Title: Intelligent Perception for Characterizing and Navigating Small Celestial Bodies

Date: Wednesday, October 11, 2023

Time:  3 PM - 5 PM EST

Location: Montgomery Knight Building, Room 317

Virtual Link: Microsoft Teams

Virtual Meeting ID: 251 716 421 883

Passcode: XvzSMH

Travis Driver

Robotics Ph.D. Student

Dynamics and Control Systems Laboratory

Georgia Institute of Technology

Committee

Dr. Panagiotis Tsiotras (Advisor), Daniel Guggenheim School of Aerospace Engineering, Georgia Tech

Dr. John Christian, Daniel Guggenheim School of Aerospace Engineering, Georgia Tech

Dr. Frank Dellaert, School of Interactive Computing, Georgia Tech

Dr. James Hays, School of Interactive Computing, Georgia Tech

Dr. Katherine Skinner, Department of Robotics, University of Michigan

Abstract

Missions to small celestial bodies rely heavily on optical feature tracking for characterization of and relative navigation around the target body. Current state-of-the-practice approaches rely on extensive human-in-the-loop verification and high-fidelity a priori information to achieve accurate results. Instead, this thesis explores the application of modern photogrammetic techniques and intelligent perception methods to increase the autonomous capabilities of missions to small bodies. First, this thesis details AstroVision, a large-scale dataset comprised of 115,970 annotated, real images of 16 different small bodies captured during past and ongoing missions. We employ AstroVision to conduct an exhaustive evaluation of both handcrafted and data-driven feature detection and description methods and for end-to-end training of a state-of-the-art, deep feature detection and description network and demonstrate improved performance on multiple benchmarks. Next, this thesis proposes a novel approach that incorporates planetary surface reflectance models into a feature-based Structure-from-Motion (SfM) system to estimate the surface normal and albedo at detected landmarks to improve surface and shape characterization of small celestial bodies from in-situ imagery.