Converted Waypoints for RCbot-AngelScript, Updated & Tested for Git Options

[madmax2]

Is it definitely a func_breakable ? Use rcbot.search and it will print all greatest entities (with developer 1)

Yes, I think it is. I duplicated the problem at the 2nd vent cover in the map. It's at the 1st humantower, bots shoot the 1st vent cover, they tower up into the vent, the bot had his crowbar out, but would not break it, just kept bumping against it and crouching back and forth in the vent. I nocliped up against it and did the debug search, it showed a func_breakable at distance 28.... At the end of this process I broke the vent an it disappeared from the list, so I would say it is func_breakble. Before I broke it I tried to run debug think on the bot, but it didn't seem to work correctly? It seemed to select the other bot. I ran into this when I tried the say command to "come" to me, a few days ago? Anyways here is what I typed and the response in the console:

CODE

] as_command rcbot.debug_bot [m00]wh3y
] as_command rcbot.debug think
[RCBOT]debugging think
L 24/11/2018 - 10:57:43: "[m00]ch33s3<10><BOT><players>" stats: frags="40.64" deaths="1" health="46"
[RCBOT]Debug '[m00]ch33s3' (if bot)

All other messages looked to be just stat updates, no debuging messages. I turned off debug & tried it on the other bot and it seemed to select that bot, but no rcbot debug messages. Again, tried the bot at the vent, still would not select it. Seems to be a bug in debug and maybe the say commands, selecting the correct bot???

Should the debug messages be on the hud? I did see it a message on the hud for enabling debug_bot and think, also when turning off debug, but nothing else but developer messages at the top of screen...

[edit] Forgot to mention, on my first run this morning, at these same vents, bots broke the problem vent cover by shooting it !! maybe he was out of ammo this time, but it doesn't seem likely??

[Edit2] Locations/images
1st Human Tower & 1st Vent Cover

2nd Vent Cover

[Cheeseh]

There's bug in the player search, you can use partial bot name

as_command rcbot.debug_bot [m00]wh3

should work

I've fixed that but yet to update git .


I will try some more debugging with vents... would be good if someone made a bot debug map for such stuff

[madmax2]

Re-ran debug with partial name on a stuck bot at same vent. It worked this time, but he was not seeing the func_breakable. I Verified it was a func_breakable.

CODE

] developer 1
] as_command rcbot.debug_bot [m00]m1l
] as_command rcbot.debug think
[RCBOT]debugging think
[RCBOT]Debug '[m00]m1lk' (if bot)
L 24/11/2018 - 19:39:53: "MadMax2<1><STEAM_ID_LAN><players>" stats: frags="27.98" deaths="0" health="100"
L 24/11/2018 - 19:40:03: "[m00]y0ghur7<2><BOT><players>" stats: frags="43.03" deaths="1" health="100"
L 24/11/2018 - 19:40:03: "[m00]m1lk<3><BOT><players>" stats: frags="49.16" deaths="1" health="100"
L 24/11/2018 - 19:41:11: "MadMax2<1><STEAM_ID_LAN><players>" stats: frags="27.98" deaths="0" health="100"
L 24/11/2018 - 19:41:21: "[m00]y0ghur7<2><BOT><players>" stats: frags="43.03" deaths="1" health="100"
L 24/11/2018 - 19:41:21: "[m00]m1lk<3><BOT><players>" stats: frags="49.16" deaths="1" health="100"

Since I was debuging, i switched to the other bot stuck in HumanTower wpt crouching. Every time other bot passed by he was thinking this

CODE

[RCBOT][DEBUG - THINK] CanHeal(player)

Then did

CODE

] as_command rcbot.debug task
[RCBOT]debugging task
[RCBOT][DEBUG - TASK]  FAILED
L 24/11/2018 - 20:05:23: "MadMax2<1><STEAM_ID_LAN><players>" stats: frags="27.98" deaths="0" health="100"
L 24/11/2018 - 20:05:41: "[m00]y0ghur7<2><BOT><players>" stats: frags="45.43" deaths="1" health="100"
[RCBOT][DEBUG - TASK]  FAILED

Then this

CODE

[RCBOT]debugging nav
L 24/11/2018 - 20:07:18: "[m00]y0ghur7<2><BOT><players>" stats: frags="45.43" deaths="1" health="100"
[RCBOT][DEBUG - TASK]  FAILED
[RCBOT][DEBUG - NAV] m_iCurrentWaypoint == 36
[RCBOT][DEBUG - NAV] NavigatorState_Fail
[RCBOT][DEBUG - NAV] CFindPathTask FAILED
[RCBOT][DEBUG - NAV] m_iCurrentWaypoint == 36
[RCBOT][DEBUG - NAV] Navigator State FOUND_GOAL...

