Sweet! Free t-shirts. What sizes are they?

Payload is kind of low at 300g (< 1 lb.)



Thomas

Like robots?

Visit:

www.Robots-And-Androids.com

<http://www.rdmag.com/News/2012/06/Life-Sciences-Engineers-Discover-Neural-Rhyth\
ms-Drive-Physical-Movement/?et_cid=2683130&et_rid=54739636&linkid=http%3a%2f%2fw\
ww.rdmag.com%2fNews%2f2012%2f06%2fLife-Sciences-Engineers-Discover-Neural-Rhythm\
s-Drive-Physical-Movement%2f>


Interesting research into the substructure of movement control.

Kevin.

Engineers discover neural rhythms drive physical movement

By Andrew Myers, Stanford University

The neurons that control movement are not a predictable bunch.
Scientists working to decode how such neurons convey information to
muscles have been stymied when trying to establish a one-to-one
relationship between a neuron’s behavior and external factors such as
muscle activity or movement velocity.

In an article published online by Nature, a team of electrical engineers
and neuroscientists working at Stanford University propose a new theory
of the brain activity behind arm movements. Their theory is a
significant departure from existing understanding and helps to explain,
in relatively simple and elegant terms, some of the more perplexing
aspects of the activity of neurons in motor cortex.

In their paper, electrical engineering associate professor Krishna
Shenoy and postdoctoral researchers Mark Churchland, now a professor at
Columbia University, and John Cunningham of Cambridge University, now a
professor at Washington University in Saint Louis, have shown that the
brain activity controlling arm movement does not encode external spatial
information—such as direction, distance, and speed—but is instead
rhythmic in nature.

Understanding the brain
Neuroscientists have long known that the neurons responsible for vision
encode specific, external-world information—the parameters of sight. It
had been theorized and widely suggested that motor cortex neurons
function similarly, conveying specifics of movement such as direction,
distance, and speed, in the same way the visual cortex records color,
intensity, and form.

"Visual neurons encode things in the world. They are a map, a
representation," said Churchland, who is first author of the paper.
"It's not a leap to imagine that neurons in the motor cortex should
behave like neurons in the visual cortex, relating in a faithful way to
external parameters, but things aren’t so concrete for movement."

Scientists have disagreed about which movement parameters are being
represented by individual neurons. They could not look at a particular
neuron firing in the motor cortex and determine with confidence what
information it was encoding.

"Many experiments have sought such lawfulness and yet none have found
it. Our findings indicate an alternative principle is at play," said
co-first author Cunningham.

"Our main finding is that the motor cortex is a flexible pattern
generator, and sends rhythmic signals down the spinal cord," said
Churchland.

Engine of movement
To employ an automotive analogy, the motor cortex is not the steering
wheel, odometer or speedometer representing real-world information. It
is more like an engine, comprised of parts whose activities appear
complicated in isolation, but which cooperate in a lawful way as a whole
to generate motion.

"If you saw a piston or a spark plug by itself, would you be able to
explain how it makes a car move?" asked Cunningham rhetorically.
"Motor-cortex neurons are like that, too, understandable only in the
context of the whole."

In monitoring electrical brain activity of motor-cortex neurons,
researchers found that they typically exhibit a brief oscillatory
response. These responses are not independent from neuron to neuron.
Instead, the entire neural population oscillates as one in a beautiful
and lawfully coordinated way.

The electrical signal that drives a given movement is therefore an
amalgam—a summation—of the rhythms of all the motor neurons firing at a
given moment.

"Under this new way of looking at things, the inscrutable becomes
predictable," said Churchland. "Each neuron behaves like a player in a
band. When the rhythms of all the players are summed over the whole
band, a cascade of fluid and accurate motion results."

Dr. Daofen Chen, Program Director, Systems and Cognitive Neuroscience at
the National Institute of Neurological Disorders and Stroke at the
National Institutes of Health, said Shenoy and team are working at the
cutting edge of the field. "In trying to find the basic response
properties of the motor cortex, Shenoy and his colleagues are searching
for the holy grail of neuroscience," said Dr. Chen. "They consistently
tackle important but tough questions in thought-provoking ways and in
ambitious proposals. NIH is proud to support this kind of pioneering and
transformative research."

Precedents in nature
In the new model, a few relatively simple rhythms explain neural
features that had confounded science earlier.

