Prototype Demo Video

So, what we have here is a video showing the basic idea behind this game. Unfortunately I don't have the most brilliant artistic skills, and my 3D artist is busy, so I used spheres and cubes as place holders. The idea behind the level you are looking at is simple: Collect 20 biosamples before time runs out. Now I'm not really trying to win here, just show off functionality, so no mocking my 1337 gaming skillz!

Here is a run down of what is what:
That guy in the middle is you - think of it as an advanced version of the Mars Rover. You have no weapons, but you can fabricate other robots to do your bidding.

Blue Spheres: What will be the "Penguin bots", the simplest defensive robots you can make.

Grey Spheres: The Pygmies - the simplest enemy robots there will be.

Green Spheres: Biosamples

Everything else should be self explainatory. Happy watching!

War! What is it good for...cool effects of course :)

So I've got to say, this point is really a milestone for this game. This is the first time you can actually "play" the game. Now obviously not all the functionality is here yet - for example I am cheating in the above picture and giving myself the resources to create "infinitely" many robots for my side, but it does give me a chance to experience the game play. It's a lot faster paced then I originally thought it would be, but that's ok, it makes for an interesting game.

Anyway - what you are looking at is my robots and the enemy robots (both controlled by variants of the same AI algorithm) having it out. The enemy robots are programmed to learn that working together is good (they keep in clusters, and fight in groups), while my robots are less sophisticated - they only care about themselves. It's interesting to watch the two armies battle it out (esp. since the enemy respawns infinitely) and is going to provide me with a few hours of enjoyment before I get back to actually making the game :)

The Swarm!

Ha! The Swarm of AI is approaching readiness :) Ok - so it's hard to show emergent behavior in a picture, but that is kinda what your currently looking at. It's not quite done yet, as not all of the rules have been implemented (nor for that matter has the learning method which allows each individual agent to change the Priority of it's rules) but it is still coming along. Also, you can see the model for the main character, the PAU (the name may change in the future, but I'm keeping PAU as a working title for now), sans color map (since it's not done yet). I think the model is pretty reasonable and looks cool even with a jupiter color map :).

Anyway, the next stage is to test each of the remaining rules, then I'll combine them with a special type of map called a Linear Integrator (LI). The advantage of the LI is that I can train it using Reinforcement Learning allowing for a wide range of experience based behaviors all coming from one personality template. The second cool thing is that I can expose the parameters of the LI to the user is friendly way (you won't need to do this to win the game, but it is cool) allowing the players to customize the AI behaviours of their minions. Personality I think it's neat, but we will see what others think once it's all put together.

Still working on the AI..

So oddly enough, the most annoying point of failure in the AI code thus far has not been the Linear Integrator I am using to direct the Wandering Algorithm, it's not the special type of ForgetFull mapping I implemented...oh no, because you know, these are the parts one would expect to screw up. What is it you ask? It's the bloody algorithmn that I used for navigation in the previous game :P . Strangely, this same algorithm works perfectly in the other game, so I am busily spending my free time (which I appear to have less of these days) tracing through the code to find out exactly where the calculations are going wrong :P.

That being said, besides the sudden failure of the trees (they still appear, but the physics engine ignores them like a ... well I can't think of a professional metaphor to place here), the game is coming along quite nicely. The 3D models are finally starting to come in, the sound guy is finally getting around to doing some of that work he claimed he could do, and I'm only going to end up taking 10 times as long to build this version as I did to make the first one! Yay efficiency on my part :).


I guess this post was pointless...I just figured I should put something here to show I'm not in fact dead, and random musings about my progress are as good as anything.

And I truck on

Ok, this isn't some kind of whiny my life sucks blog - besides my life doesn't suck :) - so I won't go into details of why I took the last month of so off from game programming. Suffice to say I've had a month away from coding for fun (not coding altogether - writing ASP code in VB really makes me miss C#) and I'm back into writing game code. I've got a couple of small bugs to fix, and I need to think of a way to code the AI. Why should this be an issue? Simple - testing! My algorithm depends on some "black boxes" - basically modified version of the idea of Linear Integrators - and I'm not sure I trust the results. I need to come up with a way to test the AI and feel confident that it's doing what it's supposed to do. This has actually turned out to be a great challenge, one that should prove to be extremely interesting to solve!