The 1st Propeller Desktop Brain
Horizontal Computational Brain Machine
1st Parallax Propeller Desktop Brain
Computational Machine in horizontal
configuration with display
Currently undergoing heavy wiring, design and testing, this latest version of the Parallax Propeller desktop computational brain machine is being developed and expanded to a total of 121 processors. Not visible is a growing back stack of 48 more processors and the side boards. I'm not sure if it counts as a blob anymore as it has taken on different course.
Humanoido, Your documentation of this is amazing. I always learn something reading this thread.
Whit, thank you! I hope the completed project will have enough information so you can build a brain for your robot. I'm following your posts too - keep up your excellent robotics work!
Brain Blob and BASIC Stamp Supercomputer Software
Similarities are striking
I loaded up the software from the BSS project to take
a look at the LCD code and something shocking
occurred to me. The code was exactly the same as
the Propeller code, allowing for PBASIC vs SPIN
programming. Somewhere in the back of my mind
I had created exactly the same concept on two
different platforms. This is good because the same
tried and true principles can be used to drive the LCD
that I have in mind for displaying the brain thoughts.
The principle is a serial LCD that connects to the bus
on the super net and eavesdrops on the traffic. The
concept is a master/boss and a series of workers/
helpers. The master talks to workers and workers
respond. The LCD hears everything...
In the BSS, the LCD has two lines and no buffer. So a small subroutine
was written to format the LCD and buffer over a couple lines. This is
initially handled by the Master each time talking takes place. It expands
the messages to two lines with 16 characters per line at a time.
In the Propeller Brain, a display with four lines of 20 characters per
line can be used. The #30058 LCD has automatic line wrapping and
scrolling so this display can greatly simply the code. It jumps to 40mA
current draw so it's best used when plugged into a power source
other than batteries, or used sparingly for portable apps.
Another idea is to simply unplug it for field operations.
Displaying Graphical Dreams will be possible as images can display and play back on a TV. A card may be needed for this app. The small TV monitor can also become portable.
Another more basic option is to use software to Vector Dream and draw dreaming as it happens in real time using Propeller LOGO language. For details about a Propeller-based LOGO, follow this link.
The third dreaming option is called Text Dreaming. This is where the dream is composed of words to represent the story line of the dream. This technique is ideal for outputting on the 4-line auto scrolling LCD as previously mentioned.
Are there other types of dreaming? Yes. Internal Dreaming takes place inside the program but does not get outputted, at least not immediately. It happens and either gets saved or does not get saved.
Will the Brain dream in black & white or color? Only time will tell..
Multi-purposed Software
Code will have several purposes. In one example,
I want to recreate the BASIC Stamp Supercomputer
using Propeller chips. In the next arrangement, code
will enhance the machine so it runs closer to super
computing stature. Another code project enables
high speed. Perhaps the best project will be an AI
program that hopefully will lead to a life form. It's
a great expectation and a high goal, I think, but could
definitely lead to greater things.
Case of the Missing Software
I searched all over looking for code samples and examples to be used with the #30058 LCD 4x20 from Parallax. This was on the web at one time but now I find nothing!
The EXO: Exoskeleton Physical Form Factor Design
Creating a new exo design
For the reason of inability to easily access and rewire
the inside row of PPPBs, the Exo is now redesigned.
A middle "inside" row of boards is now deleted and the
row moved to the outside of the form.
In the new design, the form has three sides with six boards
to a side. That makes 18 boards and 20 boards total with
the two side boards.
This creates a rectangular box with the open end
becoming the bottom as it sets on a desk top. It opens
up easy access to wire and rewire all boards. It places
the components on the outside of the boards rectangular
shaped exoskeleton.
You can connect the EXO (Exoskeleton) boards by overlapping
and bolting together at the corner holes. Unfortunately, the leading
PPPB back side edge has some protrusions - mainly the solder legs
from the switch and capacitors stick out too far.
These solder points from the power switch
and filtering capacitor on the power supply
circuit stick out on the back side of all PPPBs
and form short circuits from board to board. The text describes a method to prevent short
circuits.
