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

[Big_Ron]

Gary_P i still have a copy of 2.0.7 if you PM me i will E-mail it to you if you like

[Kosmic Debris]

It's a bug with Rendezvous Autopilot. I get that everytime I set the final distance below 100. Use the Rendezvous Planner, and click the "Match velocities at closest approach" button after it does the Hohmann Transfer. It will 'debug' the autopilot and show something like 21m.

Thanks - I had tried the planner as well and I find that works really, really well.

I have to admit that it's nice to have a flight computer - I can rendezvous manually, but it certainly takes the tedium out of getting in the ballpark. I still close and dock manually. It's a great tool!

[codepoet]

is anyone having an issue Docking with 2.0.8. It was rock solid with 2.0.7

I am using 2.0.8 and have identified that when I include mechjeb on a craft not all the rv105 thrusters fire in certain directions when you would expect them to, which means that I can;t translate without the craft being put into a roll I have raised an issue in github about it.

[codepoet]

Gary_P i still have a copy of 2.0.7 if you PM me i will E-mail it to you if you like

Big_Ron can I do a "me too" for 2.0.7 so I can test the issue I am having please? I will private message you my email address.

[codepoet]

I am using 2.0.8 and have identified that when I include mechjeb on a craft not all the rv105 thrusters fire in certain directions when you would expect them to, which means that I can;t translate without the craft being put into a roll I have raised an issue in github about it.

OK, my bad - my problem was being cased byt he experiemental RCS balancer being enabled.

[viperwolf]

It's a bug with Rendezvous Autopilot. I get that everytime I set the final distance below 100. Use the Rendezvous Planner, and click the "Match velocities at closest approach" button after it does the Hohmann Transfer. It will 'debug' the autopilot and show something like 21m.

Im going to try your method, thanks for sharing

[Thalestr]

So I was having a problem with the Docking Autopilot and ASAS doing nothing but sending my craft into a spin of death. I did a factory reset and this fixed this issue but about two days later I have encountered a new problem with docking.

I select my control docking port, my target port, and then hit Docking Autopilot and it will move my ship around so my control port faces the target port, lines up in front of it, and then just buggers off. Once the two ports are facing each other the Autopilot will completely stop working and my ship will just drift off into space. If I hit my movement keys while the autopilot is on it will correct itself and then shut off again. Resetting does not fix this issue. It will even move my ship to the right side of my target, line up the ports, and then it just stops working and my craft drifts off into space. It doesn't even try to hold the position.

I've tried various versions from the dev repository and I've tried reseting the settings and nothing works. I physically cannot get it to dock. Not even on crafts which I know worked previously.

[Benie]

Not a problem, guys. Happy to help.

Now I have a problem. I'm wondering if this modded version of the Mark 55 Radial Mount Engine that I "hacked", played a part in Landing Autopilot not doing the breaking burn during passing through the atmosphere.

I saw this on Eve, and on Kerbin. My Rover/Lander has 4 of these modded engines. All 4 are able to produce a thrust 500 pounds higher than one Mainsail (this was to get out of Eve's atmosphere).

The landing on Eve was good. Parachutes helped slow it down to like 3.4m/s, which the touchdown burn wasn't needed. But on Kerbin, when it slowed down to 110m/s, I popped the chutes, then about 5 seconds later, the chutes rip off. That has never, ever happened to me. Again, the breaking burn wasn't being used. It depended on the game physics to slow it down.

As I said,, I didn't see this (the chutes ripping) happen on Eve; only on Kerbin. The chutes ripping is a major concern for me.

The modded engines I have, each engine has 500 pounds of thrust. The engines had a "999999999" heat and damage tolerance. I'm thinking the fact they are so powerful, the Landing Autopilot didn't think they were needed to slow it down.

I'm using version 2.0.8.

Edited July 15, 2013 by Benie

[Gary_P]

Gary_P i still have a copy of 2.0.7 if you PM me i will E-mail it to you if you like

You have a pm with my email address.

[sarbian]

@Benie You didin't crash ?

