Abstract

Smart cities, characterized by their integrated technology and data-driven management, are increasingly prevalent across the United States. While these urban designs primarily focus on enhancing human experiences and efficiency, they often overlook the wildlife cohabiting within these environments. This paper addresses this gap by proposing a novel system to monitor endangered bat species in smart cities, emphasizing the cohabitation of these species in critical infrastructures like bridges. This study consisted of two parts, a novel sensor deployment survey and a cloud-based system created to aggregate data that allows users to visualize bat data across time in a cloud dashboard. Traditional survey methods for monitoring wildlife are limited, providing only temporal snapshots and often missing nocturnal activities. To overcome these limitations, we introduce a passive monitoring approach using smart camera technology. This methodology not only minimizes environmental impact, but also allows for continuous and detailed observation of bat populations. We detail the process of camera selection, deployment strategies, and the development of a virtual dashboard for data visualization, specifically tailored to the needs of the Ohio Department of Transportation (ODOT). Our findings demonstrate the effectiveness of this system in providing comprehensive, real-time data, facilitating more informed decision-making for infrastructure maintenance and wildlife conservation. The deployment of this system across various sites in the state has shown promising results in improving the accuracy of wildlife monitoring and trend analysis, ultimately contributing to the sustainable coexistence of urban development and wildlife conservation in smart cities. This makes it so that any bridge can inherently become a smart bridge and generate data that can be visualized in our cloud dashboard. Our model has shown an 88% accuracy rate in determining the lack of bat presence in an area. Further work must be done to improve the model as we only could determine bat likelihood with a 12% accuracy rate using our small dataset.