6DoF Pose: Definition and Robotics Data Requirements

Direct definition

What Is 6DoF Pose?

A 6DoF pose describes the position and orientation of a rigid body in three-dimensional space. The six degrees are translation along three axes and rotation around three axes. In robotics, the pose may describe an end effector, camera, object, robot base, human head, tool, or sensor frame.

Plain-language analogyA map location tells you where a phone is. A 6DoF pose also tells you which way the phone is facing and how it is tilted.

Operational example

Representative 6DoF pose record

translation: x, y, z in meters
rotation: Normalized quaternion x, y, z, w
parent_frame: robot_base
child_frame: end_effector
provenance: Measured, estimated, annotated, or derived with confidence

The Six Degrees of Freedom

Source-backed context[1] ROS 2 Pose message definition [2] Project Aria coordinate and data formats

Three degrees describe translation: movement along x, y, and z. Three describe rotation: orientation around those axes. Together they locate and orient a rigid body. A robot may track many 6DoF poses at once, including the base, camera, gripper, objects, and tools.

Six degrees of freedom does not necessarily mean six robot joints. A six-joint arm may control an end-effector pose, but robot configuration and rigid-body pose are different concepts. Multiple joint configurations can sometimes produce the same end-effector pose.

Position Is Simple; Rotation Needs Conventions

Source-backed context[1] ROS 2 Pose message definition [3] EuRoC visual-inertial pose and calibration data

Translation is usually represented as three coordinates in meters or millimeters. Orientation may be represented with a quaternion, rotation matrix, axis-angle value, or Euler angles. Each representation has different constraints and failure modes.

Quaternions require a documented component order and normalization. Euler angles require axis order and units and can encounter singularities. Rotation matrices require orthonormality checks. Conversions should be versioned and tested rather than assumed from array shape.

A Pose Always Exists Relative to a Frame

Source-backed context[2] Project Aria coordinate and data formats [3] EuRoC visual-inertial pose and calibration data

The same object has different coordinates in the camera frame, robot base frame, world frame, and gripper frame. Parent and child frame identifiers are therefore part of the meaning. Axis direction, handedness, and transform composition rules must also be consistent.

Calibration connects frames. Camera extrinsics can relate a camera pose to the robot. Hand-eye calibration can relate a sensor to an end effector. If hardware moves or calibration changes, the dataset needs a new calibration version and a clear validity interval.

Measured, Estimated, and Annotated Pose

Source-backed context[1] ROS 2 Pose message definition [2] Project Aria coordinate and data formats [4] DROID robot pose and trajectory data

Pose can come from robot kinematics, motion capture, visual-inertial odometry, object-pose estimation, manual annotation, or simulation. These sources have different noise, latency, drift, and failure behavior. A field named pose should not hide that origin.

Quality review should compare pose against independent evidence when possible, validate frame transforms, inspect discontinuities, test rotation normalization, and measure confidence or error. Missing or lost tracking must be represented explicitly rather than held at the last valid value.

Frame identifier missing or inconsistent
Millimeters and meters mixed
Quaternion ordering undocumented
Tracking loss represented as a valid repeated pose
Derived pose lacks calibration or software lineage

What a Buyer Should Specify

Implementation guidanceEGXO guidance for translating the definition into a project specification.

Name the rigid bodies and frames that matter to the task. Define how pose is produced, represented, timed, calibrated, validated, and aligned with images, actions, or object states. Ask for uncertainty and tracking validity as first-class fields.

Tracked body, parent frame, and child frame
Translation units and rotation representation
Timestamp, rate, interpolation, and synchronization
Measured, estimated, annotated, synthetic, or derived origin
Calibration, confidence, covariance, and validity
Evaluation tolerance for the target task

Research basis

Primary Sources and Further Reading

The sources establish the technical context. The buyer requirements are EGXO guidance for converting that context into a testable data specification.

6DoF Pose6DoF