The landing autopilot tries to keep the falling speed to what it can stop before you hit the ground. So if you have a powerful engine it won't slow down early because it can do a last second high G breaking

[rosenkranz]

The time-honored trick of telling MechJeb to keep your acceleration at terminal velocity through the ascent isn't really a very good practice, to be honest. See, the optimal ascent trajectory (in terms of delta v expended) is a zero-lift turn, which means your angle of attack is always zero. Put another way, your rocket's nose is pointed straight at the prograde marker on the navball at all times, throughout the ascent.

If you're burning such that your rocket is at terminal velocity during the early part of your ascent, you're going to end up burning way too hard. Sure, you'll save some delta v that would otherwise be lost to gravity drag, but you'll pay for it in delta v spent pushing your rocket in a direction other than along your ascent path.

So the best strategy for FAR, in my opinion, is to use MechJeb's acceleration limiter and experiment with ascent trajectories. The kinds of rockets I build tend, as a rule, to get very close to a zero-lift profile when I limit MechJeb to 18 meters per second, start turning at 500 meters, finish turning at 40 kilometers, and set the curve parameter to 60%. I got there empirically, by trying different strategies with the same rocket and dialing it in to the point where varying any parameter significantly cost me delta v. I could optimize things further, but I'm happy with where I ended up, so I'm sticking with it.

Excellent advice. So far, this and variants of this have worked beautiful.

On another note, heh, MJ is node hunting. Since installing FAR, MJ seems to have a little trouble tracking the node and is osilating for some reason (this is in space, not atmo). Anyone know what might be causing that?

[fattony4200]

I've got a quick question. I'm having an issue with the ascent autopilot where it will not throttle the liquid engines up or down. It activates the stage successfully and continues to steer but it never throttles up. It's done this on a few different rockets so far. I'm also using RemoteTech, Kethane Pack, NovaPunch, and ISA MapSat if any of those matter.

[fattony4200]

EDIT: This was supposed to be a followup but who knows where the other one went. I'm having an issue where the ascent autopilot doesn't use the liquid engines. It launches the stages correctly and steers but never throttles up. It also does not throttle down when it reaches the correct apoapsis. I have Kethane, ISA MapSat, NovaPunch, and RemoteTech installed as well.

Also, I have the box checked for Auto-Throttle of course.

Edited July 15, 2013 by fattony4200
Follow up post appeared first.

[GavinZac]

EDIT: This was supposed to be a followup but who knows where the other one went. I'm having an issue where the ascent autopilot doesn't use the liquid engines. It launches the stages correctly and steers but never throttles up. It also does not throttle down when it reaches the correct apoapsis. I have Kethane, ISA MapSat, NovaPunch, and RemoteTech installed as well.

Also, I have the box checked for Auto-Throttle of course.

Check that the engines actually have fuel supply?

[fattony4200]

Check that the engines actually have fuel supply?

Yepp. If I don't use autopilot, the engines work just fine. Only happens on some rockets and only some of the time though. Hardest problems to track down!

[mwlue]

Yep, the fps-drop gone if i removed mechjeb from the same rocket...

i think i have further nailed down where i get fps-drop. if i have radical mount rockets and the fuel is about to burn out, fps drop all the way until flame out. Any idea to fix?? Thank you.

edit: the drop is more notable if ascent AP is "off". it drop just abit less if AP is on.

[Tassyr]

Question- is it possible yet to slap MechJeb 2.0 in the code for a command pod/probe core yet? I tried back on the first release and it just sort of imploded the plugin...

[Benie]

To all that are still getting the "238y when doing 'On intercept course'" with Rendezvous Autopilot, the bug is fixed in Build #69.

3 times when I rendezvous, it worked flawlessly.

[Castun]

Question- is it possible yet to slap MechJeb 2.0 in the code for a command pod/probe core yet? I tried back on the first release and it just sort of imploded the plugin...

I've done it on all my command pods and probe cores and it works great.

