Bounded neural adaptive formation control of multiple underactuated AUVs under uncertain dynamics

Area of Science:

• Robotics
• Control Systems Engineering
• Artificial Intelligence

Background:

• Autonomous Underwater Vehicles (AUVs) face challenges in coordinated control due to uncertain dynamics and limited actuation.
• Leader-following formation control is crucial for complex AUV missions like exploration and surveillance.

Purpose of the Study:

• To develop a robust formation control law for underactuated AUVs with uncertain dynamics.
• To ensure bounded control torques and stable formation errors using advanced estimation and control techniques.

Main Methods:

• A multi-layer neural network is used for estimating unknown follower dynamics.
• The backstepping approach combined with a saturation function creates a bounded formation control law.
• Lyapunov-based stability analysis guarantees system boundedness and error convergence.

Main Results:

• The proposed control law requires only leader position, not velocity, simplifying implementation.
• Neural network update laws based on estimation errors improve transient performance.
• Simulations confirm the controller's efficiency in maintaining formations for multiple AUVs.

Conclusions:

• The developed control scheme effectively addresses leader-following formation control for underactuated AUVs under uncertainty.
• The approach ensures bounded control actions and guarantees stable formation errors, enhancing AUV operational capabilities.