Pecan Posted January 11, 2015 Share Posted January 11, 2015 Have you tried it with FAR? Non-atmosphere landings are absolutely perfect, but mechjeb does horrible with atmosphere landings as far as accuracy goes.Why are you surprised? Link to comment Share on other sites More sharing options...
Galane Posted January 11, 2015 Share Posted January 11, 2015 How is that an answer when his sole complaint is about the accuracy of atmospheric landings in the first place?Not accuracy, the impossibility of atmospheric landings in .90 with only MechJeb and Stock Drag Fix. I tried build 386 last night, still erroneously showing corrections of 100+ m/sec in random directions yet only using short thrusts. The numbers drop fast but then a new needed correction shows in a different direction. Link to comment Share on other sites More sharing options...
smjjames Posted January 12, 2015 Share Posted January 12, 2015 Why are you surprised?I'd just like reasonable accuracy, not off by kilometers. Link to comment Share on other sites More sharing options...
Pecan Posted January 12, 2015 Share Posted January 12, 2015 I'd just like reasonable accuracy, not off by kilometers.Why are you surprised that when you change the rules, MJ doesn't know how to play? Link to comment Share on other sites More sharing options...
Kowgan Posted January 12, 2015 Share Posted January 12, 2015 (edited) Have you tried it with FAR? Non-atmosphere landings are absolutely perfect, but mechjeb does horrible with atmosphere landings as far as accuracy goes.Interesting; In my experience, the opposite happens.Atmospheric landings are amusingly precise, while vacuum landings have a bit of, yet annoying offset. I'll try it with NEAR and see if I get any difference. Edited January 12, 2015 by Kowgan Link to comment Share on other sites More sharing options...
smjjames Posted January 12, 2015 Share Posted January 12, 2015 Well, actually, before the ship hits the atmosphere, it has no problem getting the targeting accurate, but once it hits the atmosphere, the actual landing spot drags way off, so, it's like it's not accounting for atmospheric drag correctly or something with FAR. Link to comment Share on other sites More sharing options...
Kowgan Posted January 12, 2015 Share Posted January 12, 2015 I think this happens with FAR/NEAR due the way it affects the game. Since it messes with Aerodynamics, your ship's position will affect how much drag it takes, and as such, how fast will you lose speed. Therefore, oscilating the Landing Prediction's position. I don't think the problem is MechJeb in this case; MJ is simply showing you the input from FAR/NEAR. Yet, even without FAR/NEAR, my issue persists. ;( Link to comment Share on other sites More sharing options...
steve_v Posted January 12, 2015 Share Posted January 12, 2015 (edited) it's like it's not accounting for atmospheric drag correctly or something with FAR.Yeah, that.The issue here, as I understand it, is that MechJeb precalculates landing coordinates based on a known aerodynamic model. This is fairly easy to do in stock, since drag depends only on mass, velocity and atmospheric density - all of which are known in advance. With FAR/NEAR however, the drag experienced by an object moving in atmosphere depends also on it's shape and orientation (or ballistic coefficient) - things MechJeb does not and, to a certain extent, can not know in advance. Hence the innaccuracy. True FAR/NEAR support would entail integrating something like the Trajectories mod into the landing AP, as well as making assumptions about the re-entry orientation of the craft and/or or making in-atmosphere adjustments, accounting for lift etc... it'll get real complicated, real quick. You'd essentailly have to teach MechJeb to fly an aircraft in FAR.IIRC, stock drag fix makes drag proportional to volume rather than mass (still very wrong ) so I suspect this would be more reasonable to implement, assuming the required values can be extracted from the game.That said, it's been so long since I played without FAR that I can't really comment on landing AP performance, except that it doesn't work too well with FAR. IIRC it worked fine with stock last time I used it. I don't do C# either, so you'll have to ask Sarbian how hard this actually is Edited January 12, 2015 by steve_v Link to comment Share on other sites More sharing options...
Starwaster Posted January 12, 2015 Share Posted January 12, 2015 IIRC, stock drag fix makes drag proportional to volume rather than mass (still very wrong ) so I suspect this would be more reasonable to implement, assuming the required values can be extracted from the game.No. It negates the mass from resources. Link to comment Share on other sites More sharing options...
sarbian Posted January 12, 2015 Share Posted January 12, 2015 It's not hard but require some significant change in the code to make the atmospheric simulation plugable. I will do it since It will most likely make the transition to the future stock areo model easier, but it won't happen overnight. Right now my focus is on making to improve the velocity / angular velocity / forward vector used by Mechjeb. The current code uses info reported by that game but those are inaccurate ( they report value for the command pod and not at the CoM ). I have the code that compute a proper velocity at the CoM but nothing yet for the angular velocity and forward vector. Once I have those this should greatly improve the handling of rocket that flex (ie anything big in KSP) Link to comment Share on other sites More sharing options...
