The site is read-only. Please transition to use Robotics Stack Exchange
| ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange |
 | 1 | initial version |
Which rtabmap and opencv versions are you using? You could try different PnP flags with Vis/PnPFlags parameter (default 0=iterative, 1=EPNP, 2=P3P). Otherwise, we can also switch to 3D->3D estimation by setting Vis/EstimationType to 0.
Pose estimation from camera is required on loop closure detection to get a relatively good guess for scan matching. Well, if you don't want visual loop closure detection, you can set Kp/MaxFeatures=-1 to avoid extracting visual features (so no motion estimation will be done with the camera).
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.