[SchroedingersHat]

So the best strategy for FAR, in my opinion, is to use MechJeb's acceleration limiter and experiment with ascent trajectories. The kinds of rockets I build tend, as a rule, to get very close to a zero-lift profile when I limit MechJeb to 18 meters per second, start turning at 500 meters, finish turning at 40 kilometers, and set the curve parameter to 60%. I got there empirically, by trying different strategies with the same rocket and dialing it in to the point where varying any parameter significantly cost me delta v. I could optimize things further, but I'm happy with where I ended up, so I'm sticking with it.

It would be interesting if MechJeb had other parameters which it would apply or replace the angle with. Such as angle between prograde and vertical (rather than raw angle) or angle between surface velocity and orbital velocity, also if max acceleration were a curve (or linear interpolation) rather than a value. Or we could just do it the old fashioned way

[Itchono]

Just a quick suggestion, you should re add the DEC coordinates options for the Mechjeb Landing Autopilot.

[Oinker]

Question- is it possible yet to slap MechJeb 2.0 in the code for a command pod/probe core yet? I tried back on the first release and it just sort of imploded the plugin...

[0.20] MechJeb and RemoteTech for All

This adds MechJeb 2.x.x to all modules and probes. It also, optionally, adds RemoteTech, Kerbal Engineer, and Protractor (any combination of the four). It supports the stock modules & probes and the following add-ons: RemoteTech, MechJeb, B9 Aerospace, NovaPunch2, KOSMOS, Firespitter, H.O.M.E, Soviet Pack, American Pack, Deep Space Mission Pack, and CleverBobCat. It uses ialdabaoth's ModuleManager.dll to inject the data into the part database so it never overwrites the part.cfg files for the stock parts or the add-on parts. KW Rocketry doesn't need it because KW doesn't contain any command modules or probes.

[Troberg]

Is it possible to use MechJebb for a non-optimal planetary transfer window?

My problem: I have a multi-ship mission going to Eve (several large base parts, rovers, airships and stuff). Now, if I launch them all into orbit and use MechJeb for the transfer, they will all be scheduled at about the same time. Since they won't run if it's not the currently active flight, all but one of them will miss their launch window. Now, the first kerbonauts will not like having to wait until the last ship arrives two years later with all their toys (rovers, airships and stuff) because they'll have to wait for the next optimum window...

If I could tell MechJeb to advance or retard the launch a number of orbits, it would be much easier. For example, on a three ship launch, first launch one ship a Kerbin orbit before optimum, the second on the optimum and the last one orbit after optimum.

[magnemoe]

Had one strange issue during an Eve landing who confused the landing autopilot a lot. Ended up missing a lot and it would go worse if I did not disable it.

Look at the speed calculation

I did land from an 120 km orbit and it might be to low but anyway.

Exiting Eve again it also behaved strange then trying to circulate, it was chasing the marker was unable to circulate. This was just the upper stage so it was agile.

[BloodyRain2k]

Is it possible to use MechJebb for a non-optimal planetary transfer window?

My problem: I have a multi-ship mission going to Eve (several large base parts, rovers, airships and stuff). Now, if I launch them all into orbit and use MechJeb for the transfer, they will all be scheduled at about the same time. Since they won't run if it's not the currently active flight, all but one of them will miss their launch window. Now, the first kerbonauts will not like having to wait until the last ship arrives two years later with all their toys (rovers, airships and stuff) because they'll have to wait for the next optimum window...

If I could tell MechJeb to advance or retard the launch a number of orbits, it would be much easier. For example, on a three ship launch, first launch one ship a Kerbin orbit before optimum, the second on the optimum and the last one orbit after optimum.

You could try to launch them seperated yourself by editing the node MJ creates to be an orbit later, the next two orbits later etc and then just do course corrections. Might waste a bit of fuel but it shouldn't be that bad.

Had one strange issue during an Eve landing who confused the landing autopilot a lot. Ended up missing a lot and it would go worse if I did not disable it.

Look at the speed calculation

I did land from an 120 km orbit and it might be to low but anyway.

Exiting Eve again it also behaved strange then trying to circulate, it was chasing the marker was unable to circulate. This was just the upper stage so it was agile.

That's a bug I haven't seen in a while, try getting the dev build from here, that should fix that.