This has a tendency to short out the two connecting boards. There's
a simple remedy and fix. Cut out strips of insulating dialectric material
(mine came from a sack distributed by the China General Technology
Group) and place it in between the board.
It was purely coincidental that blue Tyvek sack material matched the
color of the PPPB boards.
The fix of directly attaching solderless breadboards with
their original sticky backs instead of rolled tape has solved
the problem of these boards falling off when inverted or
held sideways for extended time periods.
Solderless Breadboard Positioning
The critical position is established
The exact position of the solderless breadboard on the
PPPB is critical. It fits (width-wise) in between the Prop
plug connector pins (next to the power barrel jack) and
the board's edge by the edge hole for bolting.
Positioning too far to the left will block insertion and
connection of the Prop Plug. Too far to the right and
the mounting bolt will be blocked.
I find that it will line up with the left edge of the white
breadboard with/and covering the line of holes to the
right of the printed white ink brackets in the USB2SER
marker. Photos will be added to this post later to show
the positioning.
Solderless Breadboard Inverted Testing
Making sure the breadboards do not disengage
Upside Down Test
To perform this test, the board was held upside
down for several days and nights in the rig. Some
boards held components and wiring and some did
not. The original sticky side held firm on all tested
boards.
Added Weight Test
In the next test following the first test, boards were
gripped with a several pound force. No boards
moved in position.
Inertial Motion Test
Considering the brain is designed for robotics in
motion, a test with inertia was designed. In the start
and stop action of this test, no boards were dislodged
in any amount.
Conclusion
In conclusion, the small white solderless breadboards
are firmly attached and will not fall off when wired in
full and changed in position, such as inverted or
with 90 degree rotations. This facilitates great the hybrid
function of the EXO, i.e. positioning as a desktop unit
or positioning as a tower. It also indicates the EXO is
ideal for motion robotics.
One goal with this Big Brain project is to establish
a criteria of standards. This would allow the open
source distribution of the brain and the creation of
a common platform. In terms of hardware configurations,
this could result in the ease of creating multiple
brains. More details on creating a standard will
follow.
Due to the new EXO design, all boards were disassembled
and are now waiting for reassembly with bolts, nuts, and
angle brackets. Starting all over on the design will create
a much better EXO that can serve multiple functions, such
as placements and form factors in both horizontal and
vertical positions.
Breadboard positioning & attachment
Breadboard tests
Board disassemble
Board configuration design
Removal of spacers, bolts, nuts
Locate dialectric
Cut out strips to insulate boards
Locate source of angle brackets
Test angle brackets
Mockup of form factor
Wire length test
Resistor source
Invention of Brain Wrap
When parts of your brain stick out
What do you do when parts of your brain stick out
and you're moving around in public? Something can
get caught on something or bumped and pulled out
or damaged.
To ensure the safety of all brain components that
mount on the EXO, the technology used in the SEED
Basic Stamp Supercomputer and the BSS will be used.
Sort of like a brain on seran wrap,
SEED Stamp Supercomputer was
wrapped during transport to
avoid a wire being pulled out.
This involved a stretch plastic wrap to cover the
machine and it ensured that no wires pulled out
during transport. Use Seran Wrap or similar,
available at grocery stores.
Wrapped from top down, SEED is
protected during transport.
Clear plastic wrap has many advantaged. It
weighs practically nothing. It's protectively strong.
It costs almost nothing for a section this size.
It's commonly available at a grocery store.
Design with Fewer Parts
New design is more efficient
Spacers are now obsolete in the new design
and no longer needed. The simplicity of board
to board connections on the EXO is accomplished
with bolts, angle irons and nuts.
Optical Illusion One side appears space larger than the other
If you put together a bunch of PPPB boards by spacers
while overlapping one board to the next, an interesting
optical illusion will be created. One side will appear with
spacing between boards smaller that the opposite side.
Measuring the distance between
boards proved the spacing is the
same
Round Robin Rings Schematic
Showing the Brain Blob Basic New Design Connections
Extend P8 to P20 and R2 to R21. Hand sketch
illustrates the first hybrid interface with a round
robin BUS and two wire serial full duplex
connections can be made in an outer ring.
Inner rings handle Vss and Vdd.
