Provably Correct Stigmergic Controllers for Enabling Emergent Behaviors in Swarms
Abstract
Project Abstract: Utilizing large numbers of aerial agents simultaneously to assist the warfighter is a long term goal. Having a multitude of agents assisting the warfighters offers advantages in situational awareness while enabling many new offensive and defensive capabilities. Situational awareness advantages include real-time visual, audio, and RF information supplied to the warfighters. This research project s concerned specifically with emergent behaviors in distributed robotic swarm systems to equip swarming UAVs with the ability to globally self-organize themselves through the use of only local interactions.While there is a significant body of work dedicated to emergent behaviors in swarm systems, we are not aware of any works that properly consider the physical aspect of robotic systems specifically the possibility of catastrophic collisions (in which all participating agents are immediately destroyed or otherwise disabled). More specifically, there do not exist any fully distributed control algorithms that can simultaneously guarantee both globally emergent behavior and collision avoidance. Since we are interested in scenarios in which collisions among agents are catastrophic, we require a new set of stigmergic algorithms that can be actually implemented on a real robotic swarm system and ensure collisions do not occur.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jan 09, 2020
- Source ID
- N000142012042
Entities
People
- Cameron Nowzari
Organizations
- George Mason University
- Office of Naval Research
- United States Navy