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

RTAB-MAP with xtion, rplidar, turtelbot (how to utilize lidar internally)

asked 2017-11-29 03:00:42 -0500

JunJun gravatar image

updated 2017-11-29 03:03:32 -0500


I have question about building 3D map with RTAB-MAP and lidar.

As far as I know, RTAB-MAP is SLAM based on image-features and performs loop closure using features.

3D mapping on RTAB-MAP with turtlebot, xtion is good.

But, my robot has additionally lidar and I want to utilize lidar.

I read the questions about build 3d map.

I think that my situation is similar below.

My questions are below.

  1. How do you use additional lidar internally in RTAB-MAP?

    1-1.For improving quality of map, how to use lidar in RTAB-MAP?

  2. I think this system for utilizing lidar and improving quality of map.

    I wonder my method is good or not.

    I think below system is using all sensors, so it can make good 3D map.

This link is system image I thought. link text

Thank you for reading questions.

edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted

answered 2017-11-29 08:57:02 -0500

matlabbe gravatar image

Based on this example, here is how to add lidar to rtabmap:

<node name="rtabmap" pkg="rtabmap_ros" type="rtabmap" output="screen" args="--delete_db_on_start">
   <remap from="scan" to="/base_scan"/>
   <param name="subscribe_scan" type="bool" value="true"/>
   <param name="RGBD/NeighborLinkRefining" type="string" value="true"/> <!-- Odometry correction -->
   <param name="Reg/Strategy"              type="string" value="1"/>    <!-- 1=ICP -->
   <param name="Reg/Force3DoF"             type="string" value="true"/> <!-- 2d slam -->

Make sure TF of camera and lidar are correctly set, so that 3D point clouds match with the 2D scans.


edit flag offensive delete link more


Hi Thank you for answer. I checked TF and build map.

I wonder that lidar is used for correction pose or not.

Thank you.

JunJun gravatar image JunJun  ( 2017-12-01 02:27:08 -0500 )edit

You said "Make sure TF of camera and lidar are correctly set, so that 3D point clouds match with the 2D scans."

this means that 2d scans is used for matching 3D point clouds and quality of 3D map is better than without lidar??

I want to know quality of SLAM depends on only RGB-D, image features.

JunJun gravatar image JunJun  ( 2017-12-01 02:30:06 -0500 )edit

With parameter "Reg/Strategy" set to 1 (ICP), scans are used to refine loop closure detections and for odometry correction (when "RGBD/NeighborLinkRefining" is true). In my example, lidar is used to refine the pose, e.g., the 3D clouds are added to poses refined by lidar.

matlabbe gravatar image matlabbe  ( 2017-12-01 18:01:41 -0500 )edit

In your screenshots, the TF between lidar and camera seems ok, as the lidar scans (cyan points) are aligned with their corresponding clouds created from the camera.

matlabbe gravatar image matlabbe  ( 2017-12-01 18:02:56 -0500 )edit

Thank you for your answer.

I got what you said.

As you said, I measure and set TF.

Below link is TF of my robot.

link text

JunJun gravatar image JunJun  ( 2017-12-03 22:43:13 -0500 )edit

Question Tools

1 follower


Asked: 2017-11-29 03:00:42 -0500

Seen: 1,365 times

Last updated: Nov 29 '17