Quadrotor-Boundary and Quadrotor-Quadrotor Interactions: Analysis through Reduced-Order Simulations

Quadrotors, a class of unmanned aerial vehicles (UAVs) or drones, are gaining popularity across a variety of use cases. Many proposed applications require multiple UAVs to fly in close proximity to obstacles like surfaces or other drones, where the differences in the flowfield can disrupt the UAV’s flight path. This work investigates two near-obstacle flight considerations.
In the first, a quadrotor hovers near a ground or ceiling boundary. While the increased lift in those regions may result in energy savings, the proximity to the boundary presents a risk of crashing. I explore this tradeoff, as well as the importance of the accuracy of the boundary model, via reduced- order simulations. Simulations with example parameters suggest that quadrotors can safely fly roughly one half rotor-radius closer to the ground than we would expect without having considered the ground effect. Safe near-ceiling flight should be about one rotor-radius further from the boundary than would otherwise be considered safe. This information can provide guidance for future UAV system designers.
In the second near-obstacle flight scenario, a quadrotor flies in close proximity to another flying quadrotor. This is relevant in swarming applications, where multiple UAVs coordinate their actions, often in close proximity. I investigated one such situation: a quadrotor flies horizontally closely above or below a hovering quadrotor, and its path is altered as a result. This was tested via simulations that combine potential flow solutions with a simple quadrotor dynamics model. These simulations were compared to the results of an analogous experiment conducted by Esen Yel. A variety of flow models and fit levels were assessed. The heuristic (unfitted) simulations almost always over-predicted the moving quadrotor’s deflection. There was variety in performance across flow models: the flights through flow fields informed by particle image velocimetry performed more realistically than the other models. Iterative fitting improved accuracy to a consistent level across all models.
Together, these studies highlight the importance of local airflow on UAV path-planning. The models presented here are simple physics-based simulation methods that do not require extensive experimen- tation or computational resources. Future efforts might expand upon this in several directions; work is already underway at UVA MAE to characterize the impact of a rotor’s blade twist and size on near- boundary control dynamics.