Biomorphic Explorers

7/29/2019 Biomorphic Explorers

1/50

Inventor(s):

NATIONAL AERONAUTICS AND SPACE ADMINISTRATION

CONTRACT NO.NAS 7-918

TECHNICAL SUPPORT PACKAGE

On

BIOMORPHIC EXPLORERS

I

.

Adrian Stoics

Sarita Thakoor

for September 98NASA TECH BRIEF Vol.22, No. 9, Item #

from

JPL NEW TECHNOLOGY REPORT NPO-20142

NOTICE

I

Neither the United States Government, nor INASA, nor any person acting on behalf of i~ NASA:a. Makes any warranty or representation,express or implied, with respect of theaccuracy, completeness, or usefulness of theinformation contained in this or that


2/50

Biomorphic explorers are a class ofproposed small robots that would beequipped with microsensors and wouldfeature animallike adaptability and mobility.These robots would capture key features,a specific design or function found innature, taking advantage of general animalmechanical designs and neural functionsthat have evolved to enable animals to

move through various environments.These robots are conceived for use inremote, hostile, and/or inaccessible terres-trial and other planetary environments,where they would be used to performsuch diverse functions as acquisition ofscientific data, law-enforcement surveil-lance, or diagnosis for precise, minimallyinvasive medical treatment. Depending onthe specific environment to be explored, a

biomorphic explorer might be designed tocrawl, hop, slither, burrow, swim, or fly.

The biomorphic-explorer concept is ageneralization and encompasses thenanorover concept reported in Tetherless,Optically Controlled Nanorovers (NPO-19606), NASA Tech Briefs, Vol. 21, No. 3(March 1997), page 92. Like nanorovers,biomorphic explorers would exploit theemerging technology of microelectro-mechanical structures. Biomorphic explor-ers would be enabled by a unique com-bination of direct-driven, flexible, shape-reconfigurable advanced actuators andth i d ti t l b f lt t l t bi

applied voltage; that is, they could be con-trolled photonically or electronically. Thedesired combinations of mobility andadaptability, along with fault tolerance and alimited capability for learning, would beachieved by integrating the actuators withvery-large-scale integrated (VLSI) circuits

that would implement neural-networks uti-lizing genetic algorithms.

Relative to conventional remote-sens-ing robotic vehicles, biomorphic explorers

ld b i l i i d

of insects or other groups of small socialanimals engaged in cooperative activity.

This work was done by Sarita Thakoorand Adrian Stoica of Caltech for NASAsJet Propulsion Laboratory.

In accordance with Public Law 96-517, the contractor has elected to retain

title to this invention. Inquiries concern-ing rights for its commercial use shouldbe addressed to

Technology Reporting OfficeJ PL

Biomorphic ExplorersExploratory robots would feature animallike adaptability and mobility.

NASAs Jet Propulsion Laboratory, Pasadena, California

Traditional Actuators and Motors

Conventional Controls

Designed to Move in Specific Ways

Individualistic Behavior

Paradigm Shift

Conventional Remote-SensingRobotic Vehicle (Rover) Biomorphic Explorer

Flexible, High-Efficiency Actuators

Neural Control

Evolved for Adaptation, Reconfigurable

Cooperative Behavior

The Development of Biomorphic Explorers would be consistent with a current trend away fromconventional, limited-mobility robots toward highly mobile, adaptive robots based partly on biolog-ical concepts.


3/50


4/50

1.

providing a much wider horizon of areas for

9778 20142inspectionhesting. Their dedicated sensing functions and

maneuverability would be invaluable in hazardous or difficult-to-reach territories for scouting missions.Combining the flexible actuators and biomorphic controls would offer for the first time a new direction in

ADVANCED MOBIL ITY W7TH NEW CAPABILITIES ofadapting to terrains, enhanced spatial access owingto flexibili~, and ease of multiplicity due to reduced complexi~.

