[WIN/MAC/LINUX] KSP Trajectory Optimization Tool v1.6.10 [Major LVD Improvements!]

[diomedea]

My PC has an Italian locale setting (for numbers, it is the same as French locale, that maybe easier to find).

EDIT: FYI, I did a try setting on a US locale, but KSP TOT still gives the same error. That makes me guess if some other SW component on which KSP TOT depends may be localised and doesn't change with system settings, but I have no idea of what could be (MCR is international).

Edited May 1, 2014 by diomedea
more info

[Arrowstar]

Alright, last debug version for tonight. Can you take a look, do the exact same thing you've been doing, and send me the log file? I've got a bit more output in there.

Btw, would you be willing to change the locale setting on your PC to "English (United States)" temporarily to see if that helps? I tried changing my User and System locales here to Italian, but it didn't help reproduce the issue.

[diomedea]

PLease find here the log about the last debug version (sorry for not answering before - really was deep night here).

Tried again with US locale settings (both system and user, and checked they are applied). No joy, still the same error at the same point.

[Arrowstar]

So here's the offending line:

################## Propagation Error ###################

Event Num: 4

Error Msg: Searching for SoI transition failed: initial state is already outside central body SoI. Body: Vall - SoI Radius: 2406.4015 - rNorm: 10854.923 - GM: 207.4815

#########################################################

Which makes no sense. The code is going "I'm at Vall. But hey, I'm already outside of Vall's SoI! That's not good!" And then it (intentionally) dies.

I think the best solution is to remove the check for the error condition and process the SoI transition as if nothing happened. I have no idea how you got into Vall's SoI, but shooting out of it again immediately shouldn't impact anything. I'll provide one more debug version tonight after work with this change in it and you can tell me if it works.

Oh, it also looks like you're reading in the bodies.ini file just fine. So the locale stuff probably isn't the culprit. Thanks for trying though!

One thing you could do while you wait is to try the tutorial over again, but don't let the radius of perigee get so low on Laythe. Stop is around 10 km or so, maybe even go up to 100 km to be safe. Then continue the tutorial. Let me know how that works?

[diomedea]

There is at last something different happening . As you suggested, tried the tutorial again. Couldn't really make for a minimum periapsis radius, the optimizer kept providing no valid solution (value of function higher than 10e6) and no intercept with the Jool system in a number of tests, when I made such a constraint with the initial optimization.

At the end, returned to mean values for the TIJ burn to start fresh again, with no constraints, but those mean values being just average of the admitted range (prograde=1800, normal=1600, radial=0) instead of the exact values computed by KSP TOT. Plus, I actually set Kerbin as the body for the initial Coast maneuver, instead of letting it blank. Initial optimization prrovided a "slightly different" solution (I notice one main difference, high inclination of the orbit arriving at Laythe, this time is closer to the inclination shown in the tutorial; with the orbit calculated in my previous plan inclination was close to 180°).

Introduced all three constraints (made one variant, let inclination range among 0 and 10), run the optimizer and IT WORKED (very slowly converging, but definitely no error though the solution is far off about inclination ).

Here are some screenshots, the amended MA plan and the log (that is full of weird messages, though saved after the last successful optimization, hope it is of some value):

1. after initial optimization,

2. optimizer status window at half the final optimization (with constraints)

3. final result with constraints.

Have the suspect that the old MA plan really passed through Vall's orbit, being so close to the orbital plane. Could that be the cause of the error?

It definitely seems locale settings are not responsible for this issue. Sorry for that, time lost looking at a wrong track .

[Arrowstar]

Alright, can you try this latest debug version?

Please note that I made some changes to the way constraints are handled, so you'll need to start the tutorial over from the beginning. The change I made allows you to select an applicable body on constraints. This allows you to set an inclination constraint, but only have it apply to orbits around Laythe, for example, instead of whatever orbit is at the end of the event.

If you don't run into any problems, then I think I may call that version 0.12.1 actually. Please let me know.

