Globally Optimal Hand-Eye Calibration Using Branch-and-Bound.

Area of Science:

• Robotics
• Computer Vision
• Optimization

Background:

• Hand-eye calibration is crucial for robot manipulation and task automation.
• Existing methods often require known calibration targets or prior camera information, limiting flexibility.
• Accurate calibration is essential for precise robotic interactions.

Purpose of the Study:

• To introduce a novel, targetless hand-eye calibration algorithm.
• To eliminate the need for prior knowledge of external camera parameters.
• To achieve globally optimal solutions for rotation and translation recovery.

Main Methods:

• Utilizes direct camera measurements, eliminating the need for calibration targets.
• Employs a branch-and-bound optimization strategy.
• Incorporates Linear Programming to determine bounding steps within the algorithm.
• Minimizes an objective function based on the epipolar constraint.

Main Results:

• Simultaneously recovers unknown rotation and translation parameters.
• Achieves a globally optimal solution with respect to the L∞-norm.
• Demonstrates a novel approach to hand-eye calibration without prior knowledge.

Conclusions:

• The proposed method offers a more flexible and efficient solution for hand-eye calibration.
• The algorithm guarantees global optimality, enhancing reliability in robotic applications.
• This technique advances the field of robot calibration by removing common constraints.