Muscle model prototype

CSGnet Archive CSG2014
1

Hi all,

Here is a crude prototype of a flexor-extensor muscle system that can control muscle tone and joint angle independently. It’s based on various Bill’s simulations and models of muscles and arms.

It is not a very realistic model of physical muscle systems, but the control system should be similar.

A muscle is modeled as a two-part system - one part is capable of contracting - that is a geared DC motor contracting a thread. The other part is elongating and storing elastic energy - a piece of rubber band. The rubber part is connected to a slider potentiometer and it is measuring elongation of the rubber. Elongation is proportional to tension of the rubber, so the equivalent measure in real muscles would be sensed by Golgi tendon receptors.

Joint angle is measured by a rotary potentiometer. In real muscle systems, the equivalent measure is probably sensed by muscle spindle receptors, as elongation of muscles.

In Bill’s models there is also a rate of change of muscle length component. This model moves very slowly, so it doesn’t need it, but it should be pretty straightforward to implement it, to be sensed as a derivative on joint anglular position.

In the attached video, you can see first - how the angle changes following angle reference change. Second - how muscle tone is increased, while angle is maintained. Then you can see the difference between disturbance rejection when muscle tone is high and when muscle tone is low - higher muscle tone also means higher angle gain.

This is third or fourth version of the system, first one that works, though. I tried using resistive/conductive rubber as tendon sensor, but that didn’t work very well, probably because of hysteresis of the material. I also tried using servo motors as contractive elements, but they don’t have positional feedback, so I didn’t find them very useful.

If someone would like to build one themselves, I’d be happy to help. I suggest using bigger and faster DC motors, as well as a faster Arduino board. The code is pretty simple and straightforward, here is the relevant part pasted, complete code for Arduino attached.

MuscleModelTest.ino (2.73 KB)

MuscleModel.wmv (3.87 MB)

···

cs[0].p = analogRead(senspin[0]); // left slider pot = tension of left tendon

cs[1].p = analogRead(senspin[1]); // right slider

angle.p = analogRead(sensPinAngle); // joint angle from rotary pot

tonus.p = cs[0].p + cs[1].p; // muscle tone is a sum of left and right tension

tonus.r = 2 * analogRead(refpin[0]); // reference tonus from external board

angle.r = analogRead(refpin[1]); // reference angle

angle.e = angle.r - angle.p;

tonus.e = tonus.r - tonus.p;

angle.qo = 30*angle.e;

tonus.qo = 1.5*tonus.e;

cs[0].r = tonus.qo + angle.qo;

cs[1].r = tonus.qo - angle.qo;

for (int i =0;i<2;i++) {

int e = cs[i].r - cs[i].p;

cs[i].qo = e * cs[i].Ko;

}

Best,

Adam

2

[From Rick Marken (2014.07.14.1535)]

···

On Mon, Jul 14, 2014 at 11:24 AM, Adam Matic adam.matic@gmail.com wrote:

Hi all,

AM: Here is a crude prototype of a flexor-extensor muscle system that can control muscle tone and joint angle independently. It’s based on various Bill’s simulations and models of muscles and arms.

RM: This is great Adam. I think I could even set this up myself. I think this could be the basis for a great instructional tool and academic paper demonstrating pct principles

RM: It is a very simple model because all the mathematical complexity of such models is absorbed by the physical components of your device (the motors, rubber bands, etc). The control model exists as a program in the Arduino in the context of these physical components just as the control system in a living organism exists in the nervous system, in the context of the physical components of the body (muscles/tendons) and environment (gravity, torques, etc) .

Based on the code you sent it looks like there are two higher level systems, one controlling muscle tone (tonus) and the other controlling limb angle (angle). You set the references for the perceptions controlled by these two systems – tonus and angle – by turning the poteniometers. So you are like a higher level system varying the references for angle and tonues in order to achieve your goal… The tonus and angle control systems control their perceptions by varying their outputs that set the references for two lower level tendon tension control systems. The reference for the perceived tendon tension to be produced by one of these systems is the sum of the outputs of the tonus and angle system and the reference for the other is the difference between these outputs. And it works great.

