Technical details of the VSLAM package

asked 2012-03-19 22:46:28 -0500

Divya
46 ●6 ●7 ●9

updated 2014-01-28 17:11:41 -0500

ngrennan
1 ●1 ●1

Hi, Can anyone help me with the technical details of the VSLAM package? Like What feature matching techniques were used (in the VSLAM_SYSTEM package available in ROS), what was the main uncertainty algo used - Kalman Filter? EKF? or Monte-Carlo Implementation?

Or is there any paper that I could refer to that would help me figure these out. I want to know what methods were used to implement the VSLAM package available in ROS.

I'm trying to analyze the performance of VSLAM for my thesis and I really need to find out what algorithms have been used in order for me to get started.

Thank you.

answered 2012-03-20 04:47:17 -0500

Mac

4119 ●27 ●62 ●83 http://www.cs.duke.edu/~mac

The VSLAM package in ROS (http://www.ros.org/wiki/vslam), is (a) unsupported (b) out of date and (c) very brittle, as most VSLAM systems are.

If you want to know how it works, read the code. If you're looking for a highly general VSLAM that will work on your system, for your data, we recommend you look elsewhere.

edit flag offensive delete link

Comments

Um.. I know its slightly out-dated. Unfortunately, my project is based on this particular version of VLSAM implementation and I've already done my testing using this particular set of packages! That's precisely why I need the technical details of this particular implementation! :| ..

Divya ( 2012-03-20 16:44:09 -0500 )edit

I'm also using another Visual odometry method called libviso which works with OpenCV.. I am basically trying to compare the performance of that with VLSAM.. Sorry if I am making it all the more confusing!

Divya ( 2012-03-20 16:45:25 -0500 )edit

Then you'll need to read the code; since it's not supported anymore, that's the best we can do.