Chemoreception is a critical yet underdeveloped modality for embodied agents (e.g., inspection robots). However, traditional electronic noses (e-noses) suffer from hardware vulnerability and poor anti-interference capabilities, limiting their deployment on mobile platforms. Building upon our previously proposed multi-virtual sensor array (M-VSA) based on a cascaded resonator formed by four parallel quartz crystal microbalances (QCMs), this work investigates the impact of cascade topology on sensing robustness. Specifically, we systematically compare series and parallel configurations in terms of impedance characteristics, quality factors, and fault tolerance. Theoretical and experimental analyses reveal that while the series topology achieves localized high Q-factors, it suffers from severe signal asymmetry and single-point vulnerability. In contrast, the parallel topology ensures uniform signal distribution and graceful degradation upon sensor failure—a crucial prerequisite for embodied intelligence. Adopting the parallel configuration, we extract a 24-dimensional output vector per measurement and employ a support vector machine (SVM) for data analytics. The system achieves 100% accuracy in classifying toluene/xylene risk levels across complex real-world backgrounds and 99.24% accuracy in authenticating Baijiu against adulteration. These results demonstrate the system's potential as a robust artificial olfactory module for embodied agents.

Quartz Crystal Microbalance,Cascaded Resonator,Electronic Nose,Volatile Organic Compounds