RM: Although you say it’s"crude" I actually see it as a very elegant demonstration of the control of perception approach to producing behavior – in this case, controlled variation of joint angle.

Best

Rick

It is not a very realistic model of physical muscle systems, but the control system should be similar.

A muscle is modeled as a two-part system - one part is capable of contracting - that is a geared DC motor contracting a thread. The other part is elongating and storing elastic energy - a piece of rubber band. The rubber part is connected to a slider potentiometer and it is measuring elongation of the rubber. Elongation is proportional to tension of the rubber, so the equivalent measure in real muscles would be sensed by Golgi tendon receptors.

Joint angle is measured by a rotary potentiometer. In real muscle systems, the equivalent measure is probably sensed by muscle spindle receptors, as elongation of muscles.

In Bill’s models there is also a rate of change of muscle length component. This model moves very slowly, so it doesn’t need it, but it should be pretty straightforward to implement it, to be sensed as a derivative on joint anglular position.

In the attached video, you can see first - how the angle changes following angle reference change. Second - how muscle tone is increased, while angle is maintained. Then you can see the difference between disturbance rejection when muscle tone is high and when muscle tone is low - higher muscle tone also means higher angle gain.

This is third or fourth version of the system, first one that works, though. I tried using resistive/conductive rubber as tendon sensor, but that didn’t work very well, probably because of hysteresis of the material. I also tried using servo motors as contractive elements, but they don’t have positional feedback, so I didn’t find them very useful.

If someone would like to build one themselves, I’d be happy to help. I suggest using bigger and faster DC motors, as well as a faster Arduino board. The code is pretty simple and straightforward, here is the relevant part pasted, complete code for Arduino attached.

cs[0].p = analogRead(senspin[0]); // left slider pot = tension of left tendon

cs[1].p = analogRead(senspin[1]); // right slider

angle.p = analogRead(sensPinAngle); // joint angle from rotary pot

tonus.p = cs[0].p + cs[1].p; // muscle tone is a sum of left and right tension

tonus.r = 2 * analogRead(refpin[0]); // reference tonus from external board

angle.r = analogRead(refpin[1]); // reference angle

angle.e = angle.r - angle.p;

tonus.e = tonus.r - tonus.p;

angle.qo = 30*angle.e;

tonus.qo = 1.5*tonus.e;

cs[0].r = tonus.qo + angle.qo;

cs[1].r = tonus.qo - angle.qo;

for (int i =0;i<2;i++) {

int e = cs[i].r - cs[i].p;

cs[i].qo = e * cs[i].Ko;

}

Best,

Adam

–
Richard S. Marken PhD
www.mindreadings.com

3

Hi Rick and Adam,

I am wondering whether this could be another one for YouTube, with Rick’s annotation?

Warren

···

On Mon, Jul 14, 2014 at 11:24 AM, Adam Matic adam.matic@gmail.com wrote:

Hi all,

AM: Here is a crude prototype of a flexor-extensor muscle system that can control muscle tone and joint angle independently. It’s based on various Bill’s simulations and models of muscles and arms.

RM: This is great Adam. I think I could even set this up myself. I think this could be the basis for a great instructional tool and academic paper demonstrating pct principles

RM: It is a very simple model because all the mathematical complexity of such models is absorbed by the physical components of your device (the motors, rubber bands, etc). The control model exists as a program in the Arduino in the context of these physical components just as the control system in a living organism exists in the nervous system, in the context of the physical components of the body (muscles/tendons) and environment (gravity, torques, etc) .

Based on the code you sent it looks like there are two higher level systems, one controlling muscle tone (tonus) and the other controlling limb angle (angle). You set the references for the perceptions controlled by these two systems – tonus and angle – by turning the poteniometers. So you are like a higher level system varying the references for angle and tonues in order to achieve your goal… The tonus and angle control systems control their perceptions by varying their outputs that set the references for two lower level tendon tension control systems. The reference for the perceived tendon tension to be produced by one of these systems is the sum of the outputs of the tonus and angle system and the reference for the other is the difference between these outputs. And it works great.

