Software experiments for solving engineering problems of the future

ryan@freestatelabs.com

Github

Twitter

**Projects:**

**rockets**

├─
Introduction

├─
Physics

├─
LQR Controller

├─
Simulation

├─
Results

├─
References & Future Work

`rockets`

To complete this project, I relied primarily on the following sources of information, as well as others I linked to throughout the website:

- Modern Control Engineering (2015) by Ogata
- Discrete-Time Control Systems (1995) by Ogata
- Introduction to Space Dynamics (1986) by Thomson

- Control Bootcamp by Steve Brunton (University of Washington)
- katkimshow by Katherine Kim (PEARS Lab, National Taiwan University)

- A Robust Control Approach for Rocket Landing (2017) by Reuben Ferrante
- Powered Descent and Landing of an Orbital-Class Rocket (2019) by Gunnell, Mansfield, Rodriguez, and Medina

In particular, the paper by Ferrante was very helpful. We essentially solved the same problem, but with different techniques.

I have a lot of work planned for this project. Some of the highlights include:

- Implementing a more robust method of selecting cost function weights for the LQR controller and using this to compute the flight of a Falcon 9 rocket
- Implementing model-predictive control (MPC) to replace the LQR controller; this will reduce some of the approximations made in the LQR controller and allow for a more optimal solution
- Incorporate noise, errors, perturbations in the simulation, such as wind/aerodynamics, state measurements, c.g. offset, etc. and determine additional methods of making controller more robust
- Studying the use of continuous-time reinforcement learning algorithms in solving this problem