M.A.D. Science
On World War V, M.A.D. robots fight on behalf of each country for global dominion.
Awards
Files
File | Uploader | Date |
---|---|---|
MidSaturday.png
How it looks on mid Saturday |
BlueDragon | 2012/05/12 13:10 |
Sunday.png
How it looked on Sunday |
BlueDragon | 2012/05/12 13:10 |
Diary Entries
Day One - Pyweek begins!
Theme will be: "Mad Science" - too bad! - I had an almost complete idea for
Everything Goes to Hell, while my idea for Mad Science is very vague. Oh well, it's part of the challenge I guess :)
I don't think I'm going to code anything right now, I'll just go to bed and hope my idea is automatically refined while sleeping :)
I don't think I'm going to code anything right now, I'll just go to bed and hope my idea is automatically refined while sleeping :)
Day One - New idea
This morning I woke up with a new idea!
The previous one was a vague idea about making some weird species escape from the lab of a mad scientist - those specimen would move on their own but you could somehow provide hints to help them find the way across a maze or some kind of difficult terrain.
The new idea comes from the fact that as Ernie pointed out, MAD can also stand for Mutual Assured Destruction.
So this is the incipit of my game:
"It's 2099 and the world is on the edge of a new World War. Nations strive for control of the last resources remaining on Earth. However, everybody knows that fighting an all out war with super-nuclear, plasma and graviton bombs would only bring one obvious outcome: mutual assured destruction (M.A.D).
Thus, War will be fought by single robots - each nation having its champion fighting for her.
Robots fight on their own, without external guidance, powered by artificial neural networks.
No human involved in fighting, no human will be hurt... only robots.
These robots are known as M.A.D. Robots.
As the leading scientist in your country, you will have to design and assemble the most clever and powerful robot and beat all the rest."
The game will let you design your own robot, by assembling various parts, e.g. sensors, guns, movement parts (e.g. legs or wheels) and so on.
One core feature of the game is that once you design your robot, it will fight on its own.
The robot will really have an artificial neural network deciding its action - training the neural network and deciding some its properties will be part of the game.
Every fight is 1 country vs 1 country (1 robot vs 1 robot) on a best of 5: if you win 3 times on 5 you conquer that country, earning resources.
More resources means you can build better parts for your robot, including better neural networks.
-------
So far the idea looks good (to me) but there are going to be a few challenges:
- having the robot fight on its own could be boring, especially if it fights badly
While I'm absolutely newb at graphics in general, I will try to include as many special effects as possible to make the battle entertaining.
- it could turn out to be impossible to train the neural network to fight in a nice way
- training neural networks can easily involve a lot of processing power. Having to wait 20 minutes with cpu @ 100% before your robot is trained and ready to fight would be no good.
-----
Enough talking, let's now dive into some code.
Oh, and I just started Chronolapse ( http://keeyai.com/projects-and-releases/chronolapse/ )
The previous one was a vague idea about making some weird species escape from the lab of a mad scientist - those specimen would move on their own but you could somehow provide hints to help them find the way across a maze or some kind of difficult terrain.
The new idea comes from the fact that as Ernie pointed out, MAD can also stand for Mutual Assured Destruction.
So this is the incipit of my game:
"It's 2099 and the world is on the edge of a new World War. Nations strive for control of the last resources remaining on Earth. However, everybody knows that fighting an all out war with super-nuclear, plasma and graviton bombs would only bring one obvious outcome: mutual assured destruction (M.A.D).
Thus, War will be fought by single robots - each nation having its champion fighting for her.
Robots fight on their own, without external guidance, powered by artificial neural networks.
No human involved in fighting, no human will be hurt... only robots.
These robots are known as M.A.D. Robots.
As the leading scientist in your country, you will have to design and assemble the most clever and powerful robot and beat all the rest."
The game will let you design your own robot, by assembling various parts, e.g. sensors, guns, movement parts (e.g. legs or wheels) and so on.
One core feature of the game is that once you design your robot, it will fight on its own.