RM: Although you say it’s"crude" I actually see it as a very elegant demonstration of the control of perception approach to producing behavior – in this case, controlled variation of joint angle.

Best

Rick

It is not a very realistic model of physical muscle systems, but the control system should be similar.

A muscle is modeled as a two-part system - one part is capable of contracting - that is a geared DC motor contracting a thread. The other part is elongating and storing elastic energy - a piece of rubber band. The rubber part is connected to a slider potentiometer and it is measuring elongation of the rubber. Elongation is proportional to tension of the rubber, so the equivalent measure in real muscles would be sensed by Golgi tendon receptors.

Joint angle is measured by a rotary potentiometer. In real muscle systems, the equivalent measure is probably sensed by muscle spindle receptors, as elongation of muscles.

In Bill’s models there is also a rate of change of muscle length component. This model moves very slowly, so it doesn’t need it, but it should be pretty straightforward to implement it, to be sensed as a derivative on joint anglular position.

In the attached video, you can see first - how the angle changes following angle reference change. Second - how muscle tone is increased, while angle is maintained. Then you can see the difference between disturbance rejection when muscle tone is high and when muscle tone is low - higher muscle tone also means higher angle gain.

This is third or fourth version of the system, first one that works, though. I tried using resistive/conductive rubber as tendon sensor, but that didn’t work very well, probably because of hysteresis of the material. I also tried using servo motors as contractive elements, but they don’t have positional feedback, so I didn’t find them very useful.

If someone would like to build one themselves, I’d be happy to help. I suggest using bigger and faster DC motors, as well as a faster Arduino board. The code is pretty simple and straightforward, here is the relevant part pasted, complete code for Arduino attached.

cs[0].p = analogRead(senspin[0]); // left slider pot = tension of left tendon

cs[1].p = analogRead(senspin[1]); // right slider

angle.p = analogRead(sensPinAngle); // joint angle from rotary pot

tonus.p = cs[0].p + cs[1].p; // muscle tone is a sum of left and right tension

tonus.r = 2 * analogRead(refpin[0]); // reference tonus from external board

angle.r = analogRead(refpin[1]); // reference angle

angle.e = angle.r - angle.p;

tonus.e = tonus.r - tonus.p;

angle.qo = 30*angle.e;

tonus.qo = 1.5*tonus.e;

cs[0].r = tonus.qo + angle.qo;

cs[1].r = tonus.qo - angle.qo;

for (int i =0;i<2;i++) {

int e = cs[i].r - cs[i].p;

cs[i].qo = e * cs[i].Ko;

}

Best,

Adam

–
Richard S. Marken PhD
www.mindreadings.com

4

[From Adam Matic 2014.07.15.]

Rick Marken (2014.07.14.1535)

RM: This is great Adam. I think I could even set this up myself. I think this could be the basis for a great instructional tool and academic paper demonstrating pct principles

RM: It is a very simple model because all the mathematical complexity of such models is absorbed by the physical components of your device (the motors, rubber bands, etc). The control model exists as a program in the Arduino in the context of these physical components just as the control system in a living organism exists in the nervous system, in the context of the physical components of the body (muscles/tendons) and environment (gravity, torques, etc) .
Based on the code you sent it looks like there are two higher level systems, one controlling muscle tone (tonus) and the other controlling limb angle (angle). You set the references for the perceptions controlled by these two systems -- tonus and angle -- by turning the poteniometers. So you are like a higher level system varying the references for angle and tonues in order to achieve your goal.. The tonus and angle control systems control their perceptions by varying their outputs that set the references for two lower level tendon tension control systems. The reference for the perceived tendon tension to be produced by one of these systems is the sum of the outputs of the tonus and angle system and the reference for the other is the difference between these outputs. And it works great.
RM: Although you say it's"crude" I actually see it as a very elegant demonstration of the control of perception approach to producing behavior -- in this case, controlled variation of joint angle.

