kitti2rosbag2

kitti2rosbag2 is designed to convert the KITTI Odometry dataset to ROS2 bag format, emphasizing manual control over message publishing and bag recording.

KITTI car setup

Image source: (https://www.cvlibs.net/datasets/kitti/)

Usage

1. Clone Repository

cd ~/ros2_ws/src
git clone https://github.com/bharadwajsirigadi/kitti2rosbag2.git

2. Parameters Input

Open params file

Update following tags.

kitti_pub:
  ros__parameters:
    sequence: <sequence_no>  #Integer
    data_dir: '<dataset_dir>'
    odom_dir: '<data_odometry_poses_dir>'
    bag_dir : '<bag_dir>/<bag_name>'
    odom : [True/False] 

Example: Converts kitty dataset to rosbag2.

Download data_odometry_poses from here

kitti_rec:
  ros__parameters:
    sequence: 0
    data_dir: '/home/user_name/Download/data_odometry_gray/dataset/'
    odom_dir: '/home/user_name/Download/data_odometry_poses/dataset/' 
    bag_dir : '/home/user_name/Download/00_bag'
    odom : True

3. Building Package

cd ~/ros2_ws
colcon build --packages-select kitti2rosbag2 --symlink-install

4. Converting to bag

ros2 launch kitti2rosbag2 kitti2rosbag2.launch

KITTI Odometry Dataset Folder Structure

ROS Topics Info

• car/base/odom--> Odometry of Car.
• car/base/odom_path--> Ground Truth path of Car.
• camera2/left/image_raw--> Color Images from Left Camera.
• camera3/right/image_raw--> Color Images from Right Camera.
• camera2/left/camera_info--> Left Camera Information.
• camera3/right/camera_info--> Right Camera Information.

Requirements

• Python 3.x
• ROS2 installed(tested on ROS2 humble)

Note

Ground Truth-Odometry data is available for sequences(1-10) in KITTI dataset.

Contributors

• Sai Bharadwaj Sirigadi