Jump to content

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

r4m0n

By r4m0n
in KSP1 Mod Releases

 Share

Recommended Posts

sarbian

sarbian

Posted
Posted (edited)

Galane : for now it only change the dV simulation. I may push it into the auto-staging later but not before .23. I don't want to change the staging code just before a release.

codepoet : I'll work on it this weekend if I don't go out.

MaxP : sorry, had to be sure :P Those case are a bit hard to duplicate and test. It happened to me while playing but I never get it while testing... If your ship don't use too many part I could use a save/quick save where the bug is here. I should add a big debug windows so player could take a screenshot of it and I would know a bit more.

I may find the bug at one point when I take the time to check all the logic and math of the landing AP. The Duna problem is one of those case where I am sure there is a small error in the math that make it works fine on Kerbin but not on other planet with atmo.

And testing the landing AP is sooo boring. 5 to 10 min of doing nothing :(

The others : I understand you don't want to read a 420 page thread but please read the last 2 3 page before asking a question. This one is answered at least once every 2 page.

I'll add some text in game to explain it, so we can lower the amount of traffic this question generate for the .23 release.

And about the "I can't unlock the part" : it's a KSP bug and I can't do anything about it. Most of the time it seems to happens when you have multiple copy of one part with the same name and/or you edit the science tree after unlocking a part. Check you install (again, I know :( ) and make sure you use the MJ from the OP of this thread and not one from spaceport and that you have no other mod conflicting.

Edited by sarbian
Link to comment
Share on other sites
MaxP

MaxP

Posted
Posted
And testing the landing AP is sooo boring. 5 to 10 min of doing nothing

Yeah... ))

Those case are a bit hard to duplicate and test

I think it may be a problem with tangential speed (don't sure it correct in english)) - the inclination of entry orbit was near 30 deg, so the target crossed trajectory ground projection from left to right with ~30 degrees angle.

And, probably it is imposed with planet's atmosphere parameters - the jumping blue marker jumps from the end of the trajectory (once I decided to watch what will happens - so meeting the ground on ~900 m/s was happened)) to the point near the correct (if the auto-reentry would be correct).

Link to comment
Share on other sites
codepoet

codepoet

Posted
Posted (edited)
Guys, tell me please...

I cannon land on Duna. When I press Land at target, some actions may happens (sometimes not), but, anyway, I've got "Reentry simulation timed out" in autopilot status. The vessels stays on orbit, nothing is happening.

I use last dev build (128) and RT2 (but vessel is manned, and controllable, with "Local control").

This could be to do with changes in the parachute targeting patch that was pulled in #128. In order to increase the precision of the simulation the simulation's time step is now variable, depending on how long it takes, (with a view to being able to do at least 5 simulations per second) If you have a fast computer then you will get a shorter timestep (and hence a more accurate simulation) and as you get closer to landing the simulation comes to an end sooner, and so takes less time and so becomes more accurate. However, now I think about it, the simulation times out after a certain number of time steps, so if your computer is fast, and the timestep is small then it could time out before the simulation lands / aerobreak completes. I did my development on a rubbish old laptop as I do not have lots of money for a swankie fast machine, so I have not been able to test that situation. If it is all my fault - sorry guys :(

Edited by codepoet
Link to comment
Share on other sites
Starwaster

Starwaster

Posted
Posted
Guys, tell me please...

I cannon land on Duna. When I press Land at target, some actions may happens (sometimes not), but, anyway, I've got "Reentry simulation timed out" in autopilot status. The vessels stays on orbit, nothing is happening.

I use last dev build (128) and RT2 (but vessel is manned, and controllable, with "Local control").

Cannon landing? Was that a typo or is that a euphemism for 'lithobraking'?

Link to comment
Share on other sites
sarbian

sarbian

Posted
Posted
I think it may be a problem with tangential speed (don't sure it correct in english)) - the inclination of entry orbit was near 30 deg, so the target crossed trajectory ground projection from left to right with ~30 degrees angle.

And, probably it is imposed with planet's atmosphere parameters - the jumping blue marker jumps from the end of the trajectory (once I decided to watch what will happens - so meeting the ground on ~900 m/s was happened)) to the point near the correct (if the auto-reentry would be correct).

If I understand you were doing some kind of aerobraking + landing ? You did not start from a stable orbit ? That would be something to add to my test

Link to comment
Share on other sites
MaxP

MaxP

Posted
Posted (edited)
I think he meant can not land

Trolls ))

If I understand you were doing some kind of aerobraking + landing ? You did not start from a stable orbit ? That would be something to add to my test

Orbit is stable, I tried from 40-45 to 200-300 km.

For me it happens with any vessel landing on Duna.

Edited by MaxP
Link to comment
Share on other sites
Tokay Gris

Tokay Gris

Posted
Posted

I have a problem landing on Duna.

As was mentioned before, the landing predictions jump around. It sometimes shows a "predicted orbit after aerobraking" of 120 x -10.

I would not mind if it did actually land. But it doesn't.

"Landing at target" works like this: normal deorbit burn. Then a course correction that does a "him go thataway!"

I just downloaded the latest version of MechJeb (I hope) and tried again. Same result. Broke of the test at an orbit of 80 x 1.000.

