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

[stupid_chris]

Alright I've been playing with 2.0.7 a bit and I've found a few bugs and a few things that would be useful.

- With the Rendez-Vous module, sometimes I was bringing new pieces to my space station, and the final close approach was literally at 10m. which sometimes isn't a good idea given the size of my station. Maybe giving the choice to select the distance of your final approach would be good.

-In the maneuver node planner, when I make a Hohmann transfer to a moon, it always shoots straight into the moon and I have to manually modify the periapsis. Maybe making a "transfer to a moon" planner would be useful, with the ability to select the desired periapsis.

- The landing autopilot is working pretty well, but I have been having a few bugs sometimes where it would just keep saying "burn to change low orbit plane of Xm/s" and it would stay stuck at 0m/s and never do anything. Also when selecting a target on the map, the selection cursor is in the way of some of the coordinates and we can't see them. Lastly when in the final descent, just before hitting the ground, in the case you have a large ship, it might get a tad bit off course and the autopilot tries to correct it by making 45 degree turns at 20m from the ground. Often ends in a catastrophic way, so maybe modify the code so that in the final descent it just goes straight down where it is? And adding an "emergency abort" button where it would just shoot you back up a few hundred meters if something bad is happening. Allowing to land with parachutes would be useful too.

-I noticed the ascent stats in the ascent autopilot are gone. They were quite useful so maybe bringing them back would be good. And also having a way to know your LPA if you want to save it for later would be good.

-Forgot something about the rendez-vous guidance: When the hohmann transfer to target is scheduled in a long long while and that the autopilot warps 1000x, it doesnt stop in time and usually overshoots the node of ~3mins.

That's all I think about for now although I probably forgot some things. I apologize if I repeated something someone said before, I didn't take the time to read everything that came before my post, I'm just trying to help, you have a really awesome mod with a lot of potential! Looking forward to the next update!

[HoY]

The roll field is already in the dev version. You can find a list of change in the dev version here : https://github.com/MuMech/MechJeb2/commits/dev

It's only in the Surface SASS not Orbital

Add a Roll field to the SmartASS SRF mode. (commit: 4606fc67b3e6268ee897eb32481e389a4490d49d)  gmcnew /

Edited May 10, 2013 by HoY

[sarbian]

It's relative to the surface, not surface only. Try it.

Or I don't understand what you want.

[HoY]

The smart a.s.s. module has different modes, yes surface mode will work fine while in orbit, but orbital mode has things like prograde, retrograde, +/- normal and the like. Surface mode doesn't have the rest of these options.

I simply would like to see the Roll field added to all the sub options in orbital SASS. The only reason for this is docking. Making sure your orbital constructions are all lines up properly by having them all set at Roll 0

Actually adding Roll globally inside SASS would be better IMO, that way it's available in target mode as well.

[shikarirock]

Hi! I have a problem with the latest version (2.0.7). I have installed the plugin and part correctly, yet when I attach the part to a ship, the menu on the right side of the screen doesn't show up at all. The older version works fine. I tried deleting the old mechjeb plugin but it didn't work. Here's a list of all other plugins that I currently have installed, in case any of them conflict with MJ 2.0:

http://i.imgur.com/ZVPyav7.png

Help? I would love to go ahead and delete the other plugins but I'm afraid my game wouldn't work.

EDIT: I have found the culprit by trial and error. It was the cart.dll file that was conflicting with the MJ 2.0 plugin, preventing it from running.

Maybe you should add this to the original post.

Edited May 10, 2013 by shikarirock
Problem solved.

[Frostiken]

The cart.dll is pretty far out of date...

[shikarirock]

Yeah, I guess it was left over from me copying all of my plugins into the new KSP version a couple of times.

[Paul Kingtiger]

Hello,

Is there any way with MechJeb to show the Pe and Ap of orbits planned with a maneuver node?

Currently I have to trial and error it by adjusting and hovering over the map view.

Edited May 10, 2013 by Paul Kingtiger

[HoY]

Hello,

Is there any way with MechJeb to show the Pe and Ap of orbits planned with a maneuver node?

Currently I have to trial and error it by adjusting and hovering over the map view.

There's an option in maneuver planner that lets you set your target apoapsis and/or periapsis by simply pushing a button and typing in the altitudes.

There's also an option if you are doing a capture at a planet or moon that lets you set your target periapsis, tho I don't remember off hand where it's located. It works quite well

[Mihara]

...I am not sure if this is to be considered a bug or an obscure plugin interaction problem, and won't find out for sure for a while yet, but here is my question, for what it's worth:

So I'm designing a lander for Duna, which must be able to land in any point on Duna, take off from it, and still have enough fuel left over to rendezvous with it's service module for refueling, as the mission of the day is easter egg hunting, so I'll probably have to do it a number of times. Before letting live Kerbals in, I'm running numerous simulations in a separate save file. That is, strapping it to a probe core, putting it out on launchpad, using HyperEdit to drop it in orbit around Duna and telling MechJeb 2.0.7 (latest build too) to land to see how it works.

