Re: the quest for om.
- The human eye takes pictures at approximately 25shots/second. If
there is an algorithm that can be perfected to the point where the
robot will be able to recognize a pencil in different background _A_,
_B_, _C_, _D_ ....etcetera; then, why can't the robot "facture" a
scene comprising of a moving target into multiple "still" or stagnant
pictures?....The reverse of what moviegoers sees at the cinema--slow
the scene down then execute the algorithm on it. In such a case, we
won't have to worry much about developing special algorithms for
detecting if the target is moving or not. Does this make any sense?
(II)On _depth perception_, i hypothesize two approaches: The robot
calculate the time differential of the light bouncing back from each
isolated target in its visual field--back into its optic sensors,
from that the robot can estimate distances. (b), From physics, the
lens formula points out the location or how far an _image_ is formed
from the lense, given the center of curvature and the distance of the
viewed _object_. Practically if one knows how far an _image_ is
formed from the mirror one can use the lense equation to determine
the distance of the _object_.(Clause:How do you know the
distance/position of the image beforehand?). Does this make any sense?
X + Y = Z is equivalent to Z - Y = X. Of course!!!
In my view, If a robot is ordered to find an exit out of a room; and
the robot circles about looking for a door or opening and happens to
find none--the room is walled completely. How does the robot
improvise on its own? E.g. Break down a wall. This kind of adaptive
behaviour is what i think signifies intelligence--creativity.
For the robot to accomplish this, the robot has to go through these
(a)Determine that by "get out of the room" it means getting from
point _A_ (which is the inside of the room) to point B(which is the
out of the walled enclosure).
(b)Determine that in order to get from _A_ to _B_ it(robot) needs to
transverse a sectional area not occupied by the wall.
(c) Determined that there is no such area.
[From this, can the robot then make the deduction that the only way
to get from _A_ to _B_ is for a sectional removal of the wall?, that
is, Z - Y = X, where Y can be any point in the domain(Z) of
approximately 6 feet in width.?]
(d)[Can the robot then initiate a search into its memory files to
isolate data correlating with slicing, cutting, sectioning, etcetera--
things that will aid in getting rid of the wall?]
I KNOW ALL THIS TO BE VERY CRUDE, any ideas? Does this make any sense?
[Common sense a given population is merely the notion or idea that
seems to have the most weight or gravity in opinion polls of a given
population. Feed this data into the robot, i will wager the robot
will have common sense!]
- --- In artificialintelligencegroup@y..., young_and_benevolent
> bobdeloyd wrote:like
> > Once we finally get a "real" artificial intelligence
> > the rest is evolution....
> Perhaps all we have to do is mimic something relatively simple,
> a flatworm, and then set it off on a virtual evolution. Thequestion
> is: How do we make the virtual reality that our AI lives in andDear young_and_benevolent:
> evolves through?
Yes theres that word reality again.. Are we talking about
intelligence or game theory? //bob
- young_and_benevolent wrote:
> How do we make the virtual reality that ourbobdeloyd wrote:
> AI lives in and evolves through?
> Dear young_and_benevolent:An artificial intelligence would have to exist in some form of
> Yes theres that word reality again.. Are we talking about
> intelligence or game theory? //bob
environment for it to enact and react within. This is the "reality"
that the AI is "aware" of. That space could be as simple as a command
line interface, or as complex as a three dimensional interactive
environment. (Probably the former, though.)
So when I say "virtual reality" I mean the environment that the AI is
aware of. I am using the word "aware" because we strive to build a
cognitive and sentient intelligence, in the long run.
- "red_cell_op" <RED_CELLss@H...> wrote:
> The human eye takes pictures at approximately 25shots/second.Actually, film works at 24 frames per second, and TV at 25/sec. Your
eyes, however, work at about 1000 fps. Your brain is cognitive of
these pictures in a vastly different way, so that a "frames per
second" rating becomes meaningless.
> On _depth perception_ ...Maybe the lasers that land surveyors use to judge distance would be
easier and more accurate than running an algorithm on an image.
> If a robot is ordered to find an exit out of a room;A good question. I guess that it would either be programmed to
> and the robot circles about looking for a door or
> opening and happens to find none--the room is walled
> completely. How does the robot improvise on its own?
> E.g. Break down a wall.
destroy objects in it's way, or it would have to learn to do that by
itself. How, though, I'm not sure!
