Table of Contents

Overview
#

Understanding coordinate systems is essential for sensor fusion in robotics. This guide covers LiDAR coordinate conventions and ROS Transform (TF) system.

Standard Coordinate Convention
#

RGB Axis System
#

         Z (Blue)
         ↑
         │
         │
         └────────→ X (Red)
        ╱
       ╱
      Y (Green)
AxisColorDirection
XRedForward
YGreenLeft
ZBlueUpward

This follows the right-hand rule.

TF (Transform) System
#

What is TF?
#

TF manages relationships between multiple coordinate systems:

  • Robot base to sensor frames
  • Map to robot frame
  • World to local frames

TF Tree Structure
#

       map
        │
        ↓
      odom
        │
        ↓
    base_link
    ↙       ↘
laser       camera

Common Commands
#

View TF Tree
#

rosrun tf view_frames

Generates frames.pdf showing transform tree.

Alternative (Noetic)
#

For Python 3 compatibility:

rosrun tf2_tools view_frames.py

Or use GUI:

rosrun rqt_tf_tree rqt_tf_tree

Echo Transform
#

rosrun tf tf_echo base_link laser

Output:

At time t
- Translation: [x, y, z]
- Rotation: [qx, qy, qz, qw]

Static Transform Publisher
#

For fixed sensor positions:

rosrun tf static_transform_publisher x y z yaw pitch roll parent_frame child_frame period_ms

Example:

rosrun tf static_transform_publisher 0.1 0 0.05 0 0 0 base_link laser 100

RViz Visualization
#

Launch RViz
#

rosrun rviz rviz

Add TF Display
#

  1. Click “Add”
  2. Select “TF”
  3. Configure:
    • Show Axes: Check
    • Frame Timeout: 15
    • Frames: Select relevant ones

URDF Definition
#

Sensor Position in URDF
#

Located in: ~/catkin_ws/src/your_robot_package/urdf/robot.urdf

<robot name="my_robot">
  <link name="base_link"/>

  <link name="laser_link">
    <visual>
      <geometry>
        <cylinder length="0.02" radius="0.03"/>
      </geometry>
    </visual>
  </link>

  <joint name="laser_joint" type="fixed">
    <parent link="base_link"/>
    <child link="laser_link"/>
    <origin xyz="0.1 0 0.05" rpy="0 0 0"/>
  </joint>
</robot>

Parameters
#

ParameterDescription
xyzPosition offset
rpyRoll, Pitch, Yaw (radians)
parentReference frame
childThis sensor’s frame

Transform Types
#

Static Transform
#

Fixed relationship (sensor to robot):

<node pkg="tf" type="static_transform_publisher" name="laser_tf"
      args="0.1 0 0.05 0 0 0 base_link laser 100"/>

Dynamic Transform
#

Changing relationship (robot to map):

  • Published by odometry
  • Updated by localization

Sensor Fusion Application
#

Why Transforms Matter
#

To combine LiDAR with other sensors:

  1. Know each sensor’s position
  2. Transform data to common frame
  3. Fuse in unified coordinate system

Example: LiDAR + Camera
#

import tf

listener = tf.TransformListener()

# Get transform from camera to laser
(trans, rot) = listener.lookupTransform(
    '/camera_link',
    '/laser_link',
    rospy.Time(0)
)

# Transform point from laser to camera frame
# Apply translation and rotation

Troubleshooting
#

“Could not find transform”
#

# Check if frames exist
rostopic echo /tf | grep frame_id

Old Python Error (Noetic)
#

If view_frames fails:

# Use tf2 instead
rosrun tf2_tools view_frames.py

Verify Transform Chain
#

rosrun tf tf_monitor

Shows all transforms and their rates.