📝 SummarXiv

Abstract

Indoor fungal contamination poses significant risks to public health, yet existing detection methods are slow, costly, and lack spatial resolution. Conventional approaches rely on laboratory analysis or high-concentration sampling, making them unsuitable for real-time monitoring and scalable deployment. We introduce \textbf{\textit{Scensory}}, a robot-enabled olfactory system that simultaneously identifies fungal species and localizes their spatial origin using affordable volatile organic compound (VOC) sensor arrays and deep learning. Our key idea is that temporal VOC dynamics encode both chemical and spatial signatures, which we decode through neural architectures trained on robot-automated data collection. We demonstrate two operational modes: a passive multi-array configuration for environmental monitoring, and a mobile single-array configuration for active source tracking. Across five fungal species, our system achieves up to 89.85\% accuracy in species detection and 87.31\% accuracy in localization under ambient conditions, where each prediction only takes 3--7\,s sensor inputs. Additionally, by computationally analyzing model behavior, we can uncover key biochemical signatures without additional laboratory experiments. Our approach enables real-time, spatially aware fungal monitoring and establishes a scalable and affordable framework for autonomous environmental sensing.