Impact of Traffic-Following on Order of Autonomous Airspace Operations

TL;DR

This work tackles how to achieve orderly traffic in distributed autonomous airspace by creating a dynamically updating traffic-pattern map that captures the consistency and frequency of traffic flows. Each agent selects its degree of traffic-following via the traffic-following factor $k_t$, balancing unimpeded transit costs with traffic costs derived from the map. Path planning is performed with a Dijkstra-based planner over a hexagon-edge graph, complemented by a local repulsion-based conflict-resolution scheme and quantified by entropy $H(X)$. Computational results show that at low densities, following traffic yields only modest travel-time penalties while improving order, whereas aggressive traffic-following increases travel time with limited additional order, demonstrating a tunable trade-off between efficiency and airspace order. The framework supports dynamic adjustment of $k_t$ to balance objectives and provides a basis for extending to uncertain or high-density regimes.

Abstract

In this paper, we investigate the dynamic emergence of traffic order in a distributed multi-agent system, aiming to minimize inefficiencies that stem from unnecessary structural impositions. We introduce a methodology for developing a dynamically-updating traffic pattern map of the airspace by leveraging information about the consistency and frequency of flow directions used by current as well as preceding traffic. Informed by this map, an agent can discern the degree to which it is advantageous to follow traffic by trading off utilities such as time and order. We show that for the traffic levels studied, for low degrees of traffic-following behavior, there is minimal penalty in terms of aircraft travel times while improving the overall orderliness of the airspace. On the other hand, heightened traffic-following behavior may result in increased aircraft travel times, while marginally reducing the overall entropy of the airspace. Ultimately, the methods and metrics presented in this paper can be used to optimally and dynamically adjust an agent's traffic-following behavior based on these trade-offs.

Figures (9)

• Figure 1: An ownship can decide the degree to which it follows traffic, and which traffic it follows, as it progresses towards its destination.
• Figure 2: We partition the airspace into a hexagonal grid, where each cell is assigned edges numbered from 1 to 6. This numbering system facilitates the indexing of the cost matrix associated with each cell.
• Figure 3: (a). Illustrating different trajectories that a vehicle can take to get from one edge in the grid to another. (b). Using Dijkstra's algorithm, an aircraft computes the total cost for each possible path, starting from the initial edge. (c). The least costly edge that can lead to another hexagon is picked for the next cost step calculation. (d). Again, the least costly edge that has not been explored, starting from the base edge, is picked for the next cost step calculation. This continues until the next edge to be explored is the destination edge.
• Figure 4: (From left to right) Progression of a head-on scenario diversion where aircraft enter from opposite edges of a hexagon and head towards the edge the other aircraft entered from.
• Figure 5: Demonstrating the effect of $k_t$ for a single ownship: Aircraft are introduced into the airspace at staggered intervals, originating from the bottom-left with destinations in the top right. Only the last aircraft (orange) is designated to be an ownship here, i.e., has the option to follow or not. (a). $k_t$ = 0, i.e., no traffic-following is permitted. (b). $k_t$ = 3, i.e., traffic-following is permitted.