Current rovers Biomorphic explorers

.P\a 1 flexible, high efficiency

traditional actuatordmotorsconventional controls

j ~ #=.#...-{--+..+...... Tdesigned J! I .+.+_&---. ~~individualistic behavior } *.#-~____.J+ . . ,& --

i ~paradig~ shift

neural control

evolved for adaptation

reconfigurablecooperative behavior

.

C. Detailed Report attached


5/50


6/50


7/50


8/50


9/50


10/50


11/50


12/50


13/50


14/50


15/50

JPLRECONFIGURABLE,ADAPTABLE, MOBILEBIO-MORPHIC EXPLORERS

JPLSaritaThakoor and Adrian Stoics

SYSTEMS

a+wu March 4,1997JET PROPULSION LABORATORY

CALIFORNIA INSTITUTE OF TECHNOLOGYPASADENA, CALIFORNIA 91109I Presentation to DARPA:

Dick Urban,Asst.Director, TECHNOLOGY INTEGRATION,


16/50

APLRECONFIGURABLE, ADAPTABLE, MOBILE SYSTEMSBIO-MORPHIC EXPLORERS

PURPOSE OFTHEMEETING:.

Q TO UNDERSTAND DARpAs/DoDs NEEDS INTHE AREA OF RECONFIGURABLE, ADAPTABLE,MOBILE SYSTEMS

q TO DISCUSS THE JPL CONCEPT OF BIO-MORPHICEXPLORERI


17/50

JPLDARPAS INTEREST

q SMALL INEXPENSIVE MOBILE SYSTEMS THATCAN TRAVEL OVER A VARIETY OF TERRAINS- BEING ABLE TO AUTONOMOUSLY CHANGE THEIR

PHYSICAL CONFIGURATION IN FIELD/ IN-SITU- TO ADAPT TO THE CHANGES IN THE TERRAINCONDITIONS, AS REQUIRED

q oe*oa*ee**oo q 0 0 0 0 0autonomousexplorationSynergistic withNASAs interest inrovers for scoutingmissionslplanetary


18/50

ADVANCED MOBILITY FOR BIO-MORPHICJPL EXPLORERSCILIARY/FLAGELLAR MECHANISMS FOR FLUID NAVIGATION,

MULTIPOD \ FLEXIBLE / HOVERING MECHANISMS.0 CRAWLING MECHANISMS ACTUATORS (AERIAL EXPLORATION)(INSECT ROVERWCRAWLERS) finsstraight legs --------> webbed feet


27/50

JPL

q

I

RECONFIGURABLE MOBILESYSTEMS

HOW WILL IT BE ACHIEVED?Ifsolidsutiacefor examplechanges tosottffour-/ikesoi~thelegswould change continuouslywtithe distance between them,the gait ofmotion, @m+the position inwhich theywillcontactthe grounde.g. vertical stick to arms bent at elbows inorder to change from a pogo stick configurationto a webbed feet configuration


28/50


29/50


30/50

JPLFlexible Actuators For Advanced Mobilitv

FLEXIBLEPIEZOCERAMIC SUBSTRATE\FILM

r\

OPTICALLY

DRIVEN

/PIEZOCERAMICFILM

b

ELECTRICALLY

DRIVEN

METAL

FLEXIBLE 1SUBSTRATE

FLEXIBLE

ACTUATOR

SHEETS

Flexible actuators are envisioned by depositing tailored thick (- 2-10micron)films of active materials on judiciously chosen, strong flexible substrates.

.


31/50

JPL, BIO-MORPHIC EXPLORERS

TEAM

cSaritaThakoor,JPL: Advanced Flexible Actuators,Optical ActuationIntegration of Biomorphic Explorer

qAdrian Stoics,JPL:Recontigurable,Evolvable Electronics

Biomorphic Controls

qSurampudi Rae,JPL:SmallPower SourcesqCALTECH: Mechanical Design, Telecommunication UC Berkeley: Biologically Inspired Mechanical Designq Penn State:Actuation, Amplification techniquesINDUSTRY: Manufacturing of film actuators

#

.


