U.S. patent application number 09/835323 was filed with the patent office on 2002-10-17 for programmable smart membranes and methods therefor.
Invention is credited to Bennett III, Forrest H., Dolin, Bradley E., Rieffel, Eleanor G..
Application Number | 20020152006 09/835323 |
Document ID | / |
Family ID | 25269215 |
Filed Date | 2002-10-17 |
United States Patent
Application |
20020152006 |
Kind Code |
A1 |
Bennett III, Forrest H. ; et
al. |
October 17, 2002 |
PROGRAMMABLE SMART MEMBRANES AND METHODS THEREFOR
Abstract
A programmable smart membrane and methods therefor. The smart
membrane conducts an overall function on at least one of a sorting
function, a filtering function and an absorbing function of at
least one object having an attribute. The smart membrane includes a
plurality of module units disposed adjacent each other. Each of the
plurality of module units obtains information from an environment
around each of the plurality of module units. The plurality of
module units also each perform a function based on at least a first
control method that determines the function based on the
information for each of the plurality of module units. Wherein the
plurality of module units individually perform function to
collectively perform the overall function of the membrane based on
the attribute of the object.
Inventors: |
Bennett III, Forrest H.;
(Palo Alto, CA) ; Rieffel, Eleanor G.; (Mountain
View, CA) ; Dolin, Bradley E.; (Stanford,
CA) |
Correspondence
Address: |
OLIFF & BERRIDGE, PLC
P.O. BOX 19928
ALEXANDRIA
VA
22320
US
|
Family ID: |
25269215 |
Appl. No.: |
09/835323 |
Filed: |
April 17, 2001 |
Current U.S.
Class: |
700/245 |
Current CPC
Class: |
B82Y 10/00 20130101;
Y10S 977/962 20130101; Y10T 403/341 20150115; Y10T 403/342
20150115; G06N 3/002 20130101; Y10T 403/34 20150115; Y10T 403/347
20150115; Y10S 977/70 20130101; Y10T 403/345 20150115 |
Class at
Publication: |
700/245 |
International
Class: |
G06F 019/00 |
Claims
What is claimed is:
1. A membrane for conducting an overall function of at least one of
a sorting function, a filtering function and an absorbing function
of at least one object having an attribute, comprising: a plurality
of module units disposed adjacent each other; each module unit of
the plurality of module units obtains information of an environment
around the module unit; at least a portion of the plurality of
module units each perform at least one function based on at least a
first control method; and the first control method determines the
function based on the information for each of the plurality of
module units, wherein the at least a portion of the plurality of
module units perform the at least one function to collectively
perform the overall function of the membrane based on the attribute
of the object.
2. The membrane, as recited in claim 1, wherein: the plurality of
module units are disposed adjacent each other to form at least two
layers of module units.
3. The membrane, as recited in claim 1, wherein: the plurality of
module units are disposed to form a membrane that provides the
overall function in two dimensions.
4. The membrane, as recited in claim 1, wherein: the plurality of
module units are all identical.
5. The membrane, as recited in claim 1, wherein: the plurality of
module units are robotic modules.
6. The membrane, as recited in claim 5, wherein the robotic modules
are side-sliding robotic modules.
7. The membrane, as recited in claim 1, wherein: each of the
plurality of module units includes at least one sensor that obtains
the information.
8. The membrane, as recited in claim 1, wherein: each of the
plurality of module units includes at least one object sensor that
detects the objects, at least one other module unit sensor that
detects other module units of the plurality of module units that
are adjacent the module unit and at least one wall sensor that
detects walls of a world that the membrane is disposed in all as
the information.
9. The membrane, as recited in claim 1, wherein: each of the
plurality of module units includes a processor that executes the
first control method.
10. The membrane, as recited in claim 1, wherein: each of at least
a portion of the plurality of module units performs the function
based on a second control method that determines the function based
on the information.
11. The membrane, as recited in claim 1, wherein: the first control
method is a program.
12. The membrane, as recited in claim 11, wherein: the program is
generated by genetic programming.
13. The membrane, as recited in claim 11, wherein: the program is
predefined.
14. The membrane, as recited in claim 1, wherein: the movement of
one of the plurality of module units is allowed only when the one
of the plurality of module units is adjacent another module
unit.
15. The membrane, as recited in claim 1, wherein: the movement is a
single movement of one of the plurality of module units.
16. The membrane, as recited in claim 1, wherein: the movement is a
line movement of a line of the plurality of module units.
17. The membrane, as recited in claim 1, wherein: each of the
plurality of module units includes a motor for performing the
movement.
18. The membrane, as recited in claim 1, wherein: the movement is a
sliding movement of one of the plurality of module units, relative
to other adjacent module units.
19. The membrane, as recited in claim 1, wherein: the plurality of
module units collectively perform the overall function of the
membrane based on more than one attribute of the object.
20. The membrane, as recited in claim 1, wherein: the plurality of
module units collectively perform the overall function of the
membrane for more than one object.
21. The membrane, as recited in claim 1, wherein: each of the
plurality of module units obtains the information by at least one
communication link with another module unit or the environment.
22. The membrane, as recited in claim 21, wherein: at least one
communication link is a physical contact with at least one other
module unit of the plurality of module units.
23. The membrane, as recited in claim 21, wherein: the at least one
communication link is an electrical connection with at least one
other module unit of the plurality of module units.
23. The membrane, as recited in claim 21, wherein: the at least one
communication link is a wireless communication link with at least
one other module unit of the plurality of module units.
24. The membrane, as recited in claim 1, wherein: the first control
method is reprogrammable.
25. The membrane, as recited in claim 1, wherein: the information
is information that is local to each of the plurality of module
units.
26. The membrane, as recited in claim 1, wherein: the information
is information that is based on information received by the at
least a portion of the plurality of module units.
27. The membrane, as recited in claim 1, wherein: the function is a
movement.
28. The membrane, as recited in claim 1, wherein: the function is a
change in an internal state of one module unit of the plurality of
module units.
29. The membrane, as recited in claim 1, wherein: the information
is the first attribute.
30. A self-reconfigurable robot, comprising: a first module; a
second module disposed adjacent the first module; the first and
second modules each comprising a function circuit and a sensor; and
at least one controller, wherein the at least one controller
executes a first control program to generate a function instruction
for the first and second modules based on information obtained by
the sensor, and wherein the function circuits perform a function
based on the function instructions and the function of the first
and second modules collectively perform an overall function of at
least one of a filtering function, a sorting function and an
absorbing function for the reconfigurable robot;
31. A self-reconfigurable robot, as recited in claim 30, wherein:
the function circuit operates a movement motor.
32. A self-reconfigurable robot, as recited in claim 30, wherein:
the information is local information for each of the first and
second modules.
33. A self-reconfigurable robot, as recited in claim 30, wherein:
the information is information from both the first and second
modules.
34. A method for providing a membrane for conducting an overall
function of at least one of a sorting function, a filtering
function and an absorbing function of at least one object having an
attribute, comprising: providing a plurality of module units each
capable of obtaining information of an environment around the
module unit; disposing the plurality of module units adjacent each
other; and providing a control method for controlling an operation
function for each of the plurality of module units based on the
information; wherein the plurality of module units individually
perform the operation function to collectively perform the overall
function of the membrane based on the attribute of the object.
35. An article of manufacture, including a computer-readable
memory, comprising: a first plurality of software functions for
performing a respective plurality of functions for a smart
membrane; a first software program for randomly selecting a
plurality of software functions for forming a first plurality of
software programs; a fitness function for obtaining a respective
plurality of fitness function values for the respective first
plurality of software programs; and, a second software program for
conducting a second plurality of software programs from the first
plurality of software programs based on the plurality of respective
fitness function values.
36. A method for providing software for a smart membrane,
comprising: providing a plurality of software functions for
performing a respective plurality of module functions; constructing
a first plurality of software programs for solving an objective
from the plurality of software functions; executing the first
plurality of software programs to obtain a plurality of respective
results; providing an objective fitness function for providing a
fitness value; calculating a plurality of respective fitness
function values for the first plurality of software programs;
selecting a second plurality of software programs from the first
plurality of software programs based on the respective fitness
function values; and, constructing a third plurality of software
programs from the second plurality of programs using a genetic
operation.
37. A method for controlling a membrane that performs an overall
function on an object having an attribute, the membrane having a
plurality of module units that obtain information of an environment
around the plurality of module units, comprising: determining a
function for each of the plurality of module units based on the
information, wherein each the module units perform the function to
collectively perform the overall function of the membrane based on
the attribute.