This is your Propeller brain on the napkin
sketch series of diagrams. Brain domains
show as correct orientation connectivity in
circular notation
The Brain Spans make up the sides of the EXO. Their assembly is a little more challenging since they assemble at right angles from row to row. There are currently three rows of six boards to be assembled in this manner. The problem occurs when the small bolt on one board obstructs the small bolt on the other board at right angles. I switched from half inch bolts to quarter inch bolts but once again, when I tried to assemble it, the bolt and nut interfered with the fastening.
The discovered the solution when examining the size of the bolt head and comparing it to the size of the nut. The nut is appreciable larger than the bolt's pan head. The solution is to use the smaller quarter inch bolt and place the nut on top of the board away from the fastening joint. Even when mounting the right angled hardware, the half inch bolt is not needed (use the 1/4th-inch bolt throughout).
Right Angle Hardware Selection
Angle iron connects Brain Spans together
It appears that the size of the mounting hardware, the right angle iron in particular, is also critical. I was lucky to have several variations of angle iron on hand. (lighter weight plastic is not usable due to being too flexible) As it turned out, the smallest angle iron is the best. I used a special 5/8ths to a side iron with varying slotted holes for adjustment.
Balance Focal Point How to balance three brain spans
This has become an issue for solving. The weight distribution is best when the row of the heaviest component is counter balanced. This means varying the breadboard postions. However, to vary the breadboards would require repositioning the boards where wires will need to be 50 to 75% longer. This is undesirable as we expect to use overclocking in the future. So examination of balancing may involve flipping the board so breadboards are side by side and 90-degrees to each other with the third row at the farther distance for compensation. It could be a compromise and I will know more after assembly of the two remaining Spans.
Brain Span Analysis In terms of weight and wire length
Maybe we can say there is no need to balance Brain Spans or the distance of connecting wires is relatively unimportant. But I think there is a great use for brains in robots. In a free wheeling robot, we don't want a lopsided brain shifting the weight to one side or the other. It would skew steering and navigation. For humanoid robots, the same applies. Even in humans, our brains are relatively symmetrical with left and right hemispheres of equal weight around a common axis. Our neck does not need to exert undue muscle to balance the head-brain combination. An unbalanced brain would throw the head out of sync and require special muscle to counterbalance the weight, obviously an undesirable situation.
Wiring is also important. It's desirable to have minimal length wiring from board to board, from Brain Span to Brain Span. With minimal wiring, we can overclock and make things run faster, and have more accurate data transfers.
So far, the weight and wiring distribution possibilities using three brain spans and end effectors is something like the examples below. Here we see an example from the end of the rectangle shape, represented by complete balance with no breadboards. |
|
|
|
|
| End view of three rectangular Brain Spans
|
|
|
|
In the example below, we see the introduction of breadboards (represented by the capital letter O) on all three rows of PPPBs. This is a two at right angles and one at distance configuration. If we call the wiring distance between breadboards as X, this arrangement is 2X. In the nomenclature, we will identify the locations of the heavy breadboards and their positions on the PPPBs. We will try enough examples until finding both a minimal wiring condition and a point of greatest balance condition.
For wiring and weight balance, we sum the number of point to point distances between breadboards. A larger X summation is best for balance. The smallest X summation is best for wiring.
O
O
|
| | A) Top board left, left board top, right board bottom, 2X
| |
| |
|
O
O
O
| |
| | Top board left, left board bottom, right board top, 2X
| |
| |
O |
O
| |
| | C) Top board left, left board bottom, right board bottom, 3X
| |
| |
O
O
O
| |
| | D) Top board right, left board bottom, right board bottom, 3X
| |
| |
O
O
O
O
O
| | E) Top board right, left board top, right board top, 1X
| |
| |
| |
O
|
O
| | F) Top board right, left board bottom, right board top, 2X
| |
| |
O |
O
O
O
| | G) Top board left, left board top, right board top, 1X
| |
| |
| |
In conclusion, the following is best for minimal wire length: E and G. The following are best for balance: C & D. The configuration condition must be resolved by assigning a weighted value to weight balance and wire distance. Perhaps an average of the two is best. Then we are looking at all the 2X conditions. This includes A, B, and F. Since we want wiring convenience at the forefront, throw out condition F. This leaves us with a decision of A and B. In looking at this further, we could say weighting more to the front or back is a better counter balance. If this is the case, then G would be acceptable for minimal wire length and a the most weight balance in one direction.
