[From Rupert Young (2012.04.14 00.40 BST)]
This is my robot Bill,
Can you tell what he is doing?
Regards,
Rupert
[From Rick Marken (2013.04.14.0845)]
Rupert Young (2012.04.14 00.40 BST)
This is my robot Bill,
http://www.youtube.com/watch?v=Fa6sKT3a00A
Can you tell what he is doing?
RM: That looks great Rupert!
As to what it’s doing. It looks a bit like the drunk searching for keys under the lamp post;-)
My guess is that it is controlling for perceiving a certain level of average light intensity in the two eyes and/or the relative amount of light falling on the two eyes. The fact that Bill doesn’t stay put for long suggests that its goal (reference) for average intensity and/or intensity balance across the eyes is changing (randomly?).
Nice work! Can you give a brief explanation of how it was constructed?
Best
Rick
···
–
Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com
David Goldstein (2013.04.14.1245 EDT)
Rupert Young (2012.04.14 00.40 BST)
Thanks for sharing this. I noticed that robot Bill only turns clockwise.
I am not sure what started him in motion.
I am guessing that he approaches an area of the floor that is brightest.
There are two bright spots on the floor that he seems to go between.
He is acting like he might be in conflict between them.
If we could introduce a disturbance, say a another bright light, we might be able to find support for the above hypothesis.
···
From: Richard Marken rsmarken@GMAIL.COM
To: CSGNET@LISTSERV.ILLINOIS.EDU
Sent: Sunday, April 14, 2013 11:45 AM
Subject: Re: Meet Bill
[From Rick Marken (2013.04.14.0845)]
Rupert Young (2012.04.14 00.40 BST)
This is my robot Bill,
http://www.youtube.com/watch?v=Fa6sKT3a00A
Can you tell what he is doing?
RM: That looks great Rupert!
As to what it’s doing. It looks a bit like the drunk searching for keys under the lamp post;-)
My guess is that it is controlling for perceiving a certain level of average light intensity in the two eyes and/or the relative amount of light falling on the two eyes. The fact that Bill doesn’t stay put for long suggests that its goal (reference) for average intensity and/or intensity balance across the eyes is changing (randomly?).
Nice work! Can you give a brief explanation of how it was constructed?
Best
Rick
Richard S. Marken PhD
rsmarken@gmail.com
http://www.mindreadings.com/
[From Rupert Young (2012.04.14 23.00 BST)]
I should have called it Bill the Bacterium, as it is actually an implementation of the E.coli gradient climbing strategy, with light as the attractor.
It uses a single light sensor and controls for positive change of the light value, by moving forward. When the change is zero or less then a lower control system controls for a random rotation of the robot. As David noticed the rotation is always clockwise.
The thing that looks like eyes is actually an ultrasonic sensor with a transmitter and receiver, and is not used in this case.
The reason that it doesn't stay put at the end is that it wanders out of the peak area, down gradient, and so has to turn around.
RM: Nice work! Can you give a brief explanation of how it was constructed?
Sure, do you mean that hardware or the control systems?
Regards,
Rupert
[From Rick Marken (2013/04.15.1515)]
Rupert Young (2012.04.14 23.00 BST)–
RY: I should have called it Bill the Bacterium, as it is actually an implementation of the E.coli gradient climbing strategy, with light as the attractor.
RM: Ah, I should have guessed! Yes, that’s exactly what it looks like.
RM: Nice work! Can you give a brief explanation of how it was constructed?
RY: Sure, do you mean that hardware or the control systems?
RM: Mainly hardware because I’m pretty familiar with software; hardware is the “hard” part for me.
Best
Rick
Well, the hardware is fairly easy actually, as it is the Lego
Mindstorms ()
robot kit, which comes with motors, wheels and light, touch, sound
and ultrasonic sensors. In my model I just used the light sensor.
