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

[Andon]

Andon? how do you get 125%? I didn't think you could go over 100%

With the slider, no. But you can click the box and type in whatever value you wish.

[Bobe]

I don't know what I'm doing wrong, but every time I try to use the rendezvous autopilot to transfer to another planet (so far just Mun), it just puts the ship into a collision course regardless of what I set as the desired final distance. Similarly, if I choose Hohmann transfer from the rendezvous planner, it puts me on the same course.

Also, does desired final distance refer to surface altitude or orbital radius?

I manually added a mid-transfer burn to steer away, but now that I have entered Mun's sphere of influence and disabled the autopilot, I can't set it as target. It does very briefly, I can see the nodes appear for a fraction of a second before it goes away. And now it doesn't even give me the context menu.

EDIT: Seems the rendezvous autopilot is not that great for interplanetary transfers. I'll try working with the maneuver planner instead.

Edited January 1, 2014 by Bobe

[Galane]

I've landed on Mun, Minmus, Kerbin, Duna, Ike, and one time on Gilly. It's just Eve that causes problems where MJ won't burn early enough unless there's plenty more than enough thrust available.

What was a surprise last night was how it failed to lower the landing gear soon enough. That's a thing MechJeb has never done to me before.

To optimize the lander for gasbag assisted launch it first has to be able to get down safely. Getting to an optimum fuel capacity/thrust is a bit difficult when the autopilot is being too conservative with the throttle when landing on Eve.

I've built ships intended to just land on Eve, with enough excess thrust that they can make it down safe under MJ control without parachutes or any other non-rocket landing aid. That makes them overweight for their intended purpose and they could lift off, for a while, which is why I have the engines and fuel tanks setup to jettison with sepratrons to ensure they'll fly away to crash. I may alter their design some to use smaller fuel tanks since they don't even use half their fuel going down to near Eve sea level.

Kethane drilling and converting rovers for Eve. https://mega.co.nz/#!vdtT1LIT!WsM8Zn-fdSGS4loOoIAdo0TOjs46GXXw_z9yO85Z_Z0 Requires KAS, ReStock and Kethane, possibly KSPX. Designed to stack the drilling rover (equipped with two small Kethane scanners) on top of the converting rover. Both of those are expected to stack on top of the manned lander.

MJ ought to be able to safely land any ship with a TWR greater than 1 at Eve's surface, which can be done with manual throttle control. Landing with a TWR less than 1 but with aids like parachutes should also be possible, just go full throttle starting high enough so that when combined with the parachute drag the result is a non-damaging touchdown speed. Once it's down it's down for good.

[Galane]

EDIT: Seems the rendezvous autopilot is not that great for interplanetary transfers. I'll try working with the maneuver planner instead.

That's what it's for. Use Hohmann transfer with Mun as your target. Once on your way, use fine tune closest approach. That will make you miss the Mun instead of hitting it and get on a trajectory to be able to get into a prograde orbit.

Once you have the missing trajectory, use circularize in maneuver planner and change it to periapsis. When you have that activated you should be able to sit back and watch your ship go into Mun orbit on its own.

[LameLefty]

Add me to the list of those with landing problems lately. Tried to land a ship last night at a flag I have planted about 2.5 km away from the pad at KSC that I've designated as the landing field. There was nothing unusual about the landing; it was from a regular 100 km orbit, inclined maybe 2 degrees after returning from Duna. First off, the MJ didn't deploy the chutes until I was down to under 12 km. Then when the lander was still about 8 - 9 km up, it became apparent it was going to overshoot the landing field by a couple kilometers. MJ began a "correction burn"; in doing so, the autopilot titled the craft up around 90 degrees, almost perpendicular to the velocity vector, to try to correct. At one point the it was actually tilted up greater than 90 degrees while hanging under the chutes, trying to make that correction burn. At that point I got alarmed and tried to abort the auto land but just as I did so, my craft broke apart. The lander can snapped off the rest of the lander structure. Fortunately, there were enough chutes attached to it that I was able to land safely and recover the vessel but clearly something was amiss in the landing sequence.

First of all, I don't understand why it missed the landing so badly; the vessel mass, the number of available chutes, the engine thrust and fuel load were all available, and this was a very easy west-to-east landing approach. Second, having missed by a couple kilometers while hanging from parachutes at 8 - 10 km above the targeted landing site, why was MJ so aggressive in trying to "correct" when the vessel clearly didn't have the lateral cross-range while under parachutes necessary to make it, to the extent that it broke the lander? When does MJ think to itself, "This is as close as I'm gonna get" and just put the thing on the turf rather than keep trying to correct?

