ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question
1

Fastest way to build a 3D map from 3D Scans and TF data?

asked 2013-04-15 10:47:49 -0600

madmax gravatar image

updated 2014-01-28 17:16:11 -0600

ngrennan gravatar image

Hi,

I did some 3D scans and saved them to a bag file with the corresponding tf, imu and odom frames. But I am wondering how I can build a map of my data?

Is there something like a package where I just have to input the PointCloud2, imu and tf data?
Or do I have to code something in combination with the PCL library?

edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted
1

answered 2013-04-15 11:02:45 -0600

dornhege gravatar image

updated 2013-04-15 11:03:06 -0600

Define, what a 3D map is for you and what you expect from the algorithm.

I'm assuming from the topic you want mapping from known poses and not SLAM. There are multiple valid representations. A quick and simple way would be to build an octomap using the octomap_server package. This accepts localized (via tf) PointCloud2 and produces an octree map.

edit flag offensive delete link more

Comments

Well, when I use an octomap, the movement of my robot is not recognized. That is the main problem. So the captured point clouds are interfering and my map is not "growing". Hope you understand my english... Maybe a solution is to transform a point cloud to a laser scan and do mapping with the scans.

madmax gravatar image madmax  ( 2013-04-15 12:23:17 -0600 )edit

What kind of tf data do you have? tf needs to transform to a fixed frame to work for mapping. Otherwise you need a 3D SLAM approach or in the simplest case try if odometry works.

dornhege gravatar image dornhege  ( 2013-04-15 13:04:44 -0600 )edit

We only have transforms on the robot itself and not to a static point. The odometry was recorded on another pc and has to be integrated.

madmax gravatar image madmax  ( 2013-04-17 07:09:33 -0600 )edit

Question Tools

Stats

Asked: 2013-04-15 10:47:49 -0600

Seen: 555 times

Last updated: Apr 15 '13