The dataset shows a lecture hall from Würzburg university. it contains two
registered scans with 22.3 million points each. The data was collected using a
Riegl VZ-400 laser range finder.

MD5sum of lecturehall.tar.xz: 8127a8f56a125984c19795e6cecec2b5

The dataset contains four files: two point clouds and two 6DOF poses.

lecturehall1.pose1.object1.label.csv and lecturehall1.pose2.object2.label.csv
contain point cloud data in CSV format compatible with comma and snark:

https://github.com/acfr/comma
https://github.com/acfr/snark

The first three columns are cartesian x,y,z point coordinates in a right-handed
coordinate system with the z-axis pointing up. The fourth column is the scan
number and always zero. This column only exists for compatibility with the
shell scripts for change detection provided by Underwood et al.:

http://www.acfr.usyd.edu.au/papers/icra13-underwood-changedetection.shtml

The fifth column indicates whether a point is static or dynamic. The second
scan only has static points while the first scan contains about half a million
dynamic points belonging to objects in space that it seen as free by the second
scan.

lecturehall1.pose1.csv and lecturehall1.pose2.csv contain 6DOF poses in
single-line CSV format. The first three values indicate a translation and the
last three values a rotation in radians.

To convert the data from the CSV format to the left-handed coordinate system
understood by 3DTK you can use awk. To convert a CSV file to a .3d file in uos
format use:

awk -F, '{print -$1, -$3, $2}'

To convert a CSV pose file to a 3DTK .pose file use:

awk -F, 'BEGIN{pi=atan2(0,-1)} {print -$1, -$3, $2; print ($4*180)/pi, ($6*180)/pi, (-$5*180)/pi}'