This project is aimed at creating a few general purpose tools for piloting spacecraft in Kerbal Space Program. It includes full GNC (guidance, navigation, and control). Current capabilities include:
- Orbital plotting - plots the spacecrafts current orbit including drag models (assuming coefficients provided) for arbitrary bodies
- Vision based surface analysis - Uses images of the surface to determine good landing locations
- Nonlinear burn solver - (WORK IN PROGRESS) a solver that solves an initial burn for a spacecraft to place it in a desired location
- GFOLD solver (FUEL OPTIMAL LARGE DIVERT) - A convex solver for optimal landing trajectory calculation
- Trajectory tracking - A simple linear trajectory tracking control system
Example of the orbital plotting
Vision based surface reconstruction
GFOLD solver trajectory
NOTE: This project is still a work in progress and is intended as a learning tool rather than a supported library for KSP! I am happy to answer questions on it and take feature requests though!
The ksp_autopilot module contains the AutoPilot class that has associated functions for different operations. See source for details. The general usage looks like
from ksp_autopilot import AutoPilot
autopilot = AutoPilot()
autopilot.plot_orbit()
# Position of the helipad in the rotating reference frame
goal_pos = np.array((159188.42536982114, -1012.4470751636361, -578679.892709093))
autopilot.do_landing_burn(goal_position)- Move the per-ship settings (proportial gains, drag coefficients, etc.) to config files
- Currently the GFOLD only does the position optimal solve. Need to add fuel optimal and line search for final time parameter.
- It is most likely possible to dynamically solve for a controller based on the ship's dynamics (i.e. LQR) so this doesn't need to be recomputed.