Note: the post has changed some of the spacing in the figure.
Brain Span Configuration
Making the decision to position weight and wiring
After the Brain Span analysis, it is decided to go for shortest possible wire length for greatest computational speed and ability and reliability. The brain must be given the highest considerations. For robotic balance point, the overall weight will be shifted and oriented in the best position and direction.
O
O
O
| | G) Top board left, left board top, right board top, 1X
| |
| |
| |
You can still insert a barrel plug from a transformer
according to the PPPB specs, and it will power
up the 5-volt regulator for sensor operations.
This is key as each prop's 32 pins will have
purpose and ability to drive numerous input
and output devices.
Brain Hybrid Form Factor
Establishing a dual form factor brain
The dual form factor is now established for the Brain Blob.
Taking on the new shape that's more rectangular, it can
now easily function as a vertical high rising tower sky-
scraper or as a horizontal slug fit to the robotic platform
of choice. For humanoid robotics, it's back-packable.
For wheeled robots, the hybrid offers two orientations,
one for smaller robots with limited cargo space (vertical)
and one for larger platformed carriages (horizontal).
The desktop config is for programming convenience
at a desk where it can be removed from the robot to
establish higher brain functions, add-ons, upgrades,
service and other operations.
You knew I had to try this, especially since we were
trying out different versions of EXOskeletons. It's a
pyramid EXO made from multiple Brain Spans. Since
we've established the need for three Brain Spans
with six boards per span, it seemed fitting. The extra
three boards would go under the pyramid when
flipped into vertical.
Just for fun, a tryout of a brain pyramid. The
form factor is interesting but may limit the
addition of more boards.
Comments
Horizontal Computational Brain Machine
1st Parallax Propeller Desktop Brain
Computational Machine in horizontal
configuration with display
Currently undergoing heavy wiring, design and testing, this latest version of the Parallax Propeller desktop computational brain machine is being developed and expanded to a total of 121 processors. Not visible is a growing back stack of 48 more processors and the side boards. I'm not sure if it counts as a blob anymore as it has taken on different course.
Similarities are striking
I loaded up the software from the BSS project to take
a look at the LCD code and something shocking
occurred to me. The code was exactly the same as
the Propeller code, allowing for PBASIC vs SPIN
programming. Somewhere in the back of my mind
I had created exactly the same concept on two
different platforms. This is good because the same
tried and true principles can be used to drive the LCD
that I have in mind for displaying the brain thoughts.
The principle is a serial LCD that connects to the bus
on the super net and eavesdrops on the traffic. The
concept is a master/boss and a series of workers/
helpers. The master talks to workers and workers
respond. The LCD hears everything...
In the BSS, the LCD has two lines and no buffer. So a small subroutine
was written to format the LCD and buffer over a couple lines. This is
initially handled by the Master each time talking takes place. It expands
the messages to two lines with 16 characters per line at a time.
In the Propeller Brain, a display with four lines of 20 characters per
line can be used. The #30058 LCD has automatic line wrapping and
scrolling so this display can greatly simply the code. It jumps to 40mA
current draw so it's best used when plugged into a power source
other than batteries, or used sparingly for portable apps.
Another idea is to simply unplug it for field operations.
Four types of brain dreaming
Displaying Graphical Dreams will be possible as images can display and play back on a TV. A card may be needed for this app. The small TV monitor can also become portable.
Another more basic option is to use software to Vector Dream and draw dreaming as it happens in real time using Propeller LOGO language. For details about a Propeller-based LOGO, follow this link.
The third dreaming option is called Text Dreaming. This is where the dream is composed of words to represent the story line of the dream. This technique is ideal for outputting on the 4-line auto scrolling LCD as previously mentioned.
Are there other types of dreaming? Yes. Internal Dreaming takes place inside the program but does not get outputted, at least not immediately. It happens and either gets saved or does not get saved.
Will the Brain dream in black & white or color? Only time will tell..