Description
BACKGROUND OF THE INVENTION
[0001] 1. Field of Invention
[0002] This invention is directed to apparatus and methods relating
to programmable smart membranes and methods therefor.
[0003] 2. Description of Related Art
[0004] Some related art devices include micromachined sorters. For
example, U.S. Pat. No. 5,655,665, discloses a micromachined
magnetic particle manipulator and separator for magnetic particles
suspended in fluid.
[0005] Another micromachined sorter is disclosed in U.S. Pat. No.
5,893,974, wherein a micro scale passive filter is disclosed having
sufficiently small holes to permit the passage of small desired
biomolecules and at the same time prevent the passage of all larger
molecules such as antibodies.
[0006] Another related art micromachined sorter includes the
passive particle sorter disclosed by M. Koch, C. Schabmueller, A.
G. R. Evans, A. Brunnschweiler, "A Micromachined Particle Sorter:
Principle and Technology", Tech. Digest, Eurosensors XII,
Southampton, UK, Sep. 13-16, 1998.
[0007] However, these related art micromachined sorters can only
sort particles by predefined features or are static/passive and
thus are non-programmable.
[0008] Other related art devices include nanometer scale membranes.
For example, Yoshihiro Ito, "Signal-responsive Gating by a
Polyelectrolyte Pelage on a Nanoporous Membrane", Nanotechnology,
pp 205-207, Vol. 9 No. 3, September 1998, discloses a nanometer
scale active membrane wherein the rate of water permeation through
the membrane is controlled by pH levels and ionic strength.
[0009] Another nanometer scale membrane is disclosed in K. Eric
Drexler, "Nanosystem: Molecular Machinery, Manufacturing, and
Computation, John Wiley & Sons, Inc., 1992. Drexler discloses a
design for a unidirectional active molecule sorter capable of
transporting specific molecules across a barrier. Drexler mentions
the possibility of a version of this design which can change the
specificity of the transported molecules by using an elastic
deformation of the receptor or altering the local charge
distribution around the receptor.
[0010] Another related art nanometer scale membrane with poricity
controlled by electroactive polymer actuators is described in
Toribio F. Otero, Univ. del Pais Vasco, and San Sebastian, "EAP as
Multifunctional and Biomimetic Materials", in Smart Structures and
Materials 1999: Electroactive Polymer Actuators and Devices, edited
by Yoseph Bar-Cohen, pp.26-34, 1999.
[0011] However, all of the related art nanometer scale membranes
have filtering functions that are preset and which cannot be
subsequently altered.
[0012] Related art control methods and techniques include the
system disclosed in U.S. Pat. No. 6,011,372, which is a unified
modular system of drives and controls for mobile robotic
applications. However, the robot of this modular system is not
modular and the control of the robot is hierarchical.
SUMMARY OF THE INVENTION
[0013] The invention provides a programmable smart membrane and
methods therefor that overcome the deficiencies and shortfalls of
the prior art.
[0014] The invention provides a programmable smart membrane that is
capable of being programmed so as to reject, transport, or absorb
objects depending on features of the objects.
[0015] The invention separately provides the control software for a
programmable smart membrane.
[0016] The invention separately provides a programmable smart
membrane that separates two or more regions.
[0017] The invention separately provides a programmable smart
membrane that separates a region and provides one-way filtering
into or out of that region.
[0018] The invention separately provides a programmable smart
membrane that separates a region and provides two-way filtering
into and out of that region by transporting objects in both
directions through the membrane.
[0019] The invention separately provides a programmable smart
membrane that is planar.
[0020] The invention separately provides a programmable smart
membrane that separates three-dimensional regions.
[0021] The invention separately provides a programmable smart
membrane that works with a single object at a time.
[0022] The invention separately provides a programmable smart
membrane that works with multiple objects simultaneously.
[0023] The invention separately provides a programmable smart
membrane that is programmable and reprogrammable to reject,
transport, or absorb objects based on a diverse set of properties
of the objects.
[0024] The invention separately provides a programmable smart
membrane that has diverse applications.
[0025] The invention separately provides a programmable smart
membrane that sorts parts on a macro scale level.
[0026] The invention separately provides a programmable smart
membrane that purifies substances and augments biochemical
processes on a nanometer scale level.
[0027] The invention separately provides control software for a
programmable smart membrane that includes both identical and
non-identical module units.
[0028] The invention separately provides a control software for a
programmable smart membrane that is generated by evolutionary
techniques.
[0029] The invention separately provides decentralized control
software for a programmable smart membrane.
[0030] The invention separately provides control software for a
programmable smart membrane that scales with the number of module
units in the membrane and the number of objects to be filtered,
both in sequence and simultaneously.
[0031] The invention separately provides control software for a
programmable smart membrane which is functional under various
configurations of its modules and the objects to be filtered.
[0032] The invention separately provides a programmable smart
membrane that includes a plurality of sliding robotic modules.
[0033] The invention separately provides a programmable smart
membrane that includes a plurality of rotating robotic modules.
[0034] The invention separately provides a programmable smart
membrane that includes a plurality of compressible robotic
modules.
[0035] The invention separately provides a programmable smart
membrane that uses control software found using a fitness
function.
[0036] The invention separately provides a programmable smart
membrane that can sort, filter, and/or absorb objects based on at
least one of the following properties of the objects: color; size;
shape; temperature; density; weight; stiffness; elasticity; charge;
force; reflectivity; magnetic flux; conductivity; frictionality;
compressibility; movement of the objects being filtered;
communication with the object; the way any of the above attributes
change over time; and the way any of the above attributes vary
across the object.
[0037] The invention separately provides a programmable smart
membrane that continues to perform a desired task in the presence
of various failures or noise in or inaccuracy of sensors, motors,
power, and changing or unknown environmental conditions.
[0038] The invention separately provides a programmable smart
membrane that continues to perform a desired task in spite of the
failure of a motor in one or several of the module units.
[0039] The invention separately provides a programmable smart
membrane that continues to perform a desired task in spite of the
failure of a sensor in one or several of the module units.
[0040] The invention separately provides a programmable smart
membrane that continues to perform a desired task in spite of power
loss in one or several of the module units.
[0041] The invention separately provides a programmable smart
membrane that continues to perform a desired task in spite of
inaccurate sensor readings by sensors in one or several of the
module units.
[0042] The invention separately provides a programmable smart
membrane, including 2D sliding robots, that sorts objects based on
color.
[0043] These and other features and advantages of this invention
are described in, or are apparent from, the following detailed
description of various exemplary embodiments of the systems and
methods according to this invention.
BRIEF DESCRIPTION OF THE DRAWINGS
[0044] Various exemplary embodiments of this invention will be
described in detail, with reference to the following figures,
wherein:
[0045] FIGS. 1A-1C are schematics of an exemplary embodiment of a
smart membrane according to the invention that filters objects;
[0046] FIGS. 2A-2C are schematics of another exemplary embodiment
of a smart membrane according to the invention that filters objects
bi-directionally;
[0047] FIGS. 3A-3C are schematics of another exemplary embodiment
of a smart membrane according to the invention that sorts
objects;
[0048] FIGS. 4A-4C are schematics of another exemplary embodiment
of a smart membrane according to the invention that absorbs
objects;
[0049] FIG. 5 is a block diagram of one exemplary embodiment of a
module unit of an exemplary smart membrane according to the
invention, wherein a module control circuit is disposed within the
module unit;
[0050] FIG. 6 is a block diagram of another exemplary embodiment of
a module unit of an exemplary smart membrane according to the
invention, wherein a module control circuit is not disposed within
the module unit;
[0051] FIG. 7 is a schematic of an exemplary embodiment of a
modular robotic smart membrane according to the invention;
[0052] FIG. 8 is a flowchart of one exemplary embodiment for
constructing a software program for a module unit of a smart
membrane, according to the invention;
[0053] FIG. 9 is a schematic of another exemplary embodiment of a
modular robotic smart membrane according to the invention;
[0054] FIGS. 10A-10C are schematics displaying a single movement of
exemplary robotic modules;
[0055] FIGS. 11A-11B are schematics displaying a line movement of
exemplary robotic modules;
[0056] FIGS. 12A-12C are schematics displaying the smart membrane
of FIG. 9 in an initial stage, an intermediary stage and a
subsequent stage of filtering an object; and
[0057] FIG. 13 is a schematic of another exemplary embodiment of
modular robotic smart membrane according to the invention.
DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS
[0058] The invention is a programmable membrane and methods
therefor that overcome the deficiencies and short falls of the
related art.
[0059] The invention includes a smart membrane that can be
programmed to reject, transport, and/or absorb items depending on
features or attributes of the items. The smart membrane of the
invention includes a plurality of module units. The overall smart
membrane is self-reconfigurable in that it can reconfigure without
any external aid. Each module unit performs functions based on
information about its environment. Each module unit communicates
with and moves in relation to its neighboring module units in order
to accomplish an overall task for the entire membrane, such as
filtering. The function performed by the module units are based on
a control program/method.
[0060] FIGS. 1A-1C display an exemplary embodiment of a smart
membrane 100 according to the invention. The smart membrane 100
includes a plurality of module units 110. The smart membrane 100 in
this exemplary embodiment is functioning as a one-way filter. It
should be appreciated that in other various exemplary embodiments,
a smart membrane according to the invention can also function to
filter multi-directionally, and sort and capture/absorb objects, as
shown in FIGS. 2A-2C, 3A-3C and 4A-4C, respectively.
[0061] As shown in FIGS. 1A-1C, the smart membrane 100 filters
objects to allow objects with a first attribute 120 to pass the
membrane 100 and to stop objects with a second attribute 130 from
passing the membrane 100. Objects with the first attribute 120 are
displayed as triangles and objects with the second attribute 130
are displayed as circles for discussion purposes only.
[0062] FIG. 1A displays an initial stage of the filtering, FIG. 1B
displays an intermediate stage of the filtering and FIG. 1C
displays a final stage of the filtering. As shown in FIG. 1B,
temporary transport channels 140 are created by actions of the
module units 110. The temporary channels 140 allow the objects 120
and 130 to pass through the membrane 100.
[0063] In this exemplary embodiment, the module units 110 are
identical. However, it should be appreciated that in other various
exemplary embodiments, the module units 110 are not all
identical.
[0064] In this exemplary embodiment, the module units 110 operate
based on identical sets of instructions or methods. However, it
should be understood that in other various exemplary embodiments,
less than all of the module units 110 operate based on the
identical method.
[0065] Each of the module units 110 obtain information about the
environment around the module unit. In various exemplary
embodiments, the information is local information. The local
information is information about the immediate environment around a
module unit. In other various exemplary embodiments, the
information is global information from all of the module units,
i.e. the membrane's environment.
[0066] The method of obtaining the information can vary based on
each application of the membrane. For example, in various exemplary
embodiments, the information is obtained by a physical touching
between the module and the environment, which may include
neighboring module units. In other various exemplary embodiments,
the information is obtained by an electrical connection between a
module unit and the environment, which may include neighboring
module units.
[0067] In various exemplary embodiments, the information is
obtained by sensors. The sensors may be any known or later
developed sensors or sensing devices and/or methods. In various
exemplary embodiments, the sensors are disposed on and/or in the
module units. In various exemplary embodiments, the sensors are for
detecting the objects 120 and 130 and/or the bounds of the
environment for the membrane 100. In various exemplary embodiments,
the sensors are infrared sensors.
[0068] In other various exemplary embodiments, the information is
obtained by a module unit communicating with the environment. For
example, in various exemplary embodiments, the module units
communicate with one another. In various exemplary embodiments, the
module units only communicate with immediately adjacent module
units. In other exemplary embodiments the module units communicate
with the objects 120 and 130 and/or the bounds of the environment.
Any known or later developed methods and/or devices for
communication may be used. In various exemplary embodiments, the
communication is conducted by wireless communication.
[0069] It should be appreciated that in various exemplary
embodiments, the module units may include only sensors,
communication devices or both. It should also be appreciated that
the sensors and the communication devices may be combined into a
single device. Further, it should be appreciated that in various
exemplary embodiments, the information obtained via devices for
sensing and communicating may include such properties of the
objects as, but not limited to: color; size; shape; temperature;
density; weight; stiffness; elasticity; charge; force;
reflectivity; magnetic flux; conductivity; frictionality;
compressibility; movement; communication with the objects; the way
any of the above attributes change over time; and the way any of
the above attributes vary across the object.
[0070] The information obtained by the module units is processed
based on the control method that the module units 110 are to follow
for their respective functions.
[0071] The processing of the information results in function
instructions for the respective module unit 110. The function
instructions are instructions for the module unit 110 to perform a
particular function or functions. The function instructions also
include a non-function instruction, which instructs a module unit
110 not to perform any function. In this exemplary embodiment, the
function instructions are movement, communication, and internal
state change instructions. However, it should be appreciated that
the function instructions could include a variety of
actions/functions. For example, in various exemplary embodiments,
the functions include, but are not limited to: to communicate with
the environment, to connect with another module unit, to not move,
and/or to change an internal state of the module unit.
[0072] In other various exemplary embodiments, the function
instructions are effector instructions for the respective module
unit 110. For example, a module unit having a faceplate for
latching may attach to or detach from a neighboring module unit,
for example by latching or unlatching its faceplate to the
faceplate of an adjacent module unit.
[0073] In various exemplary embodiments, the function instructions
are communication instructions for the respective module unit 110.
For example, module units may send and receive messages to other
module units. These messages can include the sharing of any type of
information, requests for information, etc.
[0074] In other various exemplary embodiments, the function
instructions are instructions for changing the internal state of a
module unit, including but not limited to changing the contents of
an internal memory for the respective module unit 110, such as,
storing the results of calculations, sensor readings, or planned
actions for future reference.
[0075] In various exemplary embodiments, wherein the functions are
movements of the module units, the movement of each module unit is
limited to a set of predefined limitations. For example, the module
units may have physical structures that limit the movement of the
module units 110. In other exemplary embodiments, the movement
limitations may be set in an attempt to keep the module units
together as a group. An example of a specific movement limitation
is that a module unit can only move when it is adjacent to at least
one other module unit.
[0076] With each of the module units 110 acting independently,
based only on their respective environmental information, the
membrane as a whole behaves as desired.
[0077] The method that the individual module units 110 are
following will determine how the module units 110 behave
collectively. The control method in various exemplary embodiments
is a computer program. In various exemplary embodiments, the
computer program is developed using genetic programming techniques,
such as but not limited to, the automatic control program
generation techniques disclosed in copending U.S. patent
application Ser. No. 09/611,395, filed Jul. 7, 2000, which is
incorporated herein by reference in its entirety.
[0078] It should also be appreciated that the first and second
attributes of objects 120 and 130 may be any type of attribute that
can be acquired or learned by the module units 110 through their
information obtaining methods and devices. It should also be
appreciated that each object may have multiple attributes and that
an attribute can change over time and across the actual object. In
various exemplary embodiments, the attributes are, but are not
limited to, any of the following: color; size; shape; temperature;
density; weight; stiffness; elasticity; charge; force;
reflectivity; magnetic flux; conductivity; frictionality;
compressibility; movement of the objects; communication with the
object; the way any of the above attributes change over time; and
the way any of the above attributes vary across the object.
[0079] In various exemplary embodiments, the membrane 100 includes
a mechanism, not shown, for activating and deactivating the
membrane 100. The activation mechanism is an "on/off" switch for
the membrane 100. In various exemplary embodiments, the mechanism
is activated through known or later developed wireless
communication methods and devices. In other various exemplary
embodiments, the activation mechanism is disposed on the membrane
100 or the module units 110 so that a user can manually activate
the mechanism. In other various exemplary embodiments, the
activation mechanism includes a plurality of devices that activate
individual module units 110.
[0080] FIGS. 2A-2C display an exemplary embodiment of a smart
membrane 200 according to the invention. The smart membrane 200
includes a plurality of module units 210. The smart membrane 200 in
this exemplary embodiment is functioning as a bi-directional filter
for objects 220 and 230. It should be appreciated that the membrane
200 is identical to the membrane 100 except that the membrane 200
is functioning as a bi-directional filter. In other words, the
objects 220 and 230 approach the membrane 200 from two directions
and the membrane 200 filters the objects 220 and 230 based on
attributes of the objects, such that the objects 220 are filtered
to be below the membrane 200 and the objects 230 are filtered to be
above the membrane. It should also be appreciated that the objects
220 and 230 are identical to the objects 120 and 130. FIG. 2A
displays an initial stage of the filtering, FIG. 2B displays an
intermediate stage of the filtering and FIG. 2C displays a final
stage of the filtering. As shown in FIG. 2B, temporary transport
channels 240 are created by the functions/actions of the module
units 210. The temporary channels 240 allow the objects 220 and 230
to pass through the membrane 200.