The brick looking thing is the NXT computer (included with
Mindstorms), into which you can plug the sensors and motors. Lego
provide a graphical programming interface to control the unit over
bluetooth. However, I prefer more precise and flexible control so
use the lejos ()
Java interface. This allows you to install a JVM on the brick and
either upload programs to the unit or run Java programs from your PC
controlling the unit over bluetooth. I use the latter as the Java
version on the NXT is a cut-down version, and I prefer to use full
Java, though there is some latency involved. I have developed my own
Java library for PCT which allows me to control the robot and also
to graphically view and plot the values of the control systems in
real-time.
In terms of the control systems I’d appreciate it you could cast
your eye over them to verify that they are valid in terms of PCT,
when I have drawn them in a diagram.
Regards,
Rupert
···
[From Rupert Young (2012.04.16 14.15
BST)]
Rick Marken (2013/04.15.1515)]
RY: Sure, do you mean that hardware or the control systems?
RM: Mainly hardware because I'm pretty familiar with software;hardware is the “hard” part for me.
[From Rick Marken (2013.04.16.0810)]
Rupert Young (2012.04.16 14.15BST)–
RY: Sure, do you mean that hardware or the control systems?
RM: Mainly hardware because I'm pretty familiar with software;hardware is the “hard” part for me.
RY: Well, the hardware is fairly easy actually, as it is the LegoMindstorms (http://mindstorms.lego.com )
robot kit, which comes with motors, wheels and light, touch, sound
and ultrasonic sensors.
RM: Ah, yes, I’ll have to get myself a kit;-)
RY:In terms of the control systems I'd appreciate it you could castyour eye over them to verify that they are valid in terms of PCT,
when I have drawn them in a diagram.
RM: Sure, that would be great. I’d like to see them. But based on observed behavior the E. coli algorithm seems to be working just fine.
Best
Rick
[FromRupert Young (2012.04.20 15.45 BST)]
Rick Marken (2013.04.16.0810)]
RM: Sure, that would be great. I'd like to see them. But based on
observed behavior the E. coli algorithm seems to be working just fine.
I've put together an overview in the attached. I'd welcome any feedback,
particularly on the validity of the control systems and the language
used to describe them.
Btw, what would be a suitable reference for the behaviour of E-coli.
Regards,
Rupert
EcoliModelAngleRotation.pdf (210 KB)
[From Rick Marken (2013.04.21.0825)]
Rupert Young (2012.04.20 15.45 BST)
I’ve put together an overview in the attached. I’d welcome any feedback, particularly on the validity of the control systems and the language used to describe them.
Thanks. The algorithm looks to be quite a bit more complex than the one Bill and I use but I;ll look it over in more detail when I get a chance.
Btw, what would be a suitable reference for the behaviour of E-coli.
I of course recommend these two papers:
Marken,
R. S. and Powers, W. T. (1989) Random-Walk Chemotaxis: Trial-And-Error as a
Control Process. *Behavioral Neuroscience,*103, 1348-1355.
Marken,
R. S. (1985) Selection of Consequences: Adaptive Behavior from Random
Reinforcement. Psychological Reports,
56, 379-383.
Chapter 7 of Living Control Systems III also discusses the E. coli navigation algorithm. At the end of that chapter is a reference to a book by Koshland, which is the original source on the navigation of E. coli itself.
Oh, and there is my on-line simulation of the E. coli algorithm at
http://www.mindreadings.com/ControlDemo/Select.html
That demo does the E. coli navigation when you press “Control”.
Best
Rick
···
–
Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com
[From Rupert Young (2013.05.17 0725) BST]
(Rick Marken (2013.04.21.0825)]
Thanks. The algorithm looks to be quite a bit more complex than the one Bill and I use
I've uploaded a video of a much improved, simpler, version.
Rather than use separate systems for forward and rotational motion it simply reverses one wheel for a random period of time. As you can see, compared to the previous version, it is much more smooth and continuous. Though, the initial version is still of interest due to the require conflict control system.
Regards,
Rupert