[RCBOT][DEBUG - NAV] touchedWpt()
L 24/11/2018 - 20:08:03: "[m00]m1lk<3><BOT><players>" stats: frags="72.45" deaths="4" health="69"
L 24/11/2018 - 20:08:11: "MadMax2<1><STEAM_ID_LAN><players>" stats: frags="27.98" deaths="0" health="100"
[RCBOT][DEBUG - TASK]  FAILED
[RCBOT][DEBUG - NAV] m_iCurrentWaypoint == 36
[RCBOT][DEBUG - NAV] Navigator State FOUND_GOAL...

[RCBOT][DEBUG - NAV] touchedWpt()
L 24/11/2018 - 20:08:21: "[m00]y0ghur7<2><BOT><players>" stats: frags="45.43" deaths="1" health="100"
[RCBOT][DEBUG - TASK]  FAILED
[RCBOT][DEBUG - NAV] m_iCurrentWaypoint == 36
[RCBOT][DEBUG - NAV] Navigator State FOUND_GOAL...

[RCBOT][DEBUG - NAV] touchedWpt()
L 24/11/2018 - 20:08:33: "[m00]m1lk<3><BOT><players>" stats: frags="72.45" deaths="4" health="69"
L 24/11/2018 - 20:08:41: "MadMax2<1><STEAM_ID_LAN><players>" stats: frags="27.98" deaths="0" health="100"
[RCBOT][DEBUG - TASK]  FAILED
[RCBOT][DEBUG - NAV] m_iCurrentWaypoint == 36
[RCBOT][DEBUG - NAV] Navigator State FOUND_GOAL...

He does a touchedWpt() everytime he stands & crouch's

It seems random, but maybe this means something?

CODE

[RCBOT][DEBUG - NAV] CFindPathTask FAILED

Again, if I crouch under him he will start moveing towards the other map goals, and he doesn't need to go thru the vent...

[edit] Added a wip/test wpt, if you want to check out the vent problem & first human tower. It's not to far from the beginning, go left to get key card in room with plank board, then first door can be opened. Wipe out room full of enemies, door, cavern, door and you are there. Or teleport bots might be faster...

This wpt is not ready for the git, parts are not working well yet...

Updated Wpt added to pack for git...

[Cheeseh]

Trying sc_doc waypoints now, got to the vent and bot didn't seem to see it.... What's really annoying is the lack of debugging ability. I can only add stuff in like messages on the fly but still have to reload the plugin, and when I do that the bots now recognise the breakable and break it !!! then I don't know what caused it in the first place... Debugging some more I think its caused by the bots being too close to the breakable and the breakable origin is too far from its field of view so it is disregarded. I added some code in before that if an entity is blocking the player it is also made visible, but didn't seem to work, so I've forced it into DoButtons() now. See if that helps.

btw I fixed the human tower bug -- when the bot finished doing the human tower task he'd find a new waypoint to start from which would always be the human tower waypoint and start the task again, so bots now no longer choose a human tower waypoint when finding their nearest waypoint to start from

I put the sc_doc waypoints into git and also added some unreachables at turrets.

[madmax2]

Thanks..

11-25 build

On first test, the bot shot the vent, but saw bot crowbar a different vent. Just getting back to this, going to run more tests. [edit] on 2nd test vent not there, 3rd test bot shot vent again...

