CABot2

Since this paper was submitted, we have finished CABot2.
You can download it from www.cwa.mdx.ac.uk/CABot/cabot2/cabot2.html
It has an improved parser that succeeds 100% of the time.
It's also a lot faster parsing in aroun 150 cycles as opposed to 1000.
Control now allows more than one goal to be set.
Vision works with multiple objects in the visual field.
It also learns goal action pairs for a particular command (Center the pyramid/stalactite.).