Sunday, March 7, 2021

Ramble: On Doing Things

I'm one of those weird folks that really loved doing research and found graduate school to be incredibly exciting. It was finally my opportunity to learn a lot about a field of interest, and really teach myself everything that I wanted to know. Every since I was a little kid, I was intrigued by the rocket landing problem. I first saw these garage built rockets competing in the Northrop Grumman Lunar Lander Challenge around 2009 on YouTube when I was 12. This really captured my imagination, and I quickly watched everything I could. I then stumbled upon SpaceX and their efforts with Grasshopper a couple years later. This led me down the Google search rabbit hole of optimal control and how these vehicles actually work. I quickly found words like Hamiltonian and Convex Optimization, but stopped there for lack of mathematical knowledge.
Well fast forward a decade and now I can derive much of that mathematics. I revisited those research papers and authors and found myself on the precipice of understanding things. So, I took it a step further and decided to make optimal control my research area. Here is a gif from results of my MS Thesis under Robert D. Braun. I primarily focused on the landing portion of Entry, Descent, and Landing (EDL). This work really isn't that novel, or that special, but I learned how to do it. I taught myself the math and implementation. Then I went out and did it. It works; pretty well too. Now, from what I have learned, I can go do some other crazy (novel?) shit. Anyway, here is my short thesis describing the algorithm.

 I am now working full time at that one company that actually lands orbital class boosters on a normal basis doing guidance and control algorithm work. Although I don't work on EDL things currently (hopefully in the future), I find this environment to be one of incredible education and adaptation. It doesn't matter if you come in a subject matter expert, you do your best, implement cool shit, test the bloody hell out of it, then fly it. Flying it is the thing that makes you the subject matter expert; because you are able to perform analysis on performance, and drive requirements/wants/needs for the next time. At other companies, they essentially require excellence a priori. But this cannot be a chicken or the egg problem, as you'd never be qualified. The difference is flying it.

A shitty unpolished thing that works is better than nothing at all. Iteration is the component that matters, and you cannot have iteration without an initial condition.

Saturday, March 6, 2021

Linear Actuator vol4

All the McMaster, eBay, and Amazon parts came in. I still have not worked out my plan to machine the rest of the parts, so they are currently 3D printed. I got access to a machine shop here in LA, which has a couple different CNC machines. They appear to be in all states if disrepair though, and it's unclear if I can really use any of them.

I was too lazy to hook up the driver, so here is a shitty video of the motor hooked up to a drill,

There are already some major flaws here... There should be two linear motion rails to minimize torsion. I also forgot to buy extra collars to keep the linear rods in place. I modified it slightly, and ordered some more components. I don't have the optical encoder in CAD right now, and still need to add the back plate. I have been pretty lazy on this one. But it seems mostly there, the rest of the stuff is small potatoes (I say now...)