ORCID

0009-0004-2956-4296

Keywords

Robotics, Control, AI, Multi-Agent Reinforcement Learning

Abstract

Multi-agent target encirclement is a formation control problem concerned with driving a team of agents to maintain a circular trajectory around a static or moving target. This thesis presents the Secure, Heterogeneous and Rotational Knowledge for Swarms, Version 2 (SHARKS V2) algorithm, a hybrid approach to decentralized multi-agent target encirclement that combines Multi-Agent Proximal Policy Optimization (MAPPO), a model-free reinforcement learning algorithm, with Control Barrier Functions (CBFs), a model-based safety constraint mechanism. SHARKS V2 builds upon the foundations of the original SHARKS algorithm, extending it with safety constraints through CBF-based obstacle avoidance. We evaluate the performance of SHARKS V2 against existing model-based and model-free baselines, and study the effects of our proposed deployment strategy. Our results demonstrate that SHARKS V2 outperforms existing baselines while significantly reducing unsafe behaviors.

Completion Date

2026

Semester

Spring

Committee Chair

Chinwendu Enyioha

Degree

Master of Science in Computer Engineering (M.S.Cp.E.)

College

College of Engineering and Computer Science

Department

Electrical and Computer Engineering

Format

PDF

Document Type

Thesis

Identifier

DP0053246

Release Date

5-15-2027

Download

Available for download on Saturday, May 15, 2027

Share

COinS
 

Accessibility Statement

This item was created or digitized prior to April 24, 2027, or is a reproduction of legacy media created before that date. It is preserved in its original, unmodified state specifically for research, reference, or historical recordkeeping. In accordance with the ADA Title II Final Rule, the University Libraries provides accessible versions of archival materials upon request. To request an accommodation for this item, please submit an accessibility request form.