allista
-
Posts
2,208 -
Joined
-
Last visited
Content Type
Profiles
Forums
Developer Articles
KSP2 Release Notes
Everything posted by allista
-
Which hangar, and what plane? Could you make a screenshot of both? Does the hangar have strict launch positioning? Hangar v3.3.1.1 for KSP 1.3 (2017.06.05) Almost no changes except for several small fixes and compatibility changes for KSP-1.3.
-
TCA v3.4.1.1 for KSP 1.3 (2017.06.05) Almost no changes except for several small fixes and compatibility changes for KSP-1.3. The big update with all the miracles of new orbit optimizations will follow...
-
Currently no, you can only edit target waypoint if it was set manually; proxy waypoints (that point to a ship) are not editable. And CPS does not work during deceleration, only when the final landing sequence is engaged. Deorbit Autopilot still needs a lot of work; I'm currently testing it massively with the same automated testing framework that I used for rendezvous. And there are some crashes even in a seemingly normal situation, and suboptimal behaviour when there's not enough fuel. Sorry, no. And honestly, I'm not planning to: it's currently out of the scope of TCA as I see it, and there's too much to be done with the existing modules anyway. Doesn't MJ's docking autopilot work for you?
-
Now you can watch the next series of tests: this time rendezvous from launchpad at KSC. This should be more interesting and eventful than orbit2orbit rendezvous. @Kobymaru, btw, for this I had to reimplement the testing framework as a standalone addon with dedicated UI and a simple interface for test classes. Have not pushed the changes up to GitHub, though. Will do it tomorrow.
-
Latest test results: 11h 10m of continuous operation 198 successful rendezvous 41 prevented collisions no crashes, stopped manually Next: rendezvous from launchpad to orbit. Oh... that's new. Sure, they play nice together, AT_Utils being just a common library for all my mods (including TCA and Ground Construction which USI MKS uses); so it's only natural that both depend on it. There shouldn't be any conflicts in CKAN either, as far as you install everything through it.
-
And another two finely captured: https://clips.twitch.tv/SullenBadKiwiNerfBlueBlaster https://clips.twitch.tv/UglyPerfectHabaneroRaccAttack
-
This is how it looks: https://clips.twitch.tv/AnimatedHilariousRatDAESuppy https://clips.twitch.tv/AdorableRichYakinikuBrokeBack
-
The stream was somehow interrupted after 35m, but the test went on for full 8h, yielding 138 successful rendezvous until I finally stopped it manually. During the test 18 cases of potential direct hitting were prevented according to the log. I would say that's a positive result
-
No, asteroid's geometry is taken into account, heuristically from afar, using actual visual mesh when it's loaded. The same for the active vessel; I even take into account exhaust plumes of all the engines. I'm not aware that stock rocket engines have any spool-up time. The thrust gauge on the UI have one, but that's only for the show. But yes, the issue is the timing: the node is created for analytical solution of a braking problem for inertial reference frame. So it works fine for low-gravity situations, like high orbits, or around small moons. But shedding off some 2km/s in a LKO takes enough time for the end orbit to be different enough from the calculated one (nodes in KSP always assume instant maneuvers). Talking about PIDs, I did have an idea of a PID-controlled non-instant maneuver executor, but that's a hard problem to solve for now... What I did (and what finally began to work after a few logic modifications) is an in-proximity course correction. It won't help with overshooting (but with a rate of ~1km of distance per 2km/s of the maneuver it seems tolerable); but it will move a trajectory a little bit to the side of the target as soon as it's loaded (to obtain its actual size); which is enough, considering that at that range remaining dV is always less than 200m/s. UPD: with the last modifications of the course correction code there were no more crashes so far (>40 rendezvous). I will launch the test/stream again in a few hours and leave it for the night. Stay tuned.
-
Nope. Still crashing... https://clips.twitch.tv/SpunkyAffluentPresidentPoooound
-
Testing (and streaming) again. Added simple course correction. Seems to be working, but need time to be sure.
