Jump to content
Pericles

Concerned over rare pathfinding problems in SF2 demo

Recommended Posts

IanL - I explained where the men I was unhappy with were going: North. Simply read what I wrote. 

On 10/23/2018 at 11:33 AM, Pericles said:

When the fire team panicked, they ran North (towards enemy forces) not South (towards friendly forces). 

Also, try looking at the compass in the videos.

1 hour ago, IanL said:

Having said that we now have a bunch of videos so even though none of them snow what the orders were I can at least go play in that same area of the map and try out some orders and see what choices the men make.

I have described - in great detail in this thread - what the orders were corresponding to the videos. 

Share this post


Link to post
Share on other sites
8 hours ago, George MC said:

Blue friendly direction in Wilcox is SOUTH - confirmed using demo version of Passage at Wilcox scenario.

Thanks. Firing again "The Passage at Wilcox" scenario for trying to reproduce this problem, I have just realized how close the buildings are from "Going to Town" scenario of CMSF1. I have played so much this scenario that I can close my eyes and see the map... 😃

Share this post


Link to post
Share on other sites
28 minutes ago, ncc1701e said:

Thanks. Firing again "The Passage at Wilcox" scenario for trying to reproduce this problem, I have just realized how close the buildings are from "Going to Town" scenario of CMSF1. I have played so much this scenario that I can close my eyes and see the map... 😃

He! Nice :)

Going to Town was the original Passage at Wilcox scenario with some changes. This is in the demo is the totally revamped and upgraded Passage at Wilcox which I think, will be a stock scenario in CMSF2 when released. The new Wilco has a heap of map changes and tweaks plus revised REDFOR OOB and new AI plans. 

Share this post


Link to post
Share on other sites
2 hours ago, Pericles said:

IanL - I explained where the men I was unhappy with were going: North. Simply read what I wrote. 

See I thought you had but when I reviewed it I could not see what was right in front of me. oops.

2 hours ago, Pericles said:

Also, try looking at the compass in the videos.

The video that you removed - that I only got to see once.

 

2 hours ago, Pericles said:

I have described - in great detail in this thread - what the orders were corresponding to the videos. 

Yeah, I cannot follow any of that to the level of detail that is needed.

I now that those other videos are there, I can now see the kind of thing you are talking about and the area on the map you are talking about so I can try to reproduce some of those goings on. A couple of other testers have read this too. If there is something wrong we'll figure it out.

That is still an if - the game is designed so that soldiers don't always make the correct decisions at all times. It is part of the game. So, when players say that their pixel troops didn't do what they wanted "therefore bug" that is just not enough.

Share this post


Link to post
Share on other sites

Perhaps your reluctance to affirm that this is a problem boils down to semantics. Instead of using the term "bug", I will just use the word "TacAI problem" from now on. 

A "TacAI problem" in the context of Combat Mission might be defined as follows: A behavior by a unit that is not believable or realistic in the context of real-life human decision-making. 

With this definition in mind, an example of a TacAI problem is as follows: A pixeltruppen panics and runs toward the enemy without being ordered to. 

Now, I will concede that it is believable that a solider in real life might do this. For example, if he didn't know the enemy was there, or if he sustained a traumatic brain injury, or if he was suicidal. But if these conditions do not hold, it is not believable that someone would do this. 

So, if this happens in-game, it can be identified as a "TacAI problem". 

Now, if you are to watch the videos I posted, and if you are to believe what I am saying in terms of the orders I had given each of the units before commencing the real-time computation of the turn, then you will agree that this is a "TacAI problem".

 

 

Share this post


Link to post
Share on other sites
1 hour ago, IanL said:

I now that those other videos are there, I can now see the kind of thing you are talking about and the area on the map you are talking about so I can try to reproduce some of those goings on. A couple of other testers have read this too. If there is something wrong we'll figure it out.

@IanL if I may help, I think the problem comes from the walls. I have created a small scenario in CMBN editor (game engine number 4.0) for reproduction. The problem is seen when the attack is perpendicular to the main axis of attack. In our case, the main axis of attack is North-South. And, in my example, like in the above videos, the attack is West-East.

Setup:

  • Allied Friendly Direction: South
  • Axis Friendly Direction: North

yL4Q9.jpg

Three times the same houses:

  1. One with walls for a West-East attack.
  2. One without walls for the same West-East attack.
  3. One with walls for a North-South attack.

Y9JnJ.jpg

I am playing hotseat.

Share this post


Link to post
Share on other sites

First attack using the houses with walls for a North-South attack axis. I have tried several times and the pixeltruppen are still fleeing in the South direction. No problem here.

