AI Pathfinding
Moderator: Arjuna
AI Pathfinding
Here I am again, jumping on my favourite 'hobby horse', the sometimes poor quality of pathfinding by the AI. My previous problems had been with British units and I thought I'd sussed that it was due to the size of the units and the number of vehicles involved leading to the AI trying to avoid 'traffic jams'. However, having just watched KG Harder take eight hours to cover two kilometres from North Arnhem to a position north-west of Koepel I'm not so sure! They were not under fire, there were no British units nearby and none of the component units were 'fatigued' to any significant degree. I didn't set any waypoints as I saw no reason to. Neither did I set a formation as I believed that the AI would treat the move in a logical manner. Fortunately, this did not affect my overall strategy as I was moving them in order that they could function as a reserve (it being overnight Day4/5 of a 'Double-Drop Late campaign).
This, for me, is the one issue with AA that destroys my enjoyment of the game whenever it rears its ugly head and is one that I sincerely hope is rectified when HTTR finally comes out.
This, for me, is the one issue with AA that destroys my enjoyment of the game whenever it rears its ugly head and is one that I sincerely hope is rectified when HTTR finally comes out.
Thankyou for using the World Wide Web. British designed, given freely to the World.
Kevin,
Can you show us a screen shot of the route? It's a bit hard to reply until I can see what happened. Alternately, and even better, can you send a zipped saved game to contact@panthergames.com?
Can you show us a screen shot of the route? It's a bit hard to reply until I can see what happened. Alternately, and even better, can you send a zipped saved game to contact@panthergames.com?
I generally find that larger collections of units (here I mean unit icons) move slower than smaller collections.
Some of it appears to be due to coordination through multiple levels of command. Also, a good part of it also appears to be due to security of using overwatch. The contrast between ordering a single unit to move some place and a group is pretty substantial.
So, I would say that if you need to get someone someplace very, very fast, then think small.
Some of it appears to be due to coordination through multiple levels of command. Also, a good part of it also appears to be due to security of using overwatch. The contrast between ordering a single unit to move some place and a group is pretty substantial.
So, I would say that if you need to get someone someplace very, very fast, then think small.
2021 - Resigned in writing as a 20+ year Matrix Beta and never looked back ...
Cheers Mark,
couldn't agree more, most of the problems I've had have been with the British brigades. I expect them now so I detach the battalions and order them to a destination separately, reattaching once the units are all where I need them to be. Since KG Harder isn't that big I didn't expect any problem.
I think the issue I have is that AA stands or falls by the quality of its AI, more so than most other games. It is really a 'General sim' and this is what sets it apart from other wargames (and which is why we all enjoy playing it right!!
). If a General loses confidence in his subordinates he loses the battle as gets he enmeshed in the detail! We, as players, can deal with 'micromanagement' but in so doing we 'lose' the game.
couldn't agree more, most of the problems I've had have been with the British brigades. I expect them now so I detach the battalions and order them to a destination separately, reattaching once the units are all where I need them to be. Since KG Harder isn't that big I didn't expect any problem.
I think the issue I have is that AA stands or falls by the quality of its AI, more so than most other games. It is really a 'General sim' and this is what sets it apart from other wargames (and which is why we all enjoy playing it right!!
Thankyou for using the World Wide Web. British designed, given freely to the World.
Sorry to dredge this thread up, but I'd like to hear something about how pathfinding has improved.
I don't have trouble with traffic jams. My problem is I sometimes get units drawing "Curly-Q" route lines. When moving off-road, some units will draw a line that double-backs on itself for no good reason. Sometimes it's a zig-zag that looks like a Z. Other times it looks like a rollercoaster loop-de-loop.
Has this been corrected?
Thanks, citizen
I don't have trouble with traffic jams. My problem is I sometimes get units drawing "Curly-Q" route lines. When moving off-road, some units will draw a line that double-backs on itself for no good reason. Sometimes it's a zig-zag that looks like a Z. Other times it looks like a rollercoaster loop-de-loop.