Great, thanks!
I agree that control systems Bill proposed for control of muscles look very elegant and they work nicely, but I am not very clear how exactly physics of real muscles and tendons corresponds to this system with rubber and a DC motor. First - is the tension adequately measured by a linear potentiometer, and second, is the force generated by the motor-rubber system similar to force generated by muscle-tendon system.
Warren, I'll put something on youtube, probably later this week.
Adam

···
5

Fab.

That’s a very important point about the physiology and essential if we are to persuade the skeptics!

Warren

···

[From Adam Matic 2014.07.15.]

Rick Marken (2014.07.14.1535)

Great, thanks!

I agree that control systems Bill proposed for control of muscles look very elegant and they work nicely, but I am not very clear how exactly physics of real muscles and tendons corresponds to this system with rubber and a DC motor. First - is the tension adequately measured by a linear potentiometer, and second, is the force generated by the motor-rubber system similar to force generated by muscle-tendon system.

Warren, I’ll put something on youtube, probably later this week.

Adam

RM: This is great Adam. I think I could even set this up myself. I think this could be the basis for a great instructional tool and academic paper demonstrating pct principles

RM: It is a very simple model because all the mathematical complexity of such models is absorbed by the physical components of your device (the motors, rubber bands, etc). The control model exists as a program in the Arduino in the context of these physical components just as the control system in a living organism exists in the nervous system, in the context of the physical components of the body (muscles/tendons) and environment (gravity, torques, etc) .

Based on the code you sent it looks like there are two higher level systems, one controlling muscle tone (tonus) and the other controlling limb angle (angle). You set the references for the perceptions controlled by these two systems – tonus and angle – by turning the poteniometers. So you are like a higher level system varying the references for angle and tonues in order to achieve your goal… The tonus and angle control systems control their perceptions by varying their outputs that set the references for two lower level tendon tension control systems. The reference for the perceived tendon tension to be produced by one of these systems is the sum of the outputs of the tonus and angle system and the reference for the other is the difference between these outputs. And it works great.

RM: Although you say it’s"crude" I actually see it as a very elegant demonstration of the control of perception approach to producing behavior – in this case, controlled variation of joint angle.

6

[Martin Taylor 2014.07.15.09.39]

I suppose it would depend on which band of skeptics you want to

persuade. Robot builders use all sorts of devices as muscles –
pneumatic, elastic, you name it. What matters is what works.
The physiologists are a different clan. I don’t know what is
necessary there, but a point to note is that in perceptual control,
linearity doesn’t matter (much). It influences absolute precision of
control at different points on the scale because it changes the
local loop gain near any particular value of the perception, but it
doesn’t affect the basic control phenomenon unless the local loop
gain gets so high as to lead to oscillation for some value of
perception. Whether a particular physiologist understand this is the
question. If so, then you may well be good. If not, the
linearity-nonlinearity issue might be a sticking point.
Martin

···

On 2014/07/15 9:22 AM, Warren Mansell
wrote:

Fab.

    That's a very important point about the physiology and

essential if we are to persuade the skeptics!

Warren

    On 15 Jul 2014, at 12:11, Adam Matic <adam.matic@gmail.com        >

wrote:

[From Adam Matic 2014.07.15.]

              I agree that control systems Bill proposed for

control of muscles look very elegant and they work
nicely, but I am not very clear how exactly physics of
real muscles and tendons corresponds to this system
with rubber and a DC motor. First - is the tension
adequately measured by a linear potentiometer, and
second, is the force generated by the motor-rubber
system similar to force generated by muscle-tendon
system.

7

[From Rick Marken (2014.07.15.1100)]

···

RM: Good point. I agree that they physics of the demo may not match the physics of the arm (at least not yet). But I still like the idea of testing the control model in the context of a real physical environment; the computer then just becomes the “organism” component of the model and the “environment” component is the actual environment in which the system does its controlling. But I see that this approach has it’s problems too; getting the lab produced environment to match the environment of the “organism” being modeled. But even with the problems, I think a model like your demonstrates the principles of perceptual control, as it occurs in the “real world” really nicely.

