search

kralf / carmen

The Carnegie Mellon Robot Navigation Toolkit.
GNU General Public License v2.0
11 stars 12 forks source link

[Question] about CARMEN module developing #1

Open kenshin23 opened 10 years ago

kenshin23 commented 10 years ago

Hi, sorry to bother you. I am developing a CARMEN module as my undergraduate thesis and was browsing Github looking for related projects, and yours were the only ones I could find.

I noticed you have developed some modules and addons for CARMEN and was wondering if you could offer some guidance as to where to begin, or how to proceed if I wanted to add a new type of laser sensor to CARMEN.

Any help or advice would be greatly appreciated.

kralf commented 10 years ago

Hi kenshin. Sorry for being so unresponsive, but other work was keeping me busy.

One of the projects you were looking at is a build system port of the original CARMEN project and includes a couple of bug fixes related to 64-bit architecture, newer package dependencies, etc. Afaik, CARMEN trunk development stalled a while ago, see also http://carmen.sourceforge.net/news.html. The other project is a collection of add-ons to the ported CARMEN trunk. Just wanted to make sure you are aware of this.

As for your new "type of laser", your intention seems to be to integrate a Velodyne LiDAR with CARMEN. Technically, there exist two approaches: a) Write a new CARMEN device driver in the style of sick_laser. The problem here would be compatibility of the message interface which is historically designed for 2d range finders. b) Take inspiration on the CARMEN Add-ons project which implements its own message format and simply interfaces libvelodyne (https://github.com/jmaye/libvelodyne).

Have you been trying to build any of those projects?

Cheers, Ralf.

kenshin23 commented 9 years ago

Hi Ralf, thank you for your reply, and sorry for not responding sooner. Work has unfortunately kept me away from my thesis but now I have to seriously get it done. Let me explain further:

The thesis consists on replacing an LMS laser rangefinder with a Kinect, in order to bring costs down and even add 3D mapping capabilities, if that can be achievable (although I will focus on 2D for now, and leave 3D to another student.) From your response, it seems that approach 'a' is definitely the way to go, as I'd have to build a device driver using OpenKinect/libfreenect to basically replace or mimic the SICK driver.

As for building any of those projects, not yet. I had to deal with having my development laptop stolen a few months ago, and I'm just getting back to work on the thesis.

If you have any additional advice, please let me know. It would certainly be much appreciated (and needed!) Thanks,

Carlos