Safe Autonomous Flight for Characterization of Hydrodynamic Environments from Infrared (IR) Imagery
Abstract
In this project, we will develop fully autonomous flight and GPS-denied navigation solutions for an Uninhabited Aerial Vehicle (UAV) operating at night using only a monocular thermal camera and IMU. The proposed methods will enable the UAV to autonomously navigate, localize its location, and segment water and other environmental features, such as hazardous obstacles at night. Further, robust IR-camera based localization, segmentation, and mapping will facilitate accurate measurements of surface velocity and bathymetry forriverine environments estimated using image processing techniques and hydrodynamic inverse modeling applied to Infrared (IR) video imagery of the water surface, collected by a small unmanned aerial system (UAS) platform. The overachieving goal of this project is to accomplish a fully autonomous nighttime flight of a UAV in a riverine environment using a minimal set of sensors such as a monocular IR/thermal camera. We will develop new tools to establish various innovative IR-centric navigation, segmentation, and motion planning methods to enable autonomous navigation and flight at night. Such methods include a novel way to adapt an RGB- trained three-dimensional (3D) segmentation network (e.g., water or shoreline segmentation and other obstacles) to target-domain thermal imagery using online self-supervision by leveraging water texture and motion cues as supervisory signals. Approved for Public Release.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- May 15, 2023
- Source ID
- N000142312518
Entities
People
- Soon-Jo Chung
Organizations
- California Institute of Technology
- Office of Naval Research
- United States Navy