README.md

Note: work in progress. Refer to TODO.md for the roadmap.

Moon*1 lander*2 guidance software.

The goal is to land on KSP's Mun. This repo comes with a simulator to verify the guidance before attempting a landing.

• *1: Moon, or any other body
• *2: Land, or takeoff, or dv maneuver

Architecture

Overview

1. The plan (From https://blog.nodraak.fr/2020/08/aerospace-sim-1-preliminary-study/):
2. The first (Python) prototype: https://blog.nodraak.fr/2020/12/aerospace-sim-2-guidance-law/
3. The final (Rust) implementation: this repo

Implementation

GNC is implemented in Rust.

I/O will be implemented via adapters (to be able to swap between the simulator and KSP):

• Simulator: direct (Rust) function calls
• KSP: calling Python from Rust (https://github.com/PyO3/pyo3), itself calling kRPC lib (protobuf to KSP) (https://krpc.github.io/krpc/)

Building and running

Dependencies

• Rust (and a few crates)
• For the ksp subcommand:
  • KSP game
  • kRPC addon: sudo pip3 install krpc

Building

cargo build

Running

Sim:

1. cargo run -- -c conf/Apollo-descent.yaml sim

KSP:

1. Start KSP and kRPC
2. cargo run -- -c conf/Apollo-descent.yaml ksp

Available scenarios:

• Apollo descent
• Apollo ascent
• Test Mission 1: 150 m hop
  • Ascent (takeoff)
  • Descent (landing)

Plotting

You can pipe moon_lander's stdout to plotter.py: cargo run -- -c conf/Apollo-descent.yaml sim | py plotter.py. It will run MoonLander, then show some curves. The graphs will also be saved as output.png.

Example landing (Apollo scenario):

Key concepts needed for landing on the Moon

• GNC -> blog
• TGO-based guidance (PID with a predesigned trajectory is flaky) -> blog
• Take into account moon_gravity and moon_centrifugal (guidance)
• PD controller for the engine gimbal, using the spacecraft's angular position (P term) and velocity (D term) -> blog
• IMU and radar altimeter sensor fusion