search

SICKAG / sick_scan_xd

Based on the sick_scan drivers for ROS1, sick_scan_xd merges sick_scan, sick_scan2 and sick_scan_base repositories. The driver supports both Linux (native, ROS1, ROS2) and Windows (native and ROS2).
Apache License 2.0
90 stars 83 forks source link

No "time" field in the PointCloud2 output using the ROS1 driver #295

Closed aubin-tchoi closed 4 months ago

aubin-tchoi commented 4 months ago

Hello,

I successfully built the ROS1 driver and have everything working very well but found out that there was no field containing the point-wise timestamps in the PointCloud2 messages output. I was wondering whether this is something that will be added in the future or can already be accessed with some minor tinkering on the driver.

Thanks in advance.

rostest commented 4 months ago

Thanks for your feedback. Most scanner transmit one 32 bit timestamp for each scan (detail from the telegram listing): image Point-wise timestamps are currently not provided.

aubin-tchoi commented 4 months ago

Thank you for your feedback, what I am looking for is indeed point-wise timestamps. Do you find this is something that you are planning to add in the future to the ROS driver?

rostest commented 4 months ago

Point timestamps are not currently planned. They could be calculated in the following way:

The point cloud time stamp is the transmission time stamp, i.e. the time stamp of the last shot converted to system time. The time step dt in seconds between two points is constant and can be calculated as follows

dt = (1 / scan_frequency) * (max_azimuth - min_azimuth) / (360 degrees)

given the horizontal scan range (max_azimuth - min_azimuth) resp. the azimuth range of a scan segment in degree and the scan_frequency in Hz.

The timestamp of the ith point can be calculated as t_i = pointcloud_timestamp - (N - 1 - i) * dt with 0 <= i < N and N points in the point cloud.

Note that this calculation works for common lidars with unsegmented point clouds, i.e. all point clouds that are not cumulated over different scan segments. In the case of point clouds concatenated from segments, i.e. the full frame point clouds of multiScan and picoScan lidars, the points are not ordered by ascending time. In this case, the calculation must be performed for each individual segment.

aubin-tchoi commented 4 months ago

Thank you for your answer, I am currently using the multiScan100, which has multiple scan segments, but based on your documentation of the sensor I can think of a way to implement point-wise timestamps when I have the time. Thank you for everything, I am closing this issue then!