[diomedea]

Well, quite a number of things to report.

1. It works . Redid the tutorial and could do it to the end.

2. I notice that the inclination and periapsis radius constraints now allow to set the celestial body. At step 8 I did a first try setting Laythe for those. The result was an error after a few iterations. Here the dialog and the log about that. Believe this new feature of allowing the applicable body needs to be checked . Leaving the celestial body empty with the constraints, it went OK.

3. I notice that when setting the objective function to minimize eccentricity or minimize inclination, the applicable body is no more selectable (gray backgrund; in the tutorial its background is white). Possibly just an improvement in this 0.12.1 version.

3. I had a different solution with the Laythe initial intercept (step7), and that reverberated in the following steps. Step 8 went fine, but when doing the optimization at step 10, the objective function could not get down enough. Did some tests, in the end allowed DV(prograde) in the range -5000 to -1000 and that made it (optimizer went for DV= -4556 m/s). When doing step 11 (inclination) again the objective function could not decrease enough, but the final solution was with DV(prograde)=-4999.99 (then again, if I had more DV to spend, the optimizer would provide a better solution). Quite probably my encounter with Laythe is head-on, while the tutorial solution is smarter. I should probably spend some time playing with the constraints to find how to get a smarter solution (requiring less DV for the injection burn).

I can now say the tutorial is very effective in teaching how to use the optimizer properly, showing how its logic works and how to make it perform better .

[Arrowstar]

1. It works . Redid the tutorial and could do it to the end.

Great!

2. I notice that the inclination and periapsis radius constraints now allow to set the celestial body. At step 8 I did a first try setting Laythe for those. The result was an error after a few iterations. Here the dialog and the log about that. Believe this new feature of allowing the applicable body needs to be checked . Leaving the celestial body empty with the constraints, it went OK.

Yes, this is a new feature and the issue you saw has actually already been resolved. It does precisely what you described, though.

3. I notice that when setting the objective function to minimize eccentricity or minimize inclination, the applicable body is no more selectable (gray backgrund; in the tutorial its background is white). Possibly just an improvement in this 0.12.1 version.

It's actually a bug fix. Those functions aren't supposed to have a body associated with them and I simply forgot that when adding them. Selecting either of those two options minimizes the selected quantity at the end of the applicable event.

3. I had a different solution with the Laythe initial intercept (step7), and that reverberated in the following steps. Step 8 went fine, but when doing the optimization at step 10, the objective function could not get down enough. Did some tests, in the end allowed DV(prograde) in the range -5000 to -1000 and that made it (optimizer went for DV= -4556 m/s). When doing step 11 (inclination) again the objective function could not decrease enough, but the final solution was with DV(prograde)=-4999.99 (then again, if I had more DV to spend, the optimizer would provide a better solution). Quite probably my encounter with Laythe is head-on, while the tutorial solution is smarter. I should probably spend some time playing with the constraints to find how to get a smarter solution (requiring less DV for the injection burn).

Your encounter with Laythe is most likely head on. What's the inclination of the Jool orbit you're in before you get to Laythe? I'm going to get it's greater than 90. Use the state log to find out. If you need more wiggle room, feel free to bump up that -5000 m/s bound all the way to -10000 just so the optimizer can have some room to move in.

I can now say the tutorial is very effective in teaching how to use the optimizer properly, showing how its logic works and how to make it perform better .

Great! I have one or two more things I need to add some 0.12.1 tonight, and then I think I can get the new version released to everyone. Thank you for your help!

[diomedea]

Great. While looking how to find the inclination with Jool's orbit (as suggested) I just discovered an Easter Egg .

I still had KSP TOT open, no log saved from the previous session and I was looking to find that. And then I found "State Log" with the Tools menu in MA (still undocumented).

And you're right, inclination with Jool is at 90.5437° . Graph here (I couln't believe how it may have happened until seen that).

