Camera Transforms and Calibration

This document explains how the O3R camera transform mechanism works, the calibration process, and how transforms are published in ROS 2.

Overview

The O3R camera system uses a multi-level transform hierarchy to accurately position sensor data in 3D space. Understanding this hierarchy is crucial for proper integration with robotic systems and point cloud processing.

Transform Hierarchy

The O3R system defines three main coordinate frames:

ifm_base_link - The reference frame after calibration
mounting_link - The physical mounting point of the camera
optical_link - The optical center of the camera sensor

Transform hierarchy overview

The complete transform chain is: ifm_base_link → mounting_link → optical_link

Calibration Reference Point

Important

The O3R calibration reference point, mounting_link, is located at the center back of the camera housing, not at the optical center. This is a key difference from some other camera systems.

Detailed view of mounting and optical frames

When performing calibration:

The mounting frame reference is at the back of the O3R camera head housing
Scale drawings for exact mounting dimensions can be found on ifm.com in the download section for your specific camera model
The optical frame is automatically calculated from factory intrinsic parameters

Additional Calibration Resources

For comprehensive information about O3R calibration routines and best practices:

General calibration guide: O3R Calibration Routines
Introduction to calibrations: Calibration Concepts

Calibration Process

Where to Calibrate

Calibration should be performed in the ifm coordinate system, not in “ROS world” coordinates. There are two primary methods:

Method 1: Using ifm Vision Assistant

The ifm Vision Assistant provides a graphical interface for camera calibration:

Connect to your O3R device via Vision Assistant
Navigate to the camera configuration section
Use the built-in calibration tools to set extrinsic parameters
The calibration defines the transform from mounting_link to ifm_base_link

For detailed instructions, refer to changing parameters using ifm Vision Assistant.

Method 2: JSON Configuration File

You can also calibrate via JSON configuration files. See the configuration documentation:

For camera nodes: Camera Parameters (see config_file parameter)
For ODS applications: ODS Configuration

Example extrinsic calibration in JSON:

{
  "ports": {
    "port0": {
      "processing": {
        "extrinsicHeadToUser": {
          "rotX": 0.0,
          "rotY": 0.0, 
          "rotZ": 1.57,
          "transX": 0.5,
          "transY": 0.0,
          "transZ": 1.2
        }
      }
    }
  }
}

Note

Independent Calibration Values: The rotation and translation values in the JSON configuration are applied independently, not sequentially. Each parameter contributes directly to the final transform matrix.

Published Transforms

After calibration, the ifm3d-ros2 node publishes the following transforms:

1. Base to Mounting Transform

From: ifm_base_link
To: mounting_link
Source: Extrinsic calibration parameters (extrinsicHeadToUser)
Purpose: Positions the camera relative to the calibrated reference frame

2. Mounting to Optical Transform

From: mounting_link
To: optical_link
Source: Factory intrinsic calibration parameters
Purpose: Provides the optical center for image and point cloud data

Both transforms can be enabled/disabled via ROS parameters. See Camera Parameters for configuration options.

Point Cloud Reference Frame

After calibration, all point cloud data is published with ifm_base_link as the reference frame. This means:

Point cloud coordinates are relative to your calibrated reference system
The calibration establishes the relationship between camera data and your robot/world coordinate system
Visualization tools like RViz will display the point cloud in the ifm_base_link frame

Transform Configuration Parameters

The transform publishing behavior can be controlled via ROS parameters:

Parameter	Type	Default	Description
`~/tf.base_frame_name`	string	`ifm_base_link`	Name for the ifm reference frame
`~/tf.mounting_frame_name`	string	`camera_mounting_link`	Name for the camera mounting frame
`~/tf.optical_frame_name`	string	`camera_optical_link`	Name for the optical frame
`~/tf.publish_base_to_mounting`	bool	`true`	Publish base→mounting transform
`~/tf.publish_mounting_to_optical`	bool	`true`	Publish mounting→optical transform

Integration with ROS Navigation Stack

For integration with ROS navigation and other robotics applications:

Robot Base Link: Your robot should publish a transform from its base_link to ifm_base_link
Static Transform: This can typically be a static transform since the camera position relative to the robot is fixed
Multiple Cameras: Each camera can have its own ifm_base_link or share a common reference frame

Example static transform publisher:

ros2 run tf2_ros static_transform_publisher 0 0 0 0 0 0 base_link ifm_base_link