Swarm Declustering

This goal of this project is to develop a behavior to efficiently and safely decluster a set of vehicles. In other words, transitioning from a large cluster of vehicles, with inconvenient and initial position and pose, to a de-conflicted set of deployed vehicles. We define a set of vehicles to be declustered when the average risk of collision for each vehicle is sufficiently low. Part of this project has been to define a way to calculate this risk. One of the motivations for this project is to streamline deployments of multi-vehicle missions by removing the need for manual declustering.

The proposed algorithm to decluster N vehicles can be summarized in 4 steps:

Figure 1.1: Declustering in four phases.

The algorithm has so far been tested on a fleet of 8 Heron USVs on the Charles River by MIT.

Figure 1.2: The Clearpath Robotics Herons used for field testing.

The below provides an example taken from one of many field tests on the Charles River in Fall of 2024.

Figure 1.3: One of many field tests on the Charles River in the Fall of 2024.

Status:	Ongoing since July 2024
People:	Filip Stromstad, Ray Turrisi, Tyler Paine, Mike Benjamin (PI)
Sponsors:	Aker Fellowship (Norway)
	Lockheed Martin
Software:	MOOS-IvP, https://moos-ivp.org
Robots:	The Clearpath Robotics Heron USV

Document Maintained by: filipts@mit.edu
Page built from LaTeX source using texwiki, developed at MIT. Errata to issues@moos-ivp.org. Get PDF