Improve (fix) the docking computer (warning, a bit technical)

[Cyberneticist]

I've been watching the docking computer at work.

Basically, it should be able to easily out-dock human pilots and set the ship on the landing pad precisely and gently in minimal time. In reality, however, it has some obvious bugs (it tried to land my Cobra side first when docking at an outpost), and the control algorithm used for ship position and velocity nervously fiddles with the controls, but still does not reduce the control error to zero (i.e. it occasionally misses the landing pad or lands roughly).

Why is this?

(Warning, this is where it gets technical. Control engineer/systems theory guy here)

I assume the docking computer uses proportional (P) control for controlling ship orientation, velocity and position. The ships have nonnegligible inertia, so this type of controller will cause oscillations and is unable to reduce the control error to zero.

Using a proportional/integrating (PI) controller would be better suited for this task. Correctly tuned, it will reduce oscillations and control error. Adding a dead zone in which no control action is taking would further reduce oscillations. Once the control algorithm is improved, the docking computer could use a higher ship velocity when docking, which reduces congestion in the stations.

 

[norcus]

I've been watching the docking computer at work.

Basically, it should be able to easily out-dock human pilots and set the ship on the landing pad precisely and gently in minimal time. In reality, however, it has some obvious bugs (it tried to land my Cobra side first when docking at an outpost), and the control algorithm used for ship position and velocity nervously fiddles with the controls, but still does not reduce the control error to zero (i.e. it occasionally misses the landing pad or lands roughly).

Why is this?

(Warning, this is where it gets technical. Control engineer/systems theory guy here)

I assume the docking computer uses proportional (P) control for controlling ship orientation, velocity and position. The ships have nonnegligible inertia, so this type of controller will cause oscillations and is unable to reduce the control error to zero.

Using a proportional/integrating (PI) controller would be better suited for this task. Correctly tuned, it will reduce oscillations and control error. Adding a dead zone in which no control action is taking would further reduce oscillations. Once the control algorithm is improved, the docking computer could use a higher ship velocity when docking, which reduces congestion in the stations.

The docking computer definitely bugs out sometimes on outposts. Not sure if its intended. The docking computer could maybe be more efficient at stations too. I stopped using it altogether because I need the internal slot for other stuff and I manually dock faster than the computer anyway.

 

[dementia]

I'm hoping it is deliberately rubbish and that down the line better module classes and ratings will become available with the features you describe. However docking is very easy so it is not worth the slot it takes up. I'm hoping that other station designs and especially the planet landing expansions will actually require a higher degree of piloting skill thus making the docking computer a worthwhile tool to have.

 

[Cyberneticist]

I'm hoping it is deliberately rubbish

The thing is, the control behavior of the docking computer does not look like intentional performance degradation. With a proper control algorithm, it is possible to degrade the performance without having obvious signs (oscillations, control error) of a poorly designed controler. Docking computers of various grades could simply use different values for allowed maximum velocity and acceleration - the low-grade docking computer would take some time, while the highest grade docking computer would work with maximum thrust but still be able to land the ship gently and accurately.