This is my final project to bachelors degree, i want to develop two things:
Radagast AI: This AI's purpose is to:
Work in all terrain types.
Develop strategies for ambushes and traps in a clever way, combining groups of different terrains and skills.
Learning in-game and a learning game to game.
Flocking/FlowField + Custom Formations: This project intends to do something similar to what already exists in Supreme Commander 2, firstly working with something like flocking, later developing the ability to organize formations in a personalized way, (something like squares, circles, that sort of thing) coupled with the ability to create internal flows of units, something like units that are in front going into the last row when their lives are lower.
I moved to a city today, and despite being a little busy with all things to be done, soon want to upgrade my projects and bring the news right here. For now I'm without internet access then I'll be a few days off. However I intend to make a small prototype still in early January. Goodbye.
Radagast AI + Flocking/FlowField + Custom Formations
Moderators: hoijui, Moderators
Re: Radagast AI + Flocking/FlowField + Custom Formations
Regarding Radagast AI, I recommend you remove the 'AI' part, else we'll have lots of confusing ( RAI? Did you mean Reths AI or Radagasts AI? ).
You'll also be able to write quite extensively on maps with those goals, and luckily there are a lot of them, so you have thousands of potential man hours or experiments at 100x gamespeed. If you need any help I can show you what I did for Shard, I ran ~10 scenarios, 100 times each at 120x gamespeed, and gathered stats in a lua widget for each so I could compile graphs etc.
Also, regarding pathfinding, you will want to talk to kloot as he's made experiments in this field regarding that algorithm
I should also point out that if you wish to publicly distribute Radagast AI before your project is completed and marked, you must be sure your supervisor agrees that it's okay to opensource, else you'll fall foul of the engine GPL license.
You'll also be able to write quite extensively on maps with those goals, and luckily there are a lot of them, so you have thousands of potential man hours or experiments at 100x gamespeed. If you need any help I can show you what I did for Shard, I ran ~10 scenarios, 100 times each at 120x gamespeed, and gathered stats in a lua widget for each so I could compile graphs etc.
Also, regarding pathfinding, you will want to talk to kloot as he's made experiments in this field regarding that algorithm
I should also point out that if you wish to publicly distribute Radagast AI before your project is completed and marked, you must be sure your supervisor agrees that it's okay to opensource, else you'll fall foul of the engine GPL license.
Re: Radagast AI + Flocking/FlowField + Custom Formations
For which game?Zyre wrote:Develop strategies for ambushes and traps in a clever way, combining groups of different terrains and skills.
-
- Moderator
- Posts: 2464
- Joined: 12 Oct 2007, 09:24
Re: Radagast AI + Flocking/FlowField + Custom Formations
If you want complexity I think Zero-K has the widest selection of options for traps and terrain types. Vehicles, hovercraft, sea, hill climbers, spiders, jumpjets (good for ambushes). As far as ambushes go melee jumpjet units are particularly good. There are 5 mine type units that can cloak and release a lot of damage on their target:
EvoRTS with it's cloaking-in-trees system is probably nice for relatively simple traps. You could judge how many units to put where and when to reveal them. There are all terrain units, hovercraft and amphibious units but I'm not sure if the terrain can work for ambushes. I only recall burst damage units for the hovercrafts but don't take my word, I'm sure Forb could say more.
Both games have AI you can test against and would be happy to have more AIs that can play the game.
- Tick - all terrain 'mobile mine', cloaks when stationary and explodes in a reasonably large AOE with medium damage emp that disables units.
- Roach - another mobile mine, more expensive than Tick and not all terrain. Explodes with decent normal damage
- Skuttle - Expensive, slow moving mine that can move while cloaked and has jumpjets. Does small AOE and very high damage.
- Blastwing - Cheap flying mine, cloaks when landed and deals low damage.
- Infiltrator - Not exactly a mine but very close, it is all terrain and cloaked and can disable a single unit with an EMP blast with very long reload.
EvoRTS with it's cloaking-in-trees system is probably nice for relatively simple traps. You could judge how many units to put where and when to reveal them. There are all terrain units, hovercraft and amphibious units but I'm not sure if the terrain can work for ambushes. I only recall burst damage units for the hovercrafts but don't take my word, I'm sure Forb could say more.
Both games have AI you can test against and would be happy to have more AIs that can play the game.
Re: Radagast AI + Flocking/FlowField + Custom Formations
...and if you want simplicity Spring Tanks has the smallest range of of unit types. It is basically all about protecting and ambushing (flag carrying) units so maybe good to test with. In most games the most effective formation is a simple "battle line" or encircling the enemy. In ST it might actually make sense to have a box formation with a flag carrying unit in the middle.
Units recharge their health so "something like units that are in front going into the last row when their lives are lower" is important too. Units can also hide in bushes to ambush or lose persecutors.
The game needs to be updated for 84.0 Spring version but until January that should be done...
Units recharge their health so "something like units that are in front going into the last row when their lives are lower" is important too. Units can also hide in bushes to ambush or lose persecutors.
The game needs to be updated for 84.0 Spring version but until January that should be done...