NSF Award
2301475
The use of uncrewed aerial vehicles in urban centers will benefit society and national security through applications such as commercial/medical package delivery, public safety, military surveillance, and infrastructure inspection. However, operating small, lightweight, aerial vehicles at building-level altitudes remains a challenge, especially in the presence of strong winds that exacerbate the risk of collision with people or property. Complex wind patterns around buildings and other structures can result in sudden unanticipated disturbances, increased power consumption, and degraded vehicle performance and stability. This Engineering Research Initiation (ERI) award aims to improve the fundamental understanding of how aerial robots with limited computational capabilities can collaboratively estimate and exploit a complex urban wind field to plan safer and more efficient flight paths. New algorithms will be designed and analyzed that combine information relating the layout of buildings with physics-based simulations of wind flows to create a realistic and computationally efficient wind field estimation algorithm. A path planning technique will also be established that uses the wind field estimate to more accurately predict vehicle motion to improve safety. The knowledge generated by this project can be adopted to enable future aerial vehicles to operate in urban wind fields that are prohibitive for existing systems. Additionally, this project will include outreach activities to inspire interest in STEM among middle-school-age children and will recruit undergraduate students in the research program with focus on underrepresented groups.<br/><br/>The specific objective of this research is to establish and analyze novel algorithms that enable multiple aerial robots to collaboratively build spatial maps of complex urban wind fields and exploit them for path planning at building-level altitudes. First, a modern data-driven estimation approach, such as Gaussian process regression, will assimilate local along-path wind measurements to predict the global wind field and its associated spatial uncertainty. Each measurement will consist of a spatial position, wind velocity, and an environment feature vector that characterizes local building morphology. Hyper-parameters of the estimator will be trained utilizing computational fluid dynamics simulations of urban wind fields to produce a physics-informed wind estimator that is cognizant of the environment geometry. Second, a chance-constrained path planning algorithm will be established to minimize energy usage subject to collision-risk constraints that are quantified using the predicted wind-field, its uncertainty, and the closed-loop path following vehicle dynamics. The wind mapping accuracy, computational efficiency, and reduction in collision risk of the planned approach will be assessed through numerical simulations and small-scale outdoor flight experiments with anemometer-equipped quadrotors.<br/><br/>This project is supported by the cross-directorate Foundational Research in Robotics program, jointly managed and funded by the Directorates for Engineering (ENG) and Computer and Information Science and Engineering (CISE).<br/><br/>This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
