What Moves On-Rails Ships, and What Causes Unexpected SOI Encounters?

[Geschosskopf]

My Remote Tech enterprise is continually being disrupted by on-rails ships being teleported hither and yon. This is only happening (so far) in the vicinity of Mun; satellites in Kerbostationary orbit are behaving themselves. And the really weird thing, this disruption is NOT happening, as you might expect it would, while controlling other ships at high warp. Instead, the disruption happens while I'm not looking at space at all, but upon first restarting the game for a new session, or while I'm in the VAB. IOW, I quit the game or go into the VAB with everything fine, but when I restart or go to launch what I just built, things are often all screwed up.

Here's an example:

Looking at Mun, the 2 things closest to it are a mapping probe in polar orbit and a station in equatorial orbit. Ignore them. Now look out ahead of Mun near the right edge of the pic. That's an RT relay satellite orbiting Kerbin in the same orbit as Mun, called MOS Leading (Mun Orbit-Sharing), designed to keep 1/2 of Mun covered with its cone. There's supposed to be an identical satellite, MOS Trailing, about the same distance behind Mun. But no. It is instead the thing in the big eccentric orbit around Mun. WTF? Something getting captured into orbit? That defies the laws of KSP physics.

And the thing is, this only just happened. The sequence of events leading up to this pic was as follows:

1. Built an interplanetary ship and parked it in LKO to await its transfer window in a few days. Did not touch any of the Remote Tech probes at all, especially not those near Mun. I did, however, cast a quick glance over my network from the safety of the map view, without switching to any of them (a reflexive habit of RT players) and all was fine.

2. I returned to KSC, went into the VAB, and build a small ship to send out to the station at Mun.

3. I launched this ship into LKO without doing anything with RT (it didn't even have an antenna on it) and then, while focused on Mun to fine-tune its transfer Pe in conic draw mode zero (the purple line just left of Mun), I couldn't help but notice MOS Trailing was now in Munar orbit. So I took the above pic.

So what causes this? It's extremely annoying. It's happened before to these satellites a couple times, but before the satellites were discovered in big eccentric orbits with Pe near Mun's and Ap near Minmus' orbit. This looks like what you'd expect if the satellite had, due to an imperfect orbit, drifted into Mun's SOI and gotten a gravity kick. So even though I was sure the orbits were good enough this shouldn't have happened (at least not within the 1st month of the game), I could rationalize it as some sort of rounding error compounded by warping. But this?!?!?! A satellite getting captured into another SOI?!?!?! In the complete absence of warping?!?!? WTF?

----------------------------------------------

Which brings me to my second question. I very frequently get "unexpected" encounters with Mun while flying ships in its vicinity. This happened while putting MOS Trailing in place to start with, in getting MOS Leading back from its big eccentric orbit, and in sending and returning many landers, probes, and stations to/from Mun. By "unexpected", I mean such encounters don't show up on the map at all, it's just POOF, you're in Mun's SOI, catching even the automatically generated SOI-change alarms of KAC by surprise. If you're going to Mun, the map will show the encounter you intend, but before your ship gets to that point, it encounters Mun. If you're leaving Mun, you've already left its SOI and are coasting down to your Ap at Kerbin, when POOF, you're back in Mun's SOI again.

These unexpected encounters almost always happen at relatively high levels of warp, because you are just trying to get to your next maneuver as quickly as possible. And of course, due to crossing the SOI boundary at such high warp, there's no telling what you'll get, and you might have precious little time to react before colliding with Mun, especially because the high warp is still going. Very annoying.

So what causes this?

Thanks.

[Red Iron Crown]

If I understand what you were doing correctly, you were faking a trojan point by putting the satellite into an orbit similar to the Mun's just outside its SoI. If so, how close were you to the SoI? If you are too close the rounding errors and impossibility of perfectly matching the Mun's orbit can put you into an encounter with it, and then all bets are off for how your orbit ends up. I suggest moving the satellites so that there is a more significant distance from the SoI, though I understand that compromises their coverage of the darkside a bit.

I've experienced the unexpected encounter bug before, not sure what causes it.

[undercoveryankee]

We don't know exactly what causes SoI jumps, beyond the obvious fact that the code that determines what SoI a ship is in doesn't always agree with the code that checks future orbit patches for SoI crossings.

The self-captures are probably because the jump took the satellite a fair distance inside the SoI boundary. You said in your mission report that an object that enters an SoI has to either exit at the same speed or impact. That's true if it enters at the actual boundary; the orbit is either hyperbolic or elliptical with an apoapsis outside the SoI, and either case is an escape. If it appears deeper inside the SoI, then it is only guaranteed to return to that altitude at the same speed, so if it ends up on an elliptical orbit an apoapsis between the appearance location and the actual boundary is possible.

I have a hypothesis that phantom SoI hops happen because when the game checks a ship's location, it is checking the Mun's position after the current world update against the ship's position before, or vice versa. If so, the danger zone is approximately the distance traveled in one update at the highest warp level encountered.

Mun and anything co-orbital with it have an orbital speed of 542.5 m/s. Assuming that positions of planets update on the "max physics delta time" setting and not on the constant 0.02s FixedUpdate clock, the tick is likely to be in the neighborhood of 0.04 s. 0.04 s * 542.5 m/s * 10,000x warp = 217 km that a ship could teleport into the SoI from outside due to update skew. As an idea of the scale, that's about halfway between 1 Mun radius and 10% of the SoI radius. Do you know how many kilometers outside the SoI your satellite was before it jumped?

