Title: Coordinating Team Tactics for Swarm-vs.-Swarm Adversarial Games

Date: Monday, May 2, 2022

Time: 2:00 PM - 4:00 PM ET

Location (virtual): https://gatech.zoom.us/j/99261253719

Laura Strickland

Robotics Ph.D. Candidate

School of Interactive Computing

Georgia Institute of Technology

Committee:

Dr. Matthew Gombolay (Advisor) — School of Interactive Computing, Georgia Institute of Technology

Dr. Jeremy Reed — Sensors and Electromagnetic Applications Laboratory, Georgia Tech Research Institute

Dr. Charles Pippin — Aerospace, Transportation, and Advanced Sciences Laboratory, Georgia Tech Research Institute

Dr. Frank Dellaert — School of Interactive Computing, Georgia Institute of Technology

Dr. Seth Hutchinson — School of Interactive Computing, Georgia Institute of Technology

Abstract:

While swarms of UAVs have received much attention in the last few years, adversarial swarms (i.e., competitive, swarm-vs.-swarm games) have been less well studied. In this dissertation, I investigate the factors influential in team-vs.-team UAV aerial combat scenarios, elucidating the impacts of force concentration and opponent spread in the engagement space. Specifically, this dissertation makes the following contributions:

(1) Tactical Analysis: Identifies the conditions under which either decentralized, implicitly-coordinating hand-scripted tactics or centralized, explicitly-coordinating hand-scripted tactics are tactically superior in swarm-vs.-swarm engagements;

(2) Learn to Leverage Tactics: Introduces and explores a deep reinforcement learning scheme to train agents to switch between the hand-scripted tactics introduced in the Tactical Analysis chapter and select coordination opportunities with teammates when such opportunities are advantageous, and test agents trained with this scheme in engagements against teams operating under one of the hand-scripted tactics; and

(3) Bio-Inspired Coordination: Investigates the force concentration implications of the guarding scheme of a specific bee species through Monte-Carlo agent-based simulations, and explains how the allocation of guard roles balances force concentration spread and preservation of the guarded resource.