This is a data loader for KITTI. It publishes the RGB image data, velodyne pointclouds and IMU data at 10 Hz. We use LCM to publish these messages. Also, we publish the tracklet information as a set of tracked objects, their position and classes (both in velodyne coordinate frame and as bounding boxes in each camera) at 10 Hz.
You will need to install LCM to use this and export a variable JAVA_CLASSPATH that gives the
location of lcm.jar on your computer. Either use lcm-gen to compile the LCM messages in
kitti.lcm or use the included lcmtypes_kitti.jar file.
You will also need to download the KITTI dataset.
- Run
python kitti2lcm.py [base_dir] [date] [drive name]
- Run
./spy.shwhich should openlcm-spyprogram that can show each message as it is being published. You can then runlcm-loggerto log these messages as an LCM file and play them back later.