Adam Matic 2014.07.15.–

AM: I agree that control systems Bill proposed for control of muscles look very elegant and they work nicely, but I am not very clear how exactly physics of real muscles and tendons corresponds to this system with rubber and a DC motor. First - is the tension adequately measured by a linear potentiometer, and second, is the force generated by the motor-rubber system similar to force generated by muscle-tendon system.

Best

Rick

–
Richard S. Marken PhD
www.mindreadings.com

8

[From Adam Matic 2014.07.15.1230]

···

Rick Marken (2014.07.15.1100

Yes, having the physics take care of itself instead of having to simulate it on a computer makes the game a lot of fun (at least for us non-experts in physics). And using a mechanical device that people can touch and turn some knobs to see what happens seems to work nicely as a demonstration.

I think it also needs a complementary visualization on a computer in the form of plots and animated changes of nervous system frequencies.

Adam

But even with the problems, I think a model like your demonstrates the principles of perceptual control, as it occurs in the “real world” really nicely.

9

[From Adam Matic 2014.07.20.14.30]

Here is a short youtube video: https://www.youtube.com/watch?v=TmN99zNRs-4
Apparently, I left a black edge around the video, not sure how that happened. Diagrams may need pausing, but they are visible.

I’ve been playing with the system, added a rate of change component to tension and angle systems, changed gains and slowings. There seem to be more ways the setup can operate. Maybe flexor and extensor need separate length controllers. Anyone know of a detailed neuronal map involving alpha and gamma efferent fibers, all the interneurons, Golgi organs and muscle spindles? I keep getting some partial maps.

Best,

Adam

···

On Wed, Jul 16, 2014 at 12:40 PM, Adam Matic adam.matic@gmail.com wrote:

[From Adam Matic 2014.07.15.1230]

Rick Marken (2014.07.15.1100

Yes, having the physics take care of itself instead of having to simulate it on a computer makes the game a lot of fun (at least for us non-experts in physics). And using a mechanical device that people can touch and turn some knobs to see what happens seems to work nicely as a demonstration.

I think it also needs a complementary visualization on a computer in the form of plots and animated changes of nervous system frequencies.

Adam

But even with the problems, I think a model like your demonstrates the principles of perceptual control, as it occurs in the “real world” really nicely.

10

[From Rick Marken (2014.07.21.1700)]

This is marvelous Adam! The charts and model are very clear. But I would suggest that you take the time to produce this you tube video like a low budget TED talk. That is, I would have you giving a talk ysing the slides and the model but with you talking it through, possibly with pointers when you are talking about the slides. And then I think you should should be in the video of the model arm, pointing out what you are doing. For example, I could tell what you were doing when you increased the muscle tone and pushed on the arm but an observer might not.

I think this demo is really valuable, both as an explanation of the presumed physiological mechanisms that move out limbs but also as a demonstration of the control model of how it might actually work based on what we know of the physiology.

I think it would be particularly cool if you could also point out the areas of controversy. For example, you could point out that, although we know a lot about the physiology, there is controversy about how it actually works. You have produced one model which works by controlling sensed muscle length and tension. It’s a feedback control model. But there are other models of how the physiology works. For example, there’s the equilibrium point model which is ostensibly not a feedback control model. You might give a quick description of how that model is presumed to work and why it is almost certainly wrong (that might wake up the audience;-)

Great work, Adam. Really nice.

Best

Rick

···

On Mon, Jul 21, 2014 at 5:38 AM, Adam Matic adam.matic@gmail.com wrote:

[From Adam Matic 2014.07.20.14.30]

Here is a short youtube video: https://www.youtube.com/watch?v=TmN99zNRs-4
Apparently, I left a black edge around the video, not sure how that happened. Diagrams may need pausing, but they are visible.

I’ve been playing with the system, added a rate of change component to tension and angle systems, changed gains and slowings. There seem to be more ways the setup can operate. Maybe flexor and extensor need separate length controllers. Anyone know of a detailed neuronal map involving alpha and gamma efferent fibers, all the interneurons, Golgi organs and muscle spindles? I keep getting some partial maps.

