SkyGrid: Energy-Flow Optimization at Harmonized Aerial Intersections

Sahand Khoshdel, Fatemeh Afghah, Qi Luo

Introduction

As Urban Air Mobility (UAM) transitions from concept to reality, the management of dense aerial traffic in urban corridors presents a significant challenge. Intersections, in particular, serve as critical bottlenecks where the risk of collision is highest and energy consumption can spike due to frequent velocity adjustments. Traditional traffic management systems, such as ground-based stoplights, are ill-suited for the dynamic, three-dimensional nature of flight. Conversely, purely decentralized collision avoidance often struggles with high vehicle densities, communication latencies, and privacy concerns.

SkyGrid proposes a "rhythmic control" framework designed to coordinate unsignalized aerial intersections. By finding a middle ground between rigid centralized control and chaotic decentralized movement, the system aims to maximize throughput while minimizing the energy drain on battery-constrained aircraft.

Methodology: Rhythmic Control and Platooning

The SkyGrid framework adapts rhythmic control—originally developed for connected and automated ground vehicles—to the constraints of aerial flight. The system replaces physical signals with synchronized arrival times, dividing right-of-way into periodic intervals (e.g., North-South vs. East-West).

Figure 4: The intersection is modeled as a 3D cube mapped to a 2D intersection graph with multiple lanes ($n_c$) and defined edge lengths ($l_e$).

Central to this approach is the concept of seat-based platooning. Vehicles are assigned to specific "seats" within virtual platoons. This fixed-slot method incorporates a "headway guard band" ($l_g$), providing a safety buffer that makes the system robust against communication dropouts or minor velocity variations.

Figure 3: Spatial parameters of the virtual platooning system, including platoon size ($l_p$), guard bands ($l_g/2$), and following distances ($d_f^p$).

The core of SkyGrid is a multi-objective optimization cost function ($J$) that balances intersection flow ($f_{int}$) and average power consumption ($P_{int}$). Using a Pareto frontier, the system identifies the optimal weight ($\alpha$) to balance these conflicting goals. Trajectories are planned using high-degree polynomials to ensure smooth movement through both straight and curved paths, minimizing energy-intensive accelerations while meeting rhythmic timing constraints.

Key Insights and Results

The researchers evaluated SkyGrid using a simulated 6x6 grid intersection model. Key findings include:

  • The Energy-Throughput Trade-off: Sensitivity analysis revealed that maximizing flow often requires vehicles to take longer paths or adjust speeds in ways that increase energy consumption. SkyGrid quantifies this trade-off, allowing for dynamic adjustments based on traffic urgency or battery levels.
  • Dynamic Path Assignment: In low-traffic scenarios, vehicles are funneled through the shortest, most energy-efficient paths. As demand increases, the system redistributes traffic to outer lanes to prevent saturation.
  • Safety Impact: Increasing the guard band ($l_g$) to account for uncertainty significantly reduces capacity. For instance, increasing the buffer from 5% to 30% of a segment length caused a measurable drop in flow and a rise in power consumption.
  • Optimization Performance: The COBYLA solver consistently reached a stable state within 100 iterations, identifying an optimal weighting of $\alpha = 0.9845$ for the tested scenarios.

Conclusion and Future Directions

SkyGrid demonstrates that rhythmic control can effectively manage dense aerial intersections without requiring ultra-reliable real-time communication. While the current study focuses on a 2D planar graph, future work will expand the framework to coordinate across multiple vertical altitude layers and account for heterogeneous fleets with varying performance characteristics. Additionally, integrating this logic into vertiport management could further optimize the complex energy-flow of take-off and landing cycles.

Summary updated

BibTeX
@inproceedings{khoshdel2024skygrid,
  author = {Sahand Khoshdel and Fatemeh Afghah and Qi Luo},
  title = {{SkyGrid: Energy-Flow Optimization at Harmonized Aerial Intersections}},
  booktitle = {VTC Fall},
  year = {2024},
  doi = {10.1109/vtc2024-fall63153.2024.10757812},
}
Figure