Code will have several purposes. In one example,
I want to recreate the BASIC Stamp Supercomputer
using Propeller chips. In the next arrangement, code
will enhance the machine so it runs closer to super
computing stature. Another code project enables
high speed. Perhaps the best project will be an AI
program that hopefully will lead to a life form. It's
a great expectation and a high goal, I think, but could
definitely lead to greater things.
I searched all over looking for code samples and examples to be used with the #30058 LCD 4x20 from Parallax. This was on the web at one time but now I find nothing!
Creating a new exo design
For the reason of inability to easily access and rewire
the inside row of PPPBs, the Exo is now redesigned.
A middle "inside" row of boards is now deleted and the
row moved to the outside of the form.
In the new design, the form has three sides with six boards
to a side. That makes 18 boards and 20 boards total with
the two side boards.
This creates a rectangular box with the open end
becoming the bottom as it sets on a desk top. It opens
up easy access to wire and rewire all boards. It places
the components on the outside of the boards rectangular
shaped exoskeleton.
Avoiding short circuits
You can connect the EXO (Exoskeleton) boards by overlapping
and bolting together at the corner holes. Unfortunately, the leading
PPPB back side edge has some protrusions - mainly the solder legs
from the switch and capacitors stick out too far.
These solder points from the power switch
and filtering capacitor on the power supply
circuit stick out on the back side of all PPPBs
and form short circuits from board to board.
The text describes a method to prevent short
circuits.
This has a tendency to short out the two connecting boards. There's
a simple remedy and fix. Cut out strips of insulating dialectric material
(mine came from a sack distributed by the China General Technology
Group) and place it in between the board.
It was purely coincidental that blue Tyvek sack material matched the
color of the PPPB boards.
Favorable results offer permanent solutions
The fix of directly attaching solderless breadboards with
their original sticky backs instead of rolled tape has solved
the problem of these boards falling off when inverted or
held sideways for extended time periods.
The critical position is established
The exact position of the solderless breadboard on the
PPPB is critical. It fits (width-wise) in between the Prop
plug connector pins (next to the power barrel jack) and
the board's edge by the edge hole for bolting.
Positioning too far to the left will block insertion and
connection of the Prop Plug. Too far to the right and
the mounting bolt will be blocked.
I find that it will line up with the left edge of the white
breadboard with/and covering the line of holes to the
right of the printed white ink brackets in the USB2SER
marker. Photos will be added to this post later to show
the positioning.
These prevent short circuits
As mentioned, dialectric insulating strips are cut
and prevent short circuits from one board to the
next.
Scissor out these strips from a discarded Tyvek
shopping bag and use as insulating strips to
prevent short circuits from board to board.
The design of the strip, for the long side of the
PPPBs is as follows:
Width = 1-inch
Length = 3.25-inch
For the short side of the boards, use 2.5
inch long strips.
Remember to meter the resistance of the strips
to confirm it is indeed insulating rather than
conducting.
Strips are held in place by the forced contact
of the two board surfaces.
Making sure the breadboards do not disengage
Upside Down Test
To perform this test, the board was held upside
down for several days and nights in the rig. Some
boards held components and wiring and some did
not. The original sticky side held firm on all tested
boards.
Added Weight Test
In the next test following the first test, boards were
gripped with a several pound force. No boards
moved in position.
Inertial Motion Test
Considering the brain is designed for robotics in
motion, a test with inertia was designed. In the start
and stop action of this test, no boards were dislodged
in any amount.
Conclusion
In conclusion, the small white solderless breadboards
are firmly attached and will not fall off when wired in
full and changed in position, such as inverted or
with 90 degree rotations. This facilitates great the hybrid
function of the EXO, i.e. positioning as a desktop unit
or positioning as a tower. It also indicates the EXO is
ideal for motion robotics.
One goal with this Big Brain project is to establish
a criteria of standards. This would allow the open
source distribution of the brain and the creation of
a common platform. In terms of hardware configurations,
this could result in the ease of creating multiple
brains. More details on creating a standard will
follow.
New time schedule in effect
Due to the new EXO design, all boards were disassembled
and are now waiting for reassembly with bolts, nuts, and
angle brackets. Starting all over on the design will create
a much better EXO that can serve multiple functions, such
as placements and form factors in both horizontal and
vertical positions.
