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

[Poodmund]

12 minutes ago, dboi88 said:

Yep, read 1 page back

Errr... I already had... and I am adding my own coordinates for the KSC Pad and VAB for Landing Guidance and the KSC Runway for Spaceplane Guidance. I think you misunderstand my issue, I don't want to add my own coordinates for the Island Runway, I want to override MechJeb's default behavior of automatically adding in the vanilla coordinates for the Island Runway if no other coordinates are specified elsewhere.

Edited January 27, 2017 by Poodmund

[dboi88]

1 minute ago, Poodmund said:

Errr... I already head... and I am adding my own coordinates for the KSC Pad and VAB for Landing Guidance and the KSC Runway for Spaceplane Guidance. I think you misunderstand my issue, I don't want to add my own coordinates for the Island Runway, I want to override MechJeb's default behavior of automatically adding in the vanilla coordinates for the Island Runway if no other coordinates are specified elsewhere.

Apologies i had misunderstood.

[Poodmund]

4 minutes ago, dboi88 said:

Apologies i had misunderstood.

No worries, bud. 

Spoofing the 'Island Runway' name entry into the custom config and then deleting the coordinates doesn't seem to override it as an invalid entry either.

		Runway	
		{
			name = Island runway
			!start
			!end
		}

It still gets positioned at the vanilla coordinates.

[DerekL1963]

2 hours ago, edemlama said:

ok I'll change my design and see if the problem persist. Thanks!

When I encounter this, I usually shut the AP down the circularize with the Maneuver Manager.

[WuphonsReach]

In dev build #685 (Jan 15th), there's possibly a bug with the Advanced Transfer function.  After picking a point on the pork chop plot and creating the node, if I stay focused on the rocket the burn time value slowly changes.  And if I sit there a long time, the value gets really silly.  So my workaround is to set the burn, then go back to the space center until KAC tells me it's time for the burn.

Other maneuver nodes don't exhibit this behavior.

I have not yet tried to create this with just stock + MechJeb build #685.

[Loren Pechtel]

There's a minor math error of some kind.

 

Rendezvous Planner:

Time until closest approach  0s

Separation at closest approach  138m

However, Distance to target  133.4m

I suspect it's simply roundoff.

[ansaman]

This may have been mentioned and I missed it. I will set up a maneuver using Mechjeb, typically an advanced transfer. The maneuver executes properly and then just as it terminates, it executes the same thrust maneuver again unless I stop it. I am not sure whether this is a Mechjeb or a KSP problem. I am further not sure if this is related to staging in the middle of executing a maneuver or not. Anyone else experience this? 

[WuphonsReach]

Staging in the middle of a maneuver is fine (I use the Utilities menu to enable it on a case-by-case basis).

Adding another maneuver node (or changing the target) while MechJeb is currently executing a maneuver typically leads to tears.  Especially if MJ is actively executing a maneuver burn.  That's been touchy for as long as I can remember.

(Not saying it hasn't wigged out at other times, usually after a few dozen scene changes as I jump between ships, but that's probably stock KSP being weird.  And a restart of KSP has always fixed it.)

[ansaman]

59 minutes ago, WuphonsReach said:

(Not saying it hasn't wigged out at other times, usually after a few dozen scene changes as I jump between ships, but that's probably stock KSP being weird.  And a restart of KSP has always fixed it.)

Thanks for the help.

[Loren Pechtel]

3 hours ago, WuphonsReach said:

Staging in the middle of a maneuver is fine (I use the Utilities menu to enable it on a case-by-case basis).

The problem I have found is that if it stages during the circularization burn to a stage with a much lower TWR the burn time estimate is all wrong--it lights the engine late (expecting to do it all with the high TWR engine) and that results in entering the wrong orbit.  It would take a fair amount of simulation for MechJeb to figure out how to handle it correctly. 

Circularization burns are by far the ones most likely to be bit by this because TWR often drops a lot and it's a large burn in a rapidly changing situation.  It can happen elsewhere, I've seen a pretty bad burn heading for a moon due to this (2nd stage still had dregs left after attaining orbit.  It lit the injection burn late because of this.)