Unfortunately it's not working. While MechJeb has no problem whatsoever landing literally on the head of a Kerbal standing on the ground when on the Mun, and lands in KSC like a pro with scant meters deviation from the launchpad itself, it behaves very bizarrely on Duna in the case described, particularly if I give it a landing location somewhere far from the equator.

First it deorbits to a highly inclined orbit, as expected, after which, for some reason, a predicted location at least a quarter of the globe away from target appears on the map.

Then it starts doing numerous course correction burns, the first typically being as high as 50m/s, which seemingly never quite stop once they go down to 0.2m/s. By that time it seemingly abandons all hope of ever getting to the target location and just settles on being 300km away - not a distance I can expect to realistically cover in a rover. Sometimes it never actually stops doing those course corrections until it runs out of fuel.

In the end I settled on landing manually with the use of parachutes and translatron, which seems to exhibit no problems and at least gets the lander down in one piece, but that required me to overbudget on fuel to ensure I can rendezvous with the service module later. Even that seems problematic as when I try that after takeoff, MechJeb appears to have issues with correctly executing maneuvers it calculated, missing warp times (always late by about 200 seconds for some reason), trying to catch drifting maneuver node targets and nonsensically spinning, and otherwise misbehaving.

What I want to know, is that a side effect of how HyperEdit sets orbits on things, and MechJeb is just trying to make the best of a bad situation, a runtime interaction of some kind, or a genuine problem MechJeb exhibits around Duna area?

And another question:

MechJeb's default ascent path curve is, as I seem to remember reading somewhere around the forum, appropriate for taking off from Kerbin. What would be ascent path curves appropriate for other planetary bodies, and what would make them appropriate?

[Nibb31]

Have you tried deorbiting from a lower altitude? I noticed that it behaves differently when deorbiting to the Mun from 150km than from 50km.

[Dexee]

Question for anyone that has already done this: For those that love the parts from MJ1 and have adapted them to MJ2, how do you go about modifying the old parts to be used w/ MJ2 without breaking the parts?

[Nibb31]

It's been explained several times in this thread. Edit the .cfg of the part like this:

Replace:

part = MuMechJebWhatever => part = Part or part = CommandPod

Add:

MODULE

{

name = MechJebCore

}

[sadron]

If it'll save the windows, at least teach me to make it so un-arbitrary x-x

[sarbian]

The smart a.s.s. module has different modes, yes surface mode will work fine while in orbit, but orbital mode has things like prograde, retrograde, +/- normal and the like. Surface mode doesn't have the rest of these options.

I simply would like to see the Roll field added to all the sub options in orbital SASS. The only reason for this is docking. Making sure your orbital constructions are all lines up properly by having them all set at Roll 0

Actually adding Roll globally inside SASS would be better IMO, that way it's available in target mode as well.

Ok, done : https://github.com/MuMech/MechJeb2/pull/89

[jackjt8]

Well this is a pain... I copied the files into the ksp folder...

loaded it up... saw all the new mechjeb parts, placed the control unit thingy and nothing! It worked fine before i changed to steam and this is the only mod i have... any help?

[Mihara]

Have you tried deorbiting from a lower altitude? I noticed that it behaves differently when deorbiting to the Mun from 150km than from 50km.

Yes, multiple times, from altitudes as low as 50km to as high as 200km. While the approach pattern was different, the symptoms of failure were more or less the same each time.

[Payload]

"Added settings window that allows switching to the MJ 1 GUI skin."

Am I the only one that can't find this or what? There is no settings window that I can find anywhere.

Problems that I have so far.

1. If you make any change to base PID values they apply to every ship. They are not saved per ship.

2. Launch profiles are not saved per ship. Again, global values are forced on every ship.

3. Landing autopilot settings are not retained ever.

Suggestions.

1. In ILS guidance control the HOLD should not be altitude specific. Hold is pretty clear in it's meaning. I want to hold! Not descend or ascend to 1000m first.

2. Do not retain information from ships named Untitled Space Craft. It's pretty clear it is a testing craft.

3. Do not retain information from my ships more than one time and name them differently. Spamming of the Plugin Data folder is out of control. Just over write the prev. We don't need a new file every time.

Alternatively, provide a save settings button that will save for that ship alone and not change default values. That way you don't have to be so indiscriminate in saving information.

4. Add a save as default button so that we can have a custom set of default values to use for our new crafts. Expose this button per window so that certain settings can be saved as default and not overwrite the whole file.

5. Expose Steering Error field in S.A.S.S. I am tired of having to make a custom window whenever I want to know if MJ is having problems steering my craft.

This business of saving random information for no reason at all is causing problems. More problems in fact than before, where only one set of settings was available and retained. The current system is not an improvement yet.

Edited May 11, 2013 by Payload

[Markarian421]

While MechJeb has no problem whatsoever landing literally on the head of a Kerbal standing on the ground when on the Mun, and lands in KSC like a pro with scant meters deviation from the launchpad itself, it behaves very bizarrely on Duna

I've had similar problems, just last night I was trying to land on Duna and after the initial deorbit burn (with a projected landing hundreds of km from the target) the lander just started spinning. I retired several times from several orbits with the same result, the landing autopilot just goes nuts. Eventually I just went back and landed manually after the initial burn, aligning the ship retrograde and burning manually with no further course adjustments, and was able to land within 0.3km of the target. It gets the initial course right but something goes completely out of control after that.

