End-Effector Pose

The position and orientation of an arm's gripper or tool in space, usually given relative to the base frame.

The end effector is the part at the very tip of a robot arm that actually interacts with the world — a gripper, a suction cup, a welding torch. The end-effector pose is its position (x, y, z) plus its orientation, 6 degrees of freedom altogether, generally expressed relative to the base frame. Orientation can be written as a rotation matrix, a quaternion, Euler angles, or axis-angle, so the end-effector pose's dimensionality can be 6, 7, or more depending on the dataset. It's the bridge between “task” and “motors”: a person or a vision model cares where the gripper is and which way it's pointing, while the motors only understand joint angles, and forward and inverse kinematics convert between the two. Many VLA (vision-language-action) and imitation-learning policies directly output an end-effector pose, or a delta to it, which a controller then converts into joint commands.

Examplerobosuite's OSC_POSE controller takes a 6-dimensional action: the first 3 values are a delta to end-effector position, the last 3 are a delta to orientation in axis-angle form. In Franka's libfranka interface, O_T_EE is the 4×4 matrix giving the end-effector's pose in the base frame.

Also called

EEF Pose, EE Pose, TCP Pose

Related

End Effector · Pose · Tool Center Point · Task Space · Inverse Kinematics (IK) · Delta (Relative) Action vs. Absolute Action

Sources

robosuite Documentation: Controllers（OSC_POSE）
libfranka robot_state.h（O_T_EE）
Wikipedia: Robot end effector