[1.12.x] Anatid Robotics / MuMech - MechJeb - Autopilot - [2.14.3] [4th March 2023]

[ChrisDishen]

9 hours ago, Brigadier said:

@ChrisDishenI would suggest that, as a  minimum, you provide some information as to how you're trying to perform your launch, i.e. screen shots of your vessel and the MechJeb parameters you're using, your KSP log file, a mod list, etc.  Your asking it to be fixed when it works successfully for others and not providing any information for the developer to help you is not going to get you much useful assistance.

To help everyone, please read the first post in this topic (no logs => no support) and the following thread:

 

Ok, I will see what I can do, maybe some screenshots. Thank you for your advices.

[New Horizons]

Do you experience a big wobble/gimble of your rockets just before stage separation? The effect is less, when gimbal range ist reduzed, but in generall you can not do that for 1st stages, which not to compensate for sometimes bade lift/weight design. 

I see the effect even at stages which burn out at a TWR of 1.5 or less.

 

[sarbian]

On 4/16/2017 at 10:32 PM, JeanRenaud said:

Any way of ripping the data by code / kOS ?  Or must i get the source code then program some CSV writer ?

It would not be hard to write but I really do not have the time ATM.

If you want some nice graph with scale (and no doubt somewhere export) : https://github.com/hudsonfoo/kerbal-openmct  + https://github.com/nasa/openmct )

 

[Starwaster]

Anyone who's had problems with premature node execution should check out build # 694 on Jenkins. I tweaked the logic a bit for burn initiation, especially for cases where the cause is MJ2 not knowing burn duration due to unstable ignition situations when used with Real Fuels.

(specifically, if you see that burn countdown or burn time read Inf. then this will  help)

 

[Bornholio]

1 hour ago, Starwaster said:

Anyone who's had problems with premature node execution should check out build # 694 on Jenkins. I tweaked the logic a bit for burn initiation, especially for cases where the cause is MJ2 not knowing burn duration due to unstable ignition situations when used with Real Fuels.

(specifically, if you see that burn countdown or burn time read Inf. then this will  help)

 

Should we use this with RO/RP-0 Dev build?

[Loren Pechtel]

Question:  Is there any defense against interplanetary transfer burns that encounter a moon?  Fortunately, I was watching the burn and saw the Minmus encounter.

Note that the problem I have previously mentioned with the burn remaining calculation suddenly getting extra time happened at the same time as the trajectory intersected Minmus's SOI.  However, this time the engines shut down.

This bird is a tanker and heading for a refueling station, wasting a bit of delta-v by circularizing outside Minmus' orbit and then heading on out wasn't a problem (other than the fact it will arrive more than two years later than intended.)  For other birds that could be quite a problem, though.

[Starwaster]

27 minutes ago, Bornholio said:

Should we use this with RO/RP-0 Dev build?

There's no reason at all why not.

[Zwaryczuk]

Not sure if this is a bug, or I am just using MechJeb wrong.

MechJeb will get me to to space, then it just will pass right by the node it needs to circularize. (I don't have skip circulation selected)

It seems as though the only function that works for me is the Assent even though I have researched Unmanned tech and all other MechJeb tech (except rovers).

I have used MechJeb in the past but on earlier versions and never experienced this, hence why I'm not sure it's a bug or not.

I'm going to try to remove all my mods and do a fresh install.

[Cucco-Master]

26 minutes ago, Zwaryczuk said:

MechJeb will get me to to space, then it just will pass right by the node it needs to circularize. (I don't have skip circulation selected)

Have you done a scene change or similar between the gravity turn and the circularization burn? Or have you set up some consecutive maneuver nodes?

Mechjeb treats the circularization burn after an ascent like any other circularization which you can manually set up via 'circularize' in the maneuver node editor (as far as I know). That means after a scene reload you will have to click 'execute next node' first, or it will just ignore it.

[Loren Pechtel]

34 minutes ago, Zwaryczuk said:

Not sure if this is a bug, or I am just using MechJeb wrong.

