Developing cities are faced with a number of challenges related to urban infrastructure and mobility. Such challenges include high traffic congestion, rapid urbanization, and increasing number of traffic accidents. Many of these cities have limited or no ongoing sources of data, which is critical to providing intelligent decision support for urban infrastructure and planning. At the same time, the roadways are highly dynamic with a wide varying number of transportation modes sharing the roads and poor road conditions. The authors’ overall goal is to determine the infrastructural and behavioral variables that are contributing to poor mobility in developing cities. In this paper, the authors present the Living Roads Framework (LRF), a framework for incorporating non-traditional data sources, specifically smartphone sensors, localized models, and to create contextually relevant transportation applications. The authors demonstrate this framework by mounting adapted smartphones on to ten garbage trucks in Nairobi, Kenya. Inertial sensors on the phones were sampling the roadways at 10-20Hz. From this data, the authors developed local models of road quality, driver behavior, and fuel consumption. The authors’ results show relationships between driver behavior and road quality for road segments in Nairobi. Moreover the authors present a preliminary analysis of the relationship between road quality, fuel consumption, and vehicle speeds.