Astronomy has entered the era of time-domain surveys, in which fast and rare astrophysical transients are discovered on a nightly basis. The most informative phases of such events, including fast blue optical transients, kilonovae, the earliest stages of supernova explosions, and gamma-ray burst afterglows, unfold on timescales of minutes to hours. Capturing these fleeting signals requires robotic telescopes that can not only react rapidly but also make intelligent, fully autonomous scheduling decisions. Static queue schedulers and human-in-the-loop sequence preparation are fundamentally too slow for this purpose, creating a bottleneck that prevents telescopes from realising their full scientific potential. This thesis develops and evaluates an integrated framework for atmospheric-aware robotic scheduling, designed and tested using the Liverpool Telescope (LT) as a platform and with direct application to the forthcoming New Robotic Telescope (NRT). Four key components are addressed. First, a real-time cloud detection and prediction system based on all-sky imaging provides actionable forecasts on ten- to twenty-minute timescales. Second, an automated extinction measurement pipeline using auxiliary Skycam imagers enables continuous, differential monitoring of atmospheric transparency. Third, empirical and machine-learning models of sky brightness are constructed, incorporating lunar and geometric parameters to deliver predictive estimates of background light levels. Finally, these data products are integrated into a signal-to-noise ratio (SNR) based exposure-time system, replacing fixed exposure assumptions with dynamic, condition-dependent calculations. Together, these developments demonstrate that robotic telescopes can achieve environmental awareness and adaptive response, optimising efficiency and ensuring data quality even under variable atmospheric conditions. The results lay the foundation for a new generation of schedulers capable of autonomously deciding not only which target to observe, but also how to observe it. By eliminating the human bottleneck and integrating real-time environmental intelligence, the framework presented here positions the NRT - and robotic observatories more broadly - to meet the demands of modern time-domain astronomy.
S Buntin (Thu,) studied this question.