Modeling Considerations for Developing Deep Space Autonomous Spacecraft and Simulators
Stanford University, NASA Jet Propulsion Laboratory
*Corresponding author. Email: cagia@stanford.edu
Extended Abstract
Over the last two decades, space exploration systems have incorporated increasing levels of onboard autonomy to perform mission-critical tasks in time-sensitive scenarios or to bolster operational productivity for long-duration missions. Such systems use models of spacecraft subsystems and the environment to enable the execution of autonomous functions (functional-level autonomy) within limited time windows and/or with constraints. These models and constraints are carefully crafted by experts on the ground and uploaded to the spacecraft via prescribed safe command sequences for the spacecraft to execute. Such practice is limited in its efficacy for scenarios that demand greater operational flexibility.
To extend the limited scope of autonomy used in prior missions for operation in distant and complex environments, there is a need to further develop and mature autonomy that jointly reasons over multiple subsystems, which we term system-level autonomy. System-level autonomy establishes situational awareness that resolves conflicting information across subsystems, which may necessitate the refinement and interconnection of the underlying spacecraft and environment onboard models. However, with a limited understanding of the assumptions and tradeoffs of modeling to arbitrary extents, designing onboard models to support system-level capabilities presents a significant challenge. For example, simple onboard models that exclude cross-subsystem effects may compromise the efficacy of an autonomous spacecraft, while complex models that capture interdependencies among spacecraft subsystems and the environment may be infeasible to simulate under the real-world operating constraints of the spacecraft (e.g., limited access to spacecraft and environment states, and computational resources).
In this paper, we provide a detailed analysis of the increasing levels of model fidelity for several key spacecraft subsystems, with the goal of informing future spacecraft functional-level and system-level autonomy algorithms and the physics-based simulators on which they are validated. We do not argue for the adoption of a particular fidelity class of models but, instead, highlight the potential tradeoffs and opportunities associated with the use of models for onboard autonomy and in physics-based simulators at various fidelity levels. We ground our analysis in the context of deep space exploration of small bodies, an emerging frontier for autonomous spacecraft operation in space, where the choice of models employed onboard the spacecraft may determine mission success. We conduct our experiments in the Multi-Spacecraft Concept and Autonomy Tool (MuSCAT), a software suite for developing spacecraft autonomy algorithms.
Quick Links
Contributing to DARE: Deep-space Autonomous Robotic Explorer
The mission, the spacecraft, and the autonomy...
The mission: Autonomous exploration of asteroids and comets
We aim to understand what "onboard models" a spacecraft needs to autonomously execute future deep space exploration missions. Future missions will require autonomous and robust reasoning, planning, and decision-making capabilities, enabled by the right choice of onboard models. These models must adequately describe and predict how the state(s) of the spacecraft, it's subsystems, and the environment change over time and as a result of actions taken to support the reliable operation of onboard autonomy algorithms.
Selected mission rationale: Autonomous small-body exploration. Many of these future missions seek to explore worlds that are not (yet) well understood by Earth's scientists. For such missions to become viable, we require a spacecraft capable of establishing and maintaining situational awareness under uncertainty and in off-nominal scenarios (i.e., if something goes wrong). Small-body exploration missions are a promising avenue [1] to mature spacecraft autonomy algorithms and architectures at relatively low costs. Due to their abundance (nearly a million to date), small sizes, and vast distance ranges from Earth's telescopes, small bodies contain several sources of uncertainty that challenge current state-of-the-art spacecraft autonomy [3, 4] while offering opportunities for incremental improvement.
Spacecraft onboard subsystems. We analyze the various choices of models that exist for onboard autonomy in four spacecraft subsystems: power, attitude GNC, navigation, and communications. We ground our analysis in a simulated small-body exploration mission based on the Deep-space Autonomous Robotic Explorer (DARE) project at NASA JPL. A visualization of DARE is shown above. While DARE consists of all mission phases ranging from cruise to proximity operations, we focus our assessment on the cruise and approach mission phases.
The spacecraft: SmallSat design & components
To explore distant small bodies, we design an autonomous SmallSat spacecraft that consists of the four main subsystems we analyze in this work. Our spacecraft is
Powered by a total of five solar arrays;
Actuated by a solar electric propulsion thruster, several microthrusters, and reaction wheel assemblies;
Equipped with six cameras and other specialized sensors (X-ray, infrared) to collect science and engineering data [2]
The autonomy: Architecture, algorithms, and models
Models are a fundamental component of spacecraft onboard functional- and system-level autonomy.
Functional-level: Models describe the behavior of the spacecraft's subsystems (in blue), which enables us to develop functional-level autonomy algorithms for planning, prediction, estimation, control tasks.
System-level: Future missions to worlds with uncertain environments will require system-level autonomy (in orange) to develop and maintain situational awareness in nominal and off-nominal scenarios. Such system-level capabilities must aggregate information across multiple functional-level subsystems, and hence, they also rely on the quality of the underlying subsystem models for robust reasoning and decision making.
Mission simulation: Rendezvous with Bennu
The simulator: MuSCAT
We conduct a rendezvous mission to asteroid Bennu in MuSCAT: a simulation software developed at NASA JPL.
MuSCAT tightly integrates low-fidelity models across multiple spacecraft subsystems to support prototyping and simulation of mission concepts that may benefit from onboard autonomy.
Bennu-Rendezvous-MuSCAT.mp4Our simulated cruise-approach to asteroid Bennu. The spacecraft's activities are orchestrated by onboard system-level autonomy software. This software dispatches tasks to the spacecraft's subsystems for fully autonomous execution.
Citation
If you found this work interesting, please consider citing:
@article{agia2024modeling,
title = {Modeling Considerations for Developing Deep Space Autonomous Spacecraft and Simulators},
author = {Agia, Christopher and Vila, Guillem Casadesus and Bandyopadhyay, Saptarshi and Bayard, David S and Cheung, Kar-Ming and Lee, Charles H and Wood, Eric and Aenishanslin, Ian and Ardito, Steven and Fesq, Lorraine and others},
journal = {arXiv preprint arXiv:2401.11371},
year = {2024}
}
Acknowledgements
The authors would like to thank Martin Cacan and Shyam Bhaskaran for their valuable advice on this paper, and Marco Tempest for producing the Deep-space Autonomous Robotic Explorer Concept of Operations (CONOPS) video. This research was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration. This work was supported by the National Aeronautics and Space Administration under the Innovative Advanced Concepts (NIAC) program and the “la Caixa” Foundation fellowship (ID 100010434, code LCF/BQ/EU21/11890112).