[edit] I think I know why this was happening, and it was only at the 2nd HT, may not need any fix for it, I'll post back later after more waypointing/testing:

Good to see they aren't stuck on HT wpt now, but they may not be holding long enough for another bot to get on top? Another 10 seconds crouching might be enough, but I want to retest at both HT locations, may need more time? Even if they do manage to get on top, I think I saw the bottom bot leave before the top bot had a chance to jump off. At the second HT in the map, with the "bugged" version, they could get on top of the computer there fairly well, now I don't think they can do it?

Anyways, will retest the vents & HT locations to be sure...

[edit] I have the 1st Human Tower working much better now. I saw you fixed bots shooting at the laser cannon, nice . I'll post more info later, maybe tomorrow.

I still need to fix the ladder in the vent, more info on that latter if I can't resolve it... Other stuff to improve at beginning & end of map too...

[madmax2]

11-25 build

Vent Covers Fixed: Bots will either shoot or crowbar them... Verified bots do crowbar both the 1st & 2nd vent covers (or shoot them)... thankyou

Human towers: A little unsure about this? I may need to try some other locations or just make a test wpt on some boxes and play with it. They do fairly well at the 1st HT in this map, I think because there are no other goals nearby. At the 2nd HT, it's ok, but I've noticed they keep running back over to the HEV/Health chargers, use it briefly, even though they are at 100%.

At the 1st HT, what worked & seemed to help, was the following:

1. made path more vertical - wanted to ask about this, should top wpt be off ledge rather than on ledge? Or does it not matter either way? I'm Looking for what works best?

Top wpt on ledge of vent


Top wpt off ledge, more vertical path (way it is now)


2. I put a staynear flag on HT wpt, to pull them in closer & away from that door there. I used the "stay close to" flags often with the old bots. With rcbot-as, at this location, staynear on the HT alone was causing some problems for the bot on top, he would end up standing looking down at the bot crouching, the bot underneath would not stand up, they would be stuck that way for a long period. Without a SN flag on the HT wpt, this wasn't much of an issue, but they couldn't tower up well plus the door was interfering too.

Before adding staynear to HT wpt



3. So the solution was to put both a SN and a Crouch flag on the HT wpt. The SN helps line them up better and the crouch helped them get into the correct sequence of crouching & standing.

In other words the sequence should be (I think?): HT bot crouch's, top bot jumps on him, at this point the top bot bot must crouch on top of the other bot (this was not happening well without combining the crouch with the SN flags), once the top bot crouch's the bottom bot will stand, then top bot will normally stand & jump up to ledge (in this case it crouch/jumps into the vent).

I tried that at the 2nd location, but it didn't seem to make any difference, probably because there is a jump wpt before the HT, so it's not as much of an issue there.

Like I say, I'm not sure I want you to change anything for HT's yet, unless you spot some bug in the code? Think I need to play with it a bit more. If you add more crouch time for HT bot, don't add more than 10 seconds, don't want them staying there to long either.

You might want to check and see why they are using chargers when at 100%? Come to think of it i've also seen them stand on health packs at 100% too.

Think I've got the problem with the bots jammed/stuck at the vent ladder fixed. I Will be uploading an updated wpts pack to include this wpt, sometime today...

[Cheeseh]

noticed bots standing on top of players at HT waypoints don't jump to allow the crouched player/bot to fully stand up, so they should do that now.

Also noticed bots trying to use health chargers at 100%, I even ignored max_health (which could go higher than 100) and just set it to 100 and they still do it??? not sure myself...

[madmax2]

HTs are working better .

Updated a couple wpts in pack, sc_doc & intruder. Added your updated sc_persia to the pack. First post lists a few of the changes.

Been away from home a few days visiting family for the holiday weekend. Hope to speed up the wpt updates this week, now that some of the bugs are fixed, and i'm getting more familiar with the new bot version.

I'll keep watching for bugs, keep it up, they are getting better...

[madmax2]

Waypointing Request :

Audio feedback for add/delete wpt, pathwaypoint_create, and pathwaypoint_remove, different tones for success or failure like you had with old bots.

Working on murks now... Fixed a bad door button issue thats been a big pain. May post some debug info later. Old rcbots had problems with some doors in this map too. Bots can open at least one door without any problems...

[Cheeseh]

Waypointing Request :

Audio feedback for add/delete wpt, pathwaypoint_create, and pathwaypoint_remove, different tones for success or failure like you had with old bots.

done some so far

[madmax2]

Thankyou, saw that on git too, will try it today .

