Robust Predefined-Time Zeroing Neural Network for Trajectory Tracking of 4WS Mobile Robots.

Area of Science:

• Robotics
• Control Systems Engineering
• Artificial Intelligence

Background:

• Four-wheel steering (4WS) mobile robots offer superior maneuverability but present complex kinematic modeling challenges due to coupled dynamics.
• Conventional controllers often struggle with precise speed tracking and robustness in 4WS systems.

Purpose of the Study:

• To simplify the kinematic model of 4WS mobile robots by establishing an equivalence to a two-wheel steering (2WS) model.
• To develop a robust predefined-time zeroing neural network (RPTZNN) controller for enhanced trajectory tracking performance and robustness.

Main Methods:

• An equivalence relation was used to transform the 4WS kinematic model into a simplified 2WS model.
• A novel robust predefined-time zeroing neural network (RPTZNN) controller was designed using new activation functions and convergence parameters.
• A cascade control framework was implemented for robot position and orientation regulation.

Main Results:

• The RPTZNN controller guarantees predefined-time convergence, with the convergence time independent of initial conditions.
• Theoretical analysis confirmed the controller's robustness against bounded disturbances.
• Simulations demonstrated that the proposed RPTZNN controller achieves faster and more robust trajectory tracking compared to conventional methods.

Conclusions:

• The kinematic transformation effectively simplifies 4WS robot modeling without altering dynamics.
• The RPTZNN controller provides guaranteed time-constrained trajectory tracking and enhanced robustness for mobile robots.
• This approach offers a promising solution for advanced control of maneuverable robotic systems.