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

[sebi.zzr]

For cases where the ship does nothing or just wobbles a little, right click a command pod part that's in line with the engine(s) you want to use then click the Control From Here button.

KSP .23 seems to have a habit of setting that to no part. Smart A.S.S. can still use reaction wheels (even without any command pod part on the ship) to change a ship's orientation. But without a command pod part or without one set to Control From Here, you get nothing or just a

jiggle from landing guidance and maneuver planner.

i incrised my stable orbit to 30 km and works as should,scenario was, my stable orbit on minimus befor was 20 km.

but foud out,mechjeb refuse to auto stage KWrocketery faring cone....ignition>start>drop fuel busters(auto stage)>drop main tank ad motor(auto stage)>drop farig(manual stage)>start orbit motor(auto stage)>...

Edited January 8, 2014 by sebi.zzr

[Starwaster]

For cases where the ship does nothing or just wobbles a little, right click a command pod part that's in line with the engine(s) you want to use then click the Control From Here button.

KSP .23 seems to have a habit of setting that to no part. Smart A.S.S. can still use reaction wheels (even without any command pod part on the ship) to change a ship's orientation. But without a command pod part or without one set to Control From Here, you get nothing or just a

jiggle from landing guidance and maneuver planner.

That sentence got away from you, didn't it.

Could you add an option in the ascent guidance module to launch when the target's orbit intersects the equator? This would make rendezvousing with a ship on an inclined orbit much easier. It doesn't have to wait for the target to be in a position to rendezvous with, just as long as the two ships end up on the same plane.

There's already an option there to launch into the target's plane. Technically that requires you to intersect the target's orbit. Same thing really.

[Bobe]

There's already an option there to launch into the target's plane. Technically that requires you to intersect the target's orbit. Same thing really.

I know you can set the inclination manually, but that requires that you also manually warp until the target's plane intersects KSC, or just before. I would like MJ to handle that automatically.

[AbeS]

That's not what he said, he talked about launch into the target's plane, which is below launch to rendezvous if I remember correctly

[Bobe]

That's not what he said, he talked about launch into the target's plane, which is below launch to rendezvous if I remember correctly

Well that's what I thought initially, but I know that launching to rendezvous does not match the target's plane, so I thought perhaps he was referring to the inclination setting.

If using launch to rendezvous, you still have to set the inclination manually, which isn't that big a hassle, but the guidance led the rocket into an a orbit with an inclination of about 20-30 degrees to the target's orbit. Obviously this still creates intersect points, but being at such an angle, rendezvousing is difficult and you waste a lot of fuel matching the planes once in orbit.

Edited January 9, 2014 by Bobe

[AbeS]

Not launch to rendezvous, let me turn on KSP

Edit: why don't you try Launch into plane of target?

Edited January 9, 2014 by AbeS

[Bobe]

Not launch to rendezvous, let me turn on KSP

Edit: why don't you try Launch into plane of target?

http://i.imgur.com/S97i1cs.png

Don't know how I missed that.

Thank you for proving my ignorance.

[federicoaa]

Suggestion: add and option in the node editor to split a node into two or more nodes on the same spot but one period away. It's useful for the long burnings required for interplanetary travels that are seldon done in one shot.

[sonicsst]

Ok, I added 1/100 second precision to orbital period to last dev

This doesn't seem to have changed anything, the orbital period display still only displays to the second.

[sarbian]

Yes, I saw that yesterday. Sorry

I'll look into it

federicoaa : the same suggestion was made a few days ago. I'll think about it. I see how I can split into an odd number of burn but not for even.

I have concerns about precision too.

Edited January 9, 2014 by sarbian

[BigD145]

First: After 1 second, it starts to count down in milliseconds - So, between 0.999 and 0.000 seconds (Except it's denoted as 999-000 ms) - Watch the actual warp indicator.

Second: As said before, make sure you've engaged your autopilot. I generally do that FIRST because the "Launch to Rendezvous" tends to start high-warp immediately.

If you're in doubt - Hit autopilot, hit LTR, then hide the autoascent box and wait. It'll take off.

I've never seen a millisecond countdown. 1 > 0 > high warp to next rendezvous. Rinse and repeat. Pick a number between 100 and 161. I've used most of the builds.

[federicoaa]

the same suggestion was made a few days ago. I'll think about it. I see how I can split into an odd number of burn but not for even.

I have concerns about precision too.

Thanks for taking it into account

BTW, could you explain what you said about breaking it into odd number?

[sarbian]

Say you have to do a 900 m/s burn at time T

If I divide it with an odd number like 3 I have to do 1 burn of 300 m/s at T - 1 orbit, an other of 300 m/s at T and the last at T + 1 orbit

If I divide with an even number like 2 I don't know when to do the burns. Doing 1 burn of 450 at T - 1 orbit, and an other at T will not put me on the right orbit. And doing one at T-1 and the other at T+1 is the same as doing 3.

[diomedea]

Say you have to do a 900 m/s burn at time T

If I divide it with an odd number like 3 I have to do 1 burn of 300 m/s at T - 1 orbit, an other of 300 m/s at T and the last at T + 1 orbit

If I divide with an even number like 2 I don't know when to do the burns. Doing 1 burn of 450 at T - 1 orbit, and an other at T will not put me on the right orbit. And doing one at T-1 and the other at T+1 is the same as doing 3.

Understand you are concerned with precision, and thinking if you split evenly the burn, one orbit before and one orbit after the correct burntime, you would still get the same. However that isn't the case. After each partial prograde burn, each orbit takes longer, and therefore the equivalence does not hold.

If possible, MJ should compute the phase angle to the destination planet with a trajectory that takes the time of all the single orbits summed up.

[sebi.zzr]

is there a way to manualy turn AR202 on/off?

