KStars AI Powered Guiding Assistant for Ekos

The AI-Powered Guiding Assistant replaces the steep learning curve of manual astrophotography guiding with a predictive, mount specific machine learning model. This project augments KStars' reactive PID controllers by utilizing an offline trained Gated Recurrent Unit (GRU) to automatically learn and counteract complex mechanical errors like gear backlash, harmonic drive signatures, and wind disturbances in real time. To ensure maximum performance on low power embedded devices like the Astroberry, the project features a native C++ inference engine built entirely from scratch using lightweight Eigen3 matrix math. This guarantees highly accurate, intelligent pulse augmentation with an absolute minimal memory footprint and without any big ML dependencies linked into the KStars ecosystem.