Best,

Adam

–
Richard S. Marken PhD
www.mindreadings.com

On Wed, Jul 16, 2014 at 12:40 PM, Adam Matic adam.matic@gmail.com wrote:

[From Adam Matic 2014.07.15.1230]

Rick Marken (2014.07.15.1100

Yes, having the physics take care of itself instead of having to simulate it on a computer makes the game a lot of fun (at least for us non-experts in physics). And using a mechanical device that people can touch and turn some knobs to see what happens seems to work nicely as a demonstration.

I think it also needs a complementary visualization on a computer in the form of plots and animated changes of nervous system frequencies.

Adam

But even with the problems, I think a model like your demonstrates the principles of perceptual control, as it occurs in the “real world” really nicely.

11

[From Adam Matic 2014.07.22 1750]

Thanks. Yeah, a video with explanations would be nice, this could be something relatively easy to make at home and get familiar with control, try different gains and slowings and hierarchical relationships.

But first I’d like to make something more similar to real muscles, some sort of artificial muscles that engineers today have a hard time incorporating into robots. Most today’s robots use electric motors as actuators, usually DC servos, and this model uses them too.

Since the term control is usually used as “control of output”, first thing people would object to when they see this model would probably be the fact that real muscles have different physical properties than a thread being rolled on by a DC motor. Rubber tendons could probably go as similar to natural ones, but an actuator would need to be something different, maybe something like this: http://io9.com/scientists-just-created-some-of-the-most-powerful-muscl-1526957560

That might be interesting and useful without many explanations.

Best,

Adam

···

On Tue, Jul 22, 2014 at 1:59 AM, Richard Marken rsmarken@gmail.com wrote:

[From Rick Marken (2014.07.21.1700)]

This is marvelous Adam! The charts and model are very clear. But I would suggest that you take the time to produce this you tube video like a low budget TED talk. That is, I would have you giving a talk ysing the slides and the model but with you talking it through, possibly with pointers when you are talking about the slides. And then I think you should should be in the video of the model arm, pointing out what you are doing. For example, I could tell what you were doing when you increased the muscle tone and pushed on the arm but an observer might not.

I think this demo is really valuable, both as an explanation of the presumed physiological mechanisms that move out limbs but also as a demonstration of the control model of how it might actually work based on what we know of the physiology.

I think it would be particularly cool if you could also point out the areas of controversy. For example, you could point out that, although we know a lot about the physiology, there is controversy about how it actually works. You have produced one model which works by controlling sensed muscle length and tension. It’s a feedback control model. But there are other models of how the physiology works. For example, there’s the equilibrium point model which is ostensibly not a feedback control model. You might give a quick description of how that model is presumed to work and why it is almost certainly wrong (that might wake up the audience;-)

Great work, Adam. Really nice.

Best

Rick

On Mon, Jul 21, 2014 at 5:38 AM, Adam Matic adam.matic@gmail.com wrote:

[From Adam Matic 2014.07.20.14.30]

Here is a short youtube video: https://www.youtube.com/watch?v=TmN99zNRs-4
Apparently, I left a black edge around the video, not sure how that happened. Diagrams may need pausing, but they are visible.

I’ve been playing with the system, added a rate of change component to tension and angle systems, changed gains and slowings. There seem to be more ways the setup can operate. Maybe flexor and extensor need separate length controllers. Anyone know of a detailed neuronal map involving alpha and gamma efferent fibers, all the interneurons, Golgi organs and muscle spindles? I keep getting some partial maps.

Best,

Adam

–
Richard S. Marken PhD
www.mindreadings.com

On Wed, Jul 16, 2014 at 12:40 PM, Adam Matic adam.matic@gmail.com wrote:

[From Adam Matic 2014.07.15.1230]

Rick Marken (2014.07.15.1100

Yes, having the physics take care of itself instead of having to simulate it on a computer makes the game a lot of fun (at least for us non-experts in physics). And using a mechanical device that people can touch and turn some knobs to see what happens seems to work nicely as a demonstration.

