Robust input design for nonlinear dynamic modeling of AUV

Area of Science:

• Robotics
• Control Systems Engineering
• Ocean Engineering

Background:

• Accurate dynamic models are crucial for autonomous underwater vehicle (AUV) control and performance.
• Traditional input design methods can be sensitive to uncertainties in AUV system parameters.
• System identification relies heavily on the quality of input signals used for model development.

Purpose of the Study:

• To develop an input design approach for AUV dynamic modeling that is robust to parameter uncertainties.
• To apply a Bayesian robust design strategy for generating informative input signals for AUVs.
• To compare the performance of robust input design against traditional optimal input design.

Main Methods:

• Bayesian robust design strategy for input signal generation.
• Incorporation of constraints on AUV system inputs and outputs.
• Particle Swarm Optimization (PSO) to solve the robust optimization problem.
• Application to dynamic modeling of autonomous underwater vehicles.

Main Results:

• The proposed robust input design method successfully generates input signals for AUV dynamic modeling.
• The developed approach satisfies both robustness to parameter uncertainties and optimality criteria.
• Estimates obtained via robust input design show comparable or improved quality compared to optimal input design.

Conclusions:

• The Bayesian robust design strategy offers an effective solution for AUV dynamic modeling under parameter uncertainties.
• This method enhances the reliability and accuracy of AUV dynamic models.
• The approach provides a practical framework for designing optimal and robust input signals for AUVs.