How to design a crowd-combat game like Dynasty Warriors

Greetings!
Dynasty Warriors has a very special place in my heart. It has a very simple, yet quite unique game play and I have spent hours just mowing through enemy lines. But why do I see little to no clones of this rather old franchise?
I think I’m pretty naive (considering my current experience with 3d and AI and stuff), but I’d like to make a crowd-combat game myself. But here is the question: where do I start? How do I move dozen of units at the same time, fighting each other, taking objectives or flee from the overwhelming enemy force?
I’d like to use this threat for brainstorming. What techniques do I need to know for this kind of game?

My current ToDo/Need-to-Know list consist of these points:
- Move a given unit from point A to point B
Quite anticlimactic, I know, but everyone has to start litte. I got the idea behind A* but I’d like to make my game without any tiles. Maybe Theta*?

- Some sort of hierarchic AI
As I understand it, calculating dozen of dozen unit behaviors at the same time is slow. My idea is it, to make a hierarchic system. An AI-overmind assigns objectives to a few “general”-AIs. These assign their given troops to tasks that will ultimately lead to the assigned goal.
This would work quite well with the Dynasty Warriors theme: When a general or lieutenant is killed in combat, his assigned troops loose their objective and flee or fight without any tactic.

- I need a general grasp on military tactics
I never served and I loose to any easy AI in RTS games. I read in an article about the game AI of Total War, that the dev’s used the book “The Art of War”. Is this a good start?

- Terrain and objective generation
This is more basic stuff. I’ll have to learn about perlin and his nousy behavior I guess.

- Animations
Well, this will be a fun one, but manageable

I’d love to hear your input ;D

Great! Same here. My friend and I used to play the series almost religiously when we were younger, and I still frequently play it to this day.

I’d think something along these lines, yes. Treat them as ‘units’, and calculate behaviour and pathfinding for a unit as a whole. Only worry about it on the sub-unit level when the unit is close enough to the player to be visible. Even then, as is evident in Dynasty Warriors, the AI is quite simple for a general’s soldiers.

No. The art of war is more of a strategic (and sometimes cryptic) text, and I don’t think that you’d get much out of it for what you want to achieve. Once again, looking back to Dynasty Warriors, the AI is relatively simple. I don’t believe the AI is employing any complex strategy (i.e. they wait for you, hunt you down, or flee from you. That’s about it). The illusion of a complex AI is essentially given by the storytelling and triggered events. If of course you wanted to build a more complex, free-willed AI, then it may be worth a thought.

If you are going with the Chinese history theme of Dynasty Warriors, then perhaps you would be better off reading something like Romance of the Three Kingdoms, or any major historical fiction or non-fiction of your choosing which fits your chosen theme.

Anyway, best of luck with this. I sincerely hope you pursue it.