[0081] FIGS. 3A-3C display an exemplary embodiment of a smart
membrane 300 according to the invention. The smart membrane 300
includes a plurality of module units 310. The smart membrane 300 in
this exemplary embodiment is functioning as a sorter for objects
320 and 330. It should be appreciated that the membrane 300 is
identical to the membrane 100 except that the membrane 300 is
functioning as a sorter. In this exemplary embodiment, the objects
320 and 330 approach the membrane 300 from one direction (the top
of the membrane 300) and the membrane 300 sorts the objects 320 and
330 based on attributes of the objects, such that the objects 320
are sorted to be one first side of a region below the membrane 300
and the objects 330 are sorted to be a second side of the region
below the membrane. It should also be appreciated that the objects
320 and 330 are identical to the objects 120 and 130. FIG. 3A
displays an initial stage of the sorting, FIG. 3B displays an
intermediate stage of the sorting and FIG. 3C displays a final
stage of the sorting. As shown in FIG. 3B, temporary transport
channels 340 are created by the functions/actions of the module
units 310. The temporary channels 340 allow the objects 320 and 330
to pass through the membrane 300.
[0082] FIGS. 4A-4C display an exemplary embodiment of a smart
membrane 400 according to the invention. The smart membrane 400
includes a plurality of module units 410. The smart membrane 400 in
this exemplary embodiment is functioning as an absorber for
absorbing either objects 420 or 430 based on one or more attributes
of the objects. It should be appreciated that the membrane 400 is
identical to the membrane 100 except that the membrane 400 is
functioning as an absorber. In this exemplary embodiment, the
objects 420 and 430 approach the membrane 400 from above the
membrane 400. The membrane 400 absorbs only the objects having a
predetermined attribute or attributes. In this embodiment, the
objects 420 have the appropriate predetermined attribute, whereas
the objects 430 do not. The membrane 400 absorbs the objects 420
and keeps the objects 430 out of the membrane based on the
attributes of the respective objects. It should also be appreciated
that the objects 420 and 430 are identical to the objects 120 and
130. FIG. 4A displays an initial stage of the absorbing, FIG. 4B
displays an intermediate stage of the absorbing and FIG. 4C
displays a final stage of the absorbing. As shown in FIG. 4B,
temporary transport channels 440 are created by the
functions/actions of the module units 410. The temporary channels
440 allow the objects 420 and 430 to pass into the membrane
400.
[0083] FIG. 5 is a block diagram of an exemplary embodiment of a
module unit 500 for an exemplary smart membrane according to the
invention. The module unit 500 includes a module control circuit,
routine or manager 510, a memory 520, a sensor 530, a communication
circuit 540 and a function circuit, routine or manager 550, each
connected to a signal/data bus 512.
[0084] The sensor 530 includes at least one sensor for obtaining
information about the environment around the module unit 500, as
discussed above. The information obtained by the sensor 530 is
provided to the module control circuit, routine or manager 510. In
various other exemplary embodiments, the sensed information is
stored in the memory 520, such that the sensed information can be
subsequently recalled for evaluation or the like.
[0085] The communication circuit, routine or manager 540, controls
the communication of the module unit 500 to obtain information
about the environment around the module unit 500, as discussed
above. The communication circuit, routine or manager 540, in
various exemplary embodiments, includes obtaining a new method for
the module control circuit, routine or manager 510 to use for
controlling the module unit. The information obtained by the
communication circuit, routine or manager 540, is provided to the
module control circuit, routine or manager 510. In various other
exemplary embodiments, the communication information is stored in
the memory 520, such that the communication information can be
subsequently recalled for evaluation or the like.
[0086] It should be appreciated that the communication circuit 540
and the sensor 530 can be elements integrated into a single
device.
[0087] The module control circuit, routine or manager 510 includes
the method for controlling the functions of the module unit 500.
The module control circuit, routine or manager 510 executes the
control method and generates function instructions based on the
information obtained by the sensor 530 and the communication
circuit, routine or manger 540. In this exemplary embodiment, the
information obtained by the sensor 530 and the communication
circuit, routine or manager 540, is information that is local to
the module unit 500. The function instructions are provided to the
function circuit, routine or manager 550. In various other
exemplary embodiments, the function instructions are stored in the
memory 520, such that the function instructions can be subsequently
recalled for evaluation or the like.
[0088] The function circuit, routine or manager 550, performs
functions based on the function instructions from the module
control circuit, routine or manager 510. In various exemplary
embodiments, the function circuit, routine or manager 550 controls
movement motors disposed on and/or in the module unit 500 based on
the function instructions.
[0089] As shown in FIG. 5, the memory 520 can be implemented using
any appropriate combination of alterable, volatile, or non-volatile
memory or non-alterable, or fixed memory. The alterable memory,
whether volatile, or non-volatile, can be implemented using any one
or more of static or dynamic RAM, a floppy disk and disk drive, a
writable or rewritable optical disk and disk drive, a hard drive,
flash memory or the like. Similarly, the non-alterable or fixed
memory can be implemented using any one or more of ROM, PROM,
EPROM, EEPROM, an optical ROM disk, such as a CD-ROM or DVD-ROM
disk, and disk drive or the like.
[0090] The memory 520 can store any information obtained by the
sensor 530 or the communication circuit, routine or manager 540,
any information generated by the function circuit, routine or
manager 550 or the function instruction from the module control
circuit routine or manager 510.
[0091] Further, it should be appreciated that the data bus 512
connecting the memory 520 to the module control unit 510 can be a
wired or wireless link to a network (not shown). The network can be
a local area network, and intranet, or any other distributed
processing and storage network.
[0092] It should also be understood that each of the circuits,
routines or managers shown in FIG. 5 can be implemented as portions
of a suitably programmed general purpose computer. Alternatively,
each of the circuits, routines or managers shown in FIG. 5 can be
implemented as physically distinct hardware circuits, routines or
managers within an ASIC, or using a FPGA, a PDL, a PLA or a PAL, or
using discrete logic elements or discrete circuit, routine or
manager elements. The particular form of each of the circuits,
routines or managers shown in FIG. 5 is a design choice.
[0093] FIG. 6 is a block diagram of an exemplary embodiment of a
module unit 600 for an exemplary smart membrane according to the
invention. The module unit 600 includes an I/O interface 620, a
modular unit controller 630, a memory 640, a sensor 650, a
communication circuit 660 and a function circuit, routine or
manager 670, each connected to a signal/data bus 612. The I/O
interface 620 is connected to a module controller circuit, routine
or manager 610 by a link 612.
[0094] The link 612 can be any known or later developed device or
system that can connect the module controller circuit, routine or
manager 610 to the module unit 600, including a direct cable
connection, a connection over a wide area network or a local area
network, a connection over an intranet, a connection over the
Internet, or a connection over any other distributed processing
network or system. In general, the link 612 can be any known or
later developed connection system or structure.
[0095] The sensor 650 includes at least one sensor for obtaining
information about the environment around the module unit 600, as
discussed above. The information obtained by the sensor 650 is
provided to the module controller circuit, routine or manager 610
via the I/O interface 620, under the control of the modular unit
controller 630. In various other exemplary embodiments, the sensed
information is stored in the memory 640, under the control of the
modular unit controller 630 such that the sensed information can be
subsequently recalled for evaluation or the like.
[0096] The communication circuit, routine or manager 660, controls
the communication of the module unit 600 to obtain information
about the environment around the module unit 600, as discussed
above. The information obtained by the communication circuit,
routine or manager 660 is provided to the module controller
circuit, routine or manager via the I/O interface 620 under the
control of the modular unit controller 630. In various other
exemplary embodiments, the communication information is stored in
the memory 640 under the control of the controller, such that the
communication information can be subsequently recalled for
evaluation or the like.
[0097] It should be appreciated that the communication circuit,
routine or manager 660 and the sensor 650 can be elements
integrated into a single device.
