Re: Podbot Beta 3.0 Updates? ^^
 

I agree with all written by mattisbogus.
These new bots are good i.e. they consume less CPU and the ping is low, but still they act stupid most of the time. Like was said before, they stop where they don't have to stop, they look where they don't have to look, and they camp where they don't have to camp.
They usually get killed when they enter doors 'cause they don't even watch where they go. It would be REALLY good if they watched the SAME direction they walk.

Re: Podbot Beta 3.0 Updates? ^^
 

When they run, it is impossible to look that the sky or only its feet. When they run, it is not possible to hear the noise of step of others. They cease running, they listen, they turn and they go there. But they never stop to aim at a wall during 1 minute as though they saw across.

Re: Podbot Beta 3.0 Updates? ^^
 

Yet another small update - V3B20f:

+Fixed problem with bots looking wrong direction at walking (in most cases they should look forward, but it may happen they are also looking at danger position - which isn't forward). That should reduce their looking at the wall.
+Fixed problem with not updating the global experience if the overflow was about to happen.
+Reverted back some function in sound slimulation (the sample of the sound is now also taken in account for 2 sec instead of 1 sec; the new sample is taken into account if the new sample's volume is higher than the old one)

Please test it and let me know if this version works better for You.

Re: Podbot Beta 3.0 Updates? ^^
 

One more time a small update - V3B20g:

+Small change for unstuck code. They should better unstuck from them-selves (faster).

Please test it and leave a feedback here (also for V3B20f changes - of course - included already in V3B20g).

Re: Podbot Beta 3.0 Updates? ^^
 

They still stuck and act stupid, sometimes even worse than before. They stuck even at free spaces, just stand still and won't do anything. Also, now they sometimes, don't watch where they get shots from, and accordingly, don't shoot back. They just keep walking like nothing's going on, or like they're more bothered about what's happening between their legs lol. Yeah, it looks like that. Sometimes...

Therefore, i moved back to V3B20e.

Re: Podbot Beta 3.0 Updates? ^^
 

At the moment, I am satisfied with improvements. Remain to notice better. Things continue surprising me:
- bots try to take back up the unidirectional ways. If I point out sense, I know that this cannot pass upside down
- they seem to favour jumps in other ways.

Re: Podbot Beta 3.0 Updates? ^^
 

This is one way of unstucking if they feel they are to close to each other. It looks much better if they go back than if they are just stupidly jumping all the time.
If they cannot unstuck other way - then finally they are trying to jump to unstuck.

Re: Podbot Beta 3.0 Updates? ^^
 

when bots, for instance, descent and cannot go back up, I put an unidirctionnel
when bots can go back up by a simple jump or a squatted jump, I put a bidirectionnel without jump.
when bots risk of falling between the surface of departure and that of arrival, I put a jump.
I create waypoints almost every day

i don't know their logic and what makes that they choose a way instead of other one..

I don't understand that a bot CT don't go to deactivate or to help to deactivate when the bomb is put down.
Of the best in the most poor skill, they should go towards the bomb to deactivate or attack terros around the bomb.
If the team risks losing, it is necessary to resist and not to go for a walk.

An idea in passage, in game, would be it possible to point out to the player the closest objective and possibly the closest point of release?

Re: Podbot Beta 3.0 Updates? ^^
 

I believe You are using some google translations or something because sometimes it's very hard for me to understand what You are talking about and if You are asking the question or You are considering some fact.
I believe in that above You wanted to say You are puting "the path one-way direction". This is at least how it is called in the Waypoint Editor.

You have to put a bi-directional path, but You have to save the jump in the backward path. I mean You have to go to the point You want to back-jump from, You have to choose from the WP editor "Waypoint Add" and type "9. Jump". Then You will see some message like obserwation on and the code is waiting for You to make a jump. Once You do the jump back - the direction of the jump and Your speed You had to make it is saved (like recording on a tape) in waypoints file and whenever the bot will be there in that place - if he wants to go that direction back - he will be "playing" Your jump like from the recorded tape (with the speed and the direction). I hope that's clear. Otherwise - the bot may get stuck there and he may try to move a bit left, right then jump (but without any specified direction nor speed!!!). So he will not do this so well like You may expect.

That shows You know how to make a jump waypoint, but You just didn't know how and when to use them. It doesn't matter if they may fall down somewhere or not. If the bot needs a jump to go somewhere - You need to put a jump waypoint. Otherwsie it takes them to much time to realise they need a jump to pass somewhere.
I was fighting with their stupid jumping everytime they could go somewhere i.e. because they got stuck with other players bot. So I had to make them using jumps less frequently (normally the human-player is not jumping in one place all the time - I had it make more human-like). But that implicated a side-effect - those places they really need a jump for a bot are forcing the waypointer to put the jump waypoint. Otherwise - as I stated above there is a risk they will be stupildy trying to wait and look only at the destination waypoint, then move a bit - finally they will do a jump.
After writing that above I started to think to add in the code something like a "smart jump" (if the vertical angle between (2 waypoints origins) and (the line paralel to the grund between those 2 WP) is higher than some certain value, the bot should try to jump in that direction no matter if You put or not a "jump" waypoint. That should help in those places the bot will not fall down somewhere, but he needs a jump to go forward there.
But - as I wrote above - the jump waypoint saves not only the direction the bot has to jump to, but it saves also the speed the bot has to have to make that jump successful. That cannot be so easy written in the code.

Well - basically their choose at their spawn some "goals" (but not only map goals - it can be also a T/CT specific WP, it can be a camp WP, too). Then they are tryng to go there with a path with lower cost. Every WP they have got hurt nearby in not so long time history gets in account some "danger" value. More hurts/deaths near a certain WP - higher "danger" value is accumulated for that WP. So they have to balance the geometrical length of the way (just the shortest path in 3D) with the smallest danger path. That makes some optimal for them path to go somewhere (maybe not the shortest 3D and maybe not the most safe - but just optimal).
There are some conditions they are not looking at the "danger" but they just choose the shortest 3D path (but I would need to search through the code to find it). That is independant of their skill. Their skill is related to possibility to kill You (words sound playing "skill" to "kill") - to aim their crosshair at You and to press the fire button in certain time. The rest of behaviour is 99% not related to their skill level.

Please - rephrase - I didn't get that at all.

Re: Podbot Beta 3.0 Updates? ^^
 

I got one question. Why the bots aren't affected by flashbangs?