Spent most of yesterday downloading software, updates, steam updates, configing software for my new (used) PC. Installed rcbot-as/sc5 to it too... so now I have 2 PC's I can wpt on...

btw, its a huge hardware upgrade for me, from a 1.5gb XP PC AMD64 with gtx275, to an i7 12gb, gtx1060 (6gb), with 2 250gb ssd, 1 1Tb & 1 750gb hard drives. Huge cooler master case with 1200W power supply.. I'm heating my house with it ...

[madmax2]

Updated first post...

Updated Quarter for faster door opening & exit from last 2 spawn rooms (important flags not needed at these doors now, bots exit faster without them). I should mention I don't see the wacky behavior the old bots had at these doors, good job Cheeseh... Bots get the HMG better/faster...

Infested2 Hotfix - disabled last long ladder so bots won't get in the way of players. I will try to fix the ladder at a later date. bots need to go down the ladder quickly...

Added Murks rcwa, Bots can play the whole map and do most everything good. There were some issues and most were resolved but one. I'll try to post about that tommorrow... I was pleased I didn't need to do much for the Human Tower at the end, they do have a little difficulty, but do as well or better than the old bots did. I didn't try to improve it any, it wasn't needed. Bots will use the shortcut through the turret room now.

sc_doc was updated a few days ago, Bots do the HT's good now, can make all the jumps at the end of the map, they may fall some but did well in my last tests, and they can end the map (player should have time to get to the end). Bots do need some help from a player, prying open a door, opening a security door, bots will turn off vent fan but player needs to help them. Recommend a small number of bots, 1 or 2 will probably work best.

[madmax2]

Updated first post...

Added desertcircle rcwa + script - For the most part it is working good. Bots can play the whole map . Bots can capture all flags, blow both towers, and open the front door. I don't have them using the health/ammo crates, for a few reasons. It effects the routes they take (i.e bots will always go to the 1st crate and not use both trenches) and causes some back tracking for health, which I don't want them backtracking much on this map. Bots can only open a crate if an important wpt is there, and bots won't do it if the important wpt is not in the script.

I did a couple last minute changes, i'm still testing. The flags in this map seem to get stuck sometimes, it has happened when I don't use the explo cheat, so maybe grenades will cause it, but seems unlikely? So I removed all the staynear flags from flag wpts, the last and not fully tested one on the 3rd flag. Out of dozens of test games, the flag finally got stuck there. I haven't seen any flag get stuck once the staynear was removed, so hopefully this will fix it?

[edit - Updated 12-13-18] After testing it more, bots were doing so poorly at capping this flag, I changed it back to something close to what I had before. I put the staynear back on the important wpt (flag) & removed the wait. I moved it slightly but its close to original spot. Ran 3 games on it and bots did all objectives, including 3rd flag perfect and no stuck flags . Still not sure if this 3rd flag will stick or not, but at least they can cap it much better than last wpt... Tried moving flag wpt away from the pole a little ways, but bots were getting slaughtered from the tower ... Still testing it...

Another last minute change is at the intersection just before the door switch & 2nd tower explosives. I added a staynear & wait_noplayer there. Not sure I like this change, but a bot was trying to go from the switch back to the explosives to blow the tower and couldn't get around that sharp corner, This fixes that but now they walk there, meh...

I may change this wpt at a later time, after more testing? Other than those gripes, it works pretty good...

sc_doc.rcwa - Temporarily removed from the pack, the latest is included with rcbot-AS git.

[madmax2]

Updated first post... 1-14-19

Updated Richard_Boderman.rcwa Mostly had to fix ladder issues...

Updated Intruder.rcwa Mostly had to fix ladder issues...

The ladder code is in flux right now, so may need to fix ladders again at some point???

[madmax2]

Updated first post... 1-14-19

Updated sc_persia from git, bots were broken at the horizontal lattice ladder. Bots would get on it, but could not move to the next wpt. Solution was to change all the ladder wpts back to normal wpts. Had to replace one wpt cause bots would get stuck there spinning in circles, moving the wpt fixed the problem, kinda wierd??? It works good with normal wpts, bots don't fall unless getting shot...

