Attitude Control, the NASA Way

[antiquark]

Has anyone considered using NASA algorithms for attitude control, in particular, in the vacuum of space? I was reading the book 'The Apollo Guidance Computer' and the authors talked a bit about attitude control. There were two interesting points: (1) in space, the control is simpler because there\'s no air drag. And (2), the control algorithm requires a 'deadband' or 'deadzone' within which the RCS thrusters are turned off. Without the deadband, you have the thrusters continually firing back-and-forth, using up fuel.

I\'ve attached two diagrams from the book. One is generic, the other one shows how the attitude control is separated into 'decision areas.' The vertical axis is the rate error (degrees per second) and the horizontal axis is the angular error (degrees).

In the Apollo crafts, you could choose two accuracy modes. Fine grained gave a deadband of 0.5 degrees, and coarse grained had a deadband of 5 degrees. The latter number seems quite large, but hey, if it worked for Apollo, it should be OK for KSP!

[Majiir]

See this thread I made on the topic. A lot of that stuff is out of date, and a lot of it is also just plain bad information. I abandoned it when I realized KSP\'s requirements are just so different from anything a real-world spacecraft deals with. In some ways, it\'s easier—for example, there\'s no instrument error to deal with. We always know exactly how fast the ship is going and exactly where it is. Great! Then again, it\'s difficult because you aren\'t designing a specialized controller for a single spaceship—you\'re building something for any arbitrary configuration of parts. It\'s not harder or easier by any obvious degree; it\'s just different, and thus requires some dedicated analysis.

If I ever get back into plugin coding, I\'ll probably work on a unified control system for engines, SAS, RCS and control surfaces, and then see how that might play into control software.

[fowlplaychiken]

It would be nice, if it puts an end to asas-enduced occilation.

[antiquark]

It would be nice, if it puts an end to asas-enduced occilation.

It would for sure. The current ASAS modules rely on PID equations for control. However PIDs are generic and have to be fine-tuned for every situation. The NASA equations above, are optimized for RCS thrusting, so you wouldn\'t have the problem of the ship oscillating like a pendulum.

[AeroPelican]

Dead zones are for joysticks, with the key board not touching anything is a dead zone.

[Mr_Orion]

Dead zones are for joysticks, with the key board not touching anything is a dead zone.

Err... we all know this. They were talking about an attitude control deadzone to stop oscillation...

[RC1062]

Dead zones are for joysticks, with the key board not touching anything is a dead zone.

It is like a margin of error, as long as the craft is pointing within this zone no more corrections need to be made.

[Romfarer]

Surprice surprice, The new Lazor Guided Flight enhancments for rockets that were implemented in the last update or Lazor System does this to a degree. It's not the NASA algorithms but it uses a similar system to set directions and keep the vessel there. It definately don't have fixed dead-zones and i am not sure how it compares to the regular ASAS in the dead-zone area but it is definately more powerfull than regular ASAS.

Feel free to test it at let me know if you want a dead-zone button added to it...