60

SERVO 09.2006

Dr. Stephen Mascaro and his

research team in the department

of mechanical engineering at the

University of Utah are making robot

muscles, working in a vein of research

called Wet Robotics. Rather than the

fascinatingly technical explanation you

might envision, they are simply actuat-

ing a metal that contracts in response

to hot water and expands in response

to cold water. But there is where the

simplicity ends.

They are making artificial muscles

for robots that work much the same as

our muscles do. The technology uses

shape memory alloys (SMAs), which

are nickel-titanium alloy strands or

poles with one of the rarest traits

among all metals — the capacity to

contract in response to heat.

Most metals tend toward expand-

ing in response to the application of

heat. Nickel-titanium does just the

opposite. “What’s going on is that

they [the nickel-titanium strands] are

actually realigning their crystal

structure when you heat them up. So,

the crystals of nickel and titanium

realign into a more compact orienta-

tion,” says Dr. Mascaro.

Experimentation is ongoing into

practical, scalable, and efficient ways

to heat up and contract SMAs to make

them act like muscles in order to apply

them as a robot muscle technology in

humanoid robots.

Some researchers (for other

applications) heat SMAs by filling

them with electric current. This is

called Joule or Resistive heating,

according to Dr. Mascaro. “You heat

them up (to temperatures) above their

transition (contraction) temperature.

This varies depending on the concen-

tration of nickel vs. titanium,” says Dr.

Mascaro.

Above their transition tempera-

ture, the SMAs are in a “super elastic

state” in which they are quite

pliable. You can actually heat these

wires up and stretch them like a

rubber band.

Figures 1 and 2. Demonstrates the Matrix idea of the Matrix Manifold and Valve system (MMV) whereby a

particular muscle can be activated based on the column and row that it’s in inside the matrix.

You can see the advantage of

being able to manufacture SMAs

with the balance of nickel and

titanium that gives you contraction

at the most desirable temperature

for a given application.

Some SMAs have transition

temperatures at “room tempera-

ture” so that you can “grab them”

barehanded and “stretch them

out” — a very unusual experience

the first time around! They use

these particular SMAs for dental

applications, “to stretch onto the

teeth to hold them in place where

your dentist wants them,” explains

Mascaro.

As actuators [potential robot

muscles], SMAs are about 1,000x as

strong as human muscles for the same

size muscle. They don’t, however,

contract as much as the human

muscle. Human muscle contracts 20

percent whereas the SMAs only

contract about four percent.

While you can heat up and con-

tract SMAs as fast as you would like

using electricity — fast enough to

approximate human muscle contrac-

tion speeds — they don’t cool down

fast enough to expand with the speed

of human muscle.

How do you cool these SMAs

down quickly enough to solve

this problem? Some researchers

accomplish this by putting the SMAs

in a cooling fluid. This introduces

another problem: “Now you have the

weight of this fluid added to your

system and you’ve lost the original

advantage of these actuators — that

they are tiny, lightweight, and give

you a lot of strength without having

the bulk of an electric motor,” says

Dr. Mascaro.

This is the problem that lead Dr.

Mascaro to his idea of how to use

SMAs as muscles. With Dr. Mascaro’s

technology, you embed the SMA

wires within tubes of flowing fluid for

cooling purposes. This gives your

robots their own sort of blood

vessels. In the human body, energy is

carried to and from the muscle

groups via blood

vessels. With the

robot muscles, you

have

cold fluid

removing heat, and

as we will discuss,

hot fluid producing

the artificial muscle

contractions.

When you use

electricity, the ener-

gy it produces is lost

in the cooling fluid. Dr. Mascaro uses

hot fluids, in this way the hot and cold

fluids are both recycled back into hot

and cold reservoirs. This produces

better energy efficiency than using

electricity.

How do you put large arrays of

these muscles into a compact area

like a robot arm? The advantage

This valve-based attempt at actuating heat-responsive

contracting nickel-titanium alloy metal strands for robot

muscles had some issues: The hot water leaked into tubing

that contracted strands the researchers didn’t want to

contract at the time, taking away some control of the

muscles; and the valves were resistive to the water flow,

costing some energy (see Figure 5).

Figures 3 and 4. Details the parts of the Matrix Vasoconstriction Device (MVD),

including the housing and water vessel tubing and the constrictors used on the

Matrix to constrict and open the flow of water.

