Python

Install the public package from PyPI:

python3 -m pip install autodataloader

The distribution is named autodataloader; import it as dataloader. The PyPI package installs only the ROS-independent conversion, reading, validation, and inspection tools.

ROS 1

source /opt/ros/noetic/setup.bash
mkdir -p ~/catkin_ws/src
cd ~/catkin_ws/src
git clone https://github.com/kimdaebeom/dataloader.git
cd ..
rosdep install --from-paths src --ignore-src -r -y
catkin_make
source devel/setup.bash

ROS 2

source /opt/ros/jazzy/setup.bash
mkdir -p ~/ros2_ws/src
cd ~/ros2_ws/src
git clone https://github.com/kimdaebeom/dataloader.git
cd ..
rosdep install --from-paths src --ignore-src -r -y
colcon build --symlink-install --packages-select dataloader
source install/setup.bash

Python CLI

dataloader-convert \
  --dataset mulran \
  --source /data/raw/mulran/KAIST01 \
  --output /data/converted

Python API

from dataloader import Dataset, convert_dataset

result = convert_dataset(
    dataset="mulran",
    source="/data/raw/mulran/KAIST01",
    output_root="/data/converted",
)
dataset = Dataset(result["sequence_dir"])
points = dataset.lidar(frame=0).numpy()  # float32 [N, 4]

ROS 1

rosrun dataloader dataloader_convert.py \
  --dataset mulran \
  --source /data/raw/mulran/KAIST01 \
  --output /data/converted

ROS 2

ros2 run dataloader dataloader_convert \
  --dataset mulran \
  --source /data/raw/mulran/KAIST01 \
  --output /data/converted

Choose a config from config/ and set its dataset root and sequence.

ROS 1

roslaunch dataloader player.launch \
  config:=$(rospack find dataloader)/config/mulran.yaml

ROS 2

ros2 launch dataloader player.launch.py \
  config:=$(ros2 pkg prefix dataloader)/share/dataloader/config/mulran.yaml

SemanticKITTI currently expects the preprocessed pcd/ plus odom_tum.txt layout described in Dataset layouts. Native KITTI/KITTI Odometry input is not yet supported.