Edited January 1, 2014 by LameLefty

[Bobe]

I have a fairly large ship that I'm trying to send to Duna. It has three LV-N rockets so the burn time is about 15 minutes. However, when using the maneuver planner for interplanetary transfer, it starts the burn at T 0, so by the time the burn is over (which I'm still waiting for), the ship is almost on the other side of the planet and I imagine the transfer trajectory might be a bit awkward.

I noticed on a test run to Mun that the burn started at T - EB/2 (where EB is estimated burn time). Why is this technique not used on interplanetary transfers and is there a way to manually compensate for it?

EDIT: I just checked on it now. At T + 11 minutes, I'm at 485km going 2200 m/s, and my remaining ÃŽâ€v, which had gotten to half the required amount, is actually going up.

Edited January 1, 2014 by Bobe

[BARCLONE]

Stripped out all but these mods:

Efflixi (mid-size engines)

FAR

Keramzit (ProcFar)

KW

MJ-dev150

NovaPunch2

... plus the Squad basics.

Using the "long and narrow" 2.5m launcher, no fins, and an SAS on both stages...

Four consecutive launches on rails, four landings on Mun, virtually flawless even in the dark...

Clearly there is a conflict between mods. Don't know yet which one(s) it is, but it'll probably be the one(s) I need for my planned evil agenda...

[dommcq]

Fairly strange wee bug....

MechJeb doesn't display the number 3 on the M/S countdown during maneuvers anymore.

It will display "22.3" as "22. "

"32.3" as " 2. "

etc.

What?

[AbeS]

is it happening in any other display? Some letters disappear sometimes when, while the game is loading on start-up, you switch to another window with Alt + Tab for example

[dommcq]

is it happening in any other display? Some letters disappear sometimes when, while the game is loading on start-up, you switch to another window with Alt + Tab for example

Nope. Just this one particular display.

I Don't Alt-tab during loading as it does weird things.

Just missing "3" in the m/s countdown.

[Virtualgenius]

Barclone if you work out which mod is screwing you over please let us know as it may be affecting us all

[longhornchris]

@sarbian - love mechjeb for a whole lot of reasons. I do have a suggestion - can you add the 'execute next node' to the maneuver node editor along with the maneuver planner. I like to use the maneuver node editor to tune my maneuvers but its annoying to need to have the other window open to execute a node once I've got it.

[codepoet]

Just reading back a few pages - I see lots of people raising general issues with the landing autopilot. I have spent a bunch of time over the last week making some changes to the landing ap, and have ended up with an even larger list if things that need to be considered. Believe me - it is not a trivial problem to solve!

However, in my own fork I can now consistantly land in the centre of the KSP pad, or VAB helipad using parachutes, land on target on laythe, eve and flat areas of duna and have fixed several other problems. However you can never be sure with this stuff if fixing one scenario has not broken it for another.

I can't promise that what I have offered will be good enough to get pulled in to the dev branch - it takes a lot of testing to be sure things are working in all the different scenarios. But I wanted to offer hope that there are people looking into this stuff.

[LameLefty]

Just reading back a few pages - I see lots of people raising general issues with the landing autopilot. I have spent a bunch of time over the last week making some changes to the landing ap, and have ended up with an even larger list if things that need to be considered. Believe me - it is not a trivial problem to solve!

However, in my own fork I can now consistantly land in the centre of the KSP pad, or VAB helipad using parachutes, land on target on laythe, eve and flat areas of duna and have fixed several other problems. However you can never be sure with this stuff if fixing one scenario has not broken it for another.

I can't promise that what I have offered will be good enough to get pulled in to the dev branch - it takes a lot of testing to be sure things are working in all the different scenarios. But I wanted to offer hope that there are people looking into this stuff.

This is excellent news to hear!

One of the things that I wish is that landing could have an option something along the lines of "Descend Under Parachute Only"- basically a user-clickable option to instruct MJ to calculate a deorbit burn, execute it, and then just use atmospheric drag and parachutes to land, saving fuel for ascent only. I can sort of kludge something similar with staging and using the "Abort" button in the auto-landing sequence but it sure would be nice if there was some way to automate this.