[sarbian]

diomedea : yes, it's an over simplification. I am sure there is some math to tell me how to do that properly. I just need to find someone to solve it or take a few hours to work it out with WolframAlpha help.

sebi.zzr : ?

[LameLefty]

I'm glad to see I'm not the only one to request support for multi-orbit transfer burns. But I don't want to drive sarbian to drink or make him rage-quit in response to demanding users!

Here's my thought on precision, so far as it matter: I have NEVER had a transfer burn of more than maybe 4 or 5 minutes that was terribly precise anyway, even when done in one go. My usual way of dealing with this, learned several KSP versions ago, is to wait until I'm out of Kerbin's SOI, then plot a correction burn some days/weeks down the line. Usually, unless it's Moho or maybe Dres, the correction burn is a relatively short burn of a a couple dozen to maybe a couple hundred m/s. Is it the most efficient? Maybe not, but it works.

[codepoet]

Say you have to do a 900 m/s burn at time T

If I divide it with an odd number like 3 I have to do 1 burn of 300 m/s at T - 1 orbit, an other of 300 m/s at T and the last at T + 1 orbit

If I divide with an even number like 2 I don't know when to do the burns. Doing 1 burn of 450 at T - 1 orbit, and an other at T will not put me on the right orbit. And doing one at T-1 and the other at T+1 is the same as doing 3.

My experience and thinking about this suggests that it is only the final burn that counts. I believe that it is necessary to do previous burns to raise the AP, but arrive at the PE in the place and at the time that a single burn would have occured. Then the final burn is the one that puts you interplanetary. This is why one of my wishlist manoeuvres is "raise Ap to arrive at Pe at a certain time."

I am on my phone, I will write more at my pc

[diomedea]

@ LameLefty: concur, both on keeping sarbian in a good mood, and in general about the "precision" thing.

More than 99% of my flights, I am just fine with a mild correction burn anyway. On the other end, I am not even sure if a correct analytical solution to this problem is computable. If not, either MJ is able to find solutions by recursion (and I have no knowledge if that's the case), or we have to accept lower precision, similar to what I may obtain by simple projection of the destination planet position at a time equal to the amount spent to fly all orbits required to complete the burn. If I know how much DV I can put at each passage, I can compute the orbital period of each next orbit, so that time is rather easily computed. My method only holds in case the planets are on low eccentric orbits and I don't even dare to make a transfer with an inclined orbit that way. But, if I want to get to Eeloo with a low-thrust engine, I may really depend on a tool like MJ to find a better solution.

[sarbian]

sarbian in a good mood

You'll have to speak with my manager for that

I am quite sure there is an "easy" solution to this. I

If something is added I'd rather have it work with most maneuver and not just the planetary transfer.

[Holyvision]

Also having landing not functioning, but even worse. It reduces time warp but then does nothing continuing to orbit without changing orientation nor engaging throttle, but the indicators for pitch, yaw, and roll bounce back and forth rapidly. Also not using FAR but am using DR.

I am having this exact same issue with the current MJ version. I have noticed that it happens 95% of the time when I am using a probe-core for control, as opposed to a command pod (which then has a failure rate of about 50%). I have tested this on all the major planetary bodies, as well as Minmus and the Mun. If I select a specific landing location, or if I hit "land somewhere", MJ will correctly perform the burn to exit it's orbital path, and will then correctly angle towards retrograde (+/- some if corrections are needed).

However, after this step (ship is currently on a sub-orbital course towards the surface) MJ will stop throttling the ship completely. The Landing Autopilot will read "final decent xxxxxkm". The Autopilot will then allow the ship to crash full-on into the surface and explode, without slowing down or correctly positioning for landing after the final decent stage. The Roll/Pitch/Yaw controls still show use as normal, however there is no thrust control again. If I cancel the auto-pilot, intentionally incorrectly orient the ship, then re-engage the autopilot -- it will correct orientation, but will still not thrust what-so-ever.

Been at this for a long time, so yes, I have fuel, the required Delta-V, electrical power, etc etc.

At first I assumed it was a conflicting mod, as MJ has almost always been very good to me; but then I removed all mods but MJ and started a new game to test it out, and got the same results. A couple times is seemed to get stuck on the initial be-orbit burn, and would just keep burning on forever past the initial Delta-V requirement until I took over with manual control.

Perhaps it's getting stuck in a loop somewhere in the landing guidance code -- but Lord knows I don't know jack when it comes to this stuff.

[sarbian]

This doesn't seem to have changed anything, the orbital period display still only displays to the second.

Fixed for real in #163.

but then I removed all mods but MJ and started a new game to test it out, and got the same results. A couple times is seemed to get stuck on the initial be-orbit burn, and would just keep burning on forever past the initial Delta-V requirement until I took over with manual control.

Get me the savegame that does that and I'll look into it.

[sebi.zzr]

is there a way to manualy turn AR202 on/off?

when you use AR202 it`s operational(green light),when all windows of MJ are closed (red light),it was just idea if posible to manually switch between states operatinal/disabled by adding MM.cfg or edit existing *.cfg?

[Galane]

Try landing on Gilly. The target marker hovers way above the surface.

Land at target doesn't want to do anything after the initial burn, which doesn't bring the periapsis down to the surface. I either have to do a manual retrograde burn or use land somewhere then stop it to get the orbit intersecting the surface before land at target will go ahead and finish landing.

[sarbian]

when you use AR202 it`s operational(green light),when all windows of MJ are closed (red light),it was just idea if posible to manually switch between states operatinal/disabled by adding MM.cfg or edit existing *.cfg?

No, there is no config for the case. If you don't like the case and its light you could add the MechJeb module to all your command pod with a Module Manager script.