How Distributed Control Enables Urban Drone Highways: A Technical Deep Dive

1. Background

The rapid growth of UAV traffic and aerial mobility has attracted major aerospace companies and tech firms. The speaker proposes an urban aerial highway framework based on spatiotemporal big data, enabling each drone to follow a pre‑planned route while maintaining safe separation for dense three‑dimensional traffic.

Key requirements include planning take‑off times and routes to avoid conflicts, ensuring mission completion without collisions, and handling uncertainties such as weather and no‑fly zones.

2. Air Highway Foundations: Network and Spatiotemporal Big Data

Fundamental infrastructure consists of communication (4G/5G, satellite), navigation (GNSS, radar, inertial, visual), and surveillance (ADS‑B, visual, acoustic). Spatiotemporal data supplies no‑fly‑zone maps, weather forecasts, and geographic obstacle information, which are essential for route planning and risk assessment.

Network quality (noise, latency, packet loss) directly impacts safety; a larger packet‑loss probability Θ requires a larger safety radius to compensate for position estimation errors.

3. Air Highway Design

The highway comprises three parts: model construction, algorithm design, and experimental validation.

3.1 Model Construction

Airway Network Model – Nodes and edges form a graph where routes must keep safe distances. Optimization targets include minimizing total network length and reducing risk based on population density and obstacle distribution. Methods used are morphological skeleton extraction, Delaunay triangulation (via Fermat points), and a hybrid approach that applies the appropriate method to dense or sparse regions.

Drone Model – Drones are controlled through a set of modal commands (power‑off, authorization wait, pre‑position, flight, obstacle‑avoidance, emergency landing). The goal is to define a standardized interface for future aerial traffic systems.

3.2 Algorithm Design

Centralized Control – Consists of offline planning (flight plan submission and approval) and online control (real‑time speed and altitude adjustments). Conflict detection and flow‑control are solved via Dijkstra‑based path planning and priority‑based conflict resolution.

Distributed Control – After take‑off, each drone makes autonomous decisions while following a common protocol for collision avoidance. The concept of a "Sky Highway" is introduced, with ring‑intersection nodes to replace traditional stop‑and‑wait traffic lights, and artificial potential fields to prevent deadlock.

3.3 Simulation and Experiments

A MATLAB‑based simulation environment integrates airway network data, pending drone requests, and no‑fly‑zone information. Physical testbeds with positioning facilities were also built to validate both centralized and distributed control algorithms.

4. Conclusions and Outlook

The work demonstrates a complete pipeline from data‑driven network modeling to algorithmic control and experimental verification, showing that both centralized and distributed low‑altitude traffic control can safely increase UAV traffic capacity.

Future directions include improving flight‑state prediction efficiency, efficient airport scheduling, fixed‑wing drone scheduling, robust algorithms for heterogeneous airspaces (rotor‑wing and fixed‑wing mix), development of semi‑physical simulation testbeds, and real‑world flight validation.