[WuphonsReach]

Yeah, for Ascent Guidance circularization burns, to deal with the issue of staging during the burn, I do a few things:

• My target orbit is 80 or 85 km (so even if I end up with a 75/90 orbit I'm still outside the atmosphere of Kerbin)
• The final flight path angle is 2-5 degrees
• Circularization stage has at least 0.5 TWR (0.8 is better)
• Sometimes I'll drop the previous stage prior to the burn (with an Alt-PRO from the smart parts mod)

For other situations, I try not to do a huge step-down in terms of TWR during a burn.  For instance, when inserting into Moho's orbit where I need ~4500 to circularize, but I only have 2500 on the current stage; I will point the rocket retrograde as soon as I enter Moho's SoI and immediate burn all of the remaining fuel in that stage.  It won't change the Pe much, and it trims off a lot of the delta-V that I'd have to burn at the actual Pe.  Then I'll setup the circularization burn at Pe like normal using the next stage.

[DerekL1963]

43 minutes ago, Loren Pechtel said:

It can happen elsewhere, I've seen a pretty bad burn heading for a moon due to this (2nd stage still had dregs left after attaining orbit.  It lit the injection burn late because of this.)

Unless I really need every meter/second of d/V, I simply drop the almost-empty stage in cases like this.
 

35 minutes ago, WuphonsReach said:

For other situations, I try not to do a huge step-down in terms of TWR during a burn.  For instance, when inserting into Moho's orbit where I need ~4500 to circularize, but I only have 2500 on the current stage; I will point the rocket retrograde as soon as I enter Moho's SoI and immediate burn all of the remaining fuel in that stage.  It won't change the Pe much, and it trims off a lot of the delta-V that I'd have to burn at the actual Pe.  Then I'll setup the circularization burn at Pe like normal using the next stage.

Another trick (if you have enough) is to use that d/V at Pe to enter an elliptical orbit, and then circularize on the next.

[Loren Pechtel]

14 minutes ago, DerekL1963 said:

Unless I really need every meter/second of d/V, I simply drop the almost-empty stage in cases like this.

Yeah, *now* I know that.  I didn't realize it at the time.

[JayPee]

I have a question about the Rover autopilot, and some forum/google searching didn't answer it.

I have a rover on autopilot with MJ, but its mission will take about 6 or 7 days.  Do I need to leave it as the active vessel this entire time?  Will the MJ autopilot keep it rolling while I'm managing my other missions?  Or will the rover come to a halt every time I go deal with something else in-game?

Thanks.

[Loren Pechtel]

I think I understand why the land-anywhere function often crashes the rocket.

A land-at-target landing "lands" the rocket 500m above the ground and then drops it.  So long as it's maneuverable enough to point where it needs to to kill any residual velocity it lands fine.

I just watched multiple land-anywhere landings--it does not appear to land the rocket 500m up, but rather right at ground level.  If the rocket had any substantial horizontal velocity (such as landing from orbit) this is catastrophic.  It also seems like it isn't always right about ground level--since it does not appear to know where it's going to land I'm guessing it is using the altitude where it was when it made the calculations but I'm not at all sure.  If the terrain is rising as it's coming in it plows into it.  If it has the altitude right it "lands" but with the rocket oriented as it was for the braking burn--which is nothing like vertical.

Furthermore, from looking at the code I think I see at least part of it.  UntargetedDeorbit.cs line 13 apparently puts the rocket into final descent mode (exactly what the landing looks like) if the apoapsis is less than 10% of the radius of the body you're landing on.  For the Mun (where I was observing) that means anything below 20km triggers this.

This is also consistent with my past observations that land-anywhere works if you're pretty much standing still, otherwise it crashes.

Also, what am I missing about the source code?  I get a gazillion type or namespace not found errors.

One is UnityEngine which would say to me that something is wrong about my install of Unity but the others look like KSP modules.

[rsparkyc]

12 hours ago, JayPee said:

I have a question about the Rover autopilot, and some forum/google searching didn't answer it.