[0098] The module controller circuit, routine or manager 610
includes the method for controlling the functions of the module
unit 600. The module controller circuit, routine or manager 610 is
disposed outside the module unit 600. In various exemplary
embodiments, the module controller circuit, routine or manager 610
is remotely located from the module unit 600. In various exemplary
embodiments, the module controller circuit, routine or manager 610
executes the control method and generates function instructions
based on the information obtained by the sensor 650 and the
communication circuit, routine or manger 660 and thus provides a
decentralized control of the module unit 600. In other various
exemplary embodiments, the module controller circuit, routine or
manager 610 executes the control method and generates function
instructions based on the information obtained by the sensor 650
and the communication circuit, routine or manger 660 as well as
information obtained by other module units, and thus provides a
centralized control of the module unit 600. The function
instructions are provided to the function circuit, routine or
manager 670 via the I/O interface 620. In various other exemplary
embodiments, the function instructions are stored in the memory 640
via the I/O interface, such that the function instructions can be
subsequently recalled for evaluation or the like.
[0099] The function circuit, routine or manager 670, performs
functions based on the function instructions from the module
controller circuit, routine or manager 610. In various exemplary
embodiments, the function circuit, routine or manager 670 controls
movement motors disposed on and/or in the module unit 600 based on
the function instructions.
[0100] As shown in FIG. 6, the memory 640 can be implemented using
any appropriate combination of alterable, volatile, or non-volatile
memory or non-alterable, or fixed memory. The alterable memory,
whether volatile, or non-volatile, can be implemented using any one
or more of static or dynamic RAM, a floppy disk and disk drive, a
writable or rewritable optical disk and disk drive, a hard drive,
flash memory or the like. Similarly, the non-alterable or fixed
memory can be implemented using any one or more of ROM, PROM,
EPROM, EEPROM, an optical ROM disk, such as a CD-ROM or DVD-ROM
disk, and disk drive or the like.
[0101] The memory 640 can store any information obtained by the
sensor 650 or the communication circuit, routine or manager 660,
information generated by the function circuit, routine or manager
670 or the function instructions from module controller circuit,
routine or manager 610.
[0102] Further, it should be appreciated that the data bus 602
connecting the memory 640 to the modular unit controller 630 can be
a wired or wireless link to a network (not shown). The network can
be a local area network, and intranet, or any other distributed
processing and storage network.
[0103] It should also be understood that each of the circuits,
routines or managers shown in FIG. 6, can be implemented as
portions of a suitably programmed general purpose computer.
Alternatively, each of the circuits, routines or managers shown in
FIG. 6 can be implemented as physically distinct hardware circuits,
routines or managers within an ASIC, or using a FPGA, a PDL, a PLA
or a PAL, or using discrete logic elements or discrete circuit,
routine or manager elements. The particular form of each of the
circuits, routines or managers shown in FIG. 6 is a design
choice.
[0104] The following description is for an exemplary embodiment of
a smart membrane of the invention. It should be appreciated that
the smart membrane of the invention may take on various embodiments
and should not be limited to the following exemplary embodiments.
Various exemplary embodiments of a smart membrane of the invention
include making the membrane a modular robot.
[0105] A modular robotic smart membrane, according to the
invention, is a self-reconfigurable modular robotic device capable
of active filtering, sorting, absorbing, or the like, of objects
based on a diverse set of properties or attributes, such as but not
limited to color, size, or shape, or any other attribute that may
be sensed or measured. A modular robotic smart membrane, of the
invention, includes a plurality of robot modules or modular
hardware.
[0106] Modular hardware that can reconfigure itself is usually
referred to as a self-reconfigurable modular robot. A
self-reconfigurable modular robot is a robot that is composed of
many modules, where each module has its own power, motors,
computation, memory, and sensors. The modules have motors and
catches that enable them to attach to and detach from each other in
order to move and perform tasks or functions. Individual modules on
their own can do little, but the robot, using just the capabilities
of the individual modules, can reconfigure itself to perform
different tasks as needed. In various exemplary embodiments, the
robot modules are identical components.
[0107] In various exemplary embodiments, it is preferred that
modular robot control software, which controls the modular robot,
be completely decentralized and completely autonomous, so that the
tasks can be performed without reference to a central controller,
whether human or machine. In various exemplary embodiments, the
modular robot control software, according to the invention, is
created using genetic programming. The incorporated U.S. patent
application Ser. No. 09/611,395, filed Jul. 7, 2000, discloses
exemplary methods of using genetic programming for creating the
control software. However, it should be appreciated that the
control software can be created by a number of different techniques
and it may also be predetermined or provided.
[0108] A smart membrane can be implemented using any of the
self-reconfigurable modular robots, including the types of
self-reconfigurable modular robots listed below.
[0109] One type of self-reconfigurable modular robot is the
compressible modular robots, such as the Crystalline Robots
developed at Dartmouth, which consist of cubical modules that can
slide past each other. Movement of these robots is possible only
through motors that contact the cubes by a factor of two in one or
another direction. Both 2D and 3D versions are possible.
[0110] Another type of self-reconfigurable modular robot includes
rotating modular robots, like Xerox PARC's Proteo robot, in which
the modules rotate with respect to each other through hinges at the
vertices (2D) or edges (3D) of the modules. The individual modules
can be different shapes, like squares or hexagons in 2D, or cubes
or rhombic dodecahedrons, as were used in Proteo, in 3D.
[0111] Another type of self-reconfigurable modular robot includes
prismatic joint modular robots, like Xerox PARC's Polybot, whose
modules are cubes which can contract into wedges, as well as attach
and detach from each other. The configurations of the modules
consist of long strings, occasionally attached to others, forming
snake-, spider-, and tank-tread-shaped robots for instance.
[0112] One type of self-reconfigurable modular robot includes
rotating joint modular robots, like USC's ISI's Spider Link robots,
which can rotate with respect to each other. Again, the
configurations of the modules consist of long strings, occasionally
attached to others.
[0113] Another type of self-reconfigurable modular robot includes
bi-unit modular robots, like CMU's I-cubes and Dartmouth's Molecule
Robots, which consist of two types of modules, cubes and links,
which can attach to, detach from, and rotate with respect to each
other.
[0114] FIG. 7 displays a schematic illustration of an exemplary
embodiment of a smart membrane 700, which is a modular robot. The
membrane 700 includes a plurality of robot modules 710. In this
exemplary embodiment, the smart membrane 700 sorts foreign objects
720 and 730 by shape. The objects 720 are square and the objects
730 are circular.
[0115] The membrane 700 is disposed in a world 740 that includes a
plurality of walls 742. The overall objective or function of the
membrane 700 is to allow the square objects 720 to pass through the
membrane 700 and to prevent the circular objects 730 from passing
through the membrane 700. Both the square and the circular foreign
objects 720 and 730 come into contact with the membrane 700 from
above the membrane, on a first side 702 of the membrane 700. The
membrane 700 configures itself so as to allow square objects 720 to
pass through to the bottom of the world 740, and so as to prevent
round objects 730 from passing through a second side 704 of the
membrane 700.
[0116] The filtering capabilities of the smart membrane 700 emerge
from the simple motions of its constituent robot modules 710. In
various exemplary embodiments the modules 710 are controlled by a
central processing unit, not shown, that determines the move that
each module 710 needs to make in order to achieve the desired
global behavior of the membrane 700. This information is then
propagated to the relevant modules. The central processing unit may
be any known or later developed processing device.
[0117] In other various exemplary embodiments, the control of the
membrane 700 is based on a distributed control paradigm. With
distributed control, each module 710 has its own processor and uses
the information provided by its sensors and interactions with other
modules to determine the next move it will make.
[0118] In various exemplary embodiments, each module 710 is running
the same control program, and the correct global filtering behavior
of the membrane emerges from the behavior of the individual modules
710 determined by this program, their sensor readings, and
communication with other modules 710.
[0119] All the modules 710 run their control programs, but behave
differently depending on individual sensor values, internal state,
and messages received from nearby modules.
[0120] The control program can be generated from genetic
programming methods, including but not limited to the methods
disclosed in the incorporated U.S. patent application Ser. No.
09/611,395, filed Jul. 7, 2000. In other various exemplary
embodiments, the control program can be provided by any known or
later developed learning programming technique. The control program
can also be any predefined program.
[0121] Genetic programming is an evolutionary computing technique
that is able to generate a computer program that solves a problem
when given a high level description of the problem to be solved. A
description of genetic programming is found in John R. Koza,
"Genetic Programming: On the Programming of Computers by Means of a
Natural Selection", Cambridge, Mass.: MIT Press 1992, which is
herein incorporated by reference in its entirety.
