Remote sensing using drone and machine learning for computation of rooftop solar energy potential

The decision to use a renewable form of energy generation was unavoidable due to the rapidly rising demand for energy and to achieve sustainability goals. Rooftop solar photovoltaic energy generation utilizes space for panel installation effectively and with the least amount of transmission losses. Using conventional methods, determining the appropriate rooftop space for solar PV systems is a laborious operation. The use of orthophotos and aerial imageries has shown to be effective for mapping effective roofs and taking accurate measurements. Utilizing automated machine learning approaches would make it easier to obtain the footprints for large geographical areas. The data collected a remote sensing imaging drone is used in this study to create an orthophoto and digital surface model. The development of deep learning and growing computational power have benefited remote sensing images and geospatial analysis. This research experiment is to extract building roofs appropriate for solar PV installations in a village setting. For the entire research region chosen, the model predicts against manually created ground truth maps. The potential for electricity generation from the existing, vacant building roofs is demonstrated via solar rooftop energy assessment. Such outputs would be used by authorities for complex spatial analysis in decision-making in coming days. The models created are intended to automatically generate vital spatial data sets incorporating geographic as well as geometric properties.