-
Result of the test: After 58 successful rendezvous, engine exhaust during a brake maneuver hit the asteroid, spun it up (because asteroids in KSP are 100x less dense than polyurethane foam), and the asteroid hit the ship, which exploded. Conclusion: this is rare. But some form of near-the-target course correction is needed to eliminate such scenario. https://clips.twitch.tv/AbstemiousEnthusiasticPandaKeepo
-
First two are automatic, the "best" transfer is used according to appropriate metric. The third, Manual, is indeed a list of transfers with info about dV, time to the target and time to the first maneuver. You can choose one, see how it goes in the map view (maneuver nodes is created automatically), choose another... then tell TCA to proceed. Well, basically it's a unit test system: testing module looks through the assembly for other modules that declare a specifically named (or marked by an attribute) method. Then it provides a ui to choose from the available tests and to launch one. Then it simply calls selected test method each frame. It is the responsibility of a tested module to perform adequate test actions in this method. The whole thing could be implemented as a part module or a vessel module, or better yet a kspaddon. P.S. Glad to meet a colleague Microbiologist, I am.
-
Sorry, i forgot to turn it back on in the game. When you're constantly testing, your get tired of it eventually. There's not much variation in the stream, because it's fully automated. I'm not even at home now Asteroid generation followed by rendezvous sequence, in an infinite loop. The main interest to me is how long it will go until some failure, if they're would be any. Actually i did, but not a standalone one - it is an integrated part of TCA, implement as yet another TCA Module It is compiled into the .dll only in debug mode. @Kobymaru, btw, is it T4, or some other phage?
-
Completely reimplemented search algorithms, improved logic for some hard cases. It now supports multi-patch orbits and can even start from a different SOI than the target. Plus, two configuration options became available: gravity turn sharpness (to overcome dense atmospheres) and trajectory search modes: pure dV minimization, fast transfer priority, and manual (in which you can choose from a set of possible transfers).
-
Sorry for my absence. Will answer everything later. But right now you can watch the streaming of continuous automatic test of the new Rendezvous Autopilot that should be out soon. https://twitch.tv/allitau
-
It's not. It's the heavy radial rocket engine which needed no resurrection. If you want to play with part config and tune the dual-engine setup of the SABER, I can also provide you with the files. Otherwise, I do remember my promise, but I also have had no time for KSP for a while now Perfect! No objections to repost, I won't remove the file until I will have time to make a decent package with those parts. Which is, alas, will not be soon. Do get well!
-
Sure thing. Tell me if it works https://www.dropbox.com/s/rtlqt7dwm3kzhq2/AT_MiscParts.zip?dl=0
-
They were from this mod. Unfortunately, I had to drop them (and many other auxiliary parts) to be able to focus on primary content.
-
Fortunately, I do it all inside my mind. Backspace key is my cat's specialty, though
-
Tonight in a dream I've actually almost implemented the whole thing! It's still not enough to just write down the code from the dream to get it working, but now I have the necessary architecture transformations. Oh, I do love when my subconscious works for me
- 738 replies
-
- 2
-
-
- resources
- fuel tanks
- (and 1 more)
-
Yep. But considering that there's no notion of volume in the stock game, I feel it's a good enough approximation
- 738 replies
-
- 1
-
-
- resources
- fuel tanks
- (and 1 more)
-
Err... you're mixing dimensions here. Density is in t/u, so dividing 1m3 by it gives you m3*u/t, which is not u/L. The calculation is as follows: let's say we have a tank with H2 under 200atm pressure. How dense it is? From the ideal gas law it would be 200 times the density of the 1atm H2 which is 0.08988g/L (at 0C), so 200*0.08988g/L or 200*0.00000008988t/L; now we divide this by the density in t/u from CRP: 200*0.00000008988t/L / 0.00000008990t/u ~= 200u/L.
-
1. It is possible, for example, via dynamic addition of the KISInventory module; the main problem is that current architecture of CC deals exclusively with part resources, so a major reimplementation of the core code is required to incorporate other volume uses. 2. Absolutely. As you can see, the TankTypes.cfg itself has a set of MM patches that allow dynamic tank type configuration depending on the installed mods. So just use :AFTER[ConfigurableContainers] directive. Or you can indeed donate your configuration and I will include it into the mod. Thus it will be available to everyone
-
Didn't quite catch the situation. You launch a vessel with solar panels that were stowed at the moment of docking into the hangar, but when it is launched, the panels appear deployed again, correct? Have you stored this vessel in the editor, or in flight? Are the panels stock or from a mod? I myself have launched the stock "ion probe" microsatellite from the inline hangar without problems, so this should be something trickier.