RGJm7.jpg

Edited by ncc1701e

Share this post


Link to post
Share on other sites
1 hour ago, Pericles said:

Perhaps your reluctance to affirm that this is a problem boils down to semantics. Instead of using the term "bug", I will just use the word "TacAI problem" from now on. 

Yeah, you can be as passive aggressive as you like - it will not change what I do or what I say (well to be fair I might loose my patience but I'll try not to) :D

So, I fired up the editor and stripped out Op for and tested some straight pathing around the first compound. There are some interesting walls in there. Some that look like they block doors. Everybody did good just moving around. The other thing I did was hit the evade button several times (which generates get away orders). Those evade orders sent the men repeatedly heading for the ditch on the South side of the compound. Seems like a good choice. The only thing that anyone could possibly say - oh yuck - was when the team was in the North building of Objective U facing the West. In that case they ran out the West door and down the road to the ditch.

Then I moved some of the Op for to the back of the map (so I could get Blue for to Objective U with out taking losses - I tried once straight up 😞 ). This time I had teams breach where your broken wall was and try to hunt along the same location and enter the South East building - like in your video. Yeah. Boring. Everyone did what I asked and took all kinds of fire like in your video. Getting disparate to see anything like what you did I sent the team that hunted to the door on a trip to the roof of the North East building onto the balcony of the North West building and onto to the second floor. This time they took a beating but executed all the orders I gave them - perfectly. Next turn I left the two survivors alone. One more casualty and eventually the last man decided "time to go". Guess where he went? Exactly where I would want him to go :) I was worried since when he decided to bail he was in the North West building facing West but he turned around and ran out the door to the South East of that building and back the way he came heading to the ditch at the south end of the building.

I'll play around with this some more since I only repeated it three or four times but indications are things are fine. I'll repeat what I said earlier since some people are hard of reading: Remember that your pixel soldiers are not guaranteed to make good choices all the time. Sometimes they will make mistakes.

Share this post


Link to post
Share on other sites
26 minutes ago, ncc1701e said:

Setup:

  • Allied Friendly Direction: South
  • Axis Friendly Direction: North

 

Three times the same houses:

  1. One with walls for a West-East attack.
  2. One without walls for the same West-East attack.
  3. One with walls for a North-South attack.

 

Great idea. I am short on time so I only skimmed - did you find a problem?

Share this post


Link to post
Share on other sites

Third attack using the houses with walls for a West-East attack axis. Usually, I am seen the soldiers going South.
ndajK.jpg

But, on rare occasions, I also see the infantry trying to flank the houses / walls going North. I suppose this is what we see in the first above video. Please note the fleeing infantry is going South in the below images. There is one guy going North but I do not think this is someone fleeing more a courageous hero trying to reach the house.

bGX4d.jpg

W3JgV.jpg

Hope this helps. I can send you the CMBN scenario is needed. 

Share this post


Link to post
Share on other sites

Sorry, I forgot to mention that I am giving two FAST orders. One to stop in front of the wall, the second to enter the building.

eVNp0.jpg

Edited by ncc1701e

Share this post


Link to post
Share on other sites

And guess what, playing once again this attack, I have seen the following. Looks like the squad has send by itself a three-man team to flank the house by the North that may be risky since this is the direction of the enemy.

XgLNX.jpg

And, after have been repulsed, the squad is fleeing in the North...

wLNn3.jpg

I have also seen that TacAI is having difficulties with destroyed walls and ignored them for moving like if there are still standing. But, this is another pathfinding subject I think. 

Share this post


Link to post
Share on other sites

ncc1701e - Great. So it seems you have replicated the problem that I identified in those three videos I posted. This is what I would call a "TacAI problem". 

6 minutes ago, IanL said:

I'll repeat what I said earlier since some people are hard of reading: Remember that your pixel soldiers are not guaranteed to make good choices all the time. Sometimes they will make mistakes.

I read you loud and clear the first time. I understand the fact that pixel soldiers do not make good choices all the time. It is a very obvious point to make. 

However, I don't believe the observed behavior - as documented in my three videos, as replicated by ncc1701e, and as confirmed in writing by others in this thread - can be explained in terms of bad decisions. And I don't believe that you can look at those videos and honestly say "The soldiers just made bad decisions. This is not a problem that could be solved (an aspect of the TacAI that could be improved)."

The fact is, this game is about simulation, believability, and realism. It is simply not believable  that soldiers would run towards an enemy when panicking rather than running away. That is why the TacAI is programmed as it is (run away from direction of enemy). This is what happens most of the time.

But when soldiers panic and run towards the enemy stupidly, having known contacts in the area, there is a significant loss of believability and realism because no one would do this in real life without some sort of aneurysm or suicidal wish.