Breadboard positioning & attachment
Breadboard tests
Board disassemble
Board configuration design
Removal of spacers, bolts, nuts
Locate dialectric
Cut out strips to insulate boards
Locate source of angle brackets
Test angle brackets
Mockup of form factor
Wire length test
Resistor source
When parts of your brain stick out
What do you do when parts of your brain stick out
and you're moving around in public? Something can
get caught on something or bumped and pulled out
or damaged.
To ensure the safety of all brain components that
mount on the EXO, the technology used in the SEED
Basic Stamp Supercomputer and the BSS will be used.
Sort of like a brain on seran wrap,
SEED Stamp Supercomputer was
wrapped during transport to
avoid a wire being pulled out.
This involved a stretch plastic wrap to cover the
machine and it ensured that no wires pulled out
during transport. Use Seran Wrap or similar,
available at grocery stores.
Wrapped from top down, SEED is
protected during transport.
Clear plastic wrap has many advantaged. It
weighs practically nothing. It's protectively strong.
It costs almost nothing for a section this size.
It's commonly available at a grocery store.
New design is more efficient
Spacers are now obsolete in the new design
and no longer needed. The simplicity of board
to board connections on the EXO is accomplished
with bolts, angle irons and nuts.
Spacers are now obsolete
One side appears space larger than the other
If you put together a bunch of PPPB boards by spacers
while overlapping one board to the next, an interesting
optical illusion will be created. One side will appear with
spacing between boards smaller that the opposite side.
Measuring the distance between
boards proved the spacing is the
same
Showing the Brain Blob Basic New Design Connections
Extend P8 to P20 and R2 to R21. Hand sketch
illustrates the first hybrid interface with a round
robin BUS and two wire serial full duplex
connections can be made in an outer ring.
Inner rings handle Vss and Vdd.
This is your Propeller brain on the napkin
sketch series of diagrams. Brain domains
show as correct orientation connectivity in
circular notation
Showing the assembly of one EXO side
The EXO has three board rows assembled in a consistent
overlapping mount held together with bolts and nuts.
This is approximately 1/3rd of an EXO.
Assembly challenge
The Brain Spans make up the sides of the EXO. Their assembly is a little more challenging since they assemble at right angles from row to row. There are currently three rows of six boards to be assembled in this manner. The problem occurs when the small bolt on one board obstructs the small bolt on the other board at right angles. I switched from half inch bolts to quarter inch bolts but once again, when I tried to assemble it, the bolt and nut interfered with the fastening.
The discovered the solution when examining the size of the bolt head and comparing it to the size of the nut. The nut is appreciable larger than the bolt's pan head. The solution is to use the smaller quarter inch bolt and place the nut on top of the board away from the fastening joint. Even when mounting the right angled hardware, the half inch bolt is not needed (use the 1/4th-inch bolt throughout).
Angle iron connects Brain Spans together
It appears that the size of the mounting hardware, the right angle iron in particular, is also critical. I was lucky to have several variations of angle iron on hand. (lighter weight plastic is not usable due to being too flexible) As it turned out, the smallest angle iron is the best. I used a special 5/8ths to a side iron with varying slotted holes for adjustment.
How to balance three brain spans
This has become an issue for solving. The weight distribution is best when the row of the heaviest component is counter balanced. This means varying the breadboard postions. However, to vary the breadboards would require repositioning the boards where wires will need to be 50 to 75% longer. This is undesirable as we expect to use overclocking in the future. So examination of balancing may involve flipping the board so breadboards are side by side and 90-degrees to each other with the third row at the farther distance for compensation. It could be a compromise and I will know more after assembly of the two remaining Spans.
In terms of weight and wire length
Maybe we can say there is no need to balance Brain Spans or the distance of connecting wires is relatively unimportant. But I think there is a great use for brains in robots. In a free wheeling robot, we don't want a lopsided brain shifting the weight to one side or the other. It would skew steering and navigation. For humanoid robots, the same applies. Even in humans, our brains are relatively symmetrical with left and right hemispheres of equal weight around a common axis. Our neck does not need to exert undue muscle to balance the head-brain combination. An unbalanced brain would throw the head out of sync and require special muscle to counterbalance the weight, obviously an undesirable situation.
