A Kalman Filter-Based Kernelized Correlation Filter Algorithm for Pose Measurement of a Micro-Robot.

Area of Science:

• Robotics
• Computer Vision
• Control Systems

Background:

• Accurate pose estimation is critical for micro-robot navigation and control.
• Existing tracking algorithms struggle with high-speed, non-linear target motion.
• Real-time performance and high precision are essential for micro-robot applications.

Purpose of the Study:

• To develop and evaluate a novel moving-target tracking algorithm for micro-robot pose measurement.
• To improve tracking accuracy and speed for fast-moving targets using an adaptive Kalman filter and kernelized correlation filter.
• To validate the algorithm's performance on a monocular vision platform.

Main Methods:

• Integration of an adaptive Kalman filter for state prediction of linear and non-linear targets.
• Application of the kernelized correlation filter (K2CF) for accurate target position detection.
• Embedding the K2CF algorithm within a micro-robot's attitude measurement system.
• Comparative analysis of K2CF against other trackers under various motion conditions.

Main Results:

• The proposed K2CF algorithm demonstrated improved positioning accuracy and maintained real-time operation.
• K2CF achieved superior accuracy compared to other algorithms on standard benchmark datasets (OTB-50, OTB-100).
• The system met precision requirements for position and attitude angles within a specified field of view and motion parameters.

Conclusions:

• The Kalman filter-based K2CF algorithm offers a robust solution for high-precision, high-speed micro-robot pose tracking.
• The algorithm's adaptive nature effectively handles complex target dynamics.
• This advancement has significant implications for micro-robotics, enhancing navigation and operational capabilities.