[PART, 1.0.2] Anatid Robotics / MuMech - MechJeb - Autopilot - Historical thread

[AlbertoKermov]

Out of curiosity, I've noticed that on 1.x there was a translate function to reach moons. But on 2.x there isn't. I've been using the Hohmann transfer nodes for this but usually leaves me on a free fall trajectory, which I then have to manually correct into orbiting. Is there a better way to do this, and/or schedule a free return trajectory?

[Starwaster]

1. 
   

Out of curiosity, I've noticed that on 1.x there was a translate function to reach moons. But on 2.x there isn't. I've been using the Hohmann transfer nodes for this but usually leaves me on a free fall trajectory, which I then have to manually correct into orbiting. Is there a better way to do this, and/or schedule a free return trajectory?

That's what you're supposed to be doing.

1. Hohmann Transfer
   
2. Fine tune (you can put in a periapsis to aim for)
   
3. Once on approach, check periapsis. If unsatisfactory, raise periapsis. Otherwise proceed to step 4
   
4. Circularize at periapsis. (for airless moons, 10km-14km)
   

[John FX]

During landing, I have noticed that if your angle is shallow the auto landing module fails to maintain altitude and lithobrakes every time. Unless it is difficult to code, could the landing module check whether it is relatively very low and very fast and so then lose most of the orbital speed before allowing the lander to start losing altitude?

It`s a much more efficient way to land for low TWR craft.

[sarbian]

Was is with "land at target" or "land anywhere" ? The code called is slightly different

[Core Xii]

The reason suggested making them still clickable with a warning about needing research is that if you just gray them out, you'll still have just as many posts but they'll be about why they're grayed out instead of missing. But if the player gets feedback from the UI as to why he2zhe can't click then they'll be less likely to come here and ignore 69,105 posts explaining that he had to research.

A simpler approach is to add a text at the bottom which reads "science unlocks more options", or something to that effect. Doesn't take as much space.

[MOARdV]

I notice that, its in a version while back. It doesn't snap to its end position, it kind of slows down towards end of roll and slowly goes into place. I didn't find it a bug, but its noticeable when a player is use to that snap into place. I know on my game I didn't have to Q or E, it got where it was going, in a matter of time.

I am aware that the ship does not snap to position. That is not I am talking about. If I select "Prograde", the ship will swing to prograde, slowing its yaw/pitch rate as it approaches, like it should do. However, once it is on prograde, the ship has a definite roll of a few degrees per second around the prograde vector. I have let it roll for a minute or more and it does not stop, even as it keeps the nose pointed to prograde. However, if I tap one of the roll keys, the roll stops. The behavior looks like what happens when I turn on SAS (the ship quickly counter rolls to reach the same roll orientation it had when I tapped the key). Same thing if I press Kill Rot for a second before switching back to Prograde: the roll stops, and it does not return. I have noticed on launch a similar thing - I typically roll the ship 90 degrees before the gravity turn so that the crew are oriented with their heads "up" relative to the surface once I reach orbit (like would happen if I could "Force Roll = 0" during launch). With the current build, I see the ship drifting through a roll during ascent such that it is oriented at least 45 degrees off of the roll=0 orientation by the time I reach orbit. With the 2.1 official release (build# whatever it is), I see MJ suppressing rolls both while holding orientation and during ascent. That behavior does not seem present in build 115, with no other changes to my vessels or the KSP environment.

[Spektyr]

When I installed the current MechJeb 2.1 on a clean install of KSP (in career mode - haven't checked sandbox) there are absolutely no autopilots. The "execute next maneuver node" functions all work, but there's no docking autopilot, no landing autopilot, and no ascent autopilot. By that I mean they do not appear in the menu whatsoever.

Any suggestions as to what's causing that?

[BigD145]

That's a feature. Research unlocks autopilots. I'm pretty sure the version history says as much in the OP.

[Spektyr]

I'm pretty sure it doesn't.

[BigD145]

"Added support for selective module unlocking for Career mode; "

[Spektyr]

... which is just vague enough that it can mean just about anything. It doesn't even say that it was implemented, only that support for implementing it was implemented.

"We made it possible to selectively unlock modules in Career mode." Great, did you do it or simply make it possible to do it? What does that mean? How was it implemented if it was in fact implemented? Does it show up in the tech tree which group unlocks a particular module?

vaguely allude to /= says as much

[Starwaster]

You were in career mode. You had modules that needed unlocking.

What exactly did you require by way of a proper explanation in the OP? Flashing neon lights?

I'm not trying to be mean btw, I'm just trying to understand how you can be better serviced.

Also, what color? My media professor always told us red was best for attracting attention.

[BigD145]

For a game that is version .22 and mods that are constantly changing, the cfg's are your readme. Know the basics of KSP cfg's. MechJeb has everything there. It is very plainly labeled "unlockTechs=____". Should some of this be on the wiki? Sure. Go do it. Make it happen.

"But I shouldn't have to..."

No. This is a free mod for an incomplete game. Just no. Don't even think about what you shouldn't have to do.

Edited November 27, 2013 by BigD145

[sarbian]

Can we avoid more than one reply each time someone ask this question ? It's only filling up the thread. Thanks

So the only feedback on the last few dev patch is the slow roll ? I'll look into it but I find it relaxing

Is the ASAS any better ? Are jet engine better handled ? Do you hate the third skin ?

@John FX : I need to see how the actual code deal with shallow angle. It may switch thru the landing steps a bit too fast.

