Asteroid capture is so broken.

[chd]

Ok, so, going to capture an asteroid, per contract, to build a station on. Found one, tracked it, rendevouz'd with it, grabbed it with a claw, and now the (not) fun begins.

How the F do you control the thing??? It's class D. I know about setting the target as the center of mass; doesn't seem to help. Any time I try to make any control input, my ship starts flapping back and forth as though it were made of silly putty, and so does the asteroid, acting like it weighs about a pound. I can set the asteroid CoM as target, but then if I tell MJ to point T+, wild oscillation.

At this point this is no longer fun, and barring a fix I'll just decline all contracts involving an asteroid go play Fallout instead...

[FleshJeb]

Lock the pivot on the claw if you haven't. If it's still floppy, I think you can turn on Autostrut: Grandparent for the 'roid.

The way I always used to do it was bring landing legs to stabilize and set up some tension against the asteroid, once it was clawed. Nowadays, i think you have to turn on Same Vessel Interaction for the asteroid and the legs.

It was always a challenging problem, the trick is to use basic principles of engineering and geometry to make the connection more robust. KSP has flaws as a simulation, but I've never encountered a situation where making something fundamentally structurally sound didn't work.

[Linkageless]

As @FleshJeb says, autostrut your main structural components to the asteroid.  I use grandparent on any significant structures, but to heaviest part on one or two will link those directly to the asteroid to stop those wobbles.  Use heaviest or root part autostrut sparingly and temporarily if possible.  Be ready to change back to grandparent on those parts if you're disconnecting from the asteroid or risk the wrath of the Kraken.

Be aware that parts near the klaw could be under tremendous strain.  Don't push them too far, or they will weirdly slip, deforming your ship, or even fail in an explodey way.  Very long vessels with lots of reaction wheel/RCS are especially prone to this.

Control point is significant when it comes to SAS oscillations.  Experiment with 'control from here' on the klaw and other places.  You may even want to turn SAS off while you slowly get it to the general attitude you want.

Larger asteroids are annoying to turn.  If you need to go from prograde to retro, you may even want to decouple and go round the other side.

The very best way to deal with the big ones I've found is to carry a small fleet of drones, each with their own klaw, reaction wheel and RCS.  Deploy those around the asteroid and you'll all be united as one vessel.  Doesn't have to be very neat as long as they're reasonably spread across the surface, the RCS will work it out from there.

Extrapolating from that, you may even want to have your vessel split into functional modules, each tacked onto the asteroid, which is fun.  Just be sure your main drive units are all equally lined up (I'd keep them in one module), you control from there, and that is perfectly aligned with the centre of mass.

 

[radonek]

On 9/10/2020 at 11:05 AM, Linkageless said:

… Be aware that parts near the klaw could be under tremendous strain. …

Also be aware that putting struts directly onto klaw is a krakenbait.

[king of nowhere]

you can also use less propulsion power. if your claw attachment is floppy, exherting less force on it is likely to strain it less.

[Linkageless]

On 9/13/2020 at 4:19 PM, king of nowhere said:

you can also use less propulsion power. if your claw attachment is floppy, exherting less force on it is likely to strain it less.

Indeed - I've on occasion needed to reduce thrust just to give reaction wheels/RCS a chance of keeping it on the straight and narrow.

Fortunately, this happens automatically when you're relying on mining the asteroid as you go, have too few drills and too many thirsty engines!

[chd]

On 9/13/2020 at 3:42 AM, radonek said:

Also be aware that putting struts directly onto klaw is a krakenbait.

This applies equally to auto-struts, I assume?

[Okhin]

Yep. But you can strut the part on which the klaw is tied.

Personnaly, on asteroid, I autostrut the first static part after the klaw to the heaviest part (which always will be the asteroid). I precise static, because I cluster claws and put them on piston, to have a bit more stability when grabbing an asteroid. My process of establishing a firm grab at asteroids goes as follow :

I fully extend the first piston, grab the asteroid, free pivot, target center of mass, retract the piston half way, and then extend the three other in sequence (extend, grab, free pivot), once everything is grabbed, I lock all klaws, retract the pistons as much as I can, lock them, disengage the engines and autostrut to heaviest part the part where all the pistons are attached. At this point, I'm aligned with center of mass, and firmly dock. The forces will be spread on the different klaws, the piston are now basically just static rods of iron. I'll just need a hell lot of reaction wheel to turn the asteroids, but 12 or more of the biggest one I can find will do it.

That's how I klawed that base on a class E :

(However, I think the engine of this base are way to underpowered to move the asteroid in any significant way. I think I can reach .01 TWR, but I'll have to check)

[Joontry]

Try turning off SAS, and using about 1.5m/s^2 of acceleration.

If you target the center of mass and attach to that with the claw, and lock pivot when you line up with the asteroid's center of mass as closely as possible, then you should be able to push it for a big burn and stay mostly pointing at the maneuver node without the SAS, assuming you have enough reaction wheels or RCS. I find that since the center of thrust and center of mass are almost never perfectly aligned when attached to an asteroid, the SAS tries to fight the imperfections too much during a burn and you end up wobbling out of control until you break apart. This worked for me for a 1km/s burn from solar orbit near Kerbin to a Duna encounter trajectory with a class D asteroid that was impossible with the SAS turned on, but successful with no SAS and 1/4 thrust. I also used only reaction wheels, not RCS. Hope that helps!

Edited February 19, 2021 by Joontry

[VoidSquid]

KER has the option to display the thrust torque offset angle, minimize that after capture, before you engage the engines.