32/50

JPL

q

q

q

#

, SUMMARYRECONFIGURABLE, ADAPTABLE,MOB/LESYSTEMSBIO-MORPHIC EXPLORERSBIO-MORPHIC EXPLORERS ARE AN EXCITING,REVOLUTIONARY APPROACH ADVANCED MOBILITY

Q FLEXIBLE ACTUATORSqBIO-MORPHIC CONTROL

REQUIRES AN INTER-DISCIPLINARY TEAMq NUCLEUS EXISTS

IS JPLsBIO-MORPHIC EXPLORER APPROACHRELEVANT TO DARPAs ADVANCED MOBILITYNEEDS?


33/50

JPLFlexible Actuators For Advanced Mobilitv

,

PIEZOCERAMICFILM

/

\

FLEXIBLE

F SUBSTRATE\

OPTICALLY

DRIVEN

ELECTRICALLY

DRIVEN

/PIEZOCERAMICFILM

((ht)

b

MHAL

FLEXIBLE iSUBSTRATE

FLEXIBLE

ACTUATOR

SHEETS

Flexible actuators are envisioned by depositing tailored thick (- 2-10micron)films of active materials on judiciously chosen, strong flexible substrates.

.


34/50

JPLFLEXIBLE ACTUATORS

Flexible actuators are envisioned by depositing tailored thick(-2-10 micron)films of active materials on judiciously chosen, strong flexible (polymeric)substrates. Flexible actuators would provide a combination of high force and,.displacement and be operable over a wide temperature range as is required for a

variety of advanced mobility applications. Potential advantages of flexibleactuators are:q

q

q

q

qU

q

q

low power (low voltage operation,


35/50

JPLFLEXIBLE ACTUATORS

q Enable a new generation ofelectro- mechanical and opto-mechanical systems

q Actuators not restricted by the clamping effect due to the rigid

substrate

q Actuation force not limited by the thickness - thicker films can bedeposited at high rate

q Results in mobile elements with

higher force to input power ratio contact-less optical activation as an option.


36/50

JPL, FLEXIBLE ACTUATORS

APPROACH

q Optimization of the Piezoceramicfilms:Optimization of the optical penetrationeffect

Optical quality enhancement

Polarization direction & intensityoptimization

COMPARISON OF BULK


37/50

ILLUMINATIONWAVELENGTH-350-400 nm

COMPARISON OF PIEZOCERAMIC BULK..AND THIN FILM ACTUATOR -

DIRECTION OFBIMORPHDEFLECTION FLEXIBLE

,dW PIEZOCERAMIC SUBSTRATE-iFILM/

+LEADLANTHANUMZIRCONATETITANATE(PLZT)

1

1

IEZOCERAMICILMMETALELECTRODE

ELECTRICALLEAD

OPTICALLY ELECTRICALLYEzlizia DRIVEN DRIVENPIEZOCERAMICBIMORPH FLEXIBLE FILM ACTUATOR

ceramic thickness:200 micron film thickness: 2 micron ~# (a) (b)

.


38/50

.

.


39/50


40/50

JPLTABLE 4:COMPARISON OF BULKPIEZOCERAMIC ACTUATOR WITH PROJECTED

, PERFORMANCE OF FLEXIBLE FILM ACTUATOR

Current Status Projected ImprovementCeramic Actuator Film Actuator

ELECTRICAL Thickness 200 microns 2 microns: Thickness reduced,ACTUATION material tailoredPARAMETERS

Operating Voltage 100 v 5 V : Operational voltage reducedEnergy Density l x 25X : Inherent advantage of

reduced thickness

Force/Volume l x 5X enhancement for film actuator

OPTICAL Optical Power 80mW/cm2 8mW/ cm 2 : Illumination IntensityACTUATIONPARAMETERS Power Ratio lox l x

Photonic to 0.1 ?40 1?40-10%: significantMechanical Conversion enhancement inEfficiency overall efficiency