"Many of the most-baffling aspects of motor-cortex neurons seem natural
and straightforward in light of this model," said Cunnigham.

The team studied non-rhythmic reaching movements, which made the
presence of rhythmic neural activity a surprise even though, the team
notes, rhythmic neural activity has a long precedence in nature. Such
rhythms are present in the swimming motion of leeches and the gait of a
walking monkey, for instance.

"The brain has had an evolutionary goal to drive movements that help us
survive. The primary motor cortex is key to these functions. The
patterns of activity it displays presumably derive from evolutionarily
older rhythmic motions such as swimming and walking. Rhythm is a basic
building block of movement," explained Churchland.

Reaching for the grail
To test their hypothesis, the engineers studied the brain activity of
monkeys reaching to touch a target. According to the researchers,
experiments show this 'underlying rhythm' strategy works very well to
explain both brain and muscle activity. In their reaching studies, the
pattern of shoulder-muscle behavior could always be described by the sum
of two underlying rhythms.

"Say you're throwing a ball. Beneath it all is a pattern. Maybe your
shoulder muscle contracts, relaxes slightly, contracts again, and then
relaxes completely, all in short order," explained Churchland.

The researchers say that although the activity may not be exactly
rhythmic, it can be created by adding together two or three other
rhythms. The team asserts that this may be how the brain solves the
problem of creating patterns of movement.

"This surprised us a bit. In decidedly arrhythmic movements, there were
these unmistakable patterns," said Churchland.

"This research builds on a strong theoretical framework and adds to
growing evidence that rhythmic activity is important for many
fundamental brain functions," said Yuan Liu of the National Institute of
Neurological Disorders and Stroke, NIH. "Further research in this area
may help us devise more effective technology for controlling prosthetic
limbs." Liu is the co-lead of the NIH-NSF Collaborative Research in
Computational Neuroscience program.

Accordingly, the seemingly complex system that is the motor cortex can
now be at least partially understood in more straightforward terms.

"The motor cortex is an engine of movement that obeys lawful dynamics,"
said Shenoy.

Source: Stanford University

Kevinr said: Friday, June 08, 2012 7:44 PM
> Engineers discover neural rhythms drive physical movement

This is a very intriguing article Kevin. It almost suggests our
brains "sings" to our muscles. Now if only we can learn the
songs...

Randy

At 11:35 AM -0500 6/9/12, Randy M. Dumse wrote:
>Kevinr said: Friday, June 08, 2012 7:44 PM
>> Engineers discover neural rhythms drive physical movement
>
>This is a very intriguing article Kevin. It almost suggests our
>brains "sings" to our muscles. Now if only we can learn the
>songs...

It probably also explains why some folks have a better sense of rhythm than
others.

(also sent to Leaf group)

I've finally done what very few people in the world have done, I've put together
an interactive life/size humanoid / android using Leaf / ALICE  on a PC for
"higher brain functions" and EZ Robot system for physical controls.

IT WORKS!

(in the words of Dr. Frankenstein)

IT'S ALIVE!

:)

(video to come.)

Thomas

Like robots?

Visit:

www.Robots-And-Androids.com

Greetings,

We have a great meeting planned for this Saturday. Check the website and this
group Friday night or Saturday AM before coming just to be sure it's happening.

Bring along robots, great finds & bargains, projects to show, etc. Bring things
give away, trade, etc. during & after the meeting. If you have pictures you'd
like to show during the Round-the-room, bring them to me on a thumbdrive (before
the meeting if possible). Bring projects to work on during the afternoon
workshop.

Mike Marzetta will present on his MINDS-i all-terrain robots.

Bring info about area or online events coming that might be of interest to club
members.

Of course we always encourage lots of mingling after meetings.

We'll do pizza & fixins for lunch ($5/person) after the meeting.

WiFi should be available in the room.

I hope lots of you can make it.

===============================
If you'd like to give a robot (or other technical subject) talk at one of our
meetings, please contact me at the secretary email link on the SRS "Contact Us"
page or talk to me at a meeting. I'm booking slots beyond October.

Come early if you can & help set up the meeting room & chat.

Steve K. - SRS Secretary & Meeting Organizer
http://www.SeattleRobotics.org

Howdy,

After about half decade of delay my 2-wheel balancing robot nBot finally has
some sensors (PING and SHARP).  The balance behavior subsystem has evolved,
painfully, to the point where it is reliable enough and stable enough that it
seemed like it would be worthwhile to add some more smarts to the robot besides
just odometry and the balance itself.

