Doctoral Dissertations

Keywords and Phrases

Disaster Management; Graph Neural Networks; Machine Learning; Reinforcement Learning

Abstract

Self-rescue during underground mine disasters is vital for miner safety. Evolving hazards and post-disaster conditions demand solutions that enable navigation under severe communication and computational constraints. Centralized systems often fail in such rugged settings, while decentralized methods—particularly Delay Tolerant Networks (DTNs), proven in battlefields and space missions—offer distinct advantages for underground applications. This research addresses five core challenges: (i) predicting miners’ next locations on low-power devices using points of interest and movement sequences; (ii) delivering timely updates on safe routes, evacuation zones, and hazardous areas; (iii) evaluating energy efficiency and comparing graph-based approaches to existing methods; (iv) enabling edge-ready frameworks, including humanoid-robot guidance, with minimal sensor input; and (v) designing scalable multi-agent self-escape strategies that adapt to evolving hazards and collective miner movement. To meet these challenges, the dissertation contributes: (a) a point-of-interest and temporally informed location predictor that improves accuracy on constrained devices; (b) a DTN-based communication protocol that combines location prediction with buffer management to enhance message delivery and conserve energy; and (c) a scalable path-planning framework that employs graph attention with hazard-aware scoring and contrastive learning to distinguish obstacles and recommend safe evacuation routes. Together, these contributions establish a foundation for real-time situational awareness, reliable communication, and adaptive path planning in underground mines. By advancing predictive modeling, communication protocols, and hazard-aware navigation, this work strengthens the feasibility of self-rescue in dynamic and high-risk underground environments.

Advisor(s)

Madria, Sanjay Kumar

Committee Member(s)

Tripathy, Ardhendu S.
Frimpong, Samuel
Chatterjee, Shubham
Nadendla, V. Sriram Siddhardh

Department(s)

Computer Science

Degree Name

Ph. D. in Computer Science

Publisher

Missouri University of Science and Technology

Publication Date

Fall 2025

Journal article titles appearing in thesis/dissertation

Paper I: Pages 11 to 50, "Miner Finder: A GAE-LSTM method for predicting location of miners in underground mines" was published in 30th International Conference on Advances in Geographic Information Systems, ACM SIGSPATIAL 2022, Seattle, Washington, USA.

Paper II: Pages 51 to 122, "Delay Tolerant Networks for Information Routing using location prediction in underground mine" was published in 25th IEEE International Conference on Mobile Data Management, Brussels, Belgium.

Paper III: Pages 123 to 194, "Disaster Information Routing using Location-prediction and Contact-graphs in Underground Mine using Opportunistic Networks" currently under review in Pervasive and Mobile Computing journal.

Paper III: Pages 195 to 237, “OGLe-Mine: Obstacle-Infused Goal-Conditioned Learning for Post-Disaster Navigation in Underground Mine" and published in the 37th International Conference on Scalable Scientific Data Management, Columbus, Ohio.

Paper IV: Pages 238 to 273, “Scalable Trajectory Prediction using Temporal Graph Attention in Underground Mines" and submitted in the 27th International Conference on Distributed Computing and Networking, Nara, Japan.

Pagination

xxiii, 286 pages

Note about bibliography

Includes_bibliographical_references_(pages 277-285)

Rights

© 2025 Abhay Goyal , All Rights Reserved

Document Type

Dissertation - Open Access

File Type

text

Language

English

Thesis Number

T 12567

Download

Share

 
COinS