Predictive Control-Based Completeness Analysis and Global Calibration of Robot Vision Features.

Area of Science:

• Robotics
• Computer Vision
• Control Systems

Background:

• Traditional vision servo algorithms face challenges with singular values, local minima, and robustness.
• Accurate robot vision features are crucial for predictive control and task completion.
• Complete macrofeature sets are essential for characterizing vision servo task constraints.

Purpose of the Study:

• To analyze robot vision features for predictive control and feature completeness calibration.
• To propose a robust, calibration-free visual servoing strategy using an interference observer.
• To develop a new scale-free vision servo method addressing traditional algorithm limitations.

Main Methods:

• Defining a complete macrofeature set based on task purpose and constraints.
• Utilizing direct image acquisition and inferred features for macrofeature set completion.
• Developing a dual closed-loop vision servo structure with a Q-filter-based interference observer.
• Estimating and eliminating interference, including model uncertainty and input disturbances.

Main Results:

• A robust, calibration-free visual serving strategy is proposed for high-performance task completion.
• The new scale-free method constructs a dual closed-loop structure for enhanced stability.
• Interference, including model uncertainty and noise, is effectively estimated and eliminated.
• An inner-loop structure presents a nominal model, enabling optimal outer-loop controller design.

Conclusions:

• The proposed dual closed-loop vision servo method significantly improves dynamic performance and robustness.
• This approach overcomes key limitations of existing scale-free vision servo algorithms.
• The strategy ensures high performance and optimal execution of vision servo tasks.