- Here's a quote from Robert O'Shea's webpage, at:
I have so many interests in human visual perception it makes my eyes
blur. Currently I'm working on projects on: Binocular rivalry
spread of rivalry, the nature of rivalry suppression, and rivalry in
split-brain observers); Early history of binocular vision; Interocular
transfer of aftereffects; Meteorological optics (including why we
perceive the bowl of the sky, perception of sun rays and their linear
perspective, and effects of height on perceived eye level); Size and
depth perception over large distances; Spatial frequency, blur,
contrast, and luminance as depth cues; Colour constancy with
reflected and emitted light; Kinetic depth effect; Perception of
contrast and blur in the peripheral visual field; Colour spreading in
the McCollough effect; and Vernier acuity with opposite-contrast and
At the same page, you can also download the PDF of an article
some of the work covered by the talk:
O'Shea, R. P., & Corballis, P. M. (2001).
Binocular rivalry between complex stimuli in split-brain observers.
Brain and Mind, 2, 151-160.
You can check out the webpage of O'Shea's collaborator, Paul
---------- Forwarded message ----------
Date: Tue, 02 Jul 2002 16:01:14 -0400
From: George Alvarez <geoalvarez@w...>
To: VisionLabTalks <geoalvarez@w...>
Subject: Harvard Vision Lab Talk Wednesday: Robert P. O'Shea,
Harvard Vision Lab Seminar Series Announcement
Binocular rivalry in split-brain observers
Robert P. O'Shea & Paul M. Corballis
Department of Psychology, University of Otago
Wednesday, July 3rd
Rm 765, William James Hall, Harvard University
33 Kirkland Street, Cambridge
A split-brain observer has had the corpus callosum, the major tract
between the left and the right hemispheres, cut to relieve epilepsy.
One can selectively stimulate the left or right hemisphere by
presenting stimuli to the right or left of fixation respectively.
Likewise, one can elicit responses from the left or right hemisphere
by requiring the observer to press keys with the right or left hand
respectively. On many tasks, these fascinating individuals behave as
though each hemisphere is acting independently of the other. Are there
differences in rivalry between the isolated hemispheres? To answer
this, we have studied two split-brain observers, VP and JW.
We first trained split-brain and intact-brain observers to respond to
real alternations between nonrival stimuli by pressing keys with the
ipsilateral hand. When we presented rival stimuli to the isolated
hemispheres of split-brain observers, their key presses showed that
their experiences of rivalry were similar to those of intact-brain
observers. When we presented stimuli to the left hemisphere of the
split-brain observers, they were also able to describe the chaotic
appearance of rivalry alternations.
Over many experiments, mainly on JW, we conclude that rivalry is
essentially normal when processed in each isolated hemisphere,
although periods of dominance are slower from the left hemisphere than
from the right. Rivalry is normal from stimuli such as sinusoidal
gratings, coloured faces, random dots, and Diaz-Caneja displays. The
distributions of periods of dominance follow the classical gamma
shape. The only case in which lacking a corpus callosum made a
difference was that the synchronization of rivalry in two regions of
the visual field did not happen when the two regions were processed by
different hemispheres. We think that the longer rivalry periods from
the left hemisphere reflect only its response bias. We conclude from
the qualitative similarity of rivalry in the two isolated hemispheres
that the rivalry mechanism is low in the visual system.
Wednesday, July 3rd
Rm 765, William James Hall, Harvard University
33 Kirkland Street, Cambridge
- Young:> Actually, film works at 24 frames per second, and TV at
25/sec. Your > eyes, however, work at about 1000 fps. Your brain is
cognitive of > these pictures in a vastly different way, so that
a "frames per> second" rating becomes meaningless.
>Borgia: I will re-check my notes about 25/sec shots(of the human
eye)....but this comparison of shots/second is moot--that is only
tangent to the point i was trying to make, which is: isolating each
frames and executing algorithms on them in _real time_. I have
forgotten about the name of a super-fast cameras that can take
thousands of shots/second--much faster, and with better resolution
than the human eye. The rate is not all that important as much as the
computing power that will process the _stills_ in real time. On how
the brain process these visual data, is irrelevant in my view--we
don't have to simulate the brain. Just let us do something that
works, brain-imitation or no brain-imitation, regardless.
