Visual Tracking and Depth Estimation of Mobile Robots Without Desired Velocity Information

Area of Science:

• Robotics
• Computer Vision
• Control Systems

Background:

• Mobile robot control often requires prior knowledge of desired velocities, complicating real-time applications.
• Accurate depth estimation is crucial for robot navigation and interaction but challenging without explicit sensors.
• Existing visual servoing methods may involve complex offline computations for controller design.

Purpose of the Study:

• To develop a visual servoing approach for trajectory tracking and depth estimation in mobile robots.
• To eliminate the need for a priori knowledge of desired velocities in robot control.
• To address nonholonomic constraints and unknown depth parameters in real-time.

Main Methods:

• Utilizing multiple images from an on-board camera for reconstructing current and desired robot poses.
• Proposing an adaptive time-varying controller for trajectory tracking.
• Employing a reduced-order observer for real-time estimation of desired velocities.
• Designing an augmented update law to compensate for unknown depth parameters and identify inverse depth constants.
• Applying Lyapunov-based stability analysis and a robust data-driven algorithm.

Main Results:

• The proposed controller achieves asymptotic trajectory tracking under nonholonomic constraints.
• The inverse depth estimate converges to its actual value under persistent excitation.
• A data-driven algorithm ensures convergence of inverse depth estimation under relaxed excitation conditions.
• Demonstrated effectiveness through simulation and experimental results.

Conclusions:

• The developed visual servoing approach enables accurate trajectory tracking and depth estimation for mobile robots without prior velocity information.
• The real-time estimation of desired velocities and compensation for unknown parameters significantly enhance control performance.
• The method offers a robust and efficient solution for real-world mobile robot applications.