View Poll Results: What should the name be?

64. You may not vote on this poll
  • FitTrack

    23 35.94%
  • OnTrack

    15 23.44%
  • TrackStar

    7 10.94%
  • OpenTrack

    19 29.69%
Page 1 of 8 123456 ... LastLast
Results 1 to 20 of 143
  1.    #1  
    I'm just starting this thread for those of us that are working on the collaborative development of PreTracker. This can be a place where we discuss ideas and share what we've worked on in the code. I'll link this thread from the other one in "Homebrew apps".

    The purpose for this thread is strictly for developers working on the application.
  2.    #2  
    Just uploaded changes to the way pace is calculated so that it can use the last 10 GPS coordinate points instead of using the GPS derived velocity. Ideally, the pace should be an average over the last few seconds, not the instantaneous pace since that will be wildly variable. I wrote it so that it follows a variable for how many GPS points to average over so that can be changed and/or made a user definable property in the future.

    I haven't tested how well it works because this is a really hard app to test in the emulator. I guess you have to load it on the device and go around with it, but I haven't been able to do that yet.

    The way I constructed this was by adding to the array that stores coordinates so that it can store distance from last point as well as cumulative distance. This will also help what I'd like to work on next which is adding the capability to calculate split times and paces for each mile (or definable distance). Having the distance in an array and timestamped will make this much easier.

    Any thoughts?
  3. #3  
    I just glanced at the code changes and it looks good. I might have seen a missing "this." or two but I can load it on soon and give it a test.

    One thing we might need to iron out before adding features is how to store not only the GPS points but also the accuracy, altitude, and other attributes in an object. We can then store it in the sqllite database as well. The basic features need to be solid which is why I liked to see the pace averaging.

    What I might work on is how to emulate the real world GPS data so we don't have to run outside and walk down the street to see if we got the code right.

    I recorded some data the other day when I was messing with the code so I will use that as a sample.

  4. clerick44's Avatar
    35 Posts
    Global Posts
    36 Global Posts
    What is the fastest the GPS can update? Right now it seems to be about one point per second. I unpacked the jar and it seems that the formulas being used to calculate distance are overly complicated for our purposes and, in a minor way, incomplete.
  5. #5  
    Quote Originally Posted by clerick44 View Post
    What is the fastest the GPS can update? Right now it seems to be about one point per second. I unpacked the jar and it seems that the formulas being used to calculate distance are overly complicated for our purposes and, in a minor way, incomplete.
    it updates like once a second. The distance does seem complicated but I think it is needed. With lat lon we are dealing with spherical coordinates. Its not just a pathagorian problem.

  6.    #6  

    Yes, you are definitely going in the right direction. We need to create an object to store an array of all the GPS data we collect at every collection point. Then we can also store that data in the database at some point. That will make your idea of exporting it to a file easier as well. We don't have to construct the file in real-time, we can export it from the object and/or database when it is needed. I might start looking at creating an object in the code tonight. I'll lock the file from SVN if I'm working on it.

  7.    #7  
    Hmmm. Just tried to test the latest build and it's not working. No distance is being calculated right now. I don't have time this morning to look at the code, but I can take a look tonight..
  8. #8  
    I noticed a few comments on this forum about the difficulty in testing in the emulator. I was reading though the SDK materials and there is a page describing how to emulate a moving GPS signal.

    The system wouldn't let me post the link because I don't have enough posts or something. From the webOS developer page, click on webOS tools, and the page I was looking at "luna-send" tool
  9. kmax12's Avatar
    315 Posts
    Global Posts
    316 Global Posts
    This initiative is AWESOME. I hope something can come of Pre Tracker. I orginally made it because I love running, but I stopped because now that XC season has started, I don't have enough ironic.
  10. #10  
    Its glad to see the originator of the code stop by! I hope we can make an app that will work well for us runners and bikers.

    Good Find!! now to figure out how to get it to work on the emulator. I have been using the Eclipse GUI and probably just need to do a little more research on now to do that luna-send command. It would make this alot easier.
  11.    #11  
    I've got a version working pretty well with an averaged pace calculation.

    I also think we should change a few small things about the UI. I don't think we need the "PreTracker" balloon at the top of the application since it doesn't communicate any information other than the title.

    I also think we should change the name. Palm's naming guidelines strongly recommend against including "Pre" in the title of the application. I kind of like "FitTrack" as a suitable title. Any other thoughts?

    Lastly, I don't think we should include the version number in the title. That isn't really standard.

