Jump to content

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


r4m0n

Recommended Posts

  sarbian said:
Each time I had this problem (resulting most of the time in a nice looping) it was not MJ but the rocket design. All MJ did was make the problem more obvious because the transition from straight ascent to gravity turn can be quite step with the default 40% turn => MJ turn a bit too hard, top heavy rocket do a loop.

Having FAR without my small extension make thing even worse since MJ think it can't turn the ship fast and it will push the control to the maximum.

Try with a different turn shape %.

Not saying that I am sure there is no bug, but I don't think there is one.

MJ definitely has issues when it comes to selecting the simplest path to a goal. At the least, in certain circumstances. Given a situation where it is already lined up such that a pitch will be sufficient to orient it 90 degrees to its long axis and then instructed to perform such a maneuver, it should not begin yawing and rolling as well. Or pitching and rolling when a yaw would suffice. I just watched that happen. If there's logic involved that makes it want to roll to some desired orientation then can't that wait until the requested orientation is done?

Link to comment
Share on other sites

  Damaske said:
snip:. . . and have my rockets attempt to slow down from 1000+ M/s when trying to do a no parachute landing.

Just to clarify what i meant, I am sick of my rockets trying to slow themselves down from 1000m/s in the last 200meters before the ground. Would there be a way to program the auto decent to apply steady thrust from 1000m until targeted landing speed is reached around maybe 200meters and then settle down slowly afterwards?

Its just i have rather large lander's and some large VTOL SSTO's that are crashing because Mechjeb is trying to slow them down in only the last 200meters from 1000m/s. I have even tried to slow em down myself by holding the throttle up and having just a little allowed thrust to happen but that's not enough.

Link to comment
Share on other sites

Can you post some picture of this lander ? In bright light and different angle please. And mod list please

The landing AP do that to lower the fuel use. How much it slow down is related to how much MJ think your lander can thrust. If you crash then MJ is overestimating your vessel thrust.

So something in your design must disrupt MJ calculation.

Link to comment
Share on other sites

My video card fried my power to the motherboard (i think), and the smaller four plug from the power to board has been melted some. I got it working once more for now. But i have been forced to remove the other card and go back to an ATI card that i have had troubles in the past with KSP. Not as smooth ect ect, But i might be forced into getting a new motherboard/psu/cpu/memory combo here as i do not think i can find a comparable board for the cpu i have. So please bear with me.

The craft i tried to land vertically I have several different engines on it, Two rockets and a Turbojet. Would the turbojet be causing it to think it has more thrust than it really does? And does MJ count engines that have been shutdown? Also Would there be a way to have a way to switch between it doing a fuel saving burn like it is now, or a longer burn as i have suggested?

I understand why the change (if it was one) to the last second burn was done now.

Link to comment
Share on other sites

I'm fairly certain MJ assumes the turbojet gets fixed thrust at a given throttle. This is easily testable with a plane and using the "current thrust" option. You'll notice, at low altitudes, MJ's thrust is way, way off the actual thrust you're getting.

Link to comment
Share on other sites

  sarbian said:
It seems I have a communication problem. I know MJ do a lot of unneeded roll. BUT it's not related to the "rocket like to do loopings" bug.

Not even when that rolling combines with pitching / yawing to throw it off course?

What do you think its response would be there if not to follow up with even more yawing and pitching to try to correct. Now it's not only pointing the wrong way but the roll it was trying to do also requires corrections. That's trouble enough in stock, let alone with FAR when you're climbing out of atmo. Fortunately I was out of atmo when it happened

It's not a communication issue btw, we clearly just don't agree on the contribution it has to the problem

Link to comment
Share on other sites

Turbojet is the word :)

It's hard for MJ to know how much a Jetengine can thrust since they take some time to go to full thrust and their thrust change with the air speed. Just try a simple jetengine VTOL with the translatron, you ll understand how bad MJ is with them. There is a patch I need to finish merging that may help, but I am not really confident.

Shutdown engine should be ok (I'll need to check this one).

For now I can't change how it works easily. I'll see what I can do when I work again on the landing AP. For now I want to solve the rolling trouble.

@Starwaster : I get your point, but even if the current rolling trouble contribute to that I don't think it's the main culprit.

I am looking at the attitude code again and I found some problem with it.

Edited by sarbian
Link to comment
Share on other sites

Ok great to hear that might be what is making it go crazy. I'll have to convert over the lander i had planed, to use the smallest rocket engine for VTOL jets. And glad to hear that it might be a fix coming out here soonâ„¢. and yes i understand about the rolling troubles as if you can't get a rocket up then you should not worry about landing it.

Link to comment
Share on other sites

  Damaske said:
Just to clarify what i meant, I am sick of my rockets trying to slow themselves down from 1000m/s in the last 200meters before the ground. Would there be a way to program the auto decent to apply steady thrust from 1000m until targeted landing speed is reached around maybe 200meters and then settle down slowly afterwards?

Its just i have rather large lander's and some large VTOL SSTO's that are crashing because Mechjeb is trying to slow them down in only the last 200meters from 1000m/s. I have even tried to slow em down myself by holding the throttle up and having just a little allowed thrust to happen but that's not enough.

