Flow-Informed Path-Planning for Safe Autonomous Flight in Cities

Autonomous flight in cities is a high-stakes open challenge. The labyrinth of buildings, antennas and trees interacts with wind to produce dynamic flow disturbances that can catastrophically perturb small drones. The proximity to people and property increases the standard for reliable flight control, even in these challenging conditions.

Most autonomous flight research focuses on inner-loop, "local" control of the flier. Relying entirely on on-board sensing and control means the flier can only respond to disturbances after being hit.

We propose incorporating the surrounding flow in path-planning, such that wind disturbances can be rejected without the flier ever encountering them. First, the flow field is estimated using the city geometry, boundary conditions, sensor feedback, past estimates and reduced-order models. Then, a path is produced on this flow estimate, optimizing for safe tracking. Paths and estimates can be updated in a real-time feedback loop.

We present a proof-of-concept instance of this idea. Our flow estimator consists of a deep-learning model capable of instantly estimating mean-flow and turbulence-intensity fields around buildings. Our path-planner uses optimal control to balance speed and safety given flow estimates and a simple model of fluid-drone interaction. We test our implementation in an obstacle course subject to strong winds and compare tracking performance against a baseline without wind and a baseline without flow estimation.