Colony

Report Submission!

2008-05-15T20:16:00.003+01:00

Finally submitted the project report. Finished it about Wednesday 7th May 2008 and submitted on the Thursday. Here is the link to the project as submitted. I hope it all scores well; I think it (the project) deserves to, I just hope the report successfully puts across what's been achieved.

Now just the viva to worry about, and then I can get back to developing properly

Pheromone Diffusion

2008-04-25T15:09:00.001+01:00

Video recorded for report;

To be of use the pheromone must diffuse in an expanding 3D Gaussian like shape. It must also dissipate in the absence of pheromone being added. The video below show a clean Gaussian shape that grows and then levels off at a constant size.

Pheromone Diffusion. Increments of 10 pheromone updates. Evaporation rate = 0.08 . Diffusion rate = 1.0. Queen phermone added at rate of 100 per Zeit tick.

Interesting project: Virtual Creatures

2008-04-13T18:52:00.003+01:00

This is a project those interested in Colony may also be interested in. Has relatively little in common with the aims of Colony but another interesting example of stigmergy in computer science and uses the same engine as Colony, Ogre3D.

About Virtual Creatures;

This application is an example of evolutionary computing that you can run on your own Windows PC at home (see the download section). It uses a process similar to biological evolution to gradually evolve a population of virtual creatures in a 3D graphical and physical environment. Much of the inspiration for this project came from the wonderful work of Karl Sims.

Video Footage... It lives!

2008-04-07T23:07:00.007+01:00

If you just want the highlights see the comments at http://video.google.com/videoplay?docid=535175355828864093&hl=en to jump to the key sections. Also see the link to the gallery at the bottom of this post to get stills if you find the video hard to see.

Another quick update with some eyecandy. The vid shows the current build. The Zeits have very simple intelligence and only one pheromone is used to build a relatively decent mound. Key points to notice;

Real-time! It can just about handle real time with just the one pheromone. The program also can be set to auto adjust the rate of pheromone diffusion/zeit update to keep things in check. For the sake of this demo zeit update and pheromone update are allowed to get out of sync.

Billboards to show pheromone (that can be switched on and off). THis runs a lot quicker than full particle affects. Also a lot easier to see whats going on

Pretty organic looking. I'm pretty proud of how organic the mound looks. I'm hoping thats pretty unique

Pretty reliable framerate. Uses three threads and always makes sure the interface is smooth, even if the diffusion/zeit calculations are having trouble

Its a little hard to see but the Zeits have to dig before they can deposit (so they have building material. This results in a small trench around the mound

They take far to long to build! I need to boost there brains a bit and get them using some more pheromone to get more impressive results

Colony Screenshots

Quick update! Threads!

2008-03-09T18:24:00.001Z

From the email to my project supervisor;

"I've been programming like a man possessed on Colony. I've hit a big problem handling the pheromones in such a way to make them fast and accurate without murdering the framerate . So far little success and as I write to you I'm in labs desperately trying to crowbar threading into the code; I figure I can get a pretty massive speed increase if I take advantage of the second core on most modern machines. It's a lot of work though trying to avoid segfaults and race conditions.

Bit worried I'm running out of time though. "

At Last!

2008-02-18T19:50:00.002Z

Finally fixed that bug... all came down to a typo :-( I was looking in the wrong place for which I blame my dodgy debug code, which I finally managed to get sorted with the help of Seth the Supervisor.

Now I have moving Zeits that stick to the ground and randomly change direction every now and then. They can climb but they do have a nasty habit of committing suicide by throwing themselves of the side of the map into the abyss.

Now working on the code for the bugs to dig and deposit soil. With any luck now I'm out of this rutt I can get some real progress.

Here's hoping

Stuck in a rutt

2008-01-31T22:44:00.000Z

Finished exams on Saturday, took a day off Sunday, and Monday got down to some serious work with Colony.... and let me tell you it is being a BITCH!

I've spent nearly a whole week now working on trying to get the Zeits moving around the environment in a sensible way.

The first part of that was to implement a SphereQuery mechanism which allows me to work out what is inside a virtual sphere on the terrain. It's not as simple as just using Ogre's built in methods, as I have to extend them to cope with the voxel engine. It's also a nightmare to test anything with the GUI as it is at the moment. By Tuesday I was sure enough that it was working correctly, and allowed me to do simple collision avoidance with the camera to prevent clipping through objects.