Kowgan Posted January 12, 2015 Share Posted January 12, 2015 Thank you, Sarbian. Link to comment Share on other sites More sharing options...
steve_v Posted January 12, 2015 Share Posted January 12, 2015 No. It negates the mass from resources.Ahh, then it's even less of a "fix" than I thought Link to comment Share on other sites More sharing options...
smjjames Posted January 12, 2015 Share Posted January 12, 2015 Yeah, that. You'd essentailly have to teach MechJeb to fly an aircraft in FAR. I thought MechJeb already knew (or pretty much) knew how to fly an aircraft in FAR? Link to comment Share on other sites More sharing options...
Starwaster Posted January 12, 2015 Share Posted January 12, 2015 Ahh, then it's even less of a "fix" than I thought It works.(10 chars) Link to comment Share on other sites More sharing options...
Pecan Posted January 12, 2015 Share Posted January 12, 2015 (edited) I thought MechJeb already knew (or pretty much) knew how to fly an aircraft in FAR?No, and this is the point:= MJ knows about stock aerodynamics, by installing FAR you've changed all the rules of atmospheric flight and all MJ's calculations are thrown-off.ETA for below: including me Edited January 12, 2015 by Pecan Link to comment Share on other sites More sharing options...
BudgetHedgehog Posted January 12, 2015 Share Posted January 12, 2015 Hell, a not-insignificant number of PEOPLE don't know how to fly aircraft with FAR.. Link to comment Share on other sites More sharing options...
smjjames Posted January 12, 2015 Share Posted January 12, 2015 Yeah, I've heard that it's easier to fly with NEAR and easier to build the plane with FAR or something? Link to comment Share on other sites More sharing options...
NihilRex Posted January 12, 2015 Share Posted January 12, 2015 Yeah, I've heard that it's easier to fly with NEAR and easier to build the plane with FAR or something?NEAR is basically FAR without the extra information windows available to you. Link to comment Share on other sites More sharing options...
Steambirds Posted January 12, 2015 Share Posted January 12, 2015 Will a module for asteroid rendezvous be available in the future? (Automatically grabbing the asteroid with your Klaw, and possibly even centering your vessel on the CoM.) Link to comment Share on other sites More sharing options...
smjjames Posted January 12, 2015 Share Posted January 12, 2015 Wouldn't the normal rendezvous work? May have to use the docking AP when close. Link to comment Share on other sites More sharing options...
WuphonsReach Posted January 13, 2015 Share Posted January 13, 2015 Yeah, I've heard that it's easier to fly with NEAR and easier to build the plane with FAR or something?FAR's big advantage is the nice flight modeling UI that you can access in the SPH in order to tune your CoM/CoD/CoL numbers. There is a simulation button thingy that you press and you get a whole lot of numbers where "green is good". Plus the stability assist window where you can lock in "keep wings level" and things like that. Link to comment Share on other sites More sharing options...
Starwaster Posted January 13, 2015 Share Posted January 13, 2015 Wouldn't the normal rendezvous work? May have to use the docking AP when close.That's all I do with asteroids... Link to comment Share on other sites More sharing options...
SilentWindOfDoom Posted January 14, 2015 Share Posted January 14, 2015 I'm impressed with the Rover Autopilot. Mun rovers have long been problematic due to their tendency to flip when you are traveling at any speed, but since i slapped the autopilot on they're actually usable as a platform. Great work. Ive been catching some serious air (with flips!) and landed safely on the wheels (or tracks) when traveling between biomes. The only thing that gets you are sudden extremely steep gradients, but that's to be expected and can be avoided with some clever waypointing. Thanks for opening up a new area of ksp for me. Javascript is disabled. View full album Link to comment Share on other sites More sharing options...
smjjames Posted January 14, 2015 Share Posted January 14, 2015 What mod are the tracks from? Link to comment Share on other sites More sharing options...
NBZ Posted January 14, 2015 Share Posted January 14, 2015 Is it just me or does MechJeb get Tylo's gravity all wrong (only twice that of Gilly)? Link to comment Share on other sites More sharing options...
Recommended Posts