I think it also needs a complementary visualization on a computer in the form of plots and animated changes of nervous system frequencies.

Adam

But even with the problems, I think a model like your demonstrates the principles of perceptual control, as it occurs in the “real world” really nicely.

12

[From Rick Marken (2014.07.23.1250)]

···

Adam Matic (2014.07.22 1750)_-

Thanks. Yeah, a video with explanations would be nice, this could be something relatively easy to make at home and get familiar with control, try different gains and slowings and hierarchical relationships.

But first I’d like to make something more similar to real muscles, some sort of artificial muscles that engineers today have a hard time incorporating into robots. Most today’s robots use electric motors as actuators, usually DC servos, and this model uses them too.

Since the term control is usually used as “control of output”, first thing people would object to when they see this model would probably be the fact that real muscles have different physical properties than a thread being rolled on by a DC motor. Rubber tendons could probably go as similar to natural ones, but an actuator would need to be something different, maybe something like this: http://io9.com/scientists-just-created-some-of-the-most-powerful-muscl-1526957560

That might be interesting and useful without many explanations.

RM: That looks great. How do you find these things?

Can’t wait to see the next iteration!

Best

Rick

Best,

Adam

–
Richard S. Marken PhD
www.mindreadings.com

On Tue, Jul 22, 2014 at 1:59 AM, Richard Marken rsmarken@gmail.com wrote:

[From Rick Marken (2014.07.21.1700)]

This is marvelous Adam! The charts and model are very clear. But I would suggest that you take the time to produce this you tube video like a low budget TED talk. That is, I would have you giving a talk ysing the slides and the model but with you talking it through, possibly with pointers when you are talking about the slides. And then I think you should should be in the video of the model arm, pointing out what you are doing. For example, I could tell what you were doing when you increased the muscle tone and pushed on the arm but an observer might not.

I think this demo is really valuable, both as an explanation of the presumed physiological mechanisms that move out limbs but also as a demonstration of the control model of how it might actually work based on what we know of the physiology.

I think it would be particularly cool if you could also point out the areas of controversy. For example, you could point out that, although we know a lot about the physiology, there is controversy about how it actually works. You have produced one model which works by controlling sensed muscle length and tension. It’s a feedback control model. But there are other models of how the physiology works. For example, there’s the equilibrium point model which is ostensibly not a feedback control model. You might give a quick description of how that model is presumed to work and why it is almost certainly wrong (that might wake up the audience;-)

Great work, Adam. Really nice.

Best

Rick

On Mon, Jul 21, 2014 at 5:38 AM, Adam Matic adam.matic@gmail.com wrote:

[From Adam Matic 2014.07.20.14.30]

Here is a short youtube video: https://www.youtube.com/watch?v=TmN99zNRs-4
Apparently, I left a black edge around the video, not sure how that happened. Diagrams may need pausing, but they are visible.

I’ve been playing with the system, added a rate of change component to tension and angle systems, changed gains and slowings. There seem to be more ways the setup can operate. Maybe flexor and extensor need separate length controllers. Anyone know of a detailed neuronal map involving alpha and gamma efferent fibers, all the interneurons, Golgi organs and muscle spindles? I keep getting some partial maps.

Best,

Adam

–
Richard S. Marken PhD
www.mindreadings.com

On Wed, Jul 16, 2014 at 12:40 PM, Adam Matic adam.matic@gmail.com wrote:

[From Adam Matic 2014.07.15.1230]

Rick Marken (2014.07.15.1100

Yes, having the physics take care of itself instead of having to simulate it on a computer makes the game a lot of fun (at least for us non-experts in physics). And using a mechanical device that people can touch and turn some knobs to see what happens seems to work nicely as a demonstration.

I think it also needs a complementary visualization on a computer in the form of plots and animated changes of nervous system frequencies.

Adam

But even with the problems, I think a model like your demonstrates the principles of perceptual control, as it occurs in the “real world” really nicely.