Raw-Data

There are two different parts of our dataset. First there are the the raw datalogs containing all the raw sensor data. Our software setup utilizes the Robot Operating System (ROS). Therefore the data is saved as ROS bagfiles. The data of each of the sensors is saved as messages, each on its own topic. Additionally we provide annotated hyperspectral data-cubes which where extracted and preprocessed from the rosbags in the form of matlab-files.

When playing the bagfiles all messages are going to be republished, while ROS adapts the old system time, so your software is able to simulate a live application.

To use these files first the ROS Middelware has to be installed. All messages published are standard ROS messages. So no additional software is needed. Then just use the following commands to play back the bagfiles.

Terminal 1:

 user@system:~$ roscore

Terminal 2:

user@system:~$ cd /path/to/downloaded/bagfiles/
user@system:~$ rosbag play --clock -r 1.0 bagfile_name.bag

For the HyKo1 dataset the following sensors were mounted on a car:

1 Hyperspectral Camera MQ022HG-IM-SM4X4-VIS with LM5JC10M lens
1 Hyperspectral Camera MQ022HG-IM-SM5X5-NIR with VIS-NIR lens from Edmund Optics
1 Basler Camera acA2040-25gc with LM6HC lens.
1 Lidar Veldoyne HDL-32E LiDAR
1 Spectrometer Qmini

The rosbags were splitted every 30 seconds to keep them small and handy.

Here is a sample output of rosbag info:


types:       
qmini_rosnode/Spectrum     [88383658b37a575733e4b2310594a823]
sensor_msgs/CameraInfo     [c9a58c1b0b154e0e6da7578cb991d214]
sensor_msgs/Image          [060021388200f6f0f447d0fcd9c64743]
sensor_msgs/NavSatFix      [2d3a8cd499b9b4a0249fb98fd05cfa48]
velodyne_msgs/VelodyneScan [50804fc9533a0e579e6322c04ae70566]
topics:      
/camera/basler_camera/camera_info   148 msgs    : sensor_msgs/CameraInfo
/camera/basler_camera/image_raw     148 msgs    : sensor_msgs/Image
/fix                                 29 msgs    : sensor_msgs/NavSatFix
/nir_cam/camera_info                149 msgs    : sensor_msgs/CameraInfo
/nir_cam/image                      149 msgs    : sensor_msgs/Image
/qmini_spectra                       15 msgs    : qmini_rosnode/Spectrum
/velodyne_packets                   298 msgs    : velodyne_msgs/VelodyneScan
vis_cam/camera_info                 149 msgs    : sensor_msgs/CameraInfo
vis_cam/image                       149 msgs    : sensor_msgs/Image

For the HyKo2 dataset the following sensors were mounted on a truck:

1 Hyperspectral Camera MQ022HG-IM-SM4X4-VIS with LM5JC10M lens
1 Hyperspectral Camera MQ022HG-IM-SM5X5-NIR with VIS-NIR lens from Edmund Optics
2 Lidar Veldoyne HDL-32E LiDAR
1 Spectrometer Qmini

The rosbags were splitted every 60 seconds to keep them small and handy.

Here is a sample output of rosbag info:

types:       
qmini_rosnode/Spectrum     [abfeda1c25189e1c164e3d6f4938e186]
sensor_msgs/CameraInfo     [c9a58c1b0b154e0e6da7578cb991d214]
sensor_msgs/Image          [060021388200f6f0f447d0fcd9c64743]
velodyne_msgs/VelodyneScan [50804fc9533a0e579e6322c04ae70566]
topics:
/nir_cam/camera_info   600 msgs    : sensor_msgs/CameraInfo    
/nir_cam/image         600 msgs    : sensor_msgs/Image         
/qmini_spectra         210 msgs    : qmini_rosnode/Spectrum    
/velodyne_packets1     599 msgs    : velodyne_msgs/VelodyneScan
/velodyne_packets2     599 msgs    : velodyne_msgs/VelodyneScan
/vis_cam/camera_info   600 msgs    : sensor_msgs/CameraInfo    
/vis_cam/image         600 msgs    : sensor_msgs/Image