Add support for standard ROS PoseStamped message #13
Conversation
| pMsg.pose.orientation.w *= -1; | ||
| } | ||
|
|
||
| pMsg.header.stamp = ros::Time::now(); |
There was a problem hiding this comment.
shouldn't this be the frame timestamp?
There was a problem hiding this comment.
@christiankerl Technically it should, but the frame timestamp does not use ROS time and the resolution is only seconds, which makes it useless anyway. Need atleast millisecond resolution on the frame stamp for it to be useful. Since I've been seeing on an average 30ms or so processing time per frame, it has to be at least that accurate so that estimators or flight control units can make use of it.
|
I added rospose output, however I am not entirely sure whether it is the correct pose (or needs to be the inverse or something). I'm using the timestamp from the original frame message (if it makes sense, i.e. is different from zero, otherwise ros::Time::now()) - as it's seconds with double precision, that amounts to more than millisecond resolution. |
|
The pose seems to be correct. One thing I found in testing, when doSLAM was enabled, the pose seemed pretty fine. On turning off doSLAM, the Y and Z axes seem to be reversed. |
This message can be used for direct vehicle state estimation or control, for people who are using LSD_SLAM onboard quadrotors, ground robots, etc. You may fuse this with metric sensor data like IMU, baro, etc. to get scaled metric measurements.
Coordinate frame : Z out of camera , Y down, X right.