Jan 4. Summary

[Tim Vanderhoek <vanderh@ecf.utoronto.ca>]

Where our discussion for our meeting Jan 4. was relevant to one of the
existing threads I have posted the summary as a response to a message
in the thread.  This may make review at a later date easier.

Ideally both of you (poonj, wongci) will get a chance to read this
before school tomorrow.  We can then immediately begin serious work
on the proposal after class Wednesday (see #3).

1)  I have put weblinks.html on the web.  Access at
    http://www.ecf.toronto.edu/~vanderh/weblinks.html

2) Earlier we agreed that one person should post a summary of each
   _significant_ meeting (ie. not of the insignificant meetings :) to
   email.  Since we all took notes at some point last night, this may
   not be necessary, but I propose it is a good idea regardless ---
   ensures that we all leave the meeting with the same perspectives and
   impressions, and (more importantly, imho) leaves a common reference
   that we can use.

3) The proposal must be done by the coming Monday.  Follows the
   general list of things we planned to include in the proposal.  You may
   ask yourself: "Does it look complete?"

	- Robot drawings ... as many as Cindy feels like drawing... :)
		- "Alternate systems" drawings:  (possibilities)
			- 'bot powered with four motors (POWER!!  YEA!!)
			- Number of mechanical sensors on our bot?
			- mount and servo for an ultrasonic sensor?
			- C'est tout?
		- Power system, power distribution
	- Circuit Diagrams
		- Diagrams for beacon sensor circuits (transmitter & receiver)
			- IR
				- modulation thereof
				- can photodiodes be used to estimate
				    how far away the enemy robot is?
			- radio?
			- ultrasonic??
		- Diagrams for motor control circuits
			- surge supressors?  fuses?  (Uh, if the fuse
			    blows during competition . . . ? ? ?)
		- Diagrams for wheel speed feedback
			- Hall sensors?  (problem with stray EMF???)
			- Optic couples?
				- throughbeam?
				- reflected?
			- counter circuit and latch?
		- Diagrams for anti-bounce of the mechanical sensors???
		- Diagrams for extra-sensory sensors
			- ultrasonic?
		- Diagrams for a stall detector (why can't this be
		    done in software?)
	- Computer Intelligence
		- Routines to translate raw sensor input into
		    something meaningful to the rest of the software
		    (eg. X differential wheel speed produces Y
		    velocity vector, eg. L mechanical switches
		    mean an object is at least size M in relative location N,
		    eg. Q sensor appears to be jammed ... ignore its input)
		- Routines to output of sensor routines and try and
		    construct a map  ....  Perhaps also maintain an
		    event memory as well??  useful when map fails
		- Routines to analyse the map to produce useful
		    information for "tactical" and "intelligence" modules
		    (eg. There is a circular obstacle here, when we
		     drive 5m forward, we will probably strike an object)
		- Intelligence module ... chooses one of several
		    "modes" or "tactics" based on current input toggle
		    switches and based on input from map analysis routines.
		    Also decides "one-time" decisions such as "drop
		    retransmitter beacon now".
		- Tactical implementations ... each module of tact
		    implements a certain behaviour, eg. "drive in circles",
		    eg. "drive to location (x,y)", eg. "move obstacle J
		    into new posity (x,y)".  Note that there is a
		    logical hierarchy of tactical routines .... module
		    "drive in circles" may use module "drive to location (x,y)"
		    to implement its goal.
		- Base motor control routines 
		- Routines to control sensors???  (eg. Ultrasonic sensors)

		- Handling certain types of anomalous input is a task
		    that must be split between the logical modules
		    labelled (*), (**), and (***) ... some type of protocol
		    for them to communicate their opinions
		    about anomalous input must be devised.

4) I glanced through the printed MIT software document you gave me,
   Joyce.  I'm not sure much of it is useful, but wow...you weren't
   kidding...  If it is fair to judge by that document (probably not...  :),
   they get it pretty easy at MIT.


-- 
Signature withheld by request of author.