About the State (be it with the MA main window, the State Log, or else), I have a request. Quite often we are dealing with hyperbolic orbits, and then the SMA is negative. I am most interested in the Periapsis radius value, but that is not shown. You certainly have it easy how to get Periapsis from SMA and eccentricity, but I would like the first to be available as well.

Edited May 2, 2014 by diomedea
Graph link added

[Arrowstar]

Great. While looking how to find the inclination with Jool's orbit (as suggested) I just discovered an Easter Egg .

I still had KSP TOT open, no log saved from the previous session and I was looking to find that. And then I found "State Log" with the Tools menu in MA (still undocumented).

And you're right, inclination with Jool is at 90.5437° . Graph here (I couln't believe how it may have happened until seen that).

Excellent. Yeah, the state log is undocumented because at the moment there is no documentation.

About the State (be it with the MA main window, the State Log, or else), I have a request. Quite often we are dealing with hyperbolic orbits, and then the SMA is negative. I am most interested in the Periapsis radius value, but that is not shown. You certainly have it easy how to get Periapsis from SMA and eccentricity, but I would like the first to be available as well.

So here's another Easter egg: hover over any state log entry on the GUI and see what you find.

[Arrowstar]

Hi everyone,

I'm happy to announce the release of KSP TOT v0.12.1. This is a bug fix and polish release of v0.12, specifically the Mission Architect application. It addresses a number of items that diomedea found in his testing as well as other things I've wanted to fix or improve.

Here's the changelog, all associated with Mission Architect:

• Fixed issue with vectorized kepler solver;
  
• Added lower periapse warning and low propellant warnings;
  
• Added zoom slider to orbit display;
  
• Added tooltips;
  
• Updated colormaps for bodies in bodies.ini:
  
  • Jool
    
  • Moho
    
  • Gilly
    
  • Vall
    
  • Bop
    
  • Pol
    
  • Eeloo
    

  [*]Added an unsaved work check upon closing MA;

  [*]Fixed bug with the Advance Script To Selected Event function;

  [*]Added scroll bar to Alerts/Warnings box;

  [*]Fixed rare display issue with Mission Optimizer;

  [*]Numeric constraints may now have an associated reference body;

  [*]Active constraint listbox tooltip now displays information on active constraints;

  [*]Arrow buttons on orbit visualization now wrap;

  [*]Fixed bug that occured when mission had more than 11 events;

  [*]Fixed bug that could cause infinite propagator recursion when going to true anomaly of 0.

The ZIP package is immediately available in the first post of this thread. Please let me know about any bugs you find or questions you have. Thanks!

[futrtrubl]

It appears KSPTOTConnect is unable to connect when the game is full-screen and not focused. I suspect it has something to do with KSP seeming to pause all activities when it's not focused and full-screen (loading is paused if it looses focus and time never progresses in the flight scene). This may not be something that can be fixed, only worked around by using KSP windowed.

Edward

[Arrowstar]

It appears KSPTOTConnect is unable to connect when the game is full-screen and not focused. I suspect it has something to do with KSP seeming to pause all activities when it's not focused and full-screen (loading is paused if it looses focus and time never progresses in the flight scene). This may not be something that can be fixed, only worked around by using KSP windowed.

Edward

Interesting. Thanks for the bug report. Could you provide some more information? I'm not clear on how the game could both be full screen (that is, not running in a windowed mode) and not focused at the same time.

[Arrowstar]

Btw, everyone, I've just added a bunch of Mission Architect examples to the KSPTOT ZIP package that I hope you all will find interesting. They are:

1) A mission to Laythe using a ballistic trajectory to Jool;

2) A mission to Laythe using a gravity assist from Eve;

3) A two burn mission to geostationary orbit around Kerbin; and

4) A swing-by of the Mun on a free return trajectory back to Kerbin.

These four examples are included with the KSP TOT download as .MAT files, as I said. The first example mission is more or less the completed tutorial. Definitely check out the Eve flyby case, that was a lot of fun to put together.

