Bot functionality is not really there to be pushing rifts that high yet. Sure the rift level's the bot handles gets much better every season, a good portion of that is the bots functionality increasing and the rest is the strength of characters is going up. S5 rift 70 is doable but any average person more or less by hand. In Season 3 some of the class leaderboards were unable to do GR 70.
Lets get an idea of what would need to happen to do 85+ rifts.
- fishing for a good density and mob set
- ignoring champion/elite packs that aren't worth killing
- grouping up mobs for good density pack kills - this involves sometimes moving 2-3 screens away in a 3by3 panel square on the map and getting these mobs into a pack
- then getting these mobs packed into a tight little corner so that they are hit by shot enough.
- moving out of aoe's correctly and only when they need to be
- again pulling the rift gaurdian into a tight little corner so that they are perma twistered
i haven't attempted twister but this what i understand to be the general concept.
from a programming logic point of view:
1: to me would be better off just letting the buy try each rift as opposed to entering, finding what kind of mobs and leaving if it doesn't validate. person opinion
2: again would have to create some sort of affix validation for each champion/elite mob type. some combinations are easy but on a different mob type that same combination may be a huge waste of time
3: trinity does a pretty good job of rounding up mobs, but not entirely in a way that a hand played high grift may. trying to create the logic so that it runs forward and grabs a few packs and kind of circles them to round up and meet a density requirement is not easy to do. Then you could also run into a problem of it doing this and not meeting a density requirement and endlessly looping while it tries to fill it. so maybe it should just round up then kill regardless of size.
4: once this size gathered then trying to get the pack nestled near a wall could be hard. im not really familiar with the nav mesh and how it can and cannot be used for tying in with combat conditions
5: correctly moving out of aoes, trinity does a great job at this but it has its flaws. it moves in and out of molten like its undecided on what to do, especially when their is a purple globe/orb in the center it wants. I would suguest trying to use a form of black spots to prevent it from doing the dance and ending casting but then the issue comes of you could completely ruin this great density combat position we created in #3 and #4 because now its a black spot that is no longer allowed.
6: getting the riftlord to do the same may be easier but that will just come down to this specific combat routine.
So now consider we get this all coded out and it works as intended, great! Now we just need to do the same thing for every other build out there. Then keep in mind all of this is being done for free by community developers.