> > On _depth perception_ ...Borgia: The _depth perception_ part of the post _the search for om_
> Maybe the lasers that land surveyors use to judge distance would be
> easier and more accurate than running an algorithm on an image.
has nothing to do with "running algorithm on an image". I do not
recall typing "running algorithm on an image to determine depth
perception", what i recalled doing was proposing two approaches from
physics for depth perception. Thanks for the thought though.
> > If a robot is ordered to find an exit out of a room;YOung:> A good question. I guess that it would either be programmed
> > and the robot circles about looking for a door or
> > opening and happens to find none--the room is walled
> > completely. How does the robot improvise on its own?
> > E.g. Break down a wall.
to > destroy objects in it's way, or it would have to learn to do
that by > itself. How, though, I'm not sure!
Borgia:From my own personal experience i think creativity =
integration. You see an apple falling down and then integrate that
visual data with other data to get Newton's gravitation. Think about
how we CREATIVELY solve problems, it seems to be one and only one
way: integrating relevant but seemily disparate data into a new
synthesis. Can a robot on seeing a woman slicing an apple on a street
corner break down this visual input into some version of this crude
formalization:"sharp object(of certain characteristics y) + force +
an object(of certain characteristics x))--> a split x + object y"??.
Then, how can one write algorithms that will attempt to match this
_solution pattern_ with a _problem pattern_(e.g. getting out of a
walled room)?. To find a solution to something, first, the problem
has to be defined. Is an algorithm capable of partially formalizing
environmental events, feasible?
After the problem of the walled room has been formalized:
(a)Get from point_A_(inside the walled enclosure) to point _B_(out of
the walled enclosure)
(b)How? Tranverse a sectional area without walls
(c)There is no such area.
[The problem has been determined: no such area. Any "~x" that
fustrates acquiring an objective "z":is a problem.
Can the robot then formalize this scenario into this format:
X(passage) + Y(robot; moving) = Z(objective--get from A to B)?
Since the room is walled, then there is no X, only ~X. "~X"
Can the robot then proceed to this stage: ~X + Y = ~Z.--this will now
be the _state of events_ in the robots' cpu, There are logically two
options: Opt for ~Z--and not leave the room(that will be against its
instruction, thus the robot can't do that) Or two, eliminate ~X--this
fits with its instructions.
How do you eliminate ~X?
Scan ~X(the walls), from the scanning the robot will gather some
scientific data from its scans--physical and chemical properties of
Then, First priority:(i)how do you eliminate walls or things bearing
close resemblance to the physical/chemical properties of walls as
determined by the robot's scans?
Second priority:(ii)How don you eliminate any object?
Inorder to answer these questions by itself(the robot), algorithms
then prompt the robot to conduct memory search for visual data
involving scenes of any form of separation involving physical
objects?--cutting, slicing, dicing, twisting, cracking, burning, ,
chemical dissolution in a degree relevantly close to the
physical/chemical objects of walls and digressing from that point
away. etcetera. Then, formalize these scenes into a X + Y = Z, partly
using _cause and effect_/physics, and attempt to implement the
_formalization_ on the walls so as to create a passage way out of a
Does this makes any sense?
- Can a chip help computers see in 3D?
09:07 Wednesday 3rd July 2002
Stephen Shankland, CNET News.com
A Silicon Valley start-up believes it can give stereo vision to video
cameras by encoding a processing scheme into a custom chip. It could
ready the way for robots with depth perception
A Silicon Valley start-up believes it can improve computer vision by
combining a custom-designed chip with the way humans see.
Human brains judge how far away objects are by comparing the slightly
different view each eye sees. Tyzx hopes to build this stereo vision
process into video cameras.
The Palo Alto, California-based start-up has encoded a processing
scheme into a custom chip called DeepSea, allowing the processor to
determine not only the color of each tiny patch of an image but also
how far away that patch is from the camera.
The technology could be a boon for surveillance systems,
strengthening the ability to track people in banks, stores or
airports. But stereo vision could have wider uses as well, helping
focus a computer's attention and cutting down on the amount of data
that needs to be crunched.
