Rapid and Automated Urban Modeling Techniques for UAS Applications

Abstract

Urban models for testing UAV path-planning algorithms commonly apply simple representations using cuboid or cylinderical shapes which may not capture the characteristics of a urban environment. To address this limitation of existing urban models, this paper presents two urban modeling techniques for an unmanned aircraft flight simulation in an urban environment. The first proposed urban modeling technique is an airborne LiDAR source-based approach that incorporates machine learning algorithms to identify the number of buildings and characterize them from the LiDAR information. The second proposed urban modeling technique is an artificial urban modeling technique without any airborne LiDAR resources that applies an adaptive spacing method, an iterative algorithm to define an artificial urban environment. Unlike the LiDAR source-based approach that creates an approximated urban model, the adaptive spacing-based urban modeling algorithm generates an artificial urban environment that is visually different from a reference city, but has similar the characteristics to it. To demonstrate the two proposed urban modeling techniques, numerical simulations are conducted using open-source datasets to construct several realistic urban models.