Force/Energy l x 2X to 20X : Multifoldenhancement in thefdm actuatorForce/Power l x 20X to 200X

#

COMPARISON ACTUATOR ATTRIBUTES


41/50

JPL COMPARISON ACTUATOR ATTRIBUTESI

h, Optical Activated BimorphstlllF RAINBOWHigh DisplacementActuation= Electrically Activated bimolph

Flex TensionDevices

/Direct ExtensionLx Devices

SAR :ITA THAKOORPIEZO02SaritaThakoor. darpa397 28


42/50

JPL~~BIO-MORPHIC ROBOTICS

q Some of the most influential recent approaches to mobile robotics are discussed inthe following. In the behavior-based approach introduced by Brooks2, the robotbehaviors are developed gradually, from simple to more complex. The approachputs an important accent on embodiment and situatedness (real robots performingin their target environments). A primary purpose in robot life is its survival in theenvironment. The neuromorphic approach illustrates the power of neural modelsusing artificial neural networks for sensory-motor control. A good example of anapplication oriented system using neural networks is the vision guided vehicleNAVLAB, which drives autonomously at 55m/h under neural contro13. Theevolutionary approach (evolve by a natural selection mechanism) makes use ofevolutionary techniques (such as genetic algorithms or genetic programming) toshape robot controls and sensory-motor maps. A variety of insect-like robots havebeen evolved4 to exhibit different gait patterns, adapted to the particularenvironments in which they behave.Bio-hybrid robots combine natural andartificial parts. One such hybrid from Shimoyamas5 lab is the roboroach, hybridcockroach-robot which has no motor or gears but two real cockroach legs actuatedby electrostimulation. Biorobotics6-1 3 is the area of robotics in which robots caneither utilize,mimic, or are inspired from, the biological functions,morpholo~y, or

# behaviors of plants and animals.


43/50


44/50

4PL. B/O-/WORPHlC CONTROLS

GENETIC ALGORITHMSimple Genetic Algorithm

{create an initial population of solutions (generation 1)evaluate individuals in current generation

for a number of generations

{select most fitted/promising individuals as parentscombine the genetic code of parents to obtain the new generation of offspringmutate the genetic code of some individual offspringevaluate/rank individuals in current generation}

}Parents Children

DmmmmmtJ+

q mm moo s mclmncl

13DIJOD ooommm noonmmAfter Afler mutationcrossover

The straightforward combination is by simple one-point crossover, in which the genetic code of 2parents gets cut atonepoint, and one of the two chunks gets swapped between the parents,resulting in 2 offspring with a hybrid genetic information. This is illustrated in Figure 3, where abinary code of 1s(black) and 0s(white) is used. Mutation is the flip of one of more bits in theoffspring.

aom


45/50

JPL, BIO-MORPHIC CONTROLS

APPROACH

Biology tells us that the control of periodic limb motion sequences are

generally delegated to a lower level controller (e.g. spinal cord in humans)than the powerful CPU (e.g. the brain). Obviously, this relieves the brain aresources to attend to the higher level cognitive functions including sensory :information processing (e.g.vision,hearing,etc.). This arrangement doesnot of course rule out a higher level command from the brain to the spinal

cord to modulatelchange the ongoing periodic motion whenever necessary,based on sensory input received and processed. For example, decision

to turn or run in a specific direction rather than walk on the sight of apreylpredator.


46/50

JPL, BIO-MORPHIC CONTROLS

APPROACH

Biological mobility systems have two important characteristics which could be used in thedevelopment of increased performance engineering systems: they are controlled by neurons,and they have evolved by natural selection to highly adapted solutions. The most popular. evolutionaryengineering mechanism are genetic algorithms (GA). GAs are a search/optimizationmethod using principles of natural selection to obtain systems adapted to their environment.