For instance, a vacuuming robot trying to discern a table leg through
pattern recognition could avoid getting caught up in examining the
wallpaper in the background. Similarly, vehicles could use the
technology to detect obstacles in their path while filtering out
"The biggest value is the segmentation. It separates out the portion
of the image that interests you," said Takeo Kanade, a stereo vision
computing pioneer at Carnegie Mellon University and a member of an
independent Tyzx advisory board. "You have not only appearance but
also distance to each point. That makes the subsequent processing,
such as object detection and recognition, significantly easier."
Tyzx's first customers are mostly research labs, with other potential
business partners evaluating the technology, chief executive Ron Buck
said in an interview. Those who have bought the systems include MD
Robotics, the company that makes the robotic arm for the Space
Shuttle and, in the future, for the International Space Station. And
ChevronTexaco is employing the equipment for "augmented reality"
work -- supplementing what ordinary people see with computer imagery
for tasks such as operating oil platform cranes in bad weather.
The company hopes to win customers in the military and surveillance
industries, and, as costs go down, to expand into
broader "intelligent environments" where, for example, doors could
open automatically or a house could send a medical alert if someone
has been sitting still for an unusually long time. But Tyzx faces a
solid challenge translating the idea into a workable product.
"I believe it's a great idea," Kanade said. "Conceptually it's easy,
but computationally it's not."
Tyzx is backed by Vulcan Ventures, the investment firm of Microsoft
co-founder Paul Allen. It has less than 20 employees, some of whom
have years of experience in the field.
John Woodfill and Gaile Gordon launched the company in early 2001,
but much of their work precedes that date. A key formula used in the
custom chip dates back to 1990, and Tyzx has had prototype chips for
about a year, Buck said. It's only recently, though, that Tyzx's
ideas have become economically feasible.
Eyes on the prize
Stereo vision may indeed be a leap ahead for computers, but there's
still a long way to go before machines can achieve the sophistication
of human sight.
"Because vision comes so naturally to us, we don't appreciate the
problem intuitively," said David Touretzky, a computational
neuroscientist at Carnegie Mellon. "I don't think we got that
appreciation until people started trying to build computer systems to
A large fraction of the brains of primates such as monkeys, apes and
humans is devoted to processing visual information, Touretzky said.
There are more than 20 different specialised areas for tasks such as
recognizing motion, color, shapes and spatial relationships between
"These areas are all interconnected in ways not fully understood
yet," Touretzky said, but together these parts of the brain can
discern the difference between the edge of a shadow and the edge of
an object or compensate for color shifts that occur when the sun
Tyzx isn't the only company trying to capitalize on stereo computer
vision. Microsoft Research is working on technology that extracts 3D
information from 2D pictures. Point Grey Research already has cameras
on the market, though its processing algorithms require a full-
In Japan, a company called ViewPlus is working in collaboration with
Point Grey Research. Its products, though, combine as many as 60
cameras into a spherical system that produces 20 simultaneous video
These other companies are taking a fundamentally different approach
to Tyzx in one respect: Their systems compare more than two images.
Carnegie Mellon's Kanade said it might seem that comparing three
images would be a harder computational task, but in fact having more
data to work with can actually make the process simpler.
The key development at Tyzx is its custom chip, which runs an
algorithm called census correspondence that quickly finds
similarities across two streams of video images broken up into a
square grid of 512 pixels, or picture elements. The chip can perform
this comparison 125 times per second with a video image measuring 512
by 512 pixels, but the 33MHz DeepSea consumes much less power than
full-fledged processors such as Intel's Pentium.
"It allows incredibly compute-intensive searching for matching pixels
to happen very fast at a very low price. It allows us to bring stereo
vision to computers," chief executive Buck said.
Another important development needed to reach Tyzx's low-price
targets is camera sensors built using the comparatively inexpensive
complimentary metal-oxide semiconductor (CMOS) technology -- the same
process used to build most computer chips, Buck said. Digital cameras
today use more elaborate -- but more expensive -- "charge-coupled
devices", or CCDs.
Kanade has an appreciation for the difficulties involved. About 10
years ago he built an expensive but pioneering stereo vision system
with many processors that could determine range information by
comparing the images from multiple cameras.
Since then, more powerful computer processing abilities have elevated
the potential of the field, which Kanade believes will take off once
stereo cameras are as cheap as today's ordinary video cameras.
"I'm very impressed with the various attempts which made real-time
stereo possible. I think the Tyzx effort may be one of the eventual
successes," Kanade said.