The TacAI will not be perfect. There will be coding problems, like with any game. This is something you seem to be incapable of admitting. 

Share this post


Link to post
Share on other sites
16 minutes ago, Pericles said:

This is something you seem to be incapable of admitting. 

I see no reason to be so aggressive against @IanL. He has tried to replicate your issue directly in the "Passage at Wilcox" scenario. Having so many variables to consider, it is perfectly normal that one cannot see the same thing. That is why those pathfinding issues are hard to understand in order to solve them. Firing the editor, I was hoping to have less variable in the loop and going directly to the issue in the less amount of time possible. I think I was lucky, that's all.

16 minutes ago, Pericles said:

So it seems you have replicated the problem that I identified in those three videos I posted.

I am not sure this is exactly the same but yes sometimes I see something weird. Hope this will guide the guys in the good direction for understanding the root cause.

Share this post


Link to post
Share on other sites
1 hour ago, IanL said:

Yeah, you can be as passive aggressive as you like - it will not change what I do or what I say (well to be fair I might loose my patience but I'll try not to) :D

 

There was nothing passive aggressive about what Pericles said.  So stop abusing Battlefront's customers, and stop your troll-like behavior.  You are not a representative of BFC, and you are not a forum administrator.  You are just a guy with to much time on your hands, based on your number of forum posts.  So cut this sh*t out!

The behavior that Pericles describes is nothing new.  I have been seeing this type of thing for a long time, across all of the CMx2 engine 4.0 games.

Share this post


Link to post
Share on other sites
43 minutes ago, ncc1701e said:

I see no reason to be so aggressive against @IanL.

If my writings in this thread are "aggressive", I would say the same about IanL and others. 

I am writing these things in the interest of furthering a reasonable conversation about known issues. 

And as an aside, I have great respect for IanL as a Combat Mission veteran and tester - his work has advanced the development of the CM engine through the years. I just disagree with him in terms of how he has approached reasoning about the issue of TacAI problems in this particular thread. 

Share this post


Link to post
Share on other sites
47 minutes ago, Chops said:

There was nothing passive aggressive about what Pericles said.  So stop abusing Battlefront's customers, and stop your troll-like behavior.  You are not a representative of BFC, and you are not a forum administrator.  You are just a guy with to much time on your hands, based on your number of forum posts.  So cut this sh*t out!

The behavior that Pericles describes is nothing new.  I have been seeing this type of thing for a long time, across all of the CMx2 engine 4.0 games.

Question. Pericles is noting a specific incident. He as far as I know has not claimed that this is something that happens a lot (the thread title says rare) nor has he as far as I can tell noted it as specific to 4.0. 

In your comment you seem to indicate it is not rare and in 4.0, but you noted you have seen for a long time which I expect means you saw it prior to 4.0. Is that a fair summation?

Share this post


Link to post
Share on other sites

sburke, I don't believe your question is relevant to the discussion. The issue here is identifying problems with TacAI in engine 4.0. Clearly, the engines before 4.0 were imperfect (that's why we have continued engine development), and one would expect to see problems with TacAI behavior in those versions as well. I seem to recall TacAI problems before 4.0 came around. 

Shock Force 2 demo TacAI problems do not happen a lot. But the fact is they happen (I was able to provide three examples from one play-through of the Wilcox scenario) and many of us would like to contribute to their solution. 

Share this post


Link to post
Share on other sites
1 hour ago, Pericles said:

sburke, I don't believe your question is relevant to the discussion. The issue here is identifying problems with TacAI in engine 4.0. Clearly, the engines before 4.0 were imperfect (that's why we have continued engine development), and one would expect to see problems with TacAI behavior in those versions as well. I seem to recall TacAI problems before 4.0 came around. 

Shock Force 2 demo TacAI problems do not happen a lot. But the fact is they happen (I was able to provide three examples from one play-through of the Wilcox scenario) and many of us would like to contribute to their solution. 

That is your assessment. Troubleshooting however relies on factual information so if it is not behavior introduced in 4.0 that would be relevant from a code perspective. I.e one does not need to look for something introduced as new code.  You are not in a position (nor am I) to say if there is a change in TAC AI behavior that Charles might want to look at but as Ian and I are two of the people you would expect to submit as a ticket to BF we do need to define what it is we might be submitting.  We do know there were changes relative to responses to arty fire for example. If we know that this behavior was evident prior to 4.0 then there is no chance that whatever was done on that isn’t having an impact here  I am certainly not going to open a ticket on behavior in 3.0 however I do need to know if this is new behavior or not    It seems from chops that it may not be therefore....... the ticket we might open won’t cite this as new behavior so it becomes a “simpler” issue of looking at the TAC AI versus the TAC AI 3.12 versus 4.0

So does it seem relevant now? 😁

or I could just say f**k it if folks are gonna blow me off and just ignore it. Honestly I think Ian was justified in his remark as your last comment makes it look like my simple question puts me in the area of people not “contributing to a solution”.  I find that fairly offensive as I was simply asking a question to see if we could clarify a bit of a discrepancy as to his comment and yours. As that is apparently not cooperating with your expectations on contributing I will bow out and leave you to it. 