MechJeb will get me to to space, then it just will pass right by the node it needs to circularize. (I don't have skip circulation selected)

It seems as though the only function that works for me is the Assent even though I have researched Unmanned tech and all other MechJeb tech (except rovers).

I have used MechJeb in the past but on earlier versions and never experienced this, hence why I'm not sure it's a bug or not.

I'm going to try to remove all my mods and do a fresh install.

I've had this happen to me, it was a bug but not what you're thinking.  Rather, the bug is that the rocket took off at all.  What's going on is that MechJeb is perfectly willing to fly the takeoff burn on a rocket that can't be controlled.  However, the circularization is done with a normal maneuver node and the game realizes that the rocket can't be controlled and doesn't light the engine.

The cure is to put a probe core or pilot on the rocket.

[Bornholio]

@Zwaryczuk Have buildings needed for maneuver nodes? MJ will not circ if nodes are not available in career

1 hour ago, Starwaster said:

There's no reason at all why not.

Working great.  Needed just a bit of acrobatics to ullage my motors otherwise only one small hiccup burning several ignitions during a circ manuever.  I'll drop a note in the golden RO/RP-0 Sheet

[Zwaryczuk]

1 hour ago, Loren Pechtel said:

The cure is to put a probe core or pilot on the rocket.

So Jeb and Val in a Salamander Command pod (I think that's the name) won't work? Or does it need to be a stock command pod to work right?

[Loren Pechtel]

57 minutes ago, Zwaryczuk said:

So Jeb and Val in a Salamander Command pod (I think that's the name) won't work? Or does it need to be a stock command pod to work right?

Shouldn't have any problem with Jeb on board.

[linuxgurugamer]

@sarbian

I'm thinking of calling Mechjeb from a mod I'm writing to control a final descent after a suicide burn. 

I haven't been able to figure it out, can you help? Essentially, control will be turned over to Mechjeb when the rocket is almost down, and, hopefully, vertical.

Thanks in advance

[sarbian]

1 hour ago, linuxgurugamer said:

@sarbian

I'm thinking of calling Mechjeb from a mod I'm writing to control a final descent after a suicide burn. 

I haven't been able to figure it out, can you help? Essentially, control will be turned over to Mechjeb when the rocket is almost down, and, hopefully, vertical.

Thanks in advance

Which part is giving you trouble ?

• Accessing MJ
• Accessing MJ modules
• Controlling the vessel with MJ
• All of the above

Edited April 26, 2017 by sarbian

[linuxgurugamer]

Just now, sarbian said:

Which part is giving you trouble ?

• Accessing MJ
• Accessing MJ modules
• Controlling the vessel with MJ
• All of the above

I looked at Gravity Turn for ideas, so I know how to get the references, but essentially, turning control over to the Final Descent module of MechJeb

[Capt. Hunt]

speaking of landings, I'm finding that while Mechjeb landing guidance works great on airless bodies, if I attempt a propulsive landing on Kerbin it puts me down as much as 20 km short of the target.  Something seems to be off with the aerobraking calculation.  I've seen this happen fairly consistently in modded and unmodded installs since they updated stock aero.

Edited April 26, 2017 by Capt. Hunt

[Starwaster]

7 hours ago, Capt. Hunt said:

speaking of landings, I'm finding that while Mechjeb landing guidance works great on airless bodies, if I attempt a propulsive landing on Kerbin it puts me down as much as 20 km short of the target.  Something seems to be off with the aerobraking calculation.  I've seen this happen fairly consistently in modded and unmodded installs since they updated stock aero.

20 km? Congratulations, that's almost Curiosity accuracy. I guess 'something seems to be off' with real life aerobraking calculations too. It may well be that MJ2 will never achieve 100% accuracy because that's simply the nature of the beast. I think to get much better than that will require a craft that can get within that area and then use powered flight to proceed the rest of the way. MJ USED to be that accurate on planets both with AND without atmosphere. But that was before our current aerodynamic model. Orientation affects drag so predictions are easily invalidated during the actual reentry.

