Spitfire
Jan 20 2003, 10:40
... but still some questions without answers!
Knowsabout, engage at will, hold fire, open fire and their effect to AI behaviour are probably the most controversial issues when it comes to mission makers' experiences about them. I decided to take the bull by the horns and do some experiments, in order to finally understand the AI. I did some research around it since I'd like to know how the AI will respond to different situations in different combat modes so I don't always have to do complex time-consuming scripting in order to make AI behave the way I want.
Especially when deciding when to fire and when to hold fire the AI seems to be very complex and seems to use higher mathematics than any of the above.
There are a couple of things I've found out for sure. Behaviour (safe, aware, combat etc) and unit skill level are both affecting AI's ability to determine the knowsabout-level at different ranges. And of course another factor is the stance of the target (prone, crouch, stand).
So far so good. Another fact is that knowsabout of 1.5 is a magical treshold value when any AI with "safe" behaviour WILL ALWAYS draw his weapon (ie, change the behaviour to combat) when confronting the enemy. What happens after that, I don't know for sure. If the AI is allowed to open fire, he will do it immediately after knowsabout reaches 1.5 and the weapon is drawn. Even if the weapon is drawn already the AI will hold fire until the knowsabout rises to 1.5. Only the combat mode of "Never Fire", of course, causes the AI to... eh... never fire. http://www.flashpoint1985.com/ikonboard301/iB_html/non-cgi/emoticons/wink.gif.
However, if the AI is ordered to Hold Fire, neither the knowsabout level nor the distance seem to bear no importance whatsoever.
I've stared at the wrong end of an AK-47 at 10 m distance with enemy knowing everything about me (knowsabout=4) but still not firing, only pointing the gun at me. However, during the next experiment the same enemy with the very same behaviour opened fire at me when the knowsabout level was only 2.5 and the distance being more than 50 m. I found out that a combination of "Hold Fire" and "Engage at Will" results in drawing the weapon at knowsabout 1.5 but not firing until I pose a threat too big for the AI, whenever that might be. When only "Hold Fire" without "Engage at Will" is issued, the weapon is not drawn until the AI is ready to pull the trigger, so it may still be holstered even at 10m with a knowsabout level of 4.
So, the only mystery about AI behaviour hides in the "Hold Fire" -command. What are the factors that the AI uses to determine when to break the command and change their GREEN combat mode to YELLOW? I guess BIS's input is needed again http://www.flashpoint1985.com/ikonboard301/iB_html/non-cgi/emoticons/smile.gif
After reading all that I'm written above I really feel like I'm splitting hairs here http://www.flashpoint1985.com/ikonboard301/iB_html/non-cgi/emoticons/wink.gif. Knowing these things about AI behaviour has really bugged me for quite some time now but I hope all this aimless sleuthing I've done is of help for someone.
Knowsabout, engage at will, hold fire, open fire and their effect to AI behaviour are probably the most controversial issues when it comes to mission makers' experiences about them. I decided to take the bull by the horns and do some experiments, in order to finally understand the AI. I did some research around it since I'd like to know how the AI will respond to different situations in different combat modes so I don't always have to do complex time-consuming scripting in order to make AI behave the way I want.
Especially when deciding when to fire and when to hold fire the AI seems to be very complex and seems to use higher mathematics than any of the above.
There are a couple of things I've found out for sure. Behaviour (safe, aware, combat etc) and unit skill level are both affecting AI's ability to determine the knowsabout-level at different ranges. And of course another factor is the stance of the target (prone, crouch, stand).
So far so good. Another fact is that knowsabout of 1.5 is a magical treshold value when any AI with "safe" behaviour WILL ALWAYS draw his weapon (ie, change the behaviour to combat) when confronting the enemy. What happens after that, I don't know for sure. If the AI is allowed to open fire, he will do it immediately after knowsabout reaches 1.5 and the weapon is drawn. Even if the weapon is drawn already the AI will hold fire until the knowsabout rises to 1.5. Only the combat mode of "Never Fire", of course, causes the AI to... eh... never fire. http://www.flashpoint1985.com/ikonboard301/iB_html/non-cgi/emoticons/wink.gif.
However, if the AI is ordered to Hold Fire, neither the knowsabout level nor the distance seem to bear no importance whatsoever.
I've stared at the wrong end of an AK-47 at 10 m distance with enemy knowing everything about me (knowsabout=4) but still not firing, only pointing the gun at me. However, during the next experiment the same enemy with the very same behaviour opened fire at me when the knowsabout level was only 2.5 and the distance being more than 50 m. I found out that a combination of "Hold Fire" and "Engage at Will" results in drawing the weapon at knowsabout 1.5 but not firing until I pose a threat too big for the AI, whenever that might be. When only "Hold Fire" without "Engage at Will" is issued, the weapon is not drawn until the AI is ready to pull the trigger, so it may still be holstered even at 10m with a knowsabout level of 4.
So, the only mystery about AI behaviour hides in the "Hold Fire" -command. What are the factors that the AI uses to determine when to break the command and change their GREEN combat mode to YELLOW? I guess BIS's input is needed again http://www.flashpoint1985.com/ikonboard301/iB_html/non-cgi/emoticons/smile.gif
After reading all that I'm written above I really feel like I'm splitting hairs here http://www.flashpoint1985.com/ikonboard301/iB_html/non-cgi/emoticons/wink.gif. Knowing these things about AI behaviour has really bugged me for quite some time now but I hope all this aimless sleuthing I've done is of help for someone.