Congestion Analysis for the DARPA OFFSET CCAST Swarm

TL;DR

This paper analyzes congestion issues in DARPA's swarm deployment, focusing on optimizing launch efficiency and safety through data-driven analysis and simulation validation.

Contribution

It introduces a congestion analysis framework for swarm deployment, combining pre-exercise constraint adjustments, real-world data collection, and simulation validation.

Findings

Identified key factors affecting congestion and vehicle blockages.

Validated simulation models against field data for congestion analysis.

Abstract

The Defense Advanced Research Projects Agency (DARPA) OFFensive Swarm-Enabled Tactics program's goal of launching 250 unmanned aerial and ground vehicles from a limited sized launch zone was a daunting challenge. The swarm's aerial vehicles were primarily multirotor platforms, which can efficiently be launched en masse. Each field exercise expected the deployment of an even larger swarm. While the launch zone's spatial area increased with each field exercise, the relative space for each vehicle was not necessarily increased, considering the increasing size of the swarm and the vehicles' associated GPS error; however, safe mission deployment and execution were expected. At the same time, achieving the mission goals required maximizing efficiency of the swarm's performance by reducing congestion that blocked vehicles from completing tactic assignments. Congestion analysis conducted before the final field exercise focused on adjusting various constraints to optimize the swarm's deployment without reducing safety. During the field exercise, data was collected that permitted analyzing the number and durations of individual vehicle blockages' impact on the resulting congestion. After the field exercise, additional analyses used the mission plan to validate the use of simulation for analyzing congestion.