Here are examples of the levels of accuracy we were capable of over the years. Curiosity only achieved that 12 x 4 zone by being able to fly a lifting reentry where it could steer towards the target zone.

 

Edited April 27, 2017 by Starwaster

[Capt. Hunt]

I figured it was something like that, why else would the issue have persisted this long, right?  I guess I should give up trying to stick a Dragon on the VAB helipad, lol. 

A warning in the GUI when attempting a propulsive landing in an atmosphere might be a good idea.  TBH, I assumed that because there wasn't a warning it was just something that wasn't working as intended.

Maybe the "VAB" and "KSC launch pad" landing coordinate presets should be replaced with just a single "KSC" preset, since MJ doesn't have the accuracy to actually hit either of those targets anymore.

Edited April 27, 2017 by Capt. Hunt

[Starwaster]

1 hour ago, Capt. Hunt said:

I figured it was something like that, why else would the issue have persisted this long, right?  I guess I should give up trying to stick a Dragon on the VAB helipad, lol. 

A warning in the GUI when attempting a propulsive landing in an atmosphere might be a good idea.  TBH, I assumed that because there wasn't a warning it was just something that wasn't working as intended.

Maybe the "VAB" and "KSC launch pad" landing coordinate presets should be replaced with just a single "KSC" preset, since MJ doesn't have the accuracy to actually hit either of those targets anymore.

One way it might be achieved is to be subsonic during the majority of your descent. Propulsive braking instead of aerobraking...

[Blackline]

Is it possible to alter the parameters of the final descent of a landing? For example, we already can change the final vertical speed before landing, but not the height at which this should be achieved. Currently, MJ starts this final phase IMHO too high, and I'd like to reduce that safety margin in favor of fuel efficiency. Another thing I'd like to change is the maximum thrust that MJ uses, again, swapping safety with efficiency. Thanks!

[Fury1SOG]

1 hour ago, Starwaster said:

One way it might be achieved is to be subsonic during the majority of your descent. Propulsive braking instead of aerobraking...

I've been attaching KRE grid fins (with buffed drag values) on my reentry vehicles.

While preparing for de-orbit, I let MechJeb calculate the descent with grid fins deployed, which usually results in a short trajectory. Then, I manually stow the grid fins during the descent, which stretches the trajectory past the intended target.

Once the target is in visual range, I manually open/close the brakes until I land near the target.

I can usually land near the VAB using this method.

Also, I've seen that a steeper descent (such as from a direct 500km orbit) results in a more pinpoint landing (although fiery and terrifying) than a more gradual descent from a 80km orbit.

Edited April 27, 2017 by Fury1SOG
Some extra observations

[Capt. Hunt]

9 hours ago, Blackline said:

Is it possible to alter the parameters of the final descent of a landing? For example, we already can change the final vertical speed before landing, but not the height at which this should be achieved. Currently, MJ starts this final phase IMHO too high, and I'd like to reduce that safety margin in favor of fuel efficiency. Another thing I'd like to change is the maximum thrust that MJ uses, again, swapping safety with efficiency. Thanks!

I have noticed this too, mechjeb tends to zero the horizontal velocity and start the suicide burn too early.  If I turn it off and just let the capsule coast, it still has plenty of altitude to stop when it passes over the VAB.

[Loren Pechtel]

1 hour ago, Capt. Hunt said:

I have noticed this too, mechjeb tends to zero the horizontal velocity and start the suicide burn too early.  If I turn it off and just let the capsule coast, it still has plenty of altitude to stop when it passes over the VAB.

I think it comes down to how maneuverable the craft is.

I've seen the 500m not be enough on things that don't turn well.  On maneuverable stuff in low gravity it's more than you need.

[Blackline]

3 minutes ago, Loren Pechtel said:

I think it comes down to how maneuverable the craft is.

I've seen the 500m not be enough on things that don't turn well.  On maneuverable stuff in low gravity it's more than you need.

Thats why I think it would make sense to (however) change the behavior based on the current situation. 

Edited April 27, 2017 by Blackline