Has this been corrected?
Thanks, citizen
Is there really no one who can answer the question:
Have zig-zags and loop-de-loops been eradicated from pathfinding?
Perhaps there's no absolute answer, so please say if they are reduced 1%, 50%, or 90%.
1%= We tried to, but failed to eradicate them.
50%= We identified the problem and fixed it somewhat.
90%= You shouldn't see them much anymore.
Arjuna? Anyone?
Thanks, citizen
Have zig-zags and loop-de-loops been eradicated from pathfinding?
Perhaps there's no absolute answer, so please say if they are reduced 1%, 50%, or 90%.
1%= We tried to, but failed to eradicate them.
50%= We identified the problem and fixed it somewhat.
90%= You shouldn't see them much anymore.
Arjuna? Anyone?
Thanks, citizen
Well, of course, there are people who know the answer to this question. However, beta testers are under an NDA, non-disclosure agreement. As such, they are not at liberty to respond to this or other questions regarding anything beyond what is already in the public domain. This is a fairly typical arrangement. The NDA contrains one from divulging information whether it be favorable or disparaging.
Thus, it's really up to Panther to respond to your question as the beta testers cannot. With regards to Panther, they have been working very hard. So, it's likely that your question and or others is not being deliberately ignored; but there is a lot to be done and limited resources to do it.
Thus, it's really up to Panther to respond to your question as the beta testers cannot. With regards to Panther, they have been working very hard. So, it's likely that your question and or others is not being deliberately ignored; but there is a lot to be done and limited resources to do it.
2021 - Resigned in writing as a 20+ year Matrix Beta and never looked back ...
citizen,
Ignore Marshot's hypothesis that we are deliberately ignoring you. Truth is I've been more than a little busy. And I knew that to provide you with a proper answer was going to take some time. Anyway here goes.
Whenever a large force moves in a formation other than road column you will get a series of "parallel" routes for each of the subGroups moving. This produces the "spaghetti" effect of lines when you select a group of units. The lines for the subGroups are based on the ofset that each subGroup has - eg. if you are moving in arrowhead and the subGroup is the left flank then its offset will be X metres to the left of the hub ( relative to its direction ). As the hub changes direction from say north to south the left flank subGroup ends up on the right side of the screen. To get there it must arc around to the north or top of the screen.
Now if it encounters an obstacle like a river, we have code which says "hey, don't cross that river if your moving in formation". So it will have to re-route around to stay on the near side of the river. This can sometime cause what you refer to as "strange loops".
Another cause for these can be when the hubs route does a U turn because it is following the road or a series of waypoints that the user has specified. Depending where the subGroups are at the time of the change in direction, some of them might take what looks like a strange route. It would not look that strange if you could see the units laid out in formation at each point.
Paul may be able to add to this. But I hope that helps explain why you see these "strange" routes.
Ignore Marshot's hypothesis that we are deliberately ignoring you. Truth is I've been more than a little busy. And I knew that to provide you with a proper answer was going to take some time. Anyway here goes.
Whenever a large force moves in a formation other than road column you will get a series of "parallel" routes for each of the subGroups moving. This produces the "spaghetti" effect of lines when you select a group of units. The lines for the subGroups are based on the ofset that each subGroup has - eg. if you are moving in arrowhead and the subGroup is the left flank then its offset will be X metres to the left of the hub ( relative to its direction ). As the hub changes direction from say north to south the left flank subGroup ends up on the right side of the screen. To get there it must arc around to the north or top of the screen.
Now if it encounters an obstacle like a river, we have code which says "hey, don't cross that river if your moving in formation". So it will have to re-route around to stay on the near side of the river. This can sometime cause what you refer to as "strange loops".
Another cause for these can be when the hubs route does a U turn because it is following the road or a series of waypoints that the user has specified. Depending where the subGroups are at the time of the change in direction, some of them might take what looks like a strange route. It would not look that strange if you could see the units laid out in formation at each point.
Paul may be able to add to this. But I hope that helps explain why you see these "strange" routes.