Robotics
Hand-eye calibration
Definition
Hand-eye calibration estimates the fixed rigid-body transformation between a robot-mounted sensor and the robot hand or end effector. The result lets sensor measurements and commanded robot motion be expressed in consistent coordinate frames.
Also known as: Hand eye calibration
Updated
Connect camera and robot frames
An eye-in-hand camera reports an object's pose in the camera frame, while a robot controller needs the pose in a base, hand, or tool frame. Hand-eye calibration estimates the fixed rigid-body transformation between the sensor and the moving robot link. Similar calibration arrangements can place the camera outside the robot and estimate its relation to the robot base.
The classical motion formulation is written AX = XB. A represents relative robot-hand motion, B represents the corresponding sensor motion, and the unknown X is the hand-to-sensor transform. Tsai and Lenz presented an autonomous method for this problem. Horaud and Dornaika analyse the classical equation and an alternative formulation for camera-based sensors.
Use several informative motions
A calibration routine observes a target or scene from several robot poses and pairs those observations with robot kinematics. Rotations about varied axes help constrain orientation; translation then constrains the positional offset. The estimated transform can support object pose estimation, point-cloud fusion, and visual servoing.
Hand-eye calibration is an extrinsic calibration between coordinate frames. It does not by itself estimate camera focal length, lens distortion, joint offsets, or time synchronization, although a larger calibration procedure may estimate some of them together.
Motion and measurement errors couple
Repeated motions about one axis or very small motions may not constrain every component of the transform. Target-detection error, robot encoder error, camera-intrinsic error, structural flex, and an incorrect timestamp pairing all affect the estimate. A sensor mount that shifts after calibration invalidates the assumed fixed transform.
Residual error in the calibration equations is not the only useful test. The transform should also be checked on held-out poses and against the downstream task, such as reprojection or end-effector placement. A low algebraic residual can coexist with a practically important spatial error.
Sources
Related terms
Rigid-body transformation
A rigid-body transformation changes a body's position and orientation without changing its shape or size. In three-dimensional robotics it consists of a proper rotation and a translation.
Pose estimation
Pose estimation determines the position and orientation of an object or robot relative to a reference frame. For a rigid body in three-dimensional space, a full pose has three translational and three rotational degrees of freedom.
Sensor fusion
Sensor fusion combines information from multiple sensors or estimation sources to produce a shared estimate. The combination must account for coordinate frames, timing, uncertainty, and dependence between inputs.
Visual servoing
Visual servoing uses visual measurements inside a feedback loop to control robot motion. The controller updates movement to reduce an error defined from image features or visually estimated pose.