[MOARdV]

So the only feedback on the last few dev patch is the slow roll ? I'll look into it but I find it relaxing

Is the ASAS any better ? Are jet engine better handled ? Do you hate the third skin ?

I can manage the roll in orbit, but I'd like to suppress roll during ascent (or, allow a "Force Roll = " option during ascent)

I don't know what you mean by "is ASAS better", but maneuvering towards a direction (such as prograde) seems better. No large circles. It looks like it is taking the shortest arc to the desired heading, which is good.

I don't use jet engines, so I can't provide feedback there.

I like the more compact skin, although some GUI elements still need adjusting (specific example, on the Maneuver Planner window, the text between the left/right arrow buttons for selecting the maneuver type, and the scheduling, is top justified, not center justified).

EDIT: Warp helper window also needs some UI elements adjusted to line up nicer.

Edited November 27, 2013 by MOARdV

[Starwaster]

So the only feedback on the last few dev patch is the slow roll ? I'll look into it but I find it relaxing

That's only because you haven't had a nauseous Billy-Boblie Kerman throw up on you. Jeb's mouth was open and he's not happy!

@John FX : I need to see how the actual code deal with shallow angle. It may switch thru the landing steps a bit too fast.

I think part of the problem is last minute course corrections. The closer to ground contact it is, the less willing it should be to deviate from a retrograde orientation. If it adheres to that and just focuses on retarding its descent it should be ok.

If it's trying to target a specific site it should be more willing to accept deviation the closer it is to contact. At a certain altitude it's just too dangerous to try to correct.

It IS possible but not by turning perpendicular to the ground, which invites disaster.

[Papa_Joe]

That's only because you haven't had a nauseous Billy-Boblie Kerman throw up on you. Jeb's mouth was open and he's not happy!

I think part of the problem is last minute course corrections. The closer to ground contact it is, the less willing it should be to deviate from a retrograde orientation. If it adheres to that and just focuses on retarding its descent it should be ok.

If it's trying to target a specific site it should be more willing to accept deviation the closer it is to contact. At a certain altitude it's just too dangerous to try to correct.

It IS possible but not by turning perpendicular to the ground, which invites disaster.

I would tend to agree. Maybe if you tied it to a function of the body's max air density and velocity (or maybe some percentage of Q) and disable course correction when exceeding that level. not sure how that might work, but seems like a consistent "hook" for the decision logic.

given the complexities of "fly-ability" in atmosphere, I would recommend making the parameters used be configurable in the module config. That way it can be tailored for a particular mission.

Edited November 27, 2013 by Papa_Joe

[drtedastro]

Hello. I am having some 'interesting' effects while using FAR with my MechJeb. When trying to use FAR with MJ and landing guidance, the initial orbit de-burn goes ok, but then the craft swaps axis and starts lots of small burns and eventually pushes itself back into orbit and never fully comes down. This only happens when FAR is in use and when Land at KSP or any other Target is selected. If I remove FAR and execute same landing (from F5 save of the flight, prior to landing) the landing goes as expected.

I have repeated this several times and it only does this behavior when FAR is in GameData and in KSP.

I tried looking for something like this in prior posts, but have missed it if it is there.

I have posted this on FAR as well, trying to see if anyone has any thoughts as to what might be going on....

Thanks for any help, ideas or anything...

[sarbian]

Don't use FAR

I have seen this bug without FAR so it may not be related. Don't post on their thread since it's MJ fault anyway. I'll add it to the list of landing fail I need to look at.

For the landing I'll take an other look at a patch I was sent a few month ago. It had some problem with the final phase of landing but it fixed other things so I may salvage some of it.

I wonder if It would be quicker to start the landing AP from scratch than trying to find out the 2 3 small bug that disturb it. It does not help that it works flawlessly each time I want to debug it...

#118 fix nauseous Billy-Boblie Kerman (aka the relaxing roll) and change the compact UI a bit.

[nlancaster]

FAR + Mechjeb has always cause me to land very short from aerodynamic changes.

[drtedastro]

What can I test or do to help???? I am happy to run various modes and keep data for you.

Just let me know...

And i will pull the post from FAR and not use it for now.

Thanks, as always, for you help and info...

[nlancaster]

I love FAR aero is so much better. I just live with missed landings, or manually land.

[Starwaster]

FAR + Mechjeb has always cause me to land very short from aerodynamic changes.

That's because FAR changes how drag works. (currently) MJ looks at stock values for drag but FAR sets those values to 0 and recalculates and implements drag itself.

Sarbian / Ferram have (I think?) done some work in helping the two mods talk to each other better so eventually we should have better integration. (presumably what's needed is for MJ to check for FAR's presence and query it about drag)

I know it's tough right now but the workaround is to let MJ do its initial de-orbit burn then abort once you have re-entry. Make sure landing prediction is enabled and check the map to see where you're going to land. Point prograde or retrograde as needed (really spellcheck? You know what retrograde is but not prograde?) and do a slow gentle burn to bring that number down. As you get better you need the map less and can just look at the landing AP GUI. Closer to the target and you'll need to use +/- radial burns to extend or reduce your landing distance. Normal burns if you're not quite lined up with the site.

[drtedastro]

sounds like a plan...

[Virtualgenius]

I would probably work on a new landing AP as you will have plenty of beta testers as we all love MJ and are only to happy to help out, but you should set a baseline ship to test it on