Figure 5. According to Dr. Mascaro, this figure shows the

MMV prototype, which exposed new challenges in the form

of fluid resistance from the solenoid valves and parasitic

behaviors such that if the device tried to deliver hot water

to one of the actuators, later on in the sequence the hot

water might leak into one of the other actuators at a time

when they didn’t want to activate that muscle.

SERVO 09.2006

61

62

SERVO 09.2006

maintains so long as you have a

high-strength to low-weight ratio.

How does this scale so that you can

have, say, 100 muscles in a robot arm,

and control them all without having

to have 100 separate controls for

each one?

This is where Dr. Mascaro’s

Matrix Manifold and Valve (MMV)

system comes into play. The system

arranges the artificial muscles into

rows and columns so that only one

switch is needed for each matrix

manifold and so for several muscles,

as well.

In production, a robot limb that

would need to extend or leverage

something would have matrix mani-

folds with a hundred muscles in 10 x

10 arrays. The current proof of concept

model has 4 x 4 arrays.

Dr. Mascaro expects that

manufacturing techniques and

processes will be key in going beyond

even the 10 x 10 robot muscle

scenario. Connecting the actuators

as an integrated array with more

than 100 SMA muscles is a scalable

manufacturing issue; something that

can’t feasibly be practical as

the work of a few human

scientists working by hand

in the lab. Manufacturing

should take the potential

muscle count to hundreds

at first, eventually even

thousands.

A scalable manufacturing

method

hasn’t

been

addressed yet, but this will

be addressed in the lab, so

that it can be taken to

production.

In the big picture, Dr. Mascaro

has figured out how to make the

muscles forceful while maintaining

their light weight. The trick here is to

make them fast while maintaining

light weight, to get high-power-

to-weight ratio, rather than just high-

force-to-weight ratio.

The cardiovascular system in the

human body has many functions:

1. Delivering chemical energy to the

muscles.

2. Thermal regulation of the body.

“When your body gets hot during

a workout, your blood starts flowing

faster, so the heat is transported by

your blood stream out from the core

of your body to the extremities where

sweating removes the heat,” says

Dr. Mascaro.

Controls in the body constrict and

open the blood vessels to regulate

your core temperature. With fluid flow

in the robot muscles, temperature

regulates the muscle contraction and

expansion.

The MMV uses a valve on

each row and column of the array.

By opening the correct row and

column valves in combination, you

send hot or cold fluid to the correct

SMA.

While the row and column arrays

passed muster, the valve system had

to go. The valves are standard

solenoid valves and while they are

common and predictable, the water

has to flow through a small hole in

the valve. “This introduces fluidic

resistance. We can turn the flow off

Constricting and opening the

vessels/tubing directly removed

the need for valves. This removed

the resistance and provided greater

control over where the water flowed

(see Figure 8).

Figures 6 and 7. Photos of the prototype MVD and the completed MVD, respectively.

Figure 8. Demonstrates the new muscle system,

which constricts the vessels transporting water to

the SMA muscles or “unobstructs” them, rather

than using solenoid valves, which presented their

own problems in the form of fluid resistance.

and on to any of the muscles, but

when it’s on, we don’t want any

resistance at all, which you can’t

do with solenoid valves,” explains

Dr. Mascaro.

Solution: Rather than putting

valves in inline with the flow of fluid,

why not try to constrict the vessels

used to transport the fluid? Dr.

Mascaro has developed a Matrix

Vasoconstrictor Device (MVD) to

replace the MMV. Rather than using

the solenoid valves, this device

constricts all the rows and columns of

the vessels using air pressure, so that

they can completely collapse the

vessel, stopping all of the fluid

flow, or open it and fluid flows

without any resistance, according to

Dr. Mascaro.

As you might assume, the

SMA muscle apparatus and MVD

constitute a fully closed system in

which the same fluid is used all the

time, a fluid that doesn’t dissipate.

“The only energy input to the

total system is keeping the hot water

hot, the cold water cold, and some

means to pressurize the system to

keep the flow moving,” says Dr.

Mascaro.

Dr. Mascaro and crew are now

working on a robotic heart to pump

the fluid in and out of the SMA

muscles. The same SMA muscles —

which are nourished by the hot

and cold fluid pumped by the heart

itself — will power the robotic heart,

explains Dr. Mascaro. However, this

work is just in the beginning stages

and does not yet appear in any of

Dr. Mascaro’s papers as of this

writing. SV

Wet Robotics, Image 1

Wet Robotics, Image 2

Mascaro paper on Wet Robotics

Mascaro paper on Wet Robotics

Other University of Utah Robotics

SERVO 09.2006

63