The next part was to get the Zeits using it in some sensible way. To do this I have a method which a Zeit can use to check if a position is valid. It does this buy querying two spheres, one small inside a larger one.
The inner sphere must be empty for the position to be valid, so that the Zeit has room to exist there.
The outer sphere must not be empty for the position to be valid, so that the Zeit is not floating in thin air.

Next tricky bit, and where I've been stuck now for far too long is getting the movement right so that they can move randomly around the terrain. A random direction is picked every few ticks. Every tick the Zeit attempts to move in this direction. To do this is check whether the position towards the direction is valid. If it is it moves there. Otherwise it tries looking just to the left or right of the direction for suitable positions, and if valid goes there instead. It also tries varying the Y height of potential destinations to go up and down slopes.

However at the moment it seems as if the Zeits make the wrong decision all the time and end up wedged in the landscape, and its blowing my tiny little mind working out why. I've got a suspicion that its the original sphereQuery method which is causing the trouble as it seems to give a different result before and after the move, which shouldn't happen as the Zeit should be ignored in the query. Anyway I hope to get that bug caught by the end of the week, or the weekend and have some Zeits moving around in my lovely 3D landscape.... I hope.

Interim report submitted

2008-01-13T08:10:00.000Z

As exciting as it sounds, but hey now its out of the way I can get back on with Colony proper. Well I can in a week anyway, just a few lil bits of coursework to get out of the way.
For anyone interested the interim report can be found here;
http://www.beastnet.co.uk/wiki/images/5/52/InterimReport.pdf

Real work tis gonna take a lil hold

2007-12-24T11:58:00.000Z

as I'm now working on the interim report. Spose its gotta be done but I much rather be doing "proper" work.

Ray queries... sounds more exciting than it is

2007-12-20T10:48:00.000Z

Been working on ray queries to give entities some way to tell what's near them. Bit trickier than I was hoping as the built in ogre methods haven't been implemented in the Polyvox Scene Manager. I don't think I have time within this part of the project to mess around in the scene manager to implement it so I'm going to have to hack one on top. I think I can get it pretty efficient, and I can still use the ogre methods to get fellow entities; the trick is going to be getting it to use both together efficiently.

Screenshots

2007-12-19T09:41:00.000Z

Current screenshot of the engine;

We now have a lil program that takes a PNG heightmap and makes a voxel map to use in game. Works like a dream!

Render of early Zeit model

2007-12-19T09:15:00.000Z

An early render of the Zeit model I knocked up to learn Blender;

The tail end glow will change color to represent Zeits current task/thought.

Colony Blog

2007-12-18T11:38:00.000Z

I'm moving any stuff I wrote in personal blog to here so that people can read about Colony without being bothered by my crap.

Initial zeit design

2007-12-17T07:27:00.000Z

Some initial sketches of Colony Zeit design;

for some initial 3D-models see bzr repo

Volumetric scene manager

2007-12-11T11:39:00.000Z

Being doing some research into how to get the terrain right. I'd like to be able to dig down and build up to get the most out of the game. This is hard to do with traditional 3D terrain as the terrain is usually one big polygon so each modification requires calculating new vertexes which is tricky. Voxel based terrains draw everything out of "voxels" ( think 3D pixel ) so that individual voxels can be removed to achieve the desired effect. Therefore voxels is wot we needs .
Found a great little scene manager for Ogre which does what I need by representing voxels with a mesh so we get the performance of a mesh but with the ability to modify. Tis a bit rough round the edges but I think I can hack life out of it. The guys code seems pretty good and he's a nice bloke.. which should help.

About Colony

2007-12-10T07:28:00.000Z

From my project brief;

Colony is a game where the player controls a nest of social insects called Zeits. A Zeit is an amalgamation of the characteristics of many different real life social insects, in particular Termites and Ants. The object of the game is to ensure the growth and success of your colony of Zeits against the elements, other insects and rival colonies. The game centers around players making subtle changes to the Zeits and their environment to implement large changes in their behaviour, nest and consequently the success of the Zeit colony. The game will also have a multiplayer aspect where two or more players can set their colonies against each other. In modelling the Zeit behaviour I will look to make use of the research done into Termite behaviour in particular the research done into nest construction.

Strategy for many years now have heavily reused the classic scheme of unit control where each unit has very little “inteligence” of its own and relies on the player for nearly all orders. With this project I intend to turn this idea on its head by giving the agents emergent intelligence. I hope that this will make a new experience which will be unique for players. It is my belief that there are few more resourseful than gamers trying to win a game, and that we can make use of this resourcefulness to further the research with agent based, self-organising, emergent behaviour.

GTD Tasklist here
Launchpad Page here (with BZR repo)