Graduate Thesis Or Dissertation

 

Informative View Planning for Autonomous Exploration in Unstructured 3D Environments Public Deposited

https://scholar.colorado.edu/concern/graduate_thesis_or_dissertations/bg257g49r
Abstract
  • This work presents a novel exploration algorithm: Informative Frontier View Exploration (IFVE). Necessary mathematical preliminaries as well as a formal problem definition are presented to provide context to understand the IFVE algorithm. The algorithm's approach and internals are explicitly detailed and separated in sub-algorithms. The IFVE algorithm optimizes an exploration-based utility function over a set of candidate goal poses sampled from a subset of frontiers at the edge of the mapped environment. Algorithm component scaling is evaluated in large simulated 3D environments and improvements to that scaling are discussed and demonstrated. The exploration and computational efficiency of the algorithm are profiled and confirmed in both simulation and in hardware experiments. The IFVE algorithm is shown to outperform state-of-the-art sample-based exploration algorithms in a 3D maze environment. The approach taken in this work is also shown to provide successful navigation for robots traverse large and often cluttered 3D obstacle environments.
Creator
Date Issued
  • 2022-04-07
Academic Affiliation
Advisor
Committee Member
Degree Grantor
Commencement Year
Subject
Publisher
Last Modified
  • 2022-07-07
Resource Type
Rights Statement
Language

Relationships

Items