[cantab]

In .23.5 at least the game fails to predict SOI changes when the ship in question doesn't cross the orbit of the satellite body, and may either show nothing or show the SOI change much too late. For example if you put a ship in a circular orbit a little smaller than the Mun's, the game will never tell you you're going to enter the Mun's SOI until it happens. (And then because the ship was moving slowly relative to the Mun it probably crashes.) KAC relies on the game's own prediction so it's subject to the same issue.

The game can also show closed orbits that extend outside the SOI of their primary. In the tracking station I've found it does this for unselected vessels, but then when you select it it correctly shows the vessel as leaving the SOI. Try that for your awry satellite.

[Geschosskopf]

If I understand what you were doing correctly, you were faking a trojan point by putting the satellite into an orbit similar to the Mun's just outside its SoI. If so, how close were you to the SoI?

The satellites were about 1.5-2x the SOI radius from Mun's center as shown on the map. You can see that in the pic by comparing the incoming purple path of the ship back at Kerbin to where the satellite at the right edge is. I figured this was enough leeway that the inevitable creep of their positions would take enough time for me to correct it before anythihg bad happened.

The self-captures are probably because the jump took the satellite a fair distance inside the SoI boundary. You said in your mission report that an object that enters an SoI has to either exit at the same speed or impact. That's true if it enters at the actual boundary; the orbit is either hyperbolic or elliptical with an apoapsis outside the SoI, and either case is an escape. If it appears deeper inside the SoI, then it is only guaranteed to return to that altitude at the same speed, so if it ends up on an elliptical orbit an apoapsis between the appearance location and the actual boundary is possible.

That seems reasonable.

I have a hypothesis that phantom SoI hops happen because when the game checks a ship's location, it is checking the Mun's position after the current world update against the ship's position before, or vice versa. If so, the danger zone is approximately the distance traveled in one update at the highest warp level encountered.

Mun and anything co-orbital with it have an orbital speed of 542.5 m/s. Assuming that positions of planets update on the "max physics delta time" setting and not on the constant 0.02s FixedUpdate clock, the tick is likely to be in the neighborhood of 0.04 s. 0.04 s * 542.5 m/s * 10,000x warp = 217 km that a ship could teleport into the SoI from outside due to update skew. As an idea of the scale, that's about halfway between 1 Mun radius and 10% of the SoI radius. Do you know how many kilometers outside the SoI your satellite was before it jumped?

But the thing is, warping isn't causing the satellite orbits to change. I've watched it at high warp and everything's fine. But the jumps happen when the game is either running at normal speed while I'm in the VAB, or isn't running at all. It's like the satellites get jumped into new orbits when, for want of a better term, "space" itself "loads". By "space", I mean being at some place in the game where you can see what's in space, like if you're flying a ship or using the tracking station.

For example, I'm flying a ship and look around space on the map view, and everything's fine. So I hit ESC, go to KSC, and either go into the VAB or hit ESC again and quit the game. Either way, next time I get to a place where I can see the map view again, a satellite might have been hosed (usually not, but way too often). So I'm thinking that this is a glitchs caused by fumbling some data either as "space" unloads when you give up access to it, or loads when you go to where you can see it again.

In .23.5 at least the game fails to predict SOI changes when the ship in question doesn't cross the orbit of the satellite body, and may either show nothing or show the SOI change much too late. For example if you put a ship in a circular orbit a little smaller than the Mun's, the game will never tell you you're going to enter the Mun's SOI until it happens.

Yeah, 0.23.5 hurt the ability to fly your ship in many ways. The most obvious thing to me was how these days your orbit can change a fair amount just because you rotated your ship. It's not as bad in 0.24 as it was in 0.23.5 but it's still there. I hadn't thought about not getting noncrossing encounters on the map, but now that I think about it, I belive you're correct.

The game can also show closed orbits that extend outside the SOI of their primary. In the tracking station I've found it does this for unselected vessels, but then when you select it it correctly shows the vessel as leaving the SOI. Try that for your awry satellite.

I've noticed this in connection wtih asteroids in the tracking station. A free-range asteroid will show as captured in the tracking center as long as it's in Kerbin's SOI, but on the map as seen from a ship it shows hyperbolic. I've heard this is just a limitation of the tracking center. But this isn't my problem. The pic above wasn't taking in the tracking center, it was taken on the map view of the ship I was flying at the time, back in LKO. And the satellite really is in Munar orbit--it said so itself when I jumped to it.

[Mr Shifty]

I've noticed this in connection wtih asteroids in the tracking station. A free-range asteroid will show as captured in the tracking center as long as it's in Kerbin's SOI, but on the map as seen from a ship it shows hyperbolic. I've heard this is just a limitation of the tracking center.

Under the game's patched conics model, captured and hyperbolic are not exclusive opposites. It's possible (easy) for an object to enter or leave a body's SOI on an elliptical trajectory. This applies particularly to moons, since their SOI is relatively small. In other words, escape trajectories are not necessarily hyperbolic. (In the real world, these trajectories wouldn't be hyperbolic either; they'd be complex Newtonian trajectories.)