[Galane]

If it can be more self adjustable to the gravity of the body being landed on, that would be very nice. For high gravity like Eve, start the final burn about 2x higher. For really low gravity it could wait until the last little bit of distance.

One thing it shouldn't ever do is wait until it's too late to lower landing gear. Is that event directly connected to starting the final landing burn? I'd rather that be set to a certain minimum altitude, no matter when it's going to start the burn.

Yet another thing, how about an option to not cut the chutes until it's actually landed? That'd be a help for those times when it runs out of fuel right at the last second. Wayyyyy back up thread, or possibly in another thread, someone had some really massive vehicles they were trying to land but they'd smash to pieces because all the chutes would cut the instant any part of the vehicle touched ground instead of waiting for all vertical velocity to stop. If that's built into KSP, could it be patched over by MechJeb?

[Virtualgenius]

Just reading back a few pages - I see lots of people raising general issues with the landing autopilot. I have spent a bunch of time over the last week making some changes to the landing ap, and have ended up with an even larger list if things that need to be considered. Believe me - it is not a trivial problem to solve!

However, in my own fork I can now consistantly land in the centre of the KSP pad, or VAB helipad using parachutes, land on target on laythe, eve and flat areas of duna and have fixed several other problems. However you can never be sure with this stuff if fixing one scenario has not broken it for another.

I can't promise that what I have offered will be good enough to get pulled in to the dev branch - it takes a lot of testing to be sure things are working in all the different scenarios. But I wanted to offer hope that there are people looking into this stuff.

Totally awesome Codepoet really appreciate the effort the MJ Team puts into our troubles and whims, thing I have been finding out lately is it doesnt seem to work out what its actual altitude is from the ground maybe it needs a radar altimeter or something to work this out have trouble landing with the Duna polar caps if it it does get pushed to DEV you will have plenty of volunteers

Edited January 2, 2014 by Virtualgenius

[codepoet]

This is excellent news to hear!

One of the things that I wish is that landing could have an option something along the lines of "Descend Under Parachute Only"- basically a user-clickable option to instruct MJ to calculate a deorbit burn, execute it, and then just use atmospheric drag and parachutes to land, saving fuel for ascent only. I can sort of kludge something similar with staging and using the "Abort" button in the auto-landing sequence but it sure would be nice if there was some way to automate this.

This is pretty much what I want to, and is the reason that I started fiddling with the code in the first place. The changes I have on my development system are able to land a craft with parachutes to within a few meters on a body with a thick atmosphere (ie not duna at high terrrain attitudes) using nothing but a deorbit burn (and perhaps a course correction on first entering the atmosphere.) I can parachute right onto the VAB helipad or the centre of the KSC pad.

However, other scenarios are harder (with a parachute assisted landing on a high duna mountain the hardest) and like I say it is difficult to make changes to improve one typeof landing without influencing a different scenerio. I am full of ideas for code changes to address lots of the issues as I understand them, however I am afraid that I am not flush with time.

Obviously encourangment is welcome - thank you. Those who keep commenting "can't you make it just......" are welcome to fork the code and make the changes themselves

[codepoet]

Totally awesome Codepoet really appreciate the effort the MJ Team puts into our troubles and whims, thing I have been finding out lately is it doesnt seem to work out what its actual altitude is from the ground maybe it needs a radar altimeter or something to work this out have trouble landing with the Duna polar caps if it it does get pushed to DEV you will have plenty of volunteers

In fact it is the discrepencies between the ASL (used by KSP to semi-deploy the chutes) the AGL (used by KSP to fully deploy the chutes) the AGL at the target location (used by mechjeb to activate the chutes) and the AGL at the predicted landing location (our best guess when you are about to smash into the ground) that make the problem of landing with parachutes in a hilly area so challenging.

I am afraid I have not testing landing to the Duna poles (my bad). If you are willing to give it a go, send me a PM and I can give you something to play around with.

[sarbian]

Those who keep commenting "can't you make it just......" are welcome to fork the code and make the changes themselves

+1

And again, thanks for your work

Happy new year to you all.

I wasn't as productive on the 1st as I hoped I would be. That may be related to some wine I drank earlier ^^

Codepoet's patch will most likely be merged in the dev branch tonight. And I'll open the 2014 bug hunting season.

