SLAM Algorthms

SLAMfinal <- PDF link to the paper we wrote.

First and foremost about these SLAM algorithms  They, in general, seem a hair platform specific, and need to be calibrated to the application. Which means applying them to an arbitrary dataset can be a bit of a challenge. That, or I suck at picking SLAM algorithms to run experiments on. Either is a valid possibility.

But, REALLY!:

lmap08

this is the result of about  20 passes of the laser sensor in our data. And it is completely and almost utterly useless… And this is after refinements to the robot model. There is a reason I said that this algorithms  DP-SLAM, was useless to us due to accuracy, not to mention it took 3x longer then tinySLAM to run. At least tinySLAM has the decency to only give us walls that it is certain that exist.

lmapb

 

I’ll take this any day over the above dataset. Not to mention it ran in a reasonable amount of time. At least it has the decency of accuracy. Not to mention speed. Both ran with about the same amount of memory consumption, which is about 8/10ths a gig of ram.

 

 

Leave a Reply

Your email address will not be published. Required fields are marked *