ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question

Navigation + 3D SLAM + Kinect

asked 2013-10-09 08:16:29 -0500

MartinW gravatar image

updated 2016-10-24 09:09:38 -0500

ngrennan gravatar image

Hello all,

I am currently working towards developing a navigation system on a non-pr2 robot. So far my colleagues and I can control the robot using keyboard teleop in ROS and obtain various sensor readings like odometry from the encoders, rgb+d data from the kinect, and some sonar, IR readings from the base of the platform. Now we are working towards developing the navigation, path planning, map building portion of the robot system. I have found the navigation tutorials here at, but we would like to implement some 3D-SLAM using the rgbdslam or 2D-SLAM using the hector_slam.

Therefore, we were wondering which of these is the best to implement with the hardware we currently have? Ideally, I think we would like to use both the 3D SLAM and 2D SLAM (or is this wildly computationally expensive/redundant?).

Additionally, what is the difference between hector_slam and the gmapping slam?

Thank you and Kind Regards, Martin

edit retag flag offensive close merge delete


as far as I know, hector_mapping needs no odometery info as input, and it publishes odom. while gmapping needs odom as input, you should get that from sth like a wheel encoder. but kinect seems not able to satisfy hector mapping because of short sight(laser scan sensor might be ok)

vdonkey gravatar image vdonkey  ( 2014-05-19 19:15:24 -0500 )edit

2 Answers

Sort by ยป oldest newest most voted

answered 2013-10-21 06:44:20 -0500

Ideally, I think we would like to use both the 3D SLAM and 2D SLAM (or is this wildly computationally expensive/redundant?).

If your robot is restricted to motion in the plane you can get the pose of the kinect from 2D SLAM and the extrinsic calibration of the Robot and Kinect. This allows to correctly backproject the point clouds. This approach is computationally cheaper and makes it easy to align the origins of 2D and 3D map than mapping 2D and 3D seperately.

If your floor is not level, consider using 3D SLAM and downprojection of the map to 2D.

edit flag offensive delete link more

answered 2015-03-29 19:11:26 -0500

I m also currently working on slam3D with kinect.... But i have a hetrogenous team of robots.... Namely a quad And a ground rover....and i m lost.... Like really lost...... All of u kind people are requested to please kik me @ redhotdragon2000

edit flag offensive delete link more

Question Tools



Asked: 2013-10-09 08:16:29 -0500

Seen: 4,170 times

Last updated: Oct 21 '13