[0122] The following is a description of one exemplary embodiment
of a method according to the invention, wherein the control program
is generated by genetic programming. In order to use genetic
programming to create a computer program, it is necessary to
provide the algorithm with an objective function for the problem to
be solved, a set of primitive functions with which to build the
programs, a few parameters, and for each primitive function, the
return type, the number of arguments and the type of arguments.
[0123] The genetic programming method begins by generating a
population of grammatically legal random programs in the target
language. The fitness of each program in the population is
determined by executing each program and measuring how that program
scores on the objective function.
[0124] Then the existing population of computer programs is used to
create a new population of computer programs using genetic
operators. The genetic operators are applied in proportion to the
fitness of the computer programs so that more fit programs are used
more often to create new computer programs. That is, more fit
programs are selected more often to be "parents" for the creation
of new computer programs in the next generation.
[0125] The genetic operations used are a crossover operation, a
mutation operation and a cloning operation.
[0126] The crossover operation takes parse tree representations of
two fitness selected "parent" computer programs and creates two new
"child" programs by exchanging a randomly selected subtree between
the two parents.
[0127] The mutation operation takes a parse tree representation of
one fitness selected computer program, randomly selects one of its
subtrees, and replaces that subtree with a new randomly created
subtree.
[0128] The cloning operation takes one fitness selected computer
program and creates a copy of it for inclusion in a new population
of programs.
[0129] After an entirely new population has been created by
applying the genetic operations to the previous population, the
fitness of each program in the new population is measured using the
fitness/objective function as before.
[0130] The process of creating new populations of programs and
measuring their fitness is repeated until a computer program is
discovered that gets an optimal score on the fitness function or
the process is terminated. The computer program with the best score
on the fitness function is the result of this process.
[0131] In an exemplary embodiment, computing the fitness function
involves running the evolved computer program on each module unit
to control a simulation of the modular robotic smart membrane in a
simulated world. The computed value of the objective function is a
measure of how well the modular robot performed on the goal task.
For example, the fitness function might measure how many foreign
objects are correctly sorted across the membrane. It should be
appreciated that there may be many features of each program that
the fitness function measures. The programs are compared to
determine which program has the best solution for the goal based on
the evaluated features. Further, it should be appreciated that any
known or later developed method of multi-objective optimization may
be used to determine the best program for the desired goal.
[0132] A fitness function (or objective function) must be provided
to the genetic programming system for the specific task or goal
that is being solved. In various exemplary embodiments, the
function is usually defined so that the better an individual
performs, the lower its fitness value. An example of a fitness
function is the number of bad objects that are mistakenly accepted
plus the number of good objects that are mistakenly rejected by the
membrane. The fitness function can also include penalties for
things like the time taken to solve the problem, the amount of
power used to solve the problem, etc.
[0133] Each decentralized controller in each of the modules is
evaluated by running a simulation of the modular robot in a world,
where each module is controlled by its own copy of the
decentralized controller under evaluation. The simulation tracks
the state of the world, including all walls and objects, and the
location of each robot module in the world.
[0134] In one example embodiment, on each turn of the simulation,
each of the robot modules is executed one at a time. In another
embodiment, all of the robot modules are executed concurrently. The
simulation is run for either a pre-specified maximum number of
turns, or for a pre-specified maximum amount of computer time, or
until some other predefined stopping condition becomes true.
[0135] The primitive functions from which the evolved computer
programs are constructed are provided to the genetic programming
system. In various exemplary embodiments, a designer provides the
primitive functions. In other various exemplary embodiments,
another computer program provides the primitive functions. It
should be appreciated that there are other numerous ways that the
primitive functions could be provided, including, but not limited
to a program that queries the modules to determine the functions of
the module and generates the primitive functions on the results of
the query.
[0136] In various exemplary embodiments, implementations for each
primitive function are supplied to the genetic programming system.
In various exemplary embodiments, the return type and parameter
types of each primitive function are also given. Examples of
primitive functions are arithmetic functions, Boolean functions,
looping constructs, conditional functions, etc. Primitive functions
for accessing the robot module's sensors, internal states, and
motors are also used in various exemplary embodiments.
[0137] In various exemplary embodiments, the language for the
evolved computer programs is defined. In various exemplary
embodiments, the language is defined by supplying an explicit BNF
grammar for the language. The language can also be defined by
specifying the return type and parameter types for each of the
primitive functions in the language. In this case, the type
specifications for the entire set of primitive functions together
create an implicit grammar definition for the language.
[0138] In various exemplary embodiments, the parameters that are
supplied to the genetic programming system include:
[0139] a. the size of the population (i.e. the number of programs
that are being considered);
[0140] b. the percentage of crossover, mutation, and cloning
operations to perform;
[0141] c. the maximum depth of the computer programs to create
(i.e. after the initial population size, this constraint controls
the size of the number of created programs);
[0142] d. the maximum depth of the computer programs to generate
randomly in the beginning of the run;
[0143] e. the maximum depth of the sub programs to generate for
mutations; and
[0144] f. the maximum amount of computer time or simulation steps
to use during the fitness evaluation.
[0145] It should be appreciated the parameters a through f, set
forth above do not have to be set to particular values. Further, it
should be understood, that an algorithm may be used to determine
these parameters automatically without any human intervention.
[0146] The method outputs a program that will be run in each of the
modules and will enable the entire programmable smart membrane to
achieve the desired sorting behavior.
[0147] The following description is an exemplary embodiment of a
modular robotic smart membrane according to the invention. However,
it should be appreciated that modular robotic smart membranes
according to this invention may include numerous other
embodiments.
[0148] FIG. 8 displays a flowchart for an exemplary embodiment of a
method for generating a control program according to the invention.
It should be appreciated that FIG. 8 illustrates specific functions
of an embodiment of the present invention, but that in alternate
embodiments, more or fewer functions may be used. In an embodiment
of the present invention, steps/functions set forth in FIG. 8 may
represent software programs, software objects, software functions,
software subroutines, code fragments, hardware operations, user
operations, singly or in combination.
[0149] The method begins at step S800 and proceeds to step S805,
wherein a plurality of software functions are obtained as described
herein. At step S810, a set of the plurality of software functions
are randomly selected to provide a first plurality of software
programs for solving a problem as described herein.
[0150] At step S815, a fitness function is used to evaluate the
first plurality of software programs by calculating respective
fitness function values as described herein. In various exemplary
embodiments, a fitness function is provided by a user.
[0151] At step S820, a second plurality of software programs is
selected from the first plurality of software programs based on the
first plurality of software programs respective fitness function
values, as described herein. In various exemplary embodiments, the
second plurality of software programs includes duplicate software
programs of the first plurality of software programs.
[0152] At step S825, the second plurality of software programs are
modified by genetic operations, such as crossover, mutation and
cloning, singly or in combination, as described herein.
[0153] At step S830, a decision is made whether another plurality
of programs are to be constructed, or a new generation or
population is to be formed.
[0154] If a new generation of software programs is desired, the
method proceeds back to step S815, where steps S815 through S825
are repeated. Otherwise, the method proceeds to step S835, wherein
the software program for a module which has the lowest fitness
function is selected as the control program, as described
herein.
[0155] FIG. 9 displays a schematic of an exemplary embodiment of a
modular robotic smart membrane 900 that includes a plurality of
robot modules 910. The membrane 900 filters objects based on color.
Objects 920 have a first color and objects 930 have a second color.
The second color can be any color other than the first color. The
second color for all of the objects 930 does not have to be the
same color.
[0156] It should be understood that any set of properties that can
be computed by the membrane using its sensor, processing, and
memory capabilities can be used as the basis of filtering.
[0157] In this exemplary embodiment, the objects 920, 930 are the
size of a single robot module 910. However, it should be
appreciated that the objects 920, 930 may be larger or smaller than
a single robot module 910.
[0158] Initially, the objects 920, 930 are placed on top of the
smart membrane 900. The membrane 900 will reconfigure itself so
that the objects 930 with the second color remain above the
membrane 900 and the objects 920 with the first color are moved to
the bottom of the world 940. In various exemplary embodiments, the
membrane 900 is required to end up, after filtering, close to its
initial position, though the individual modules 910 within the
membrane do not need to end up close to their original
positions.
[0159] In this exemplary embodiment, the simulated world 940 used
for evolving the control software for the modular robotic smart
membrane 900 is square with 10 units on a side. In this exemplary
embodiment, the wall units are the same size as the module units
910. However, it should be understood that in other various
exemplary embodiments, the size of the wall units may vary. In this
exemplary embodiment, the world 940 is divided up into 10.sup.2=100
grid locations. The perimeter grid locations of the entire world
contain wall boundary objects that the robot modules cannot pass
through.