Let me know if you have any questions!

[futrtrubl]

Interesting. Thanks for the bug report. Could you provide some more information? I'm not clear on how the game could both be full screen (that is, not running in a windowed mode) and not focused at the same time.

IE I'm running it full-screen but I have alt-tabbed out to get KSPTOT to connect with it.

[Arrowstar]

IE I'm running it full-screen but I have alt-tabbed out to get KSPTOT to connect with it.

Understood. Do you have the "Simulate in Background" option enabled in the KSP Options?

[futrtrubl]

Understood. Do you have the "Simulate in Background" option enabled in the KSP Options?

Didn't realize that option existed but I just checked and it is enabled.

[Arrowstar]

Okay. Can you try running the game in windowed mode for me and tell me how KSPTOTConnect works then?

[futrtrubl]

Okay. Can you try running the game in windowed mode for me and tell me how KSPTOTConnect works then?

Oh it definitely works fine in windowed mode which is what I have been doing.

Edit: Well actually, some oddities in the Rendezvous Maneuver Sequencer. I am able to import orbit data fine via Connect or from .sfs. Would I be right in assuming that "Search Initial Epoch" would be the earliest time to look for a good initial burn? If so could we get "get UT from KSP" like the main window?

Now to the oddities.

1. If I manually input the burn info into KSP I get good results though there is ambiguity as to which periapsis to burn before/after for DV Maneuver 1. That's not a problem when I am actually rendezvousing with an object since I can see my closest approach and shift the time to get it to match. When I'm aiming for a trailing/leading orbit it becomes a problem.

2. When I upload the maneuver to KSP via Connect (using UT) the maneuver dV and time values look correct in MJs Node Editor, however the games dotted line orbit prediction and the dV remaining bar to the right of the nav ball are completely different (in my case for a 787m/s burn it shows about 0.4m/s). If I hit Update in the Node Editor then it puts me in the correct transfer orbit but.....

3. it seems the time is wrong because the object I am rendezvousing with is not in the position indicated in the Maneuver Sequencer graphic for the burn and I don't end up close to it after the second burn.

Edward

Edited May 4, 2014 by futrtrubl

[Arrowstar]

So the only reason this would happen is if you didn't set the "Search Initial Epoch" correctly. Can you try pulling it from the game (set the current time to the Search Initial Epoch) and see if that helps any.

[futrtrubl]

I did that, used the Earliest Departure Time box on the main window to get the time from KSP via connect then copy pasted to the rendezvous window as it doesn't have the right click option to get from the game as above. The time it gave me for the burn was in the near future as it should be. I also double checked the time it gave against the UT time in the sfs and it was roughly the same as it should be.

Edward

[Arrowstar]

Can you take a screenshot of the application and post it here so I can take a look?

Thanks!

[futrtrubl]

No problem. They are up at http://edowner.net/files/screenshots.rar In it are screen shots of the KSPTOT main screen where I got the current time from KSP, the rendezvous sequencer, the map in ksp after uploading the maneuver and after hitting update in mechjeb node editor. I noticed MJ had a bug with the sign of the normal dV, I updated but got the same results and that wouldn't have put the orbit off as much as it was.

Edward

[Arrowstar]

No problem. They are up at http://edowner.net/files/screenshots.rar In it are screen shots of the KSPTOT main screen where I got the current time from KSP, the rendezvous sequencer, the map in ksp after uploading the maneuver and after hitting update in mechjeb node editor. I noticed MJ had a bug with the sign of the normal dV, I updated but got the same results and that wouldn't have put the orbit off as much as it was.

Edward

Got it, I'll take a look when I get home from work tomorrow.

[Arrowstar]

Hey Edward, can I get the savegame you're using to test RMS with? I'd like to try something within KSP itself. Thanks!

(If it's not a stock savegame, can you duplicate it and remove the non-stock vessels, or duplicate the issues on a stock only install?)