I have a rover on autopilot with MJ, but its mission will take about 6 or 7 days.  Do I need to leave it as the active vessel this entire time?  Will the MJ autopilot keep it rolling while I'm managing my other missions?  Or will the rover come to a halt every time I go deal with something else in-game?

Thanks.

pretty sure you need to stay focused on the rover, though you can use physics timewarp. that way you only need to have it running for 2 days IRL

[Loren Pechtel]

1 hour ago, rsparkyc said:

pretty sure you need to stay focused on the rover, though you can use physics timewarp. that way you only need to have it running for 2 days IRL

That's why you use the Bon Voyage rover autopilot.  It's not perfectly faithful to the terrain as it simulates rather than actually drives but it runs in the background, including in timewarp.  It makes sending a rover all over a planet practical.

[Mystique]

Can someone enlighten me how to use landing predictions from landing guidance window for bodies with atmosphere? I set up my orbit so that I should land in some sport after aerobraking, as shown by predicted trajectory. But when I actually enter atmosphere, predictions suddenly starts changing as I start to actually lose speed, which breaks all plans

P.S. Had same problem with Trajectories mod, removed it as soon as noticed that feature in MJ, since they seem to produce same results.

[WuphonsReach]

1 hour ago, Mystique said:

Can someone enlighten me how to use landing predictions from landing guidance window for bodies with atmosphere? I set up my orbit so that I should land in some sport after aerobraking, as shown by predicted trajectory. But when I actually enter atmosphere, predictions suddenly starts changing as I start to actually lose speed, which breaks all plans

For Kerbin returns from LKO (usually 200km), I'll set the target as KSP (red marker), then setup a "change Pe" maneuver to drop my Pe down to 30km.  I'll drag that forward around the body until the blue marker is just east of the red marker (a bit offshore).  It usually puts me down within 10-50km of KSC.

That's about as accurate as it gets from what I've seen.  And the longer that you are in coast phase inside the upper atmosphere, the worse the accuracy will be.  So it's a tradeoff between entering at too steep an angle in exchange for accuracy, or doing a shallow entry (for more slow-down time) and being very inaccurate.

I used to find that the Trajectories mod produced more accurate predictions.

[sarbian]

4 minutes ago, WuphonsReach said:

I used to find that the Trajectories mod produced more accurate predictions.

It should since it handle lift form some part better and should account for your ship orientation. I need to work on that for MJ....

[Kerbital]

On 1/15/2017 at 8:41 PM, Basto said:

Curious if you have a roadmap for what your priorities are. Love MJ, curious about where you plan on taking it. 

I find KSP impossible to play without MJ. I'm just too stupid:)  It saved the game for me.  Thanks a lot for the work!

I'm also curious about the future roadmap for MJ:) 

[Eugene Moreau]

On 1/27/2017 at 8:59 AM, dboi88 said:

I've had this issue too. But it happens to me with or without mechjeb. It seems the manoeuvre node isn't being calculated properly when it's a long way away(in time). And drifts during timewarp back towards the correct location. This seems to be a bigger issue the further away the manoeuvre node is. Which would fit with you eventually being able to warp properly, which would be as you had come closer to the manoeuvre node rather than anything to do with the number of times it had tried to timewarp. It is the manouvre node that is drifting rather than the orientation of the ship.

I assume that Mechjeb doesn't do anything with the manoeuvre node other than to set one so I had assumed that it was a stock issue and not a Mechjeb issue.

FYI, I removed the Persistent Rotation Mod and it all works fine now.

[dboi88]

1 minute ago, Eugene Moreau said:

FYI, I removed the Persistent Rotation Mod and it all works fine now.

Cheers for the update, i don't run that Mod though.

[Eugene Moreau]

Just now, dboi88 said:

Cheers for the update, i don't run that Mod though.

Neither do I any more.   Not even sure why I started with it.

[dboi88]

@sarbian I've finally settled on a design for the template page. I'll stick with this if you are happy with it?

I'm about half way though the content, won't be long now.