GAs perform a parallel search, operating on a population of individuals, each individual havinga unique genotype (information coded in chromosomes) and fitness, which is a measure ofphenotypes (external behavior determined by genotype) success in meeting some objectives.GAs aim to improve the general fitness of the population by selective breeding, merging thegenetic material of most fitted individuals,(and also some random mutations) to obtain thegenotypes of offspring in a new generation. The fittest member of the final generation istaken to be the optimum. In evolving robots, our approach would be behavior-basedlin the sense that the robot behaviors would be developed gradually, from simplest tomore complex.

SaritaThakoordarpa397 34


47/50

JPL/ BIO-MORPHIC CONTROLS

APPROACH/n afirststage, simple mobility mechanisms, such as gait would be evolved. As an example thecontrol of the leg motion can be evolved into several types of gait pattern. Consider the schematic ofan artificial insect, modeled after a spider, with 8legs. A neural network controller is associated witheach leg, the controllers being interconnected. The neural weights are randomly initiated andaccordingly a sequence of leg movements is produced. Scoring the walk of the robot in several trialswith different parameters (for example giving higher score to those who move further), and applying. .a genetic algorithm reproducing the genetic code with a selectionist bias on most fittest individualslead to a insect-like gait pattern.An example of a gait pattern evolved in simulation is: 3:LJ-24At time=tl , simultaneously trigger legs 1,2,5, and 6;At time=t2, simultaneously trigger legs 3,4,7, and 8. A5 6A variety of insect-like robots have been evolved to exhibit different gait patterns, adapted to theparticular environments in which they behave. Examples of evolved locomotion controller for insect-Iikerobots are given earlier4, 8. The evolutionary approach is scale-invariant. The same evolutionmechanisms apply for both big and small robots. Miniature robots can be ideally paired with smallsetmors and built at lower cost./nasecond stage, simple behaviors would be coordinated into morecomplex ones (such as moving in some direction, while actively looking for the home-target andbalancing a load).Athird stage would address the development of collective behaviors of severalindividual robots tasked with the same mission.

IL


48/50

JPIL, BIO-MORPHIC CONTROLS

APPROACH

For developing mobilitymechanisms, evolutionary algorithms would shape the type andposition of actuators, and the neuro-motorcontrolsignalsto the actuators and sensory-motor maps. Previous evolutionary robotics approaches consisted of determiningcontrol commands to electric motors that powered legs or wheels. Based on advancedactuators, our approach eliminates motors from the system, the controlled electricsignals to the active limbs, directly produce forces used for locomotion. Tolerances in $mthe characteristics of the actuator materials are accepted by the method, evolutionthrough trial and error compensates for them.

A successful conclusion of the first years activity therefore would lay an

excellent foundation for undertaking actual development of a biomorphic explorer using

flexible actuators by providing a full quantitative 3D visuallmechanical design andkinematic model of the explorer mobility and its controls. Demonstration of the Bio-morphic explorer concept feasibility insimulationwill be the key milestone of Year 1.


49/50

JPLWHY THE WHEELS WONT GO?

Michael LaBarbera, The American Naturalist,Vol.121,#3,(1983)p395-408believe an all-terrain roller skate is a physical impossibility;certainly an antlike organism with wheels 2 mm in diameterwould be unable to leave its nest without stalling on sand grains,pebbles, and fallen grass blades. On natural substrates, thereare thus scaling limitations to the utility of wheels, with smallerwheels being at an increasing disadvantage in mobility. Theseproblems may be reduced by decreasing vehicle weight to makeit easier to surmount obstacles, since the total weight which mustbe raised in elevating the center of mass will be lower.Stephen Jay Gould, Natural History,Vol90,#3,(1981)p42-48

ADVANCED MOBILITY FOR BIO-MORPHICEXPLORERS


50/50

JPL EXPLORERSCILIARY/FLAGELLAR MECHANISMS FOR FLUID NAVIGATION

SMSON)..m-

CRAWLING IVIFUI-I.

u- FIBERSI/\LEGS HEET/\ w

HOPPING MECHANISMS

(OPTICAL/ELECTRICAL BURSTTECHNIQUE INNOVATIONS)

#

o

Documents

Biomorphic Explorers