| ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange |
Recently, I am participated an ARM navigation program. I obtain point cloud data from stereo vision, and I want to obtain octomap of the environment based on the stereo vision. so i want to know that does octomap library suit for stereo vision? please help me, thank you.
OctoMap is more than suited for stereo vision and RGB-D data (e.g. from Kinect or Asus Xtion), in fact it's mainly used on the PR2 to build collision maps from stereo (in the packages collider and octomap_server). Just use the PointCloud2 message type to build maps.
The only limitation might be the amount of data you can insert, I have seen rates of about 10 Hz for tabletop (arm navigation) data. If the insertion / map update gets too slow, you can downsample your data, limit its maximum range in octomap, or increase the octomap resolution (voxel size).
thank you for your help. As your words, If the insertion / map update gets too slow, I can do something like limiting its maximum range in octomap. However, I think that the maximum range in octomap is a constant after the octomap resolution have been set. For example, if we set the octomap resolution with 0.05m, the maximum range will be equal to 1638.4m (0.05*2 ^ 15). So how can I limit the maximum range in octomap?
Today, I have found another library (point cloud library) in which 3D environmet can also been described with octree. does there exist any apparent advantages for octomap library compared with pcl_octree library?
Asked: 2012-10-15 21:18:13 -0600
Seen: 1,137 times
Last updated: Oct 17 '12
ROS Answers is licensed under Creative Commons Attribution 3.0 Content on this site is licensed under a Creative Commons Attribution Share Alike 3.0 license.