[0160] In this exemplary embodiment, each robot module 910 is in
the shape of a square that occupies one grid location in the world
940. The state of a robot module includes its location in the
world, the values of the most recent messages received from
adjacent robot modules in each of eight directions, the values of
its registers, the current values of the sensors, and the facing
direction of the module.
[0161] In this exemplary embodiment, directions are encoded as real
values in [0.0, 1.0), where direction 0.0 is a positive x axis of a
robot module 910 and direction values increase going around counter
clockwise. The direction values used by each robot module 910 are
local to a frame of reference 912 of each module 910. The direction
that a robot module 910 is facing is always direction 0.0 where
direction 0.25 is 90 degrees to the left of where it is facing,
etc. In other various exemplary embodiments, the direction values
are not discrete, but rather continuous.
[0162] There are three types of sensors, one for detecting walls,
one for detecting other robot modules, and one for detecting
foreign objects. Each module has eight sensors of each type,
pointing in different directions. The eight directions for the
sensors are: 0.0=east, 0.125=northeast, 0.25=north,
0.375=northwest, 0.5=west, 0.625=southwest, 0.75=south, and
0.875=southeast. The values of the sensors are given in units of
intensity in [0.0, 1.0], where the intensity is the inverse of the
distance to the thing sensed, 0.0 means that the thing was not
sensed at all, and 1.0 means that the thing was sensed in the
immediately adjacent grid location.
[0163] The robot modules 910 are able to move relative to one
another. In this exemplary embodiment, the modules 910 move in four
directions: east, north, west, and south. In this embodiment, an
individual robot module 910 moves by sliding against an adjacent
robot module 910. In this exemplary embodiment, a robot module 910
moves by pushing itself along another robot module 910 that is
adjacent to it at 90 degrees from a direction of motion. Similarly,
a robot module 910 can pull against another robot module that is
diagonally adjacent to it in the direction of motion. Modules with
this type of movement have been implemented at Johns Hopkins
University. See Amit Pamecha, Imme Ebert-Uphoff, Gregory S.
Chirikjian, "A Useful Metrics for Modular Robot Motion Planning,"
IEEE Transactions on Robots and Automation, pp 531-545, Vol. 13,
No. 4, August 1997, which is incorporated herein by reference in
its entirety. For this exemplary embodiment, sensing,
communication, and processing capabilities are added to the modules
disclosed by Pamecha.
[0164] In this exemplary embodiment, the robot modules 910 initiate
different types of moves to reconfigure the membrane 900. In this
exemplary embodiment, the movement types of the modules 910 include
a single move and a line move.
[0165] FIGS. 10A-10C display, in three separate steps, a single
move over time for two modules 910a and 910b, respectively. The
single move is the movement of a single module. The single move is
successful if the moving robot module 910 is moving into an empty
grid location identified as "x" and "y".
[0166] FIGS. 11A-11B display, in two steps, a line move over time
of a line of robot modules 910c. The line move is the movement of
the entire line of robot modules 910c. The line movement is
successful if the front-most module 910d in the line is moving into
an empty grid location, identified as "w". A line move can be
initiated by any robot module 910 in a line of robot modules.
[0167] In this exemplary embodiment, the objects 920 and 930 are
inactive modules that are the same size as the robot modules 910.
However, it should be understood that the objects do not have to be
the same size and may also vary in size. The robot modules 910 can
push and pull these inactive modules just as they can with other
robot modules.
[0168] The following discussion is for an exemplary embodiment for
evolving a control program for the forgoing exemplary membrane. In
this exemplary embodiment, the fitness for each individual program
is calculated as a weighted average of fitness over 24 simulated
worlds. It should be appreciated that in other various exemplary
embodiments, the fitness of an individual program may be based on a
single simulated world.
[0169] In this exemplary embodiment, each simulation world is
initialized with the modules of the smart membrane arranged in a
vertically centered rectangle extending 8 units from the left wall
946 to the right wall 948. In twelve of the 24 simulated worlds for
this embodiment, the membrane 900 includes a height of three
modules 910; in the other twelve, the membrane 900 includes a
height of four modules 910. Each module 910 has a facing direction.
In this exemplary embodiment, the facing direction of each of the
modules 910, and the order in which the programs of the modules 910
are executed, is determined randomly for each world. It should be
appreciated that these factors may be predefined or preset.
[0170] Each simulated world used to measure fitness also includes
an object, either an object 920 with the first color or an object
930 with the second color, to be filtered, which is placed on top
of the membrane 900. The goal of the membrane 900 is to filter the
objects so that objects 930 with the second color remain above the
membrane 900, while the objects 920 with the first color are
transported through the membrane 900 to end up below the membrane
900.
[0171] In this exemplary embodiment, a representative sample of
possible horizontal positions for the objects is contained in the
24 fitness worlds. In six of the worlds, the object that appears
initially above the membrane is an object 930 with the second
color, in which case the object 930 does not need to be moved at
all. In the other 18 worlds, the object that appears initially
above the membrane is an object 920 with the first color, and the
membrane 900 should transport the object 920 through the membrane
900 to the other side.
[0172] In order to evaluate the fitness of an individual program
for a particular world, the program is installed on each module in
the simulation world. Then the robot modules 910 execute the
program one at a time, in turn. In this exemplary embodiment, the
order of execution of the program by the individual modules 910 is
selected randomly for that particular world. The sequence of all
robot module program executions is a simulation "turn." In this
exemplary embodiment, the fitness of a particular program is
determined by averaging the value given by the fitness after each
of ten simulation turns. In this embodiment, lower fitness values
are better, and make it more likely that this individual program
will have offspring in the next generation of evolution.
[0173] For this exemplary embodiment, the fitness value is a
weighted sum of two measurements. The first measurement is the
location of the object 920 or 930 relative to the membrane 900. The
second measurement is the current absolute position of the membrane
900. It should be appreciated, that in other various exemplary
embodiments, the fitness value may be based on more or less
measurements.
[0174] The first measurement is determined by the average offset in
the vertical dimension of the modules 910 from the object 920 or
930, as a fraction of the least desirable displacement. For
example, when the object is an object 920 with the first color,
this term can be minimized by reconfiguring so that the object 920
ends up below all of the membrane modules 910. This term has the
effect of making sure that each object ends up in the appropriate
final position.
[0175] The second measurement is determined by the average absolute
displacement of each module 910 from the horizontal central axis
950 of the starting position of the membrane 900. This measurement
penalizes the program for failing to keep the original structure
and position of the membrane 900 intact, while the object 920 or
930 is being filtered.
[0176] In an exemplary embodiment, each decentralized controller
program generated during the evolutionary process is a legal
instance of the language defined by the BNF grammar given below. In
this exemplary grammar, the start symbol is <program>.
[0177] Language Grammar:
[0178] <program>:=<realExpression>
[0179] <realExpression>:=
[0180] "("<realFunction>")"
[0181] .vertline. <realConstant>
[0182] <realFunction>:=
[0183] sendMessage <realExpression>
<realExpression>
[0184] .vertline. readMessage <realExpression>
[0185] .vertline. rotate <realExpression>
[0186] .vertline. setRegister <realExpression>
<realExpression>
[0187] .vertline. getRegister <realExpression>
<realExpression>
[0188] .vertline. readSensorSelf <realExpression>
[0189] .vertline. readSensorWall <realExpression>
[0190] .vertline. readSensorExternalObject
<realExpression>
[0191] .vertline. getTurn
[0192] .vertline. protectedDivide <realExpression>
<realExpression>
[0193] .vertline. protectedModulus <realExpression>
<realExpression>
[0194] .vertline. if <booleanExpression>
<realExpression>
[0195] <realExpression>
[0196] .vertline. progN <realExpression>
<realExpression>
[0197] <realConstant>:=0.0 .vertline. 0.25 .vertline. 0.5
.vertline. 0.75 .vertline. 1.0 .vertline. -1.0
[0198] <booleanExpression>:="("<booleanFunction>")"
[0199] <booleanFunction>:=
[0200] moveLine <realExpression>
[0201] .vertline. moveSingle <realExpression>
[0202] .vertline. and <booleanExpression>
<booleanExpression>
[0203] .vertline. less <realExpression>
<realExpression>
[0204] .vertline. isFirst color <realExpression>
[0205] .vertline. isSecond color <realExpression>
[0206] The following is a description of the foregoing exemplary
functions.
