Workaround for this?

[Vector]

When I grab an asteroid, I notice significant swings in velocity while I attempt to (slowly) rotate. The variation in velocity is about 0.5 m/s, and is presumably because the physics is computed relative to the command pod and not the center of mass.

This makes it exceedingly difficult to pilot a captured asteroid through gravity assists to other planets because 0.5 m/s is huge relative to the accuracy I need. Making fine adjustments should be easier, not harder with a huge mass.

Any known workarounds? I think I am going to try hacking the grabber to see if that helps..

Edited April 26, 2014 by Vector

[Sirrobert]

No, the physics are computed relative to the center of mass

I have no idea how to fix it though, sorry. I only just started catching my own asteroids

[luizopiloto]

try to turn of SAS and RCS... this will help...

[cantab]

Let it settle down fully after rotating (or cheat with a burst of timewarp) then make your burn with any corrections needed.

[Kasuha]

The variation is due to a chain of facts:

- velocity is calculated from CoM motion

- CoM is calculated iteratively relative to root part of the ship (including direction relative how the part is rotated)

- as your ship-asteroid connection bends while your ship is trying to rotate the asteroid, this iterative CoM follows motion of the root part (command pod) wildly and is only slowly updated towards real CoM, causing wild apparent moving of CoM

Workaround:

You need to rotate your ship by the asteroid, not the other way around. Install torque or RCS probes around the asteroid and have only limited amount on the ship itself

Another workaround:

Make sure the root part is connected to the asteroid passively. Break your ship in two, and connect separately your command module and the main part of the ship including engines and SAS wheels.

If you prevent movement and especially rotation of the root part relatively to the asteroid, things will get much easier.

[Dust]

If the center of thrust of your RCS is away from your center of mass, you will get a net velocity change. To get around this you'd need to attach an RCS module on the opposite side of the asteroid to place the RCS center of rotation as close as possible to the CoM.

[mellojoe]

FYI: I have had this problem with my KSP. I no longer do asteroid missions. I have never been able to perform asteroid missions without massive wobbles, and that makes it impossible to do anything with them.

[Vector]

The variation is due to a chain of facts:

- velocity is calculated from CoM motion

- CoM is calculated iteratively relative to root part of the ship (including direction relative how the part is rotated)

- as your ship-asteroid connection bends while your ship is trying to rotate the asteroid, this iterative CoM follows motion of the root part (command pod) wildly and is only slowly updated towards real CoM, causing wild apparent moving of CoM

Workaround:

You need to rotate your ship by the asteroid, not the other way around. Install torque or RCS probes around the asteroid and have only limited amount on the ship itself

Another workaround:

Make sure the root part is connected to the asteroid passively. Break your ship in two, and connect separately your command module and the main part of the ship including engines and SAS wheels.

If you prevent movement and especially rotation of the root part relatively to the asteroid, things will get much easier.

Wonderful! This actually works. Props to Kashua for understanding the problem.

I tested it with this craft and I can see now how having the torque applied through the asteroid to the control point (didn't realize it had to also be the root part) will produce extremely low strain on the linkage between the control point and the asteroid. In that sense both your workarounds are similar. I see now, regardless of the shape of the part tree, a single vessel (meaning single claw) can never apply torque without straining the links between the root part and the asteroid. For my test I also disabled the reaction wheel in the probe itself so the root part / center of control is 100% passive.

In VAB the root part is on the right, just below the claw.

I still need some work on aligning the engines and control point to the center of mass, but at least now it is possible, whereas with the maneuver indicator and velocity wildly there's no way. Maybe with some docking ports, a good understanding of the quirks, and some cleverness I can get everything aligned.

Thanks again!

[pxi]

I usually get pointed in the right direction, free the pivot on the claw, and start with a very low thrust.

Too much thrust too quickly is usually where it goes wrong for me.

EDIT: If the last part sounds like anything other than piloting advice, wash your mind out with soap.