We now have the beginnings of a manual. It’s bare bones at the moment, but please feel free to send me questions or requests for things to further document.
After way too much travel and administration, I’m finally able to switch over to summer research mode starting today and I seem to have managed to start by squashing the last bug I was aware of in the 0.2 release, so I’m posting everything. Please feel free to send documentation and feature requests.
See the source code page. Twig now has the beginnings of a manual. To simplify the documentation process and generally make it easier for people other than me to extend the class hierarchy, the APIs for the Physics, TwigObject, and TwigGame classes have been refactored and rationalized. Feedback is very welcome.
This version adds tables and chairs, automatic subgoaling using role-passing, an example client for the server, and a number of bug fixes.
The source code can be downloaded from here.
I haven’t posted much recently other than papers. The school year tends to be hard for me because I teach what’s referred to as an overload (i.e. more courses than I’m normally supposed to). The good news is that the quarter is over and I’m back from GDC and the AAAI Intelligent Narrative Technologies Symposium. The bad news is that the new quarter starts tomorrow.
I’m behind on my self-imposed schedule for the next release of Twig. The new release has the fast subgoaling code from the FDG paper, as well as new prop types, new actions, grabbing that’s integrated with reaching, sitting that’s integrated with locomotion, and a bunch of fixed and extensions to the RPC system. It’s all done, I just haven’t finished porting some of the old controllers over to the new goal system. And I need to do regression testing on the old episodes of the web comic.
However, I have a lot to think through and blog about from my week of conferencing, so I’ll try to do that first while it’s still fresh in my mind. But I want to be sure to put it up before I’m sucked into further travel. So one way or another, I’ll be posting it before my talks in Singapore mid-month. If nothing else, I have 24 hours of flying time with just me and a laptop …
This is a paper I wrote for ICFDG that takes some work from my previous life as a roboticist and applies it to Twig. However, it’s situated in a tradition that the games community, and even the present-day AI community isn’t very familiar with, so much of the paper is an attempt to survey and explain the literature on using combinational logic for inference – why you’d want to do it, why it’s limited, and what you can do to overcome some of the limitations. And even then, I’m worried that it won’t be transparent to the reader; not because it’s technically difficult, but just because the motivations won’t be clear. Anyhow, any advice would be welcome.
This is a longer, journal paper on Twig, submitted to the new journal IEEE Transactions on Computational Intelligence and AI in Games (full disclosure: I’m an associate editor). This paper documents a fair amount more than the AIIDE paper, including object manipulation scripting, and a little bit about the authoring tools.
The new source release is up. See the new source link above. This version is built on XNA 3.0. Thanks to Bill Manegold for selfless trailblazing on earlier versions.
I’ve compiled Twig under XNA 3.0 and fix the minor gotchas in the project files and it seems to work. I’ll put a new cut of the source tree up tomorrow or Wednesday. In the mean time, I’ve rebuilt the webcomic episodes using ClickOnce deployment, which requires somewhat more than one click for the user, but still prevents you from having to search out and install the various libraries that are needed. The first time you try a comic episode, you’ll have to wait for it to download the redistributables for .NET 3.5 and XNA 3.0, but after that it will run fairly automatically without your having to manually open zip files, etc. Let me know if you have any problems.
I’ve decided to beef up the internal AI somewhat by porting over some of the role passing work that my group and I did on the Cerebus robot. That will let the characters do certain kinds of reactive planning and epistemic reasoning very very fast and essentially continually replan at frame rate. Depending on the number of cycles I can get to do hacking, that ought to take a few days. So I’m assuming it will take me three weeks. So here’s the current release schedule:
After that, I’ll probably move to working on some simple social behaviors, but I don’t have a release schedule for those.
The good news is that I’ve made a source release that ought to be easy for other people to install. The bad news is that it turns out that Microsoft deprecated the version of XNA that I’m using a week ago and you can’t download it anymore. So I’m designating this version -0.1. When I get back to Chicago, I’ll install VS 2008 and XNA 3.0, do the port (which in theory will be easy) and will release that as version 0.0. Since I hope to have that version in a couple of days, I’m not going to bother putting the 2.0 version up on the web, but if you just drop me a note, I’m happy to send it to you.
Recent Comments