Neural sensing and control in a kilometer-scale gravitational-wave observatory

Aligning suspended optics in gravitational-wave observatories is crucial for detecting astrophysical phenomena⏤and it is also challenging, due to several environmental factors. This research implements neural-network-based sensing and control deployed at the GEO600 detector, utilizing a sensor with a convolutional neural network and long- and short-term memory, plus a deep-reinforcement-learning-based agent for enhanced alignment. The method demonstrates sensitivity improvement over traditional schemes, thus offering a practical approach for AI-based real-time control of gravitational-wave interferometers, to aid the detection of cosmic events like merging compact neutron-star binaries.