[From Richard Kennaway (980623.1151 BST)]
I didn't take complete notes, so this is just some random recollections.
My apologies to anyone I may seem to have forgotten.
There were 14 participants:
The venue, Schloss Kroechlendorff, is a large 19th century country house,
now modernised and converted to a conference centre, in a secluded corner
of north-east Germany, a few miles from the town of Prenzlau. An ideal
setting offering an environment free of distractions.
Franz Plooij gave an overview of the work he has done on baby development
over the last 11 years. In his research he has found that babies go
through "difficult" periods in the first 18 months of life at predictable
times, following which they make significant advances in their abilities to
perform various tasks. He considers that these periods correspond to the
development of new levels of control in the hierarchy speculatively
proposed in B:CP (or more precisely, the current 11-level version). He has
written a book for parents based on this work (English title: "Why they
Franz also gave an interview for German television, for a popular science
programme. (The rest of us weren't present.)
Bill Powers presented a simulation of a five-level control system for an
inverted pendulum. He also presented another simulation which demonstrated
a learning algorithm, whereby a control system can learn to control just by
observing its own error signal. It looks like magic, if only I understood
it. Bill, is this written down anywhere in the literature? Or anywhere
not in the literature?
I presented a simulation of a six-legged insect with a two-level control
hierarchy, the bottom level for controlling joint angles in its legs and
the top level for controlling body position. Variable disturbances can be
applied (including shooting off some of the bug's legs), and the bug
maintains control. Future developments will be to make the bug walk. It
would also be nice to build it.
Wolfgang Zocher presented a model of eyeball tracking, with a comparison to
the observed behaviour of real eyeballs.
Christian Wenner presented results of experiments on muscle activity and
Stefan Balke spoke about experience of applying the RTP.
Rupert Young spoke about his model of how the eye can bring a selected part
of the retinal image onto the fovea.
There was a demo of Scilab, a public domain mathematical calculation tool
similar to Matlab, available from INRIA for PCs and unix machines.
Dag Forssell is setting up a PCT web site at forssell.com. He also
distributed pre-publication copies of Bill Powers' new book, my copy of
which is at home and I can't remember the title. But I can say it's a good
There was a discussion about the possibility of setting up a PCT journal,
but I'll leave it to the prime movers in that project to announce something
when they're ready to. Its first issue would be the proceedings of the
There were also informal discussions around dinner/beers/barbecue. A
frequent topic was, why is (almost) no-one interested in PCT? How do we
get them interested? How do we get PCT recognised?
Lastly, my thanks to Wolfgang and Marion Zocher for organising the meeting.