Introductions

[LAEL PARROTT]

Hello all,

My name is Lael Parrott.  I am a PhD candidate in the Department of 
Agricultural and Biosystems Engineering at McGill University.  My 
colleagues are:  Grant Clark (PhD student), Robert Molenaar (PhD 
student) and Robert Kok (Professor).  Collectively, we are working to 
learn how to engineer autonomous ecosystems.  To accomplish this, we 
are simulating what we call an EcoCyborg: a cognitive ('cyborged') 
ecosystem.  Our ultimate aim is to learn how to specify the initial 
assemblage of components necessary to construct an ecosystem meeting 
a set of pre-defined functional specifications -- i.e., we are 
tackling the age old question of the relationship between structure 
and function in living systems.

This project stems from earlier work by Robert Kok in which he has 
studied the design of cognitive control systems for self-directed 
greenhouses and space-based agricultural compounds.  Our main premise 
is that by supplying a complex adaptive system, such as an ecosystem, 
with a sufficiently sophisticated control system, the system 
should display increased autonomy (through intentional self-guidance, 
increased information processing and decision-making 
capabilities, self-awareness, etc.). 

We are developing a the simulation as a multi-process session under 
OS/2 Warp.  Models of a Pavlovian controller, a cognitive controller, 
an ecosystem and external forcing functions reside in four separate 
processes that share variable values and pass messages through shared 
memory segments using OS/2 API calls.  Robert Molenaar has written the
Pavlovian controller module and a rudimentary cognitive controller.  
I am working on the ecosystem model, and Grant Clark is developing a 
toolbox of investigative tools to analyse the system's behaviour.

The ecosystem model is individual-based and object-oriented in 
structure.  The model includes the possibility of 1000 different 
species and up to 100,000 individuals that interact in a 3 
dimensional environment.  Each individual will have, on average, 100 
attributes and 100 methods (possible actions).  On recently hearing 
of Swarm, I thought it sounded like it was written to accommodate just 
such scenarios....

I am presently writing the model in C (IBM C Set++) and I expect that 
as the system grows and execution speeds begin to approach realtime 
(!) I will be compiling it for our mainframe and running the 
ecosystem process on that.  I am interested in seeing how Swarm works 
and perhaps porting some or all of my code to the Swarm system both 
for fun and to compare the two systems.

I don't have a Unix machine, so I will be running Swarm as a telnet 
session from my Macintosh computer using MacX to connect to a HP/UX 
(at least I think that's what it is) down the hall....

If anyone has any questions or comments about our project, I would be 
pleased to entertain them (address@hidden).  

Happy Swarm.

- Lael


~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Lael Parrott

The EcoCyborg Project
Artificial Intelligence Lab II
Department of Agricultural & Biosystems Engineering
Macdonald Campus of McGill University
21,111 Lakeshore Rd.
Ste Anne de Bellevue, Quebec
CANADA    H9X 3V9
Phone: (514) 398-7590
Fax:  (514) 398-8387
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~