At first, it seemed normal. Deorbit burn, a bit off (around 150km from target). And then the course correction, which gets me probably on a course to Alpha Centauri. But not on the ground on Duna.

Link to comment
Share on other sites
Tokay Gris

Tokay Gris

Posted
Posted
If you are using FAR the landing guidance won't work

I am not using FAR. Just threw out most of the mods I am using.

Almost the same result. Slightly different, though.

Breaking burn, jumping predicted landing sites. Then it says "course correction" but does nothing for a while. And then.... a course correction aiming for the center of the galaxy.

Link to comment
Share on other sites
Tokay Gris

Tokay Gris

Posted
Posted
I've had various dev build versions flip out and burn for the sky during landing. They've always gone to proper behavior after reloading the last quicksave then setting up the landing again.

Tried that. Countless times. Same result.

I now do have a "solution" of sorts.

It seems Mechjeb doesn't like orbital heights below 120 km.

I did set up an orbit of 200 km and attempted again. It DID land. About 8 km off. Which is not that bad.

Ok, so it did work in the end. BUT the landing procedure was funky. Used a LOT of fuel. Dropped down the last few meters on the 'chutes. Will test again, this time with a different site.

Link to comment
Share on other sites
Galane

Galane

Posted
Posted

Huh. Weird stuff. Just for the heck of it, hyperedit something to a 100KM orbit around Eve and try to land. For Eve, but only for Eve, I'd like MechJeb to really run through the fuel. I'm having issues with it not slowing down quite enough unless the ship is well over 1 on TWR for Eve.

On any body, it should be possible to land a ship with any TWR greater than 1, if it has enough fuel. Taking off and getting back to orbit is another matter for higher gravities, but the landing should always be possible with TWR>1+enough fuel to burn.

Link to comment
Share on other sites
Tokay Gris

Tokay Gris

Posted
Posted
Huh. Weird stuff. Just for the heck of it, hyperedit something to a 100KM orbit around Eve and try to land. For Eve, but only for Eve, I'd like MechJeb to really run through the fuel. I'm having issues with it not slowing down quite enough unless the ship is well over 1 on TWR for Eve

Really? I did my sandbox testing for Eve just like that. A landing fully without engines. I need all the fuel on the ground and even with that, I need to drive to an altitude of about 4000m.

It worked just fine on Eve... but that was not the current build.

Link to comment
Share on other sites
Tokay Gris

Tokay Gris

Posted
Posted

Just did four more tests.

The first from before was roughly acceptable. 8 km off. Burning all fuel on descent.

The second one was 300 km off.

The third one said "warping to start of braking burn" but did not warp (yes, autowarp was on). It might have landed, but was waaaaaay off.

The fourth was from an orbit of 50 km. (The others 200km) "Him go thataway!" Alpha Centauri again. So, MechJeb flips out if orbit is too low.

I now tried of everything I could think of. It seems, MechJeb does not like Duna.

One last try with an orbit of 300.

Link to comment
Share on other sites
Galane

Galane

Posted
Posted

I had some issues landing on Duna. http://forum.kerbalspaceprogram.com/threads/58167-Duna-mission-planning-and-R-D One where it failed to cancel horizontal velocity and one or more where it decided to flee the area rather than land.

Now I'm trying to find things to do to fill the time until the Duna to Kerbin transfer window comes along. I lucked out a bit when I decided it was time to go to Duna, had 3 or 4 days to the window so I did most of the R&D in another save. Was pushing the clock to orbit and assemble the ship "for real" in time to go to Duna.

As for landing, *in theory* it should be possible to design a ship with exactly the amount of fuel required to reach the surface, even with a starting TWR<1, as long as by the time it hits the final landing burn enough fuel has been used to shift to TWR>1.

The Apollo 11 LEM came >thisclose< to running out of fuel on its landing due to Armstrong having to make a last minute move to avoid landing amongst some rocks. The fuel was very tightly calculated with little extra for going off the descent path.

Link to comment
Share on other sites
Tokay Gris

Tokay Gris

Posted
Posted

For now, I give up testing this and conclude that MechJeb does not like Duna.

The test from 300 km did land, did burn all the way down and might be short on fuel for ascent.

BUT it is about 4 km off target. Which forbids landing close to any cliffs.

Ok, I can get it to land, even if it is far off. Good enough for now. Maybe this gets fixed. Hopefully. It did work in the past. Accurately.

Link to comment
Share on other sites
BigD145

BigD145

Posted
Posted
Greetings

When i try and Launch to rendezvous it seems mechjeb has misplaced his protractor

This feature used to work, but now it seems it only will take a manual input and not work it our automatically

any help would be apriciated

Career mode? You'll need to unlock mechjeb's modules.

Can someone please make a readme for the zip?

Link to comment
Share on other sites
Awesome_Guy

Awesome_Guy

Posted
Posted
Career mode? You'll need to unlock mechjeb's modules.

Can someone please make a readme for the zip?

no its in sandbox

also when eve i try a intercept to dock in space, it sets the sun as target even though i have a different ship set as the target.

Link to comment
Share on other sites
Guest
This topic is now closed to further replies.
 Share
Go to topic listing
×
×
  • Create New...