Also fixed the rope ladder there so bots can get up it and back onto the lattice if they fall & land on a balcony. That won't happen often, if ever!

Added a number of wait_noplayer flags to help bots at the peg jumps, works really great at the 2nd group of pegs. Couldn't do the same thing at the 1st pegs, or it would mess up the success rate, but I did add some in the area to spread the bots out...

Some unreachables were added and more crouch/cover spots and more movement wpts/paths.

[Cheeseh]

didn't think about the horizontal ladder - I will test these later. Thanks!

[madmax2]

Hey Cheeseh,

I'll be adding sc_doc to my pack soon, probably tonite. Had to fix a couple ladders, the vent ladder & that wall ladder near the end (in the cavern). That works ok, but bots do have a little trouble there, can't seem to find the right position for the top ladder wpt without having to put a staynear on it. I Might try moving the bottom wpt, but bots probably make it 70% of the time up the ladder, so probably good enough.

I significantly improved the cavern jumps all the way to the pipe/ladder, nearly 100% success rate! So the only places they may fall is at the wall ladder and getting on top of the vent entering the cavern. I'd say the overall success getting all the way across is better than 70% or 80%, I'm very pleased with that...

A few other small tweaks, bots were having trouble with the ceiling switch at the 2nd HT, i think it's fixed now. I'm going to try one more time to get bots to wedge that security door reliably. I can get them to do it, but not very good, or fast. Might just be better to leave that for players. No one is going to want to wait for bots to do it unless they are fast, 1st or 2nd try...

So by the time you get up in the morning i should of added it to the pack with your script. BTW, great script... I think I need to study it, I think you were checking a door using ,x,<,-1660? no button entity maybe?

I'll be adding a desertcircle update soon, just not sure I'll get it done tonite ? I'm adding a couple more scripted wpts to it, just not sure it will make it better?

[madmax2]

Updated first post... 1-17-19

Added updated sc_doc wpt & Cheeseh's script to the pack last night. Mostly a fix for the ladders & much better cavern jumps. Tried to get bots wedging the security door good, but just put it back to the way it was before. Bots will use the wheel & player can wedge the door... I may go back and move the important wpt for the ceiling switch, bots tend to stick to it, but it works ok.

Guess I'll test my script additions for desertcircle today, and add that to the pack if it works good?

[Cheeseh]

Updated first post... 1-17-19

Added updated sc_doc wpt & Cheeseh's script to the pack last night. Mostly a fix for the ladders & much better cavern jumps. Tried to get bots wedging the security door good, but just put it back to the way it was before. Bots will use the wheel & player can wedge the door... I may go back and move the important wpt for the ceiling switch, bots tend to stick to it, but it works ok.

Guess I'll test my script additions for desertcircle today, and add that to the pack if it works good?

Great I'll add them to the git later after I check them.

For the wedge in the script of sc doc maybe you can find the entity of the wedge itself, i tried but couldn't find it, maybe you can try the visible check or active check

A lot of waypoints could do with scripts to be honest cos bots don't really know where they are going without one

[madmax2]

Updated sc_doc again ... Fixed the ceiling switch, bots usually press it the first attempt. 2 bots can pry open that door, sometimes quickly but usually takes a few times for bots to get in sync, player & a bot can do it fast now. Had to remove the wait_noplayer in the door to do it though, and do some other things. Bots would jam up in the narrow passage way from the spawn room right at start (just sometimes), so slightly moved a wpt, I think it fixed it (I got stuck to them too, all of us trying to attack that first room at the same time ). [Edit] didn't fix it completely so added jumps to passage way, seems good now...

[Edit - 1-20-19] Tweaked the pry door area again, I think this is the best, and should be final rework. Bots may still hang on the narrow door opening, but not long.

For the wedge in the script of sc doc maybe you can find the entity of the wedge itself, i tried but couldn't find it, maybe you can try the visible check or active check

I'll try...

A lot of waypoints could do with scripts to be honest cos bots don't really know where they are going without one

Yeah, many maps could benefit from a script, but they do good on most maps I've tried without scripts, especially linear walk through maps.