| ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange |
Hello,
I am looking for a SLAM that does not require odometry to perform, only laser scans. I have noticed 2 solutions so far: gmapping (+laser_scan_matcher) and hector_slam.
If I am in the right direction, what are the differences between both?
Both gmapping/laser_scan_matcher and hector_slam are viable options that have been demonstrated to work well in different scenarios (see also this Q/A). It might make sense to just try both for your setup and see what works best. Depending on LIDAR type, size/characteristics of the enviroment, available computing resources and other factors you might get better results with one approach or the other.
You can generate fake odometry by using the laser_scan_matcher, so using a combination of gmapping/laser_scan_matcher is method not requiring (real) odometry.
Hi, First thing that comes into my mind is RGBDslam. Because it works for a hand held camera and doesn't require odometry. But gmapping uses odometry. After going through the wiki of the hector_slam it seems that it also provides SLAM without odometry. So it depends on the sensor you are using.
For Kinect sensor you may use either RGBDslam or use hector_slam after converting the kinect pointclouds into laser pointclouds.
For Laser sensor I can think of only the hector_slam.
Hope to hear some answers from people who have already used hector_slam.
Thanks, Karthik
I have used hector_slam for mapping using only the netao laser scanner. It was pretty straight forward. I found gmapping required moving the base much slower to get a good map than with hector.
The tutorial I used for hector http://meetjanez.splet.arnes.si/2015/...
Asked: 2012-06-07 02:20:58 -0600
Seen: 14,029 times
Last updated: Jun 07 '12
ROS Answers is licensed under Creative Commons Attribution 3.0 Content on this site is licensed under a Creative Commons Attribution Share Alike 3.0 license.
