Sensor networks have emerged from research on low power, networked, and embedded systems and are now being applied to important scientific and societal issues. However, current sensor network architectures are unable to determine their most fundamental attribute, the quality of their data return, or sensing uncertainty. This poster will introduce networked infomechanical systems (NIMS), a new distributed embedded computing architecture that exploits a smart infrastructure and sensor diversity, is self-aware of its sensing uncertainty, and is able to adaptively reconfigure itself to maximize sensing fidelity. This will also include a description of the first NIMS prototypes and deployments in environmental science, public health, and physical security.