Heh, you'd like how it's flying my kethane landers at Duna. Slow burn almost all the way down, wasting a huge amount of fuel and taking a longer time to get down when I built them to be able to do a short burn to deorbit then a full throttle braking blast and a gentle touchdown. They never get going fast enough to heat up even a little. MJ 2.0.9 did that on my first Mun landing (which I did from 200KM orbit) and the lander almost didn't have enough fuel to make even a 10~15KM ascent to orbit.

A choice between the two methods would be nice to have. Slow Burn or Fast and Blast. Test each of your landing ships in another sandbox game, placing them in orbit with hyperedit, to see which way is the most efficient for each lander. Enough fuel has to be burned with the engines available to 100% counter the vehicle's velocity by touchdown no matter how the descent is conducted, but there's no need to burn one drop extra.

  Quote
Grabbing #106, will give er a spin....
But you hope MechJeb doesn't give your rocket a spin.

Another option I'd like to see in maneuver planner for going to another planet is one to go NOW instead of waiting for the most efficient time. Is it going for least time or least fuel? Whichever it is, I'll put together a rocket with enough fuel to make it there, leaving soon as it's in orbit and the fuel topped off.

My kethane landers at Duna have turned out to be terribly inefficient, making barely more fuel than they use. It's like sending a truck that burns 20 gallons of gas on the trip to bring back one gallon. Make it liters, that's even worse. I'm filling a swimming pool with two Dixie cups. I need to get something better on its way right away, not half a Kerbin year from now.

Link to comment
Share on other sites

Downloaded the newest dev model and thanks to the way it slows late it slammed a smaller ship into my LARGE fueling station that took approximately a week to build it is now broken up and scattered all over the place. This smaller ship is a Craft that has both a turbojet and rockets they are the Rockomax 48-7S'es. This is the second time this has happened but the first time was due to (i think ) the craft running out of Oxidizer thus unable to slow down and slammed into the station.

Just thought you would like to hear about that NJ MAY not be working with the engines that are shut off thinking that they are on and they are not. Something is happening that is messed up that is for sure.

Link to comment
Share on other sites

Greetings.

Just got KSP (SWEET!), and I decided to install MechJeb and check it out, but am having issues. Only other mod I've installed is the Kethane mod.

So I installed MJ, and went into the VAB, and started building a new rocket, but to MJ tab on the side appeared. Looked around and found out about the AR202 part. Modified the parts file to put in in the first R&D spot, researched it, and was able to now access it. So I slapped the AR202 on my rocket, and now I have access to the MJ tab. Now my problem is when I go to the launchpad and open the tab, the only options I have available is Settings, Delta V, Orbit Info, Surface Info, and Vessel Info. None of the other options are listed (Smart A.S.S, the autopilots, ect). I must be missing something, but what, I don't know.

This is with KSP .22 and MJ 2.1.

Any help appreciated.

Link to comment
Share on other sites

I have problem with the landing guidance it wont fire the breaking burn hence slamming into the ground and the little green guys get turned to mush using version 102 if i go back to release works perfectly

Edited by Virtualgenius
Link to comment
Share on other sites

  Virtualgenius said:
I have problem with the landing guidance it wont fire the breaking burn hence slamming into the ground and the little green guys get turned to mush using version 102 if i go back to release works perfectly

I've had the same issue with 102, especially using cbbp's Dragon capsule. Stock ships seem to work, but the deceleration burn never occurs during landing with this capsule - they "coast" to an untimely demise. Grabbing 106 now to see if that fixes things and I'll let you know how that works out.

Link to comment
Share on other sites

106 testing:

(done with MJ + KWR)

Occasionally opens throttle to full after executing a maneuver node.

landings: 2 out of 3 successful with stock Kerbal X. Failed attempt it turned 90 degrees (approx +Normal) and executed a continuous burn Until it reached eject velocity out of Kerbin orbit. At this point for ****s and giggles I turned on infinite fuel mode to see if it would ever correct the error. Landing status indicated it was making a course correction. The DV of the course correction continued increasing until I got tired and hit F9 for landing attempt #3

All 3 attempts were KSC targeted landings.

Edit: During landing, when orbit went hyperbolic, an error was generated by OrbitExtensions.NextTimeOfRadius(). The landing AP should trap errors like that and either abort the landing or correct the orbit. Although realistically, by that time it's probably too late and you're using F9 to get back to your quick save. This isn't really a 106 specific problem, not new at all but probably intermittent.

MechJeb module MechJebModuleLandingAutopilot threw an exception in Drive: System.ArgumentException: OrbitExtensions.NextTimeOfRadius: given radius of 603783.286865588 is never achieved: o.PeR = 603796.610583531 and o.ApR = 674048.814738914

at MuMech.OrbitExtensions.NextTimeOfRadius (.Orbit o, Double UT, Double radius) [0x00000] in <filename unknown>:0

at MuMech.MechJebModuleLandingAutopilot.ComputeCourseCorrection (Boolean allowPrograde) [0x00000] in <filename unknown>:0

at MuMech.MechJebModuleLandingAutopilot.DriveCourseCorrections (.FlightCtrlState s) [0x00000] in <filename unknown>:0

at MuMech.MechJebModuleLandingAutopilot.Drive (.FlightCtrlState s) [0x00000] in <filename unknown>:0

at MuMech.MechJebCore.Drive (.FlightCtrlState s) [0x00000] in <filename unknown>:0

[/code}

Edited by Starwaster
Link to comment
Share on other sites

Guest
This topic is now closed to further replies.
×
×
  • Create New...