Sensing challenges in roaming sensor systems stem from the physical process to be monitored. Depending on physical process to be monitored; more than one sensor type may be needed to gain a complete view of the measured process. As an example to assess a road surface status a camera image showing surface cracks may need to be supplemented with a ground penetrating radar image to examine the roads subsurface health condition. This drives the need for heterogeneous or multi-modal sensors. To form a complete view, data across the sensing domain needs to be fused into a single coherent picture. Since roaming sensing systems survey areas of interest, it is of important to maximize the coverage in a single pass, which can be achieved by mounting multiple sensors of the same type, e.g. in an array configuration. æThis work describes the system-level design of a roaming multi-modal multi-sensor system and puts the general problems into a real-life context. Based on the case study of VOTERS (Versatile Onboard Traffic-Embedded Roaming Sensors) — a sensing system to monitor the structural heath of roads and bridge decks. VOTERS utilizes a fleet of survey vehicles to gather information of a larger geographic area (city, state). VOTERS addresses the complexity challenge by hierarchical distributing the work across fleet management, vehicle management, subsystems and sensors. Each vehicle contains multi-modal sensors mounted to the vehicle. It offers geo-spatially and temporally correlated survey data with a convenient central access and provides invaluable information for timely infrastructure investment decisions.