BUG Community

Welcome! Log In

Forums BUG SDK Using integration build

Subscribe to Using integration build  15 posts, 4 voices

Log in to reply to this topic
 
May 21, 2008 12:26am
Medium Bug Labs team finsprings 268 posts

I switched to the latest, as of tonight, SDK: 1.0.0.1621. I can now get Angel's BUGlevel to compile but it fails when I deploy it to a real bug with:

[Wed May 21 04:12:04 GMT 2008] [ERROR] Failed to start application bundle: /opt/concierge/apps/BUGLevel.jar
Exception: Unsatisfied import com.buglabs.bug.accelerometer.pub for bundle Bundle [30]: file:/opt/concierge/apps/BUGLevel.jar


Is there an update I can do to my bug to make it compatible with 1.0.0.1621? I looked for a suitable jar in my plugins dir, and in SVN, but didn't find one obviously suitable.
May 21, 2008 12:30am
Img_missing_medium afritz 24 posts

I’m actually working on the same problem and am at the same state. It looks like the set of libs on the bug root FS are the same as the production release. They need to be upgraded to the integration level. Of course, the rootfs image that is available has production libs, not the experimental libs…

I’m about to try bulk replacing the /opt/concierge directory on the bug with the lib set eclipse is using now (after backing it up).

May 21, 2008 12:38am
Img_missing_medium afritz 24 posts

Well, copying the 1621 kernel jars didn’t have much effect that I can tell… Guess I’ll keep trying… tomorrow.

May 21, 2008 8:29am
Medium Bug Labs team kschultz 107 posts

I think that you will need to update the rootFS on the MMC card. I believe the guys are working on getting the info on how to build a new rootFS from source somewhere public, I don’t have an ETA though. They are switching the build system from LTIB to OpenEmbedded so that might have something to do with it, what is the use of posting instructions if they will obsolete in 5 days?

May 21, 2008 8:36am
Img_missing_medium afritz 24 posts

Fair enough… I can’t get my MMC reader on my laptop to work anyway so it will be tomorrow or the next day until I can replace the root image whole sale (i.e. NOT just replace certain files via scp).

It seems that there is also a need for instructions on how to upgrade JUST the java libs that are being used. In this case (and most) I would guess that would be enough to solve the problem without replacing everything. Does that exist?

May 21, 2008 9:09am
Medium Bug Labs team kschultz 107 posts

I think that to get it to run on the BUG you will need the Java libs, the JNI code below that, and the kernel with the new driver. The Java libs alone would be listening and nothing would pushing data up to it.

I’m not sure if it is IN the kernel or a kernel module, and I’m not sure if there is a layer between the JNI and the kernel or not. I never really got involved with the kernel so I don’t know.

To be honest I don’t know if it is worth the trouble, from what I can tell I think we are talking a few days away from the next release of the rootFS. I would assume that the accelerometer stuff is included in that since the BUGlevel has been out for a while and it has been working on real BUGs in house for a while.

May 21, 2008 9:18am
Img_missing_medium afritz 24 posts

That would be awesome. The first real app I wanted this thing for is basic race car telemetry. I’ve got a few week before my next track day, but it would be great to see how many G’s my 2500 can pull (not many).

Anyway, I look forward to it.

May 21, 2008 9:57am
Medium Bug Labs team kschultz 107 posts

What kind of car? I am looking to write a similar app for my Dad, he races vintage cars in the VSCCA, a 1961 MGA with a race motor/suspension.

I was thinking lap times and plotting your line over a google earth model of the track, but G force is a good idea too.

Hopefully sometime soon I can get out on a track too, so far I have been relegated to country roads, which are rapidly disappearing in my area.

May 21, 2008 10:02am
Img_missing_medium afritz 24 posts

87 Supra. If I wanted to race door to door it would race in SCCA SP (super production – open class with GT-1 safety rules). I’m more of a track day/race against the clock guy. I just don’t like the idea of some yahoo getting excited and taking out my car for a $2 trophy.

The problem with GPS in my experience is update times. The fact is 2 second or even .5 second updates just aren’t fast enough for track use. However using inertial tracking with GPS backing should be pretty good for in car segment timing as well as checking driver consistency throughout the line.

Once I get the GPS working (mine has issues) and the accelerometer working (come on new root fs image) I’ll get busy working on the app.

May 21, 2008 1:06pm
Medium Bug Labs team kschultz 107 posts

Yea, my dad stated out doing time trials (first on the road, then on a skid pad, then on the track) and ultimately got comfortable on the track. The VSCCA is much easier on the cars than the SCCA since all the cars are pretty irreplaceable, truthfully the newest car allowed in the VSCCA is a 1959, but since the MGA was 59-61, we squeezed in with a 1961.

I haven’t seen too many guys get hit, although I’ve seen plenty spin out, and I saw a 1920s Bugatti (price 300,000+) break an axle and go flying into a barrier, that was rough. I’ve seen SCCA on the other hand, and it is definitely more competitive.

It is true about the GPS polling on most consumer devices is insufficient for the task, but I think if we are writing the code expressly for the purpose it can work. Combining the GPS and accelerometer data would yield a much better result than either on its own, I believe most people go with a Kalman filter based approach, I’m looking into it for robotics at the moment but it would apply equally in both cases. I’m having trouble downloading the paper I read about it recently, but when the link works again I’ll post it.

May 21, 2008 4:08pm
Img_missing_medium afritz 24 posts

When you get the paper, I’d love to read it. I’m probably going to start by just writing a bunch of logic to do the motion tracking and test then walking, driving, etc. It should be a blast.

May 21, 2008 6:19pm
Medium Bug Labs team kschultz 107 posts

http://www.igvc.org/design/reports/dr166.pdf

This paper is written by the winner of the intelligent ground vehicle competition, a collegiate autonomous robot navigation challenge. The team comes from Virginia Tech, and wrote a report about it. Their report is not the most detailed, but on pages 9-11 you will see what I’m talking about. They do not talk about actual implementations, but my school’s team uses a similar approach with a Kalman filter and I was going to use that algorithm.

Jun 17, 2008 10:55pm
Img_missing_medium afritz 24 posts

Just curious if there has been any movement on the new version of the root fs?

Jun 18, 2008 10:23am
Medium Bug Labs team agordon 74 posts

afritz, the new rootfs has been released. You can find a link to it at http://bugcommunity.com/downloads. There are also upgrade instructions here: http://bugcommunity.com/wiki/index.php/Flash_Your_MMC_Card.
Jun 18, 2008 10:31am
Img_missing_medium afritz 24 posts

Will you be releasing anything on using open embed to build your own root image?

Does this image include accelerometer support?

Just curious!

Thanks!

Log in to reply to this topic
Forums BUG SDK Using integration build

Powered by Community Engine

Top
Login
Close
<