Wiring is also important. It's desirable to have minimal length wiring from board to board, from Brain Span to Brain Span. With minimal wiring, we can overclock and make things run faster, and have more accurate data transfers.
So far, the weight and wiring distribution possibilities using three brain spans and end effectors is something like the examples below. Here we see an example from the end of the rectangle shape, represented by complete balance with no breadboards.
|
|
|
|
|
| End view of three rectangular Brain Spans
|
|
|
|
In the example below, we see the introduction of breadboards (represented by the capital letter O) on all three rows of PPPBs. This is a two at right angles and one at distance configuration. If we call the wiring distance between breadboards as X, this arrangement is 2X. In the nomenclature, we will identify the locations of the heavy breadboards and their positions on the PPPBs. We will try enough examples until finding both a minimal wiring condition and a point of greatest balance condition.
For wiring and weight balance, we sum the number of point to point distances between breadboards. A larger X summation is best for balance. The smallest X summation is best for wiring.
O
O
|
| | A) Top board left, left board top, right board bottom, 2X
| |
| |
|
O
O
O
| |
| |
| |
| |
O |
O
| |
| | C) Top board left, left board bottom, right board bottom, 3X
| |
| |
O
O
O
| |
| | D) Top board right, left board bottom, right board bottom, 3X
| |
| |
O
O
O
O
O
| | E) Top board right, left board top, right board top, 1X
| |
| |
| |
O
|
O
| | F) Top board right, left board bottom, right board top, 2X
| |
| |
O |
O
O
O
| | G) Top board left, left board top, right board top, 1X
| |
| |
| |
In conclusion, the following is best for minimal wire length: E and G. The following are best for balance: C & D. The configuration condition must be resolved by assigning a weighted value to weight balance and wire distance. Perhaps an average of the two is best. Then we are looking at all the 2X conditions. This includes A, B, and F. Since we want wiring convenience at the forefront, throw out condition F. This leaves us with a decision of A and B. In looking at this further, we could say weighting more to the front or back is a better counter balance. If this is the case, then G would be acceptable for minimal wire length and a the most weight balance in one direction.
Note: the post has changed some of the spacing in the figure.
Making the decision to position weight and wiring
After the Brain Span analysis, it is decided to go for shortest possible wire length for greatest computational speed and ability and reliability. The brain must be given the highest considerations. For robotic balance point, the overall weight will be shifted and oriented in the best position and direction.
O
O
O
| | G) Top board left, left board top, right board top, 1X
| |
| |
| |
You can still insert a barrel plug from a transformer
according to the PPPB specs, and it will power
up the 5-volt regulator for sensor operations.
This is key as each prop's 32 pins will have
purpose and ability to drive numerous input
and output devices.
Establishing a dual form factor brain
The dual form factor is now established for the Brain Blob.
Taking on the new shape that's more rectangular, it can
now easily function as a vertical high rising tower sky-
scraper or as a horizontal slug fit to the robotic platform
of choice. For humanoid robotics, it's back-packable.
For wheeled robots, the hybrid offers two orientations,
one for smaller robots with limited cargo space (vertical)
and one for larger platformed carriages (horizontal).
The desktop config is for programming convenience
at a desk where it can be removed from the robot to
establish higher brain functions, add-ons, upgrades,
service and other operations.
Days of experimenting result in various EXO configurations
Several horizontal variations include rows from three Brain Spans
coupled to extra frontal boards.
Another variation
is this horizontal
stack.
Powerful stacked, loaded, expanded high rise skyscraper versions are towers with various supporting boards attached.
Another high rise vertical EXO version shows more symmetry.
You knew I had to try this, especially since we were
trying out different versions of EXOskeletons. It's a
pyramid EXO made from multiple Brain Spans. Since
we've established the need for three Brain Spans
with six boards per span, it seemed fitting. The extra
three boards would go under the pyramid when
flipped into vertical.
Just for fun, a tryout of a brain pyramid. The
form factor is interesting but may limit the
addition of more boards.