Link Search Menu Expand Document

rockets

Next: Introduction »

rockets is a test-bed and 2D simulation of landing rockets using optimal control techniques. The simulation is written in Julia and source code is located on GitHub.

rocky-landing

This project was of course inspired by the success of SpaceX landing their Falcon first stages: https://youtu.be/sf4qRY3h_eo?t=440.

The code currently includes a 3-degree of freedom simulation of a rocket, an LQR controller, and a module for visualization of the results. The LQR controller has significant limitations, but under certain conditions enables excellent control of flight and landing of “scale-model”-sized rockets. Current work is focusing on adding:

  • Properly tuned LQR controller for landing a full-size Falcon 9
  • Model-predictive control as replacement/supplement to LQR
  • Continuous-time reinforcement learning algorithms to compare against classic LQR/MPC

Table of contents


Copyright © 2022 ryan@freestatelabs