This paper envisions a future where autonomous robots perform scientific experiments in an open, trustworthy, and transparent manner, while also being accurate and repeatable. To achieve this, we present a semantic execution tracking framework and a cloud-based platform, the AICOR Virtual Research Building (VRB), which ensures transparency and reproducibility of automated experiments by co-recording sensor data with semantically annotated robot belief states. VRB is used to share, replicate, and validate robot task execution at scale. These tools integrate deterministic execution, semantic memory, and open knowledge representation to enable reproducible robot-driven science, laying the foundation for autonomous systems to participate in scientific discovery.