At this point it has a pretty good wall-following, or more accurately
perimeter-following, behavior working.  I'm still playing with the positioning
and alignment of the sensors but it's getting close.  Here's a video of the
robot doing right hand perimeter-following in one of the cluttered basement labs
in the geology building where I work:


<http://geology.heroy.smu.edu/dpa-www/robo/nbot/20120614_nbot_04.mpg

In wall following for a real live wall, the IR detectors can very quickly find a
parallel path a few inches away from the wall.  For more complex environments
the ultrasonic sensors keep the robot at a greater (safer) distance from the
objects defining the perimeter.

The robot is also pretty robust when bumping into things it doesn't otherwise
detect, while maintaining it's balance and stability. About halfway through the
video the robot hangs a wheel and must wiggle itself free.  Shortly thereafter
it bangs pretty hard into a closed door, and recovers.  Actually you can see
that a little blue LED lights up just inches before the collision.  It sees the
door, just not in time.  Still some more adjustments to make...

cheers!
dpa

On 6/18/2012 7:22 PM,  dpa_io wrote:
> .. the ultrasonic sensors keep the robot at a greater (safer)
> distance  from the objects defining the perimeter.

I've had problems on small wall huggers with Maxbotix ultrasonic sensors
in trying to correct for the angle of the robot wrt to the wall and the
cosine error in the distance.  The behavior tends to exhibit
oscillations and/or serious instability.

Any good references on compensating for the approach angle?  I made a
few ill-fated attempts at setting up a PID model.

--
Paul

Excellent.  Are those wheel chair motors?  So quiet!

Can the Kinect see the floor?  How far out?

dpa


--- In SeattleRobotics@yahoogroups.com, Xandon Frogget <xandon@...> wrote:
>
> That is awesome!  You have been a huge inspiration. I am glad to see you are
still pushing forward.
>
> I wanted to share this with you as well, since your inspiration led to his
design. Following your instruction closely, I was able to create Felix. I
wouldn't have been able to do it so well without your video's and articles. 
After fixing some motor controller problems in the beginning he is now
incredibly stable and perfect for working through different problems. Here he is
getting out of a canyon in the kitchen. He is so quiet, you can't hear him at
all. Except for the creaking floor every once in a while you wouldn't know he
was there.
>
> http://www.youtube.com/watch?v=-NyBIRMo7sE
>
>
> Here he is outside just wandering around. The Kinect works outdoors in fairly
cloudy weather, but goes blind when the sun comes out. Fortunately this is
Seattle so the weather works in his favor.
>
> http://www.youtube.com/watch?v=gwhzspRR_XE
>
> -Xandon
>
>
>
> On Jun 18, 2012, at 7:22 PM, dpa_io wrote:
>
> >
> > <http://geology.heroy.smu.edu/dpa-www/robo/nbot/20120614_nbot_04.mpg
> >
>

Paul,

I've had a hard time getting good wall following to work with just ultrasonic
sensors, especially the low power ones.  I think the problem is that there is no
return signal for very oblique angles, when the robot is close to the wall.  So
the detection comes and goes depending on the angle of the detector/robot.  I
got wall following to work pretty well on the jBot outdoor robot using just
ultrasonics, but those are much higher power (POLAROID) sensors, and there are
four of them.

The current nBot behavior uses the PING ultrasonic sensors to find and approach
the wall, and the sharp IR distance sensors to actually do the wall following.
They are angled 45 degrees off-center, and divided into four ranges:

close range: turn SLIGHTLY away from the wall
middle range: do nothing (dead zone)
far range: turn SLIGHTLY toward the wall
out of range: turn AGGRESSIVELY towards the wall

So the PING ultrasonic sensors and the "out of range" state combine to steer the
robot towards a right-hand (in this case) perimeter while avoiding obstacles,
and the SHARP IR detectors steer the robot parallel to the wall when it finds
one.

The exact approach angle is a combination of the mounting angle of the sensor
and the speed of the turn(s).  Hope this helps.

dpa

PS: the above technique of using a deadzone with boundaries is really a
quick-and-dirty way of emulating a PID type controller. PID probably really is
the right way to do it.  The deadzone method is much easier to adjust, though.



