Date of Award

Spring 2025

Project Type

Thesis

Program or Major

Mechanical Engineering

Degree Name

Other

First Advisor

May-Win Thein

Second Advisor

Se Young Yoon

Third Advisor

Wayne Smith

Abstract

Rescue and recovery missions on extraterrestrial sites are an inevitable necessity, especiallywith increased extraterrestrial exploration. Robotic and crewed missions provide significant challenges due to budgetary restrictions and time constraints. Particle Swarm Optimization (PSO) techniques leverage shared swarm knowledge for solution convergence. However, traditional PSO exhibits performance degradation in dynamic scenarios. The author presents Time-Variant Particle Swarm Optimization (TVPSO), which weighs newer measurements higher than previously held values. This is achieved through implementation of a configurable comparison function as part of the survival of the fittest logical calculations. Three versions of this comparison function are evaluated (exponential, linear, and logistic) against the base PSO performance in a variety of dynamic and static source movement searches through a Monte Carlo analysis. It is shown that the TVPSO algorithm variant produces reliably superior simulation results against the base PSO framework and that an exponential base comparison function is preferred for the cases analyzed in this work. Since this implementation is intended to focus upon experimental viability, experimental verification of simulation results are conducted for the preferred TVPSO variant against three case scenarios which represent realistic search and rescue constraints. The TVPSO algorithm demonstrates implementation plausibility but sees degraded results in experimental trials due to increased iteration delays stemmed from communication protocols. Future paths to mitigate or capitalize on these challenges are presented as potential areas for improvement. However, the foundational work presented in this study produces significantly improved performance results in dynamic source rescue scenarios compared to that of classical PSO methodology without substantial increases in system complexity or cost requirements.

Download

Share

COinS