3d slam with 2d sensor

Hi guys,

I am wandering if any software exists to perform 3d SLAM with a 2d laser scanner. In principle, I have a 2d hokuyo lidar attached to my UAV. I am able to calculate the odometry of the UAV thanks to the Intel Relsense tacking camera, so I can reconstruct the 6D pose of the UAV. However, I want to fuse the odometry information with a SLAM software in order con refine the localization (adjusting the odometry) and create a map of the environment.

Of course I know software like gmapping or hector_slam, but If I well understand, such software don't consider variation over the altitude of the robot. In fact the generated my is a 2d occupancy gridmap. I also used the pose of the uav to create my own octomap, but only using raw odometry data. I have wront result if I use the output of such SLAM software.

I checked out other packages like rtabmap... but again depth + image data are needed.

Thank you for any suggestion.

Asked by pacifica on 2019-12-11 08:03:48 UTC

Comments

Answers

You can use 3d slam package, if you use laserscan_to_pointcloud node and convert sensor_msgs/LaserScan into sensor_msgs/PointCloud2.

Asked by Masaya Kataoka on 2019-12-11 21:24:23 UTC

Comments

Thank you for your comment. Sure I could try using a classical 3d slam algorithm converting Laser data in point cloud. I was wandering if you were aware about a SLAM ROS package that consider point cloud and odometry to generate the map.

Asked by pacifica on 2019-12-16 09:13:09 UTC