Methods for Multiloop Identification of Visual and Neuromuscular Pilot Responses

Area of Science:

• Human-computer interaction
• Systems neuroscience
• Control theory

Background:

• Simultaneous identification of neuromuscular and visual responses is crucial for understanding pilot control.
• Conventional cross-spectral density methods rely on a noninterference hypothesis that is often unmet in practice.
• This limitation hinders accurate estimation in real-world experimental designs.

Purpose of the Study:

• To propose novel identification methods for estimating neuromuscular and visual responses in a multiloop pilot model.
• To address the limitations of the conventional cross-spectral density technique.
• To provide reliable estimation methods applicable even when the noninterference hypothesis is violated.

Main Methods:

• Development of two new identification methods: one based on autoregressive models with exogenous inputs (ARX), and another combining cross-spectral estimators with frequency domain interpolation.
• Mathematical justification for the necessity of the noninterference hypothesis in traditional methods.
• Validation through offline simulations and application to experimental data from a closed-loop control task.

Main Results:

• The classic cross-spectral density method fails when the noninterference hypothesis is not met.
• The two proposed methods (ARX-based and frequency domain interpolation) provide reliable estimates.
• Simulations and experimental data confirmed the superiority of the proposed methods over the classic approach when assumptions are violated.

Conclusions:

• The proposed identification methods offer a robust alternative to the classic technique for estimating human neuromuscular and visual responses.
• These methods enable accurate simultaneous estimation in scenarios where traditional approaches fail.
• The findings are significant for advancing the understanding of human-pilot interaction and control systems.