ORCID

0000-0003-2875-8414

Keywords

Active Vision, Social Navigation, Unstructured Terrain

Abstract

Autonomous mobile robots increasingly operate in complex real-world environments characterized by uneven terrain, dynamic obstacles, and human presence. This dissertation addresses the fundamental challenge of developing adaptive navigation strategies that enable robots to safely and effectively traverse unstructured outdoor environments including construction sites, agricultural fields, and disaster areas while maintaining social compliance around people. We advance this capability through five complementary contributions that collectively target dynamic, unstructured, human-occupied environments. First, we demonstrate evolutionary optimization of legged mobility for uneven terrain. Second, we propose using a spatiotemporal autoencoder to learn the temporal dynamics of moving people and objects. Third, we introduce DUnE, a versatile simulation framework that supports social navigation scenarios for systematic evaluation of navigation algorithms across diverse terrains. Fourth, we develop a social navigation system that fuses spatiotemporal attention-based human occupancy prediction with model-based reinforcement learning, delivering traversability-aware, socially compliant control. Fifth, we extend this system with active vision using a wide-angle camera, jointly learning viewpoint selection and locomotion to proactively acquire informative views of nearby pedestrians and obstacles under partial observability.  Collectively, this body of work advances socially compliant navigation in unstructured terrain navigation for safety-critical settings such as agriculture, search and rescue, and construction environments where traditional approaches often fail because of social constraints, terrain variability, and dynamic hazards.

Completion Date

2026

Semester

Spring

Committee Chair

Gita Sukthankar

Degree

Doctor of Philosophy (Ph.D.)

College

College of Engineering and Computer Science

Department

Computer Science

Format

PDF

Document Type

Dissertation

Identifier

DP0053135

Download

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.