Iterative Learning Control for Pareto Optimal Tracking in Incompatible Multisensor Systems.

Area of Science:

• Control Engineering
• Optimization Theory
• Signal Processing

Background:

• Multisensor systems often face tracking conflicts due to differing sensor inputs.
• Resolving these conflicts is crucial for effective system operation.
• Existing methods may not adequately address incompatible multiobjective tracking problems (IMOTP).

Purpose of the Study:

• To propose a novel iterative learning control strategy for resolving sensor conflicts in multisensor systems.
• To address the incompatible multiobjective tracking problem (IMOTP) as a multiobjective optimization problem (MOOP).
• To ensure that proposed control updates lead to Pareto improvements and convergence.

Main Methods:

• Formulating the sensor conflict problem as a multiobjective optimization problem (MOOP).
• Elaborating on the Pareto optimal solution (POS) set for the MOOP.
• Deriving gradient-based update directions for Pareto improvement.
• Establishing a learning control algorithm for convergence to a POS.

Main Results:

• The proposed iterative learning control strategy effectively resolves tracking conflicts.
• Each update in the algorithm guarantees Pareto improvement.
• The system converges to a Pareto optimal solution (POS).
• Simulations validate the theoretical advancements.

Conclusions:

• The developed iterative learning control strategy provides an effective solution for multisensor tracking conflicts.
• This approach enhances system performance by resolving incompatibilities.
• The method ensures convergence to optimal solutions, validated by simulations.