AI Developments
Posted: Thu Aug 26, 2004 2:14 am
Hi all,
This is just a note to update you on how things are going with the AI changes for COTA. I've been a busy boy in this department.[:)]
I am just finishing the last of the changes for our new Delay code. In HTTR the Delay code simply creates a single blocking position and falls back when either the blocking position is threatened or the route between it and the final objective is threatened. This often results in the delaying force being bounced from its fallback position before it has had time to deploy, let alone dig in.
The new code will, where the delaying force comprises at least two line units, split the force into two. Each blocking force will man one of two blocking positions, one behind the other. They will then leapfrog back down the delay route, either to specified waypoints or if none then at AI determined intervals. The upshot of this is that there is always "one foot on the ground" ( well most times anyway [;)] ). So far the testing I have done looks great. It will be going out to the Panther Prowlers soon.
In the process of this I have also tweaked the firebase code. It is now a lot less likely to try and move through enemy controlled territory on its way to what it thinks is a good firebase location. I also tweaked this for Attacks so that it it is now better at choosing a good location to support the assault.
That's all for now. Stay tuned.[:)]
This is just a note to update you on how things are going with the AI changes for COTA. I've been a busy boy in this department.[:)]
I am just finishing the last of the changes for our new Delay code. In HTTR the Delay code simply creates a single blocking position and falls back when either the blocking position is threatened or the route between it and the final objective is threatened. This often results in the delaying force being bounced from its fallback position before it has had time to deploy, let alone dig in.
The new code will, where the delaying force comprises at least two line units, split the force into two. Each blocking force will man one of two blocking positions, one behind the other. They will then leapfrog back down the delay route, either to specified waypoints or if none then at AI determined intervals. The upshot of this is that there is always "one foot on the ground" ( well most times anyway [;)] ). So far the testing I have done looks great. It will be going out to the Panther Prowlers soon.
In the process of this I have also tweaked the firebase code. It is now a lot less likely to try and move through enemy controlled territory on its way to what it thinks is a good firebase location. I also tweaked this for Attacks so that it it is now better at choosing a good location to support the assault.
That's all for now. Stay tuned.[:)]