CABot3
The Cell Assembly roBot version 3
- We had an EPSRC grant a few years ago entitled.
Natural Language Parsing with Cell Assemblies: computational
linguistics with attractor nets.
- What we wanted to do was build agents that did all their processing
in simulated neurons.
- They'd:
- take Natural Language commands
- view the environment
- plan actions
- learn about the environment
- It was a three year grant and each year we made a CABot.
- They all did all of these things, but the last,
CABot3, was
the best.
- Vision had a retina, and primary visual cortex.
- Planning implemented a Maes net.
- It learned a (really simple) cognitive map of the environment
(3D Crystal Space).
- It had a pretty good model of natural language parsing.
- It had an overall finite state automata to manage the switch and
collaboration between tasks.