Date of Award
Program or Major
Master of Science
May-Win L Thein
Extraterrestrial prospecting missions will be an important first step in preparation of the harvesting and extraction of natural resources in space. Additionally, knowledge from a prospecting mission will allow the subsequent harvesting or mining mission to efficiently extract resources. Due to the risk and cost of manned space-flight, these prospecting missions will be robotic in nature, and will enable investors to weigh risk and return of a harvesting mission.
The research presented here explores a hypothetical automated prospecting mission on the Martian surface using landscapes simulated from the Mars Global Surveyor satellite. In this mission, a swarm of autonomous exploration rovers have been sent to the Martian surface to cooperatively search for some natural resource. This study examines optimization and path planning algorithms that would enable such an autonomous mission to accomplish its goals.
Particle Swarm Optimization is the search algorithm that is used to identify high concentrations of the desired resource. However, no maps exist of subterranean resource distributions on Mars. Therefore in this study (without a loss of generality), the search algorithm identifies minimum elevations in the search area.
Conflict Based Search is the multiple agent path planning algorithm used for planning the paths of each search rover. This algorithm is responsible for planning optimal paths from one search location to the next, while avoiding obstacles and other rovers in the area.
This study demonstrates the plausibility of using these algorithms for such a prospecting mission, and sets a performance benchmark for the algorithms used. This research also presents a novel derivation of the state-space form of the Particle Swarm equations. This state-space representation allows for common analysis techniques from Control Theory and Signal Processing like Root Locus and Bode plots to be used when studying the system. This study also combines a modern single agent path planning algorithm with Conflict Based Search and shows a significant improvement in performance.
Johnson, Michael Andrew, "Extraterrestrial Resource Prospecting Using Particle Swarm Optimization and Conflict Based Search" (2016). Master's Theses and Capstones. 1098.