Wow! It's been a busy two months since I joined our Hong Kong project
and it's time for an OpenCog Recap! The HK project is picking up
steam, and we also have other several other people making progress
around the world.
Troy Huang and Cai Zhenhua, from our Xiamen robotics lab, transferred
to the Hong Kong project in January. Zhenhua has been working on
integrating the MicroPsi motivation system into OpenCog. The OpenCog
version is called "OpenPsi" and you can see in the trunk now (though
it's still in active development so it might not work that well yet).
Cord and Lester are working on creating a virtual world for our AIs,
using Unity. Right now Cord is working on importing the characters
from the existing Multiverse demo to Unity. Lester and Troy are
writing the code to connect OpenCog AI's to Unity.
I'm creating a planning system (based on PLN) to combine with Psi.
That is, Psi will give the AI feelings and goals and the planning
system will work out how to achieve them.
Joel's leading the project and also does some of the coding. He got
the event loop atomspace working, so it's now possible to run multiple
threads within OpenCog. There is also now a separate attention bank
(for Economic Attention Networks). And the PLN logs are now
colour-coded, which makes them much easier to read (which will make it
easier to develop PLN).
Lots of improvements to DeSTIN. Amrit has been integrating SIFT with
DeSTIN as a preprocessor.
David Crane improved the dimensional embedding code and did some
experiments with clustering (on wordpairs data). It seems to be
finding good clusters, where the words are obviously related to each
If there's anyone else I missed out, sorry. Feel free to add anything
else in the replies.