[sarbian]

If it can be more self adjustable to the gravity of the body being landed on, that would be very nice. For high gravity like Eve, start the final burn about 2x higher. For really low gravity it could wait until the last little bit of distance.

MJ does a suicide burn and adapt it to the gravity. Stop saying it does not...

Most of the landing problem comes from :

- small bugs in MJ (using ASL instead of AGL and the like)

- using mods MJ does not understand yet (FAR, mods that use or change the engine module)

- strange design that MJ fail to evaluate

- mods interaction that create "funny" bug.

One thing it shouldn't ever do is wait until it's too late to lower landing gear. Is that event directly connected to starting the final landing burn? I'd rather that be set to a certain minimum altitude, no matter when it's going to start the burn.

Yes, I use the wrong altitude to do that. One of those bug I always forget to fix when I am in front of the code

Yet another thing, how about an option to not cut the chutes until it's actually landed? That'd be a help for those times when it runs out of fuel right at the last second. Wayyyyy back up thread, or possibly in another thread, someone had some really massive vehicles they were trying to land but they'd smash to pieces because all the chutes would cut the instant any part of the vehicle touched ground instead of waiting for all vertical velocity to stop. If that's built into KSP, could it be patched over by MechJeb?

MJ does not cut the chute. The stock chute cuts themselves under .5 m/s. Try to set MJ to land at .6

[Galane]

Better is better. With so many different kinds of conditions to land in, it's a difficult thing to program.

MJ does a suicide burn and adapt it to the gravity. Stop saying it does not...

Then why is it waiting until it's almost hit the ground on Eve to do the burn?

Here's the latest fail to land on Eve. There were three Cirrus bags on that, and three LV909 engines. Modified from the ship in the second picture. I put the gear down manually. MechJeb waited until an even lower far too late altitude to start the final burn and smacked down at 9 meters per second instead of the 0.5 it was set to. Broke 4 of the 10 legs, fell over and blew up.

Previous version with only two Cirrus bags and without the side tanks and LV909 engines.

Perhaps if I adjusted the four radial chutes to open higher? Or more chutes? I can't put anything on top of the lander because that's where other things will be docked (with the bags turned straight down) for the trip from Kerbin.

I think I'll try some Landertrons the next time to see if they can do what MechJeb isn't.

Edited January 2, 2014 by Galane

[Kryten42]

Hey there,

I did not find anything useful about my problem so far, so I came here to rant

No, not really rant, but I can't get MechJeb to work at the moment and that bugs me.

I did a fresh install when .23 came out, so I was not using any mods or other stuff.

Installed MJ, started a new game and played until I came to the research for the AR202 module. After researching it, I went to build a rocket of course...just to notice that the module is still kind of "greyed out" and the text says I need to go to R&D and research it....hmm...well ...maybe I missed it or so? So I went back to R&D where I now saw another module to be researched...besides the first one in state "owned".

So I repeated the process, and the "bug" or whatever it is repeated itself also.

Long story short, I now have 3 "owned" AR202 in my R&D view and every time I try to use the module I can go back to R&D after and click on the third module, to get the research confirmation window....but I can never use the module itself.

I tried reinstalling MJ after another build came out a few days ago, but nothing changed...except that I now have an annoying transparent little grey/balck box showed on my screen when I fly something.

I checked my savegame file to see if something is weird, but couldn't find any obvious errors. Is there any way to fix this, or is it at least a known problem?

Thx in advance,

Kryten

[sarbian]

If you see 2 part then you have a rogue .cfg file somewhere that create a new part with the same name.

You say it's a fresh install ? So you got the install package from the store and installed it in a new dir ?

Where did you get your MJ zip ? If on spaceport on which page ?

[Kryten42]

If you see 2 part then you have a rogue .cfg file somewhere that create a new part with the same name.

You say it's a fresh install ? So you got the install package from the store and installed it in a new dir ?

Where did you get your MJ zip ? If on spaceport on which page ?

I use it in steam, so a fresh install there. I did not check if the directory was removed before installing again, it could be that there were leftovers from the previous install.

The first MJ I did get from spaceport yes, the second try with the newer build I got directly from jenkins.mumech.com.

I only see two or three parts after trying to research it again though, not right after install. And I only see them in R&D, not in the build view. There I only have one greyed out.

[sarbian]

Steam does not clean anything. You may have old MJ in Gamedata or worse : in the old parts & plugins folder.

The view in the editor will not show 2 part with the same internal name but the R&D view does.