Identification of optimal feedback control rules from micro-quadrotor and insect flight trajectories

Area of Science:

• Control Engineering
• Robotics
• Biophysics

Background:

• Optimal control theory is crucial for designing efficient feedback systems.
• Identifying controller optimality directly from experimental data remains challenging.

Purpose of the Study:

• To present a novel framework, "optimal identification," for inferring optimality conditions of feedback control laws.
• To validate this framework using both engineered systems and biological organisms.

Main Methods:

• Comparing closed-loop trajectory measurements with reduced-order open-loop models.
• Employing linear matrix inequalities to solve an inverse optimal control problem as a convex optimization.
• Applying the technique to a micro-quadrotor and freely flying Drosophila hydei.

Main Results:

• Successfully recovered performance indices used in optimal flight control design for a micro-quadrotor from simulated flight data.
• Demonstrated that insect flight control primarily regulates pitch rate through sensing and feedback.

Conclusions:

• Optimal identification provides a viable method for estimating controller optimality conditions.
• The framework offers insights into both engineered and biological control systems.