Future Work
- We have just gotten a grant to do a games agent, that
senses the environment, has goals, and takes natural
language instructions all based on CAs.
- This is really a huge step forward.
- The three year grant starts in August and we're looking for a
research fellow and a PhD student. Send mail if you're
interested.
- This is really the full whack.
- We have to figure out sensing.
- This gives the system grounded symbols.
- We then have to integrate this into a parser.
- The parser will use rules, and will have a LR-N stack algorithm.
- We will use an activation net to get it to follow a goal
strucute.
- Getting this to work will take a lot of engineering, but
it gives us a great system to explore all kinds of learning,
sensing, and action problems.
