ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question
0

Google Cartographer 3D map data interpretation

asked 2018-05-02 04:08:24 -0600

Clemens gravatar image

updated 2018-05-03 02:19:54 -0600

Hi,

currently I'm working on custom robot platform and I would like to map an outdoor bumpy environment. I have tested different kinds of sensor configuration (horizontal and/or vertical mounted laser scanner, 3D laser scanner like the Velodyne VLP-16) via the usage of GAZEBO. Additionally an IMU and odometry topic is available.

My problem is, that the generated map only visualizes the surrounding objects but not the ground (there is a lot of unknown space). If I try the 2D version the ground is visualized as free space.

If also tried the provided examples from the cartographer ros page, and I'm getting the same result.

Have anyone in the community some hints, how I can map the whole 3D environment? or an example how I can process further?

Thanks, Clemens

Update:

If I run e.g. this example (Link to example), I get a map which contains some obstacles in a light red color (values inside the map topic are round about 50 ). But I get no information about the ground/floor. All other spaces are represented inside the map topic as unknown space. Is it possible to change e.g. a parameter to recognize the floor?

edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted
0

answered 2018-05-02 09:38:51 -0600

In a nut shell to create a 3D map it's much easier (although not 100% required) to use a 3D sensor. A good place to start would be something like RGB-D slam which has a working implementation in ROS already.

If you can give us some more information about your specific problem we may be able to give you a better recommendation.

edit flag offensive delete link more

Comments

If think with the RGB-D SLAM I could get Problems under bad light conditions. Therefore I would like to use laser scanners instead of cameras to exclude this problem.

I have updated my primary question, so I hope you could help me further.

Clemens gravatar image Clemens  ( 2018-05-03 02:19:34 -0600 )edit

But these 2D Lidar scanners are very different sensors, which are much harder to use for 3D mapping than a direct 3D sensor. It would probably be easier to use RGB-D slam and find a robust RGB-D sensor that worked in the lighting conditions you're expecting.

PeteBlackerThe3rd gravatar image PeteBlackerThe3rd  ( 2018-05-03 06:30:41 -0600 )edit

Question Tools

Stats

Asked: 2018-05-02 04:08:24 -0600

Seen: 953 times

Last updated: May 03 '18