Learning Reachability for Hazard Detection Avoidance in Planetary Landing

Thumbnail Image
Files
aas23_461.pdf (1.46 MB)
Author(s)
Tomita, Kento
Jo, Beyong-Un
Advisor(s)
Editor(s)
Associated Organization(s)
Organizational Unit
Organizational Unit
Daniel Guggenheim School of Aerospace Engineering
The Daniel Guggenheim School of Aeronautics was established in 1931, with a name change in 1962 to the School of Aerospace Engineering
Series
Collections
Research Publications
Supplementary to:
Permanent Link
https://hdl.handle.net/1853/74916
Abstract
Autonomous hazard detection and avoidance (HD&A) poses a stochastic perceptionaware guidance problem, where the visible surface depends on the trajectory, and the safest target locations are kept updated. For the concurrent optimization of the target and trajectory, evaluating the reachable surface under guidance constraints in real-time is critical, but it requires solving optimization problems multiple times. To bypass the optimization-based computation of the reachable surface, we propose to learn the parameterized reachable surface by a neural network, which ultimately enables the reachability-aware guidance algorithms. This paper presents the proposed parameterization method and validation results by numerical simulations.
Sponsor
Date
2023-08
Extent
Resource Type
Text
Resource Subtype
Paper
Rights Statement
Unless otherwise noted, all materials are protected under U.S. Copyright Law and all rights are reserved
Rights URI
https://rightsstatements.org/page/InC/1.0/?language=en