Share this post


Link to post
Share on other sites
4 hours ago, ncc1701e said:

XgLNX.jpg

This is the kind of thing that is caused by teams trying to squeeze through the single soldier opening and a team that is last in the queue decides to go around. This  does not match what @Pericles is describing since in his case it is after they have taken fire and they want to withdraw.

4 hours ago, ncc1701e said:

And, after have been repulsed, the squad is fleeing in the North...

Now that part sounds suspecious. Can you share the scenario - I would appreciate it.

Share this post


Link to post
Share on other sites

sburke - I didn't realize TacAI problem identification was so complicated. I thought that identifying TacAI problems in the most current engine version 4.0 was all that was relevant in improving the engine. The developers would take your ticket submission regarding an identified TacAI problem and make changes to the most current engine version 4.0. 

If you are correct, and it is actually relevant to know whether or not TacAI problems identified in the current engine version were also present in the previous engine versions, then my belief that your question is not relevant to the discussion is wrong. 

That being said, it would have been constructive if you would have given your opinion on the topic of the post, which is TacAI problems with the Shock Force 2 demo. You've had a chance to look at the videos and to experiment with the demo. What do you think? 

Edited by Pericles
Needed to name person to whom I was addressing comment

Share this post


Link to post
Share on other sites
4 hours ago, Pericles said:

The fact is, this game is about simulation, believability, and realism. It is simply not believable  that soldiers would run towards an enemy when panicking rather than running away. That is why the TacAI is programmed as it is (run away from direction of enemy). This is what happens most of the time.

But when soldiers panic and run towards the enemy stupidly, having known contacts in the area, there is a significant loss of believability and realism because no one would do this in real life without some sort of aneurysm or suicidal wish.

Yeah I don't think that is true. I agree that running towards the enemy should not be common - it is not desirable for sure. But the problem is the simulated pixel troops do not have the god view that we do. In some case they don't even have the contacts that we think they should be avoiding. The FoW is part of the simulation. The confusion is part of the simulation.

 

4 hours ago, Pericles said:

The TacAI will not be perfect. There will be coding problems, like with any game. This is something you seem to be incapable of admitting. 

LOL I get it man I'm not bending to your will and agreeing that you are correct. I tried to reproduce your experience and I could not. I find bugs in software every day as a developer that's what I do :) I am actually happy when people find bugs - because it means they can be fixed and others will not find them.

What I am saying is a) I cannot make your scenario happen with any regularity and b) I'm not convinced that any occurrence of such a mistake is even a bug at all. That is not the same as refusing to admit to a bug :D

Share this post


Link to post
Share on other sites
4 hours ago, Chops said:

So stop abusing Battlefront's customers, and stop your troll-like behavior.

That's uncalled for. I admit to being sarcastic at times - I might even go so far as to say that a few times it was inappropriate - but only a few times :). My behaviour has certainly not been troll like in this thread at all. I called @Pericles on his passive aggressive tone becasue I find it annoying and usually with people if you point out they are coming across that way they will accommodate you and tone it back. I literally made that comment in order to continue having this thread be constructive and exploratory not to troll him - or you for that matter.

Share this post


Link to post
Share on other sites

I find passive/aggressive behavior quite tiring.  Point out a problem, document it, and be prepared to have it questioned or don't post about a problem in the first place.  Questioning susses out nuances, limitations, compounding circumstances, etc.  It is not an attack on someone's manhood to poke and prod at a hypothesis, but it is a sign of weakness to react negatively to poking and prodding.

We're always interested in "corner cases" in the TacAI's pathing.  It's how it gets better.  However, we have been at a point now, for a while, where it's getting harder and harder to find reproducible, addressable corner cases.  That's good.

FWIW, there was no explicit changes to the TacAI's pathing in Engine 4 that I'm aware of.  Perhaps tweaking here or there for a corner case, but nothing more than that. Sometimes fixing a corner case produces a new corner case.  Which is why corner cases in games are often left alone by game developers.  Trying to fix a 1 in 1000 event could introduce a 1 in 100 event.

Steve

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×