What do you think about automated launches?

[Jim DiGriz]

Here's a kOS script that I worked on for awhile trying to reimplement GravityTurn in kOS:

https://gist.github.com/lamont-granquist/e26279492ec020f711d2

Its a bit fugly because I was learning kOS at the same time as I was trying to write it.  The computations in the readouts are still buggy.  There's no code to bring up the attitude if the TWR drops too low.  The codestyle is a mix of how I started writing kOS and how I wound up writing kOS so its nice and inconsistent.  There's some half attempts at pulling out functions into re-usable library code that never got completed, etc...

I think where I want to go with it all next would be using kRPC instead of kOS (kOS attempts to be an "in game" computer that has some limitations -- kRPC is more of an "open up an API that lets you write your own mechjeb in some other language").

It was very cool to see my ****tty little badly tuned PID controller feathering the throttle to keep the apoapsis time pinned, and it taught me what 'integral windup' is.

EDIT: oh i didn't like the code in GravityTurn that guesses the initial angle and turn altitude, but i never found a better replacement, so that was left a bit hardcoded, you have to tweak that manually to find a good match for your rocket, etc...

Edited February 26, 2016 by Jim DiGriz

[bonyetty]

15 hours ago, Wallygator said:

^^^  This.

I think I'll take a crack at kOS.  The idea of planning and implementing a detailed launch sequence in conjunction with an over-engineered flying monstrosity seems like it would feel more like a "real space program" program!

Hmmm... Obtaining kOS now... stay tuned...

Will be doing the same.

As for an automated system I would like. An execute manover node button that only has one button and one function to execute the next node. Dividing by 2ish and staring at a green loading screen bar waiting to hit X gets a bit dull for myself. 

But now that I have been reminded of kOS and have a new interest in it, thanks to the slick video by glen.mack. My one button mod desire is probably not needed. 

[glen.mack]

11 hours ago, bonyetty said:

Will be doing the same.

As for an automated system I would like. An execute manover node button that only has one button and one function to execute the next node. Dividing by 2ish and staring at a green loading screen bar waiting to hit X gets a bit dull for myself. 

But now that I have been reminded of kOS and have a new interest in it, thanks to the slick video by glen.mack. My one button mod desire is probably not needed. 

Thank you for the kind words! kOS is probably really good for node execution, I know I saw some code for that sort of thing somewhere.

Edited February 27, 2016 by glen.mack

[Nightshift83]

On 21.2.2016 at 6:24 AM, jhousen said:

Edit: what's this mechjeb I keep hearing about?

I thought it's vanilla shipped inside this SQUAD folder, insn't it ?

[zuuhis79]

When I had played for about 600 hours of KSP and had done manually just about everything there is to do in this magnificent game, I got a little bored. Mechjeb kinda extended my interst to the game. I still enjoy playing and trying some stuff out, partly automated. I have clocked about 130 hours with mechjeb, and still going.

[322997am]

I don't want an autopilot for the stock game. It ruins the point for beginners especially. Just get bdarmory or mech jeb if you're that guy

[Madscientist16180]

If i was giving à new player advice, it would be to learn the game without automation, then install MechJeb as some "proof of concept" tool. Play loads with its fonctionallity, but learn from what you see. Then go back to non-automated.

I followed this path and can say that even if one think he is brillant and know everything about orbital mechanics, watching MechJeb WILL teach you a thing or two.

[Beowolf]

On 2/29/2016 at 3:19 PM, Madscientist16180 said:

If i was giving à new player advice, it would be to learn the game without automation, then install MechJeb as some "proof of concept" tool. Play loads with its fonctionallity, but learn from what you see. Then go back to non-automated.

I followed this path and can say that even if one think he is brillant and know everything about orbital mechanics, watching MechJeb WILL teach you a thing or two.

I did it in just the opposite order and ended up at the same place you did.  I let MJ teach me to rendezvous and land on airless worlds, but considered it training wheels and was always eager to take over manually. Now, years later, it's practically an even split between using MJ, making a maneuver node, or just winging it completely by-hand, watching the MJ instruments to see what I'm doing.

The main maneuver I still use MJ heavily for is launches. That's because I first built a standard booster for each weight class, and tuned them with dozens of launches until they practically fly into orbit without intervention. By the time I'm launching real payloads they're downright boring, so I just let MJ handle most of them on 2x speed. I like to build huge constructions in orbit, so for me the fun starts after the parts are up there.

