AI is on top of my list

[buddye]

Originally Posted by Fall_Pink?

buddye,

Just out of curiosity, but what if you 'commercialize' this knowledge you have on flight sim AI? How hard would it be and how much time would be needed?

Now every game has it's own AI (mostly flawed and buggy at first) and each developer seems to make the same mistakes over and over again. First person shooter AI is different from flight sim AI, I understand that, but BoB2 and Cod are essentially the same: ww2 flight sims. They have the same number of planes in the air, same type of fighters and bombers.

Even more modern flight sims would be able to make use of the AI tables/matrices and set of maneuvers you made for bob2, right? AI pilots remain AI pilots, whatever the time period, so mimicking human behavior is the constant here. It's just the capabilities and tactics that change when new weapons platforms emerge.

It would be really nice to see some real progress in this field, that's all

Rgs,
FP

The AI maneuvers are flown in BOBII using commands (AI do not use a joystick to fly). I would expect that the commands are tailored for BOBII (of course I have no knowledge of the COD AI design and data base). The data base for the AI is also BOBII specific as well as all the many support programs.

The basic design and approach to AI could be transferred/reused but the new code, integration, and testing would be very significant, IMHO. It would take one person years to complete (if they understood the BOBII AI) and it also assumes a good Player test team to test the AI maneuvers so the project could be coded and tested in small increments.

[Fall_Pink?]

Originally Posted by buddye

The AI maneuvers are flown in BOBII using commands (AI do not use a joystick to fly). I would expect that the commands are tailored for BOBII (of course I have no knowledge of the COD AI design and data base). The data base for the AI is also BOBII specific as well as all the many support programs.

The basic design and approach to AI could be transferred/reused but the new code, integration, and testing would be very significant, IMHO. It would take one person years to complete (if they understood the BOBII AI) and it also assumes a good Player test team to test the AI maneuvers so the project could be coded and tested in small increments.

Buddye,

Maybe naive thinking here, but I'd expected a more exchangable and compatible AI, for example a set of dll's and a SDK I wasn't really thinking of adding it all quickly in CoD; that's not going to happen.

A package that could be marketed and sold to flight sim makers and expanded into other AI areas as well. Unsure if there's a market for it, but given the state much of the AI is in today, I'd say "yes".

Rgs,
FP

[buddye]

Originally Posted by Fall_Pink?

Buddye,

Maybe naive thinking here, but I'd expected a more exchangable and compatible AI, for example a set of dll's and a SDK I wasn't really thinking of adding it all quickly in CoD; that's not going to happen.

A package that could be marketed and sold to flight sim makers and expanded into other AI areas as well. Unsure if there's a market for it, but given the state much of the AI is in today, I'd say "yes".

Rgs,
FP

The list of data interfaces (flight and FM data) required from the Simulator is so large and time critical that unless the simulator is written like FSX (which is not the case for todays Combat Flight Sims) where the flight and FM data is made available to a external application then an external AI application can not be implemented (at least that is where my current analysis of the possibility of implememting an external AI maneuver stands).

I need to look at the possibility of implementing smaller functions of the AI to see if it is cost-effective.