Distributed Path Following of Multiple Under-Actuated Autonomous Surface Vehicles Based on Data-Driven Neural...

Area of Science:

• Robotics
• Control Systems
• Marine Engineering

Background:

• Coordinated control of multiple under-actuated autonomous surface vehicles (ASVs) is challenging, especially with unknown kinetic models.
• Achieving dynamic formations and distributed path following requires robust guidance and control strategies.

Purpose of the Study:

• To develop an integrated distributed guidance and learning control architecture for multiple under-actuated ASVs.
• To enable path following and time-varying formation control without prior knowledge of vehicle kinetic models.

Main Methods:

• A robust distributed guidance law using consensus, path-following, and an extended state observer was developed at the kinematic level.
• A model-free kinetic control law employing data-driven neural predictors with integral concurrent learning was designed for the kinetic level.
• The control architecture learns vehicle dynamics from recorded data, eliminating the need for explicit model parameters.

Main Results:

• The proposed controllers successfully achieved various time-varying formations without requiring neighboring vehicle velocities.
• The model-free control law demonstrated effective learning of unknown kinetic models using only recorded data.
• Simulation results validated the effectiveness of the robust distributed guidance and model-free control for ASVs with unknown dynamics.

Conclusions:

• The integrated distributed guidance and learning control architecture provides a robust solution for multi-ASV path following and formation control.
• The model-free approach significantly advances the capabilities of autonomous systems operating in uncertain environments.
• This method offers a practical approach for deploying multiple ASVs in complex, dynamic scenarios.