SPACESHIP: synthesizable parameter acquisition via closed-loop exploration and self-directed, hardware-aware intelligent protocols for autonomous lab

Autonomous laboratories hold great promise for accelerating material discovery but are often restricted by static, predefined experimental constraints. We present SPACESHIP, an AI-driven framework for dynamic, constraint-free exploration of synthesizable regions in chemical parameter spaces. SPACESHIP integrates probabilistic models and uncertainty-aware acquisition to learn from both successful and failed experiments to adaptively refine feasible synthesis conditions. A central feature is the Autopilot strategy, which dynamically selects between models and achieves up to 4.3× faster exploration than does random sampling. When applied to gold (Au) nanoparticle (NP) and nanorod (NR) synthesis, SPACESHIP identified synthesizable regions with 90% accuracy in only 23 experiments and reached 97% accuracy within 127 experiments—substantially fewer than the 625 experiments required to establish the ground truth. This method also enables the ternary classification of Au NRs on the basis of their optical properties, revealing distinct growth regimes. Compared with literature-based maps, SPACESHIP expanded the validated synthesizable space by a factor of 8 for NPs and 4 for NRs, supporting scalable discovery and mechanistic insight.