The robot will really have an artificial neural network deciding its action - training the neural network and deciding some its properties will be part of the game.
Every fight is 1 country vs 1 country (1 robot vs 1 robot) on a best of 5: if you win 3 times on 5 you conquer that country, earning resources.
More resources means you can build better parts for your robot, including better neural networks.
-------
So far the idea looks good (to me) but there are going to be a few challenges:
- having the robot fight on its own could be boring, especially if it fights badly
While I'm absolutely newb at graphics in general, I will try to include as many special effects as possible to make the battle entertaining.
- it could turn out to be impossible to train the neural network to fight in a nice way
- training neural networks can easily involve a lot of processing power. Having to wait 20 minutes with cpu @ 100% before your robot is trained and ready to fight would be no good.
-----
Enough talking, let's now dive into some code.
Oh, and I just started Chronolapse ( http://keeyai.com/projects-and-releases/chronolapse/ )
Day 3 is still Day 1
"Life" got in the way: before even encountering the difficulties I had in my mind for the code, I simply found out that I had almost no time to code during this week.
Haven't coded anything since Sunday: the project goals will need to be tuned down to something like:
- learn to use tiles using Tiled/tmx library
- make just 1 level with only a few options: just a prototype to check if the idea was feasible.
And that's assuming I find enough time in the coming days.
Oh, and I just realized I will probably be away without internet access on Saturday/Sunday, probably coming home on Sunday night just before the upload deadline (I hope).
On the positive side, I got a Robot must DESTROY award for the idea :)
Haven't coded anything since Sunday: the project goals will need to be tuned down to something like:
- learn to use tiles using Tiled/tmx library
- make just 1 level with only a few options: just a prototype to check if the idea was feasible.
And that's assuming I find enough time in the coming days.
Oh, and I just realized I will probably be away without internet access on Saturday/Sunday, probably coming home on Sunday night just before the upload deadline (I hope).
On the positive side, I got a Robot must DESTROY award for the idea :)
My Pyweek is a Ludum Dare :)
I've finally found time to resume working on the game yesterday evening!
So, if you add today + yesterday evening + Sunday morning => 1 game in 48h..just like a Ludum Dare :)
This is how it looked on Sunday:
This is how it looks now:
The blocks on each side are "spawn" points, your robots are created there.
By staying alive on the field, your robots harvest resources from the field, until it's depleted (in the screenshot, the blue zero says there are no more resources to harvest).
Each time a spawn point creates a robot, it uses stored resources (the numbers on the left and right, one resource counter for each faction).
So far, both factions share a basic hard-coded AI (seek nearby and destroy!) and the game basically runs by itself.
This afternoon I will have to make the player able to influence his robots behaviour and finally make it a game instead of just something nice to look at.
Ideally, the behaviour will involve neural networks as I wanted originally. But I'm still far from that.
Still, it's nice to be able to work on this in the end instead of just giving up :)
So, if you add today + yesterday evening + Sunday morning => 1 game in 48h..just like a Ludum Dare :)
This is how it looked on Sunday:
This is how it looks now:
The blocks on each side are "spawn" points, your robots are created there.
By staying alive on the field, your robots harvest resources from the field, until it's depleted (in the screenshot, the blue zero says there are no more resources to harvest).
Each time a spawn point creates a robot, it uses stored resources (the numbers on the left and right, one resource counter for each faction).
So far, both factions share a basic hard-coded AI (seek nearby and destroy!) and the game basically runs by itself.
This afternoon I will have to make the player able to influence his robots behaviour and finally make it a game instead of just something nice to look at.
Ideally, the behaviour will involve neural networks as I wanted originally. But I'm still far from that.
Still, it's nice to be able to work on this in the end instead of just giving up :)
The End, No game!
I have to give up, the game..is not a game :)
The concept isn't fun enough as it is... too bad because the "swarm" of robots looked nice; maybe it'll work better next time.
The concept isn't fun enough as it is... too bad because the "swarm" of robots looked nice; maybe it'll work better next time.