humanoidsdata.com

Search

Search companies, datasets, articles, and glossary terms for humanoids and embodied AI.

← All glossary terms

Hardware & control

Forward kinematics

Forward kinematics computes the pose of robot links or end effectors from a specified robot configuration and kinematic model. It propagates known joint transformations through the kinematic chain; it does not determine the forces, torques, or motion that produced that configuration.

Also known as: direct kinematics, forward robot kinematics

Updated

Configuration maps to pose

Modern Robotics presents forward kinematics using products of rigid transformations. Given the joint variables, link geometry and frame definitions, the calculation returns where a chosen link is and how it is oriented.

For a branching humanoid model, the same principle applies along each chain from the base to a hand, foot, camera or other link. A floating-base pose must also be supplied when poses are required in a world frame.

Forward and inverse kinematics solve opposite questions

Forward kinematics answers, “Where is the hand for these joint values?” Inverse kinematics asks, “Which joint values place the hand here?” The forward map is normally deterministic for a fixed model and configuration, while the inverse problem may have multiple solutions, no solution or singular cases.

Neither calculation is a dynamics simulation. It does not predict acceleration under torque, contact forces, actuator delay or structural flex. Those effects can make measured hardware poses differ from ideal kinematic predictions.

Model conventions determine the result

Incorrect joint axes, origins, signs or link dimensions produce incorrect poses even when the implementation is mathematically sound. Calibration offsets and the exact robot revision matter.

When a dataset provides both joint state and link pose, state whether the pose was measured independently or computed by forward kinematics. Include the robot-description version and coordinate frames. Computed poses should not be presented as independent ground-truth evidence for the same kinematic model that generated them.

Sources