Oh, and almost forgot Kerbin reentry! Even when I reenter manually, I always have MJ's landing autopilot projecting my landing point, because I'm still terrible at judging that myself. Without MJ, I feel proud if I even get within 200km of my target!

[Octa]

16 hours ago, Beowolf said:

The main maneuver I still use MJ heavily for is launches. That's because I first built a standard booster for each weight class, and tuned them with dozens of launches until they practically fly into orbit without intervention. By the time I'm launching real payloads they're downright boring, so I just let MJ handle most of them on 2x speed. I like to build huge constructions in orbit, so for me the fun starts after the parts are up there.

Oh, and almost forgot Kerbin reentry! Even when I reenter manually, I always have MJ's landing autopilot projecting my landing point, because I'm still terrible at judging that myself. Without MJ, I feel proud if I even get within 200km of my target!

That's exactly how i'd do it.

Ideally, i have sub-assemblies for several weight/size-classes of payload, which are designed to bring the payload in a 100km orbit and have just enough fuel left for de-orbiting the launch stage. Or the final orbit is like 100x65km, so the payload just needs a few m/s to stay in space.

I have not yet taken time to actually design such launchers, because i still have to think about some basic ideas before i start, but after the "proof-of-concept" is tested manually, i switch to automation tools like mechjeb or more recently gravity turn. My thought: I can bring 10t satellites into space. I know i can do it. I don't need to proof that to myself 100x times when setting up a satellite network. In space, i do 95% of the stuff manually, But for landing and launches, i switch to automation as soon as it gets "routine"...

[Curveball Anders]

On 2016-02-29 at 9:19 PM, Madscientist16180 said:

If i was giving à new player advice, it would be to learn the game without automation, then install MechJeb as some "proof of concept" tool. Play loads with its fonctionallity, but learn from what you see. Then go back to non-automated.

I followed this path and can say that even if one think he is brillant and know everything about orbital mechanics, watching MechJeb WILL teach you a thing or two.

This is quite the way I've done it too.

I've always tried to do things manually first, to understand what's happening. Then I've used MJ to automate repeating stuff that I've already done.

I also use MJ to test various variants of the same rocket. Like running identical launches with several different lifters to understand how different designs work without having to manually repeat launch after launch trying to keep exactly the same profile.

[pandaman]

Automation is not something that really appeals to me much, but I can see a place for it, both as a teaching tool or to perform repetitive tasks.

I know if there were stock 'create a circularisation node' and 'auto execute burn' type functions I would quite likely use them a fair bit, but It's not really something that I think is 'needed' in the stock game.

Maybe an option to automatically do a routine 'supply run' to a station would be an option, just click on the recipient and select 'Routine Fuel Delivery' then after a suitable delay and appropriate funds cost, when the target ship is not focused, the delivery arrives.  There would be no need to actually 'fly' these missions, they could be assumed to happen in the background.  I do think that the player would need to do the run, maybe three or four times, with the same launch vessel before this became an option as a 'routine' flight, this would also give a benchmark for time and cost.  Again, not something I think I would use much, but for some play styles it could be useful

[Aladran]

I Think it would be a marvelous addition, I like to fly rockets by my self, but once the testing phase is done, I like to automate the launch.

[jonrd463]

I play my game a lot like Sumghai described a page or so back. I look at it from a macro point of view, and plan out scope, overall goals, and rocket design. Some launches I do by hand, but most of the time I use Mechjeb. I do have a few rules, though. "Shakedown" flights of new rockets are always by hand at first. I want to know how it handles so I can then tweak the launch profile in MJ. I also made it a point to learn how to do everything I ask MJ to do myself, as I feel like I need to experience it and be confident that I can do it myself if MJ decides it wants to revolt. I read a lot about how there is concern in real world aviation that pilots are becoming so reliant on automated systems that the art of actual hands-on pilotage is suffering, and it's creating safety concerns. Refer to the botched landing of an Asiana 777 in San Francisco for example. So yeah, I like to know that I could do it myself, but MJ alleviates the tedium after X number of launches or maneuvers. 

I'd really like to learn kOS as that would make the automation a thing of my own creation, but I find it hard to wrap my head around some of the more advanced features. A coder I am not.