    Any thoughts?
  12.    #12  
    I've got a version ready to go that has the changes I mentioned above. I think it would be nice to release it as a 0.6 version sometime this week to get people testing it. The other advantage to renaming it is that people can try it out and still keep PreTracker 0.5 as a relatively stable version if we screw things up.

    Troy, do you have anything else you want added before a 0.6 release?

    BTW, I've toyed around with the luna-send command and can't quite get it to work. I'll investigate a little more at some point, but for now, I've been walking around my neighborhood to test.
  13. clerick44's Avatar
    35 Posts
    Global Posts
    36 Global Posts
    so far I'm impressed with your ideas and can't wait to try out your changes.

    I've been toying with various formula ideas to see if there is a more efficient way to do it. I'm working off the idea that no formula is going to be truly accurate due to the Earth not being a true sphere. So, my goal is to come up with the most efficient way to calculate distance and not give up too much in terms of the inherent margin of error.

    I'm trying to get out of polar forms and into vector forms. The only measurement that I'm having difficulty with is longitude since a simple conversion from seconds to feet doesn't work like it does with latitude.

    I'm happy to keep tinkering with it if you're interested.

    - Jesse
  14.    #14  
    I don't think tweaking the distance calculation is worth the effort at this point. It's currently using the Haversine formula according to this webpage: Calculate distance and bearing between two Latitude/Longitude points using Haversine formula in JavaScript.

    I don't know anything about it, but from what I've seen online, it looks like determining distance between two GPS coordinates is a well thought out problem and the current solutions are as good as we can hope to come by.
  15.    #15  
    What I'm struggling with now is a good way to handle the variability of the GPS coordinates that come from the Pre. Even when I'm averaging the pace calculation over 30 seconds or so, the result still varies by up to a minute with every new calculation. I'd like to see a more steady pace, but perhaps that's not possible.
  16. clerick44's Avatar
    35 Posts
    Global Posts
    36 Global Posts
    Haversine is great if you want to know the distance from LA to New york "as the crow flies." However it can create large margins of error in higher altitude or hilly terrain since it doesn't take base altitude and altitude variations into account. For example, I have a bike route that I have taken the Pretracker on a few dozen times. Routinely, the program comes up 1/3 of a mile short of my bike computer (verified by car). I found that the variation is almost perfectly accounted for by adding the distance traveled in the vertical over the hilly terrain. Also, on more level terrain it is almost completely gone. A third of a mile over a 10 mile ride is a lot and, If we can find a way to eliminate it, I think we should.

    I'm not trying to reinvent the wheel, just want to make sure that we have the right wheel in the first place.

    What pace are you running where you are seeing a full minute? Depending on speed it may not be much to worry about. I'll take a look tomorrow and see if i can lend any insight.
  17.    #17  
    wow. I didn't know there were such problems with the distance calculation with hilly terrain. If you can find some way to address that, great! I'm definitely out of my league there. Keep us updated.
  18. Ikyo's Avatar
    290 Posts
    Global Posts
    292 Global Posts
    To calculate the distance with altitude, take a look at this thread:

    Google Answers: Using latitude, longitude, and altitude to determine distance
  19. #19  
    Wow, Lots of activity.

    I think we should release a version soon. For me the 0.5 did not register distance well and my current build does register well. It would be good to get a functional version out there and then focus on objectifying the incoming data from the GPS.

    We then can do some cleanup functions to smooth out the data to the display. I think we can come up with something that will work well because I don't know if averaging for 30 seconds is the greatest plan. What if someone wants to speed up the pace? It wont reflect the current pace unless that pace is kept for 30 seconds.

    Implementing the altitude adjustment will work well after we properly store the data as well.

    I agree we need a new name based on Palm's guidelines. Lets get a bunch of ideas and put it up for a vote with developers and users.

    Lets take out the balloon at the top because it does not give us anything useful.

    I have access to a Garmin GPS watch and I will be testing the pre results with a professional product to see how they compare. I think that will give us some hints on how to clean an interpret the results.

  20. kmax12's Avatar
    315 Posts
    Global Posts
    316 Global Posts
    how are you planning to release this? As of now only I have access to edit PreTracker's entry in the homebrew catalog. Seems like there are three possibilities:

    1. Send me the ipk and I can post it
    2. contact mods to give one of you guys access
    3. delete the pretracker entry and resumbit under the new name

    I have no problem with 1 or 2, but I think 3 is probably the best option going forward.
Page 1 of 8 123456 ... LastLast

Posting Permissions