ROUTE DETAILS
I3. Indoor. A long route inside the Innova Building. Few moving people. 
Robot commanded manually. The robot covers almost the whole floor plant. 

Sensors: ODO, IMU, LIDAR (points)



Distance travelled: 80 m

Sensor frequencies:
Sensor frequencies Hz:
	LiDAR:  0.95 (approx. 1 Hz)
	Odometry:  9.99
	GPS:  NULL
	IMU:  19.39

GPS mode: None

GPS transform: None

MAP
The solution is referenced to an arbitrary initial pos/orient=0.



COMMENTS
LiDAR is captured at 1 Hz. No hops observed or changes in delta_timestamp. The rate at which the LiDAR is captured remains constant.
Caution: the floor of the building is both shiny and dark. Most of the floor readings are not correct and the floor is not correctly represented in the map.
  
 
SCANMATCHING PARAMETERS
# caution, this is needed to remove initial LiDAR scans with no other data associated to it
start_index = 0
# sample LiDAR scans with delta_time in seconds (of course, depends on available data)
delta_time = 0.3
# voxel size: pointclouds will be filtered with this voxel size
voxel_size = None
# select the simple scanmatcher method. Recommended: icppointplane
method = 'icppointplane'


SLAM PARAMETERS    
perform_loop_closing = True
method = 'icppointplane'
skip_loop_closing = 20
skip_optimization = 50
# try to add at most, number_of_candidates_DA
number_of_candidates_DA = 5
# try to find a loop closure, distance_backwards accumulated distance from the last pose
distance_backwards = 7.0
# no uncertainty is considered. Trying to close a loop with all candidates poses within this radius (includes the error)
radius_threshold = 2.0
# visualization: choose, for example, 1 out of 10 poses and its matching scan
visualization_keyframe_sampling = 20