Edited May 11, 2013 by Markarian421
typo

[Deltac]

"Added settings window that allows switching to the MJ 1 GUI skin."

Am I the only one that can't find this or what? There is no settings window that I can find anywhere.

Download the latest version. He forgot to add the settings window in that particular version... and a few version afterwards.

[HoY]

Ok, done : https://github.com/MuMech/MechJeb2/pull/89

Sweet thanks

[Orphican]

If it'll save the windows, at least teach me to make it so un-arbitrary x-x

For anybody having the "MJ2 is not saving my windows" problem I'd suggest deleting MJ2 entirely (including the MJ2 plugin data folder) and reinstalling the latest dev build.

Just tried this and it seems to have fixed the problem of windows not saving. I wonder if the preset file gets corrupted somehow when KSP crashes. Anyway, thanks for the tip!

As far as I remember KSP currently deletes and recreates craft upon docking and undocking. I think this is why MJ2 windows (settings, position, and customizations) only appear to save for one craft. I could be wrong though.

[BloodyRain2k]

I've had similar problems, just last night I was trying to land on Duna and after the initial deorbit burn (with a projected landing hundreds of km from the target) the lander just started spinning. I retired several times from several orbits with the same result, the landing autopilot just goes nuts. Eventually I just went back and landed manually after the initial burn, aligning the ship retrograde and burning manually with no further course adjustments, and was able to land within 0.3km of the target. It gets the initial course right but something goes completely out of control after that.

...I am not sure if this is to be considered a bug or an obscure plugin interaction problem, and won't find out for sure for a while yet, but here is my question, for what it's worth:

So I'm designing a lander for Duna, which must be able to land in any point on Duna, take off from it, and still have enough fuel left over to rendezvous with it's service module for refueling, as the mission of the day is easter egg hunting, so I'll probably have to do it a number of times. Before letting live Kerbals in, I'm running numerous simulations in a separate save file. That is, strapping it to a probe core, putting it out on launchpad, using HyperEdit to drop it in orbit around Duna and telling MechJeb 2.0.7 (latest build too) to land to see how it works.

Unfortunately it's not working. While MechJeb has no problem whatsoever landing literally on the head of a Kerbal standing on the ground when on the Mun, and lands in KSC like a pro with scant meters deviation from the launchpad itself, it behaves very bizarrely on Duna in the case described, particularly if I give it a landing location somewhere far from the equator.

First it deorbits to a highly inclined orbit, as expected, after which, for some reason, a predicted location at least a quarter of the globe away from target appears on the map.

Then it starts doing numerous course correction burns, the first typically being as high as 50m/s, which seemingly never quite stop once they go down to 0.2m/s. By that time it seemingly abandons all hope of ever getting to the target location and just settles on being 300km away - not a distance I can expect to realistically cover in a rover. Sometimes it never actually stops doing those course corrections until it runs out of fuel.

In the end I settled on landing manually with the use of parachutes and translatron, which seems to exhibit no problems and at least gets the lander down in one piece, but that required me to overbudget on fuel to ensure I can rendezvous with the service module later. Even that seems problematic as when I try that after takeoff, MechJeb appears to have issues with correctly executing maneuvers it calculated, missing warp times (always late by about 200 seconds for some reason), trying to catch drifting maneuver node targets and nonsensically spinning, and otherwise misbehaving.

What I want to know, is that a side effect of how HyperEdit sets orbits on things, and MechJeb is just trying to make the best of a bad situation, a runtime interaction of some kind, or a genuine problem MechJeb exhibits around Duna area?

And another question:

MechJeb's default ascent path curve is, as I seem to remember reading somewhere around the forum, appropriate for taking off from Kerbin. What would be ascent path curves appropriate for other planetary bodies, and what would make them appropriate?

I've been encountering that problem too alot recently, it seems it fails for some reason to truly put the throttle to 0, doing a short warp up and down fixes it weirdly.

[dimovski]

I have fixed my landing issues by simply clicking on the body of the planet for a landing location,it would spin horribly when I tried landing a rover on the Mun with typed in coords from ISA mapsat,maybe thats the problem for you aswell?

[Mihara]

I've been encountering that problem too alot recently, it seems it fails for some reason to truly put the throttle to 0, doing a short warp up and down fixes it weirdly.

...er... "warp up and down" meaning you do what exactly?

I have fixed my landing issues by simply clicking on the body of the planet for a landing location,it would spin horribly when I tried landing a rover on the Mun with typed in coords from ISA mapsat,maybe thats the problem for you aswell?

I am pretty sure I have always clicked on the planet for target first, even when I want to directly input numeric coordinates. Because if I don't, it often will attempt to land on Kerbin regardless of where I am, with disastrous results.