example: I have 70+ samples of the exact same 4.1 mile course recorded by Forerunner, and the number of trackpoints created by the Smart Recording various quite a bit while the course always measures out as 4.15 - 4.17 miles.

_{Acceptable at a dance, invaluable in a shipwreck.}

jEfFgObLuE

I've got a fever...

posted: 2/20/2008 at 2:44 PM
modified: 2/20/2008 at 2:44 PM

Quote from milktruck on 2/20/2008 at 2:20 PM:

Smart Recording doesn't actually sample location at only the longer interval, it still uses 1 second samples for the current trackpoint. Then it looks back over the past X trackpoints and discards those that it feels aren't adding information. Smart Recording is algorithm about which trackpoints to retain, not a change in sampling rate.

Well, that really is Smart. So when you set it to 1-second sampling, you're simply forcing it to keep all of the data rather than chuck the unnecessary stuff. Got it. I'm almost afraid to ask, because it means I will waste inordinate amounts of time goofing around with it, but how do you know how many trackpoints were recorded on a given run?

On your deathbed, you won't wish that you'd spent more time at the office. But you will wish that you'd spent more time running. Because if you had, you wouldn't be on your deathbed.

milktruck

an amazing likeness

posted: 2/20/2008 at 4:19 PM

Quote from jEfFgObLuE on 2/20/2008 at 2:44 PM:

I'm almost afraid to ask, because it means I will waste inordinate amounts of time goofing around with it, but how do you know how many trackpoints were recorded on a given run?

I don't have a Forerunner right in front me to see if the information is available directly from the device, however the motionbased.com data for each uploaded event shows the number of trackpoints under the "equipment" section of the dashboard view. Since you're an engineer and live for facts (g), here's some Forerunner data I just pulled off the last 5 times I've used a particular route: Distance recorded / # Trackpoints Smart Recording retained 4.17 / 352 4.17 / 378 4.17 / 359 4.16 / 339 4.17 / 360

_{Acceptable at a dance, invaluable in a shipwreck.}

Trent

Good Bad & The Monkey

posted: 2/20/2008 at 4:29 PM
modified: 2/20/2008 at 4:33 PM

Quote from Gopher Snail on 2/20/2008 at 12:41 PM:

Where do you get this 5-7 second number from anyway? Did you pull up some raw GPS data to analyze it?

This info is on the Garmin website. Also, on the Garmin wiki there is this:

When the device is set to Smart Recording, course/track points are recorded whenever necessary. The device is constantly monitoring changes in direction, speed, and elevation, and captures a course/track point when it detects change. It could be every second if you're travelling fast on tight turns, or minutes if you're stopped or moving very, very slowly. Unless you're doing short short track activities or super fast downhill runs, Smart Recording is the most accurate and recommended recording setting.

http://wiki.motionbased.com/mb/Smart_Recording This basically says that on a curvy course, smart recording grabs more data points than on a straight course, compensating for the curves. Just as others above have said, it looks every second but only records when it needs to. And correct, when I am jogging or running a marathon, I do not cover much ground in 5 seconds. I do cover more ground in 5 seconds when I am running an all out sprint. Yes. I have looked at the data. Here is one example of the XML, parsed only to show the timestamp: <time>2007-08-06T17:23:14-05:00</time> <time>2007-08-06T17:23:16-05:00</time> <time>2007-08-06T17:23:22-05:00</time> <time>2007-08-06T17:23:26-05:00</time> <time>2007-08-06T17:23:30-05:00</time> <time>2007-08-06T17:23:32-05:00</time> <time>2007-08-06T17:23:38-05:00</time> <time>2007-08-06T17:23:45-05:00</time> <time>2007-08-06T17:23:50-05:00</time> <time>2007-08-06T17:23:56-05:00</time> (time to seconds is in bold underlined text in the first line) You will notice that the time between each point varies from about 2-8 seconds in this sample.

I'm running somewhere tomorrow. It's going to be beautiful. I can't wait.

Poor baby

Another Garmin Rant (Read 1057 times)