[0207] A. (sendMessage message direction)
[0208] This function sends the real valued message message from the
sending module to an adjacent module (if any) in the direction
direction, relative to the frame of reference of the sending
module. The direction is interpreted mod 1, and then rounded to the
nearest eighth to indicate one of the eight adjacent grid
locations. This function returns the value of the message sent.
[0209] B. (readMessage direction)
[0210] This function reads the real valued message from the
adjacent module (if any) that is location at the relative direction
direction from the sending module. The direction is interpreted mod
1, and then rounded to the nearest eighth to indicate one of the
eight adjacent grid locations. This function returns the value of
the message read.
[0211] C. (rotate direction)
[0212] This function rotates the robot module by the amount
direction. This does not physically move the module, it just resets
the internal state of the robot module's internal facing direction.
This function returns the value of direction.
[0213] D. (setRegister index value)
[0214] This function sets the value of the robot module's register
numbered index to value. This function returns value.
[0215] E. (getRegister index)
[0216] This function gets the current value of the robot module's
register numbered index. This function returns the value of the
register.
[0217] F. (readSensorSelf direction)
[0218] This function reads the intensity value of the self sensor.
Intensity is the inverse of the distance to the closest robot
module at the relative direction direction. The intensity is zero
if there is no robot module in that direction. This function
returns the intensity value.
[0219] G. (readSensorWall direction)
[0220] This function reads the intensity value of the wall sensor.
Intensity is the inverse of the distance to the wall or obstacle in
the relative direction direction. The intensity is zero if there is
no wall or obstacle in that direction. This function returns the
intensity value.
[0221] H. (readSensorExternalObject direction)
[0222] This function reads the intensity value of the external
object sensor. Intensity is the inverse of the distance to external
object in the relative direction direction. The intensity is zero
if there is no external object in that direction. This function
returns the intensity value.
[0223] I. (getTurn)
[0224] This function returns the current value of the "turn"
variable for the simulation. The "turn" variable is set to zero at
the beginning of the simulation and is incremented each time all of
the robot modules are executed.
[0225] J. (protectedDivide operand1 operand2)
[0226] This function evaluates both operands, and returns operand1
divided by operand2. If operand2 is zero, then this function
returns 1.
[0227] K. (protectedModulus operand1 operand2)
[0228] This function evaluates both operands, and returns the real
valued remainder of operand1 divided by operand2. If operand2 is
zero, then this function returns 1.
[0229] L. (if condition expression1 expression2)
[0230] This function evaluates the Boolean condition. If condition
is true, then expression1 is evaluated and its value is returned.
If condition is false, then expression2 is evaluated and its value
is returned.
[0231] M. (progn expression1 expression2)
[0232] This function evaluates expression1 and expression2,
returning the result of expression2.
[0233] N. (moveLine direction)
[0234] This function causes an entire line of robot modules to move
in the relative direction direction if they can move. The line of
robot modules to move is the line of modules collinear with the
current robot module in the relative direction direction of the
current robot module. The line of modules can move if and only if
the front-most robot module in the line is not blocked by an
immovable wall or obstacle and if there is another robot module
adjacent to the line that it can push or pull against. This
function returns true if the line of modules is able to move, and
returns false otherwise.
[0235] O. (moveSingle direction)
[0236] This function causes the current robot module to move in the
relative direction direction if it can move. The module can move if
and only if it is not blocked by any other object, and if there is
another robot module adjacent to it that it can push/pull against.
This function returns true if the line of modules is able to move,
and returns false otherwise.
[0237] P. (and operand1 operand2)
[0238] This function evaluates both operands and returns true if
and only if both of the operands are true, and returns false
otherwise.
[0239] Q. (less operand1 operand2)
[0240] This function evaluates both operands and returns true if
operand1 is less than operand2, and returns false otherwise.
[0241] R. (isFirst color direction)
[0242] This function returns true if there is a first color object
immediately next to this module in direction direction, and returns
false otherwise.
[0243] S. (isSecond color direction)
[0244] This function returns true if there is a second color object
immediately next to this module in direction direction, and returns
false otherwise.
[0245] Experimental Results
[0246] The exemplary embodiment of the membrane displayed in FIG. 9
was evaluated with the following parameters:
[0247] the size of the population=72,000 individuals;
[0248] the percentage of crossover operations to perform=99.0%;
[0249] the percentage of mutation operations to perform=1.0%;
[0250] the percentage of cloning operations to perform=0.0%;
[0251] the maximum depth of the computer programs to create=9;
[0252] the maximum depth of computer programs generated randomly in
the beginning of the run=5;
[0253] the maximum depth of the sub programs to generate for
mutations=4;
[0254] the maximum amount of computer time to use during the
fitness evaluation=unbounded; and
[0255] the number of turns to simulate=10.
[0256] A working solution for the control program for this
embodiment emerged in generation 32 of the evaluation/run. The
solution was composed of 49 primitive function references and 32
real valued constants. The behavior of this solution is shown in
FIGS. 12A-12C for a single initial world condition at three points
in time, namely at an initial stage, an intermediate stage and a
subsequent stage, respectively, during the simulation. As shown in
FIG. 12C, a temporary transport channel 980 is formed by the
movements of the modular units. A string representation of the
control program is given here:
[0257] (ProtectedMod (SendMessage (If (IsBlue (ProtectedDiv 0.5
1.0))
[0258] (If (Move -1.0) (ProtectedDiv 0.5 (Rotate
[0259] (GetRegisterEntry (SetRegisterEntry 0.5 0.25)))) 0.75)
(Rotate (ProtectedDiv (Rotate (ProtectedDiv (Rotate 0.25)
(ReadSensorSelf 0.5))) (ProtectedMod 0.25 (SetRegisterEntry 0.75
0.0))))) (If (MoveOne (ProtectedDiv (If (IsRed 0.25)
(ReadSensorSelf 0.0) (GetRegisterEntry 0.5)) getTurn)) 0.0 0.75))
(SendMessage (If (IsBlue (ProtectedDiv 0.5 1.0)) (If (Move -1.0)
(ProtectedDiv 0.25 (Rotate (ReadSensorBrokenAlien -1.0)))
(ReadSensorWall 0.75)) (Rotate (ProtectedDiv (Rotate (ProtectedDiv
(Rotate 0.25) (ReadSensorSelf 0.5))) (ProtectedMod 0.25
(SetRegisterEntry 0.75 0.0))))) (If (MoveOne 0.0) 0.0 (ProgN 0.0
(GetRegisterEntry 0.0)))))
[0260] Although this program was evolved for an exemplary
embodiment including a 10.times.10 world, with membrane heights of
3 and 4, and one object filtered at a time, it should be
appreciated that the software scales so that it works for membranes
of other exemplary embodiments consisting of hundreds of modules
arranged in rectangles of greatly varying heights and widths. The
software also successfully filters multiple objects at once. For
example, as shown in FIG. 13, the program functioned successfully
when installed on each of 496 robot modules 1210 of a membrane
1200, in a 64.times.64 world 1240, with a membrane height of eight
modules and six objects to be filtered (three second color objects
1230 and three first color objects 1220 spaced evenly along the top
of the membrane. FIG. 13, displays a schematic representation of
this exemplary embodiment after the filtering was completed.
[0261] It should be appreciated that in various exemplary
embodiments, a smart membrane can be programmed to be actively
moving while performing its desired function or while waiting for
objects to come into contact with it. For example, the module units
of a smart membrane may be programmed to cause the top surface of
the membrane to be agitated so that it rises and falls in an effort
to come into contact with objects to be filtered, sorted or
absorbed. In various exemplary embodiments, the agitated movement
is a sinusoidal wave motion.
[0262] It should also be appreciated that there are other exemplary
embodiments of smart membranes and types of module units other than
robotic modules according to the invention. Further, the invention
may be used in a variety of applications from a macro scale, such
as but not limited to a parts sorter, to a nanometer scale, such as
but not limited to chemical application of purifying substances and
biochemical and biomedical applications.
[0263] It should also be appreciated that a smart membrane of the
invention is applicable to other uses such as assembling and
disassembling objects.
[0264] While this invention has been described in conjunction with
the specific embodiments outlined above, it is evident that many
alternatives, modifications and variations will be apparent to
those skilled in the art. Accordingly, the preferred embodiments of
the invention, as set forth above, are intended to be illustrative,
not limiting. Various changes may be made without departing from
the spirit and scope of this invention.
* * * * *