--- In SeattleRobotics@yahoogroups.com, "Paul J. Ste. Marie" <robotics@...>
wrote:
>
> On 6/18/2012 7:22 PM,  dpa_io wrote:
> > .. the ultrasonic sensors keep the robot at a greater (safer)
> > distance  from the objects defining the perimeter.
>
> I've had problems on small wall huggers with Maxbotix ultrasonic sensors
> in trying to correct for the angle of the robot wrt to the wall and the
> cosine error in the distance.  The behavior tends to exhibit
> oscillations and/or serious instability.
>
> Any good references on compensating for the approach angle?  I made a
> few ill-fated attempts at setting up a PID model.
>
> --
> Paul
>

Nice 'bot Dave!

Alan  KM6VV

On 6/18/2012 7:22 PM, dpa_io wrote:
> Howdy,
>
> After about half decade of delay my 2-wheel balancing robot nBot finally has
some sensors (PING and SHARP).  The balance behavior subsystem has evolved,
painfully, to the point where it is reliable enough and stable enough that it
seemed like it would be worthwhile to add some more smarts to the robot besides
just odometry and the balance itself.
>
> At this point it has a pretty good wall-following, or more accurately
perimeter-following, behavior working.  I'm still playing with the positioning
and alignment of the sensors but it's getting close.  Here's a video of the
robot doing right hand perimeter-following in one of the cluttered basement labs
in the geology building where I work:
>
>
> <http://geology.heroy.smu.edu/dpa-www/robo/nbot/20120614_nbot_04.mpg
>
> In wall following for a real live wall, the IR detectors can very quickly find
a parallel path a few inches away from the wall.  For more complex environments
the ultrasonic sensors keep the robot at a greater (safer) distance from the
objects defining the perimeter.
>
> The robot is also pretty robust when bumping into things it doesn't otherwise
detect, while maintaining it's balance and stability. About halfway through the
video the robot hangs a wheel and must wiggle itself free.  Shortly thereafter
it bangs pretty hard into a closed door, and recovers.  Actually you can see
that a little blue LED lights up just inches before the collision.  It sees the
door, just not in time.  Still some more adjustments to make...
>
> cheers!
> dpa
>

I made a mistake at the last SRS meeting.  I mentioned the Greenwood Car Show
and I believe I reported that it was this weekend.  I rechecked the calendar and
it is not this weekend but the next Saturday the 30th of June.  Sorry for the
inconvenience.  I must getting old :(

There are still plans to display over 23 cars that run on batteries.  These cars
run on American made electrons as opposed to a toxic brew of chemicals that has
a 60% chance of being imported from countries that don't seem to like us very
much.

P.S. There will plenty of other cars to look at from the Ford Model T to the
latest jet propelled dragster.  Lots of shiny chrome and other customizations to
look at too.


----------------------------------------------------
"What the detractors and critics of electric vehicles
have been saying for years, is true. The electric
vehicle is not for everybody, given the limited range
it can only meet the needs of 90% of the population."

Ed Begley Jr.
----------------------------------------------------

On 6/19/2012 11:25 AM,  dpa_io wrote:

> close range: turn SLIGHTLY away from the wall middle range: do
> nothing (dead zone) far range: turn SLIGHTLY toward the wall out of
> range: turn AGGRESSIVELY towards the wall

I think that might have been similar one of my first attempts that did
work well. The Maxbotix sensors I'm using have a fairly narrow beam, and
the problem I ran into is that turning either towards or away from the
wall causes the apparent distance to increase (1/cos theta).  That
didn't play well with straight proportional control (turn rate
proportional to distance error), and PID was just a mess.

Somehow the math needs to look at the rate of approach/retreat and the
current turn rate and feed that into the PID.  The other thing may be to
use the derivative of the PID output as the turn rate.

Other things I've considered are using two side sensors or putting the
sensor on a servo and scanning left/right to locate the closest direction.

-- Paul

Who says technology and nature don't mix?

Fun video below shows how National Geographic photographer Steve Winter
put together a fairly simple robot in order to get an up-close view of
tigers.

http://www2.electronicproducts.com/Robot_vs_tiger_National_Geographic_gets_up_cl\
ose_and_personal_with_fierce_jungle_feline-article-fajb_robot_vs_tiger_jun2012-h\
tml.aspx