How to land a SpaceX rocket, Part II: Using AI algorithms to land the booster

Wherein Dr. Know-it-all continues examining how SpaceX actually lands their boosters, and their test vehicles like the Starship serial number 5 vehicle. What is a convex optimization algorithm? What is a loss function and how is it used to land a rocket? How do these algorithms deal with noisy, incomplete, or inconsistent data? Watch to find out!

Useful links:

[flwi]

It still boggles my mind. 🤯 How cool that they set this ambitious goal, failed (publicly) several times and now perfected it in a way that it's not even newsworthy.

[biovmr]

I knew 5 years ago when they first landed a F9 booster that it was an incredible achievement without knowing the level of math needed in any detail. I stopped my math education at differential equations (or you might say, my differential equations courses stopped ME). Excellent video and thanks for all of your in-depth topic coverages. Looking forward to the day (soon) when we will have FULL reusability of spacecraft, and can then focus on increasing payload and speeding up interplanetary travel.

[ronbyrd]

Thanks for another very informative video. Keep them coming.

[juhankaspre]

Wow nice episode Mr. When I was showing SN8 landing to my friends. There where like mmm cool, but it exploded. 😃

[okjhum]

Thanks for your great videos. I too want to know it all.
1. Which it's more difficult to achieve, a safe and accurate rocket landing, or a safe and accurate car full self driving?
2. How on earth (pun attempted) could manual calculations suffice for moon landing back in the pioneering days?

[diogenez]

What is your opinion to momentus stock at the moment acquirable per SRAC ? Thanks a lot

[michaelfink]

Actually, Elon calls the "flappy things" "flaps'. Tim Dodds, the Everyday Astronaut, calls them "Elonerons" which I think is really cool.

[hectorarcelus]

Spacex seems to had the knack for hiring really good people. The math you describe is know by a lot of people, bot the application of that knowledge it appears to be the trick or ULA or any other rocket company would accomplish the same goals. Spacex as a company applies the principle ‘okay let’s try it’.

[JamesTJoseph]

What bout Stephen Boyd’s CVXGEN software?

[ashere]

Can you tell us your thoughts about "momentus space", a strategic partner with SpaceX ?
They going Public Through Merger SPAC With Stable Road Acquisition Corp.( Nasdaq: SRAC )

[iDarekZ]

6:57 you said there is only one way. There are infinite numbers of „roads” and only „post factum” there is the possibility to evaluate if one you have chosen is the optimal one for your priorities. Very complex task.

[bnoel]

Please avoid sudden jump cuts in your videos like 1:33. They don't add much, and they just trigger a fight or flight response in my brain as it prepares to search for the skip ad button. Those who watch YouTube regularly all know the torture of the random mid-roll ads that plague YouTube these days. Please don't add to the problem by intentionally adding more needless interruptions within your own video.

[andrewpaulhart]

Thanks .... but . All I got was spaceX use AI stuff. Would have liked something with a little more explanation in there.

[Jimmy_Jones]

So Starship will have a 99.9....% of guessing the landing correct. Fun for humans.

[joecazana]

This is not really any different than landing an autonomous unmanned jet fighter on an aircraft carrier which is coming into fruition. But fighter pilots do it most of the time without failure. If there were a human pilot onboard the rocket, would he/she be able to manage such a landing?

[juanesvilladaofficial]

All I know is that Elon musk stole the idea of landing the rockets from Sheldon .