Skip to content

Latest commit

 

History

History
32 lines (27 loc) · 1.9 KB

README.md

File metadata and controls

32 lines (27 loc) · 1.9 KB

LLMSat

Implementation of the proposd spacecraft controller in LLMSat: An Large Language Model-Based Goal-Oriented Autonomous Agent for Space Exploration. See paper: https://arxiv.org/abs/2405.01392

As spacecraft journey further from Earth with more complex missions, systems of greater autonomy and onboard intelligence are called for. Reducing reliance on human-based mission control becomes increasingly critical if we are to increase our rate of solar-system-wide exploration. This work explores the application of LLMs as the high-level control system of a spacecraft. Using a simulated space mission as a case study, the proposed architecture's efficacy is evaluated based on key performance indicators. The popular game engine Kerbal Space Program is leveraged as the simulation environment. This researc h evaluates the potential of LLMs in augmenting autonomous decision-making systems for future safety-critical space applications.

Setup

  1. Clone the repository
  2. Install KSP with mods as per RO & RP 1 Express Installation
    1. When selecting the compatible game versions in CKAN, select all from 1.8-1.12 as per source
  3. Install Poetry as per Poetry Instructions
  4. Install Python dependencies using:
Poetry install
  1. Copy .env.example to new file .env and populate keys
  2. Run tests to verify setup

KSP Setup

  1. Start with a fresh install
  2. Install Realism Overhaul and its dependencies through CKAN
  3. Install: KRPC KRPC-Mechjeb Parallax (for nice planet textures)
  4. (optional) Disable limited ignitions by navigating to \GameData\RealFuels\RealSettings.cfg and setting limitedIgnitions=false

Test

To run tests:

pytest -v -s