Laboratory Demonstration of Optimal Identification and Control of Tip-Tilt Systems

TL;DR

This paper demonstrates the successful implementation of optimal LQG control for tip-tilt wavefront correction on a testbed, showing improved wavefront stability through real-time control and disturbance modeling.

Contribution

It introduces a physics-informed LQG control approach for tip-tilt correction, considering hardware delays and vibration effects, validated on the SEAL testbed.

Findings

LQG control reduces wavefront RMS error more effectively than integrator control.

Characterization of bench dynamics informs control model accuracy.

Real-time control implementation demonstrates practical feasibility.

Abstract

We present the results of testing optimal linear-quadratic-Gaussian (LQG) control for tip and tilt Zernike wavefront modes on the SEAL (Santa cruz Extreme AO Lab) testbed. The controller employs a physics model conditioned by the expected tip/tilt power spectrum and vibration peaks. The model builds on similar implementations, such as that of the Gemini Planet Imager, by considering the effects of loop delays and the response of the control hardware. Tests are being performed on SEAL using the Fast Atmospheric Self-coherent camera Technique (FAST), and being executed using a custom Python library to align optics, generate interaction matrices, and perform real-time control by combining controllers with simulated disturbance signals to be corrected. We have carried out open-loop data collection, characterizing the natural bench dynamics, and have shown a reduction in RMS wavefront error due to integrator control and LQG control.