The document discusses the development of robotic fingers and hands. It outlines the need for robotic fingers that can perform delicate tasks like human hands. It then details the development of small, powerful direct-drive motors by SMAC Moving Coil Actuators to enable robotic fingers that can generate forces similar to humans. The motors have been integrated into prototype robotic fingers and hands. Future work involves further miniaturizing the motors and developing control systems to enable dexterous, programmable robotic hands and fingers.
3. The Need for
Robotic Fingers and Hands
• There is a need for robotic fingers and hands that can
perform the same work as human hands do.
• Why?
• Because most work done by humans is done with their
hands.
4. Foxconn’s Disappointments
• Foxconn spent $20,000-$25,000 per robot (according to Singularity HUB) in hopes of replacing
1.5 million assembly workers. According to The Wall Street Journal, Foxconn would have to
spend anywhere from $2.1 billion to over $10 billion for fully automated plants, depending on
the type of robots used. Foxconn's traditional capital spending is below $3 billion.
• Since these robots were equipped with grippers they could not manipulate tools and parts in a
way that could emulate human hands. According to the news stories the robots turned out to
be unusable.
5. Underwriters Labs
• UL - a recent visit to UL
(Underwriter's Labs) resulted in their
asking for a Robot finger that can
swipe a touch screen like a human
and can feed back information about
the work.
• SMAC moving coil actuators can
Soft-Land on a surface and then
lightly push but not in the way a
finger does.
• Most important is the slide action.
• They said such a device does not
exist - but is needed.
6. Why No Robotic Fingers?
• This inability to copy the work done by
human hands has slowed considerably the
expansion of robots in the factory.
• What has been the problem?
7. Forces
• The human finger
can push with a force
at the finger tip of
around 10N. The
finger is small - with
joints in the 20 -
30mm diameter
range. The finger
also can vary forces -
including delicate
ones.
8. Flexibility
• The finger is also flexible and is able to
absorb and adjust to external forces
without breaking - until it reaches the end
of motion.
9. Geared Motor Solution?
• A very good solution for a robot finger
would be a very powerful - small - direct
drive servo motor. A geared motor is not a
good solution since it can be easily
damaged by external forces operating on
it. Direct drive is compliant.
10. Developing Powerful, Small Motors
• The problem is developing a torque high enough in a small package. The “D” motor
shown here is a very good commercial direct drive motor - Swiss made - that puts
out 60mNm at 48 volts and 3 amps. The only problems are:
– The lowest torque joint - the 3rd (or PIP) - requires about 200mNm in order to put
out a resultant 10N force at the finger tip - approximately 30mm from the joint.
– The motor needs to be in the 20 - 25mm diameter range.
A
B
C
D
11. Developing Powerful, Small Motors
• A solution to this problem has been in development at
SMAC. We have progressed starting from a 35mm motor
“A” generating the same torque as the large Swiss
manufacturer of the 3rd motor from the left.
A
B
C
D
12. Developing Powerful, Small Motors
• The “C” motor shown here has a diameter
of 25mm. It runs on 48 volts using a max
current of 1.5 amps.
A
B
C
D
13. Development
• It has a torque of 140mNm and so exerts a resultant force of about
7N at the finger tip.
• The motor is the 3rd integration and is based on SMAC's moving
coil design. It has a working arc of 90 degrees.
14. Motors
• The motor is a servo "partial" motor and is equipped with an SMAC rotary encoder
with 15.5K counts per revolution. It achieves a high torque/diameter ratio due to the
proprietary magnetic circuit design as well as the coil design. Intellectual property
rights have been addressed.
• Another - larger motor (35MM) acts as the 2nd joint (MCP FLex). It puts out a torque
of 630 mNm and thus can exert a resultant force of 7N at the tip. Both motors
move +/- 45 degrees - as the human joints do.
• The 3rd motor operates as the first joint (MCP Abd). This is the joint that moves side
to side. This motion is based on SMAC’s moving coil linear design and already meets
the required parameters.
• These motors have been integrated into a structure that allows them to operate and
cover the same movement capability as a human’s finger. The structure weighs
approximately 350 grams and is physically about 1.5 x larger than the average male
finger.
15. SMAC Corporation
• SMAC Corporation has had a long history of development of its MCA (Moving Coil Actuator)
technology. During this time patented "Soft-Land" and Programmable force technology has
been developed and are now being used around the world by major Consumer Electronic -
Electronic automation, automotive, packaging, and medical companies.
• This technology can be directly applied to the finger and later the hand. So soft bumping into
materials can be realized as well as programmed applied force.
16. Smaller and Smaller
• What happens next? Toyota's robotics group has viewed the motor
and mentioned that Japanese joints are more on the order of
20MM. (I was a manager at a large Japanese components supplier
to Toyota - so this comment was expected).
• That is a challenge.
• We are in the midst of the 4th and 5th FUMO (functioning model)
designs which will again increase torque by 50-60% (actually, the
fourth iteration just gave us another 35%). This should allow us to
shrink the motors so that we can approach the actual size of even a
Japanese woman's finger. (We have several of these in my family by
the way). We also must drop current to the 500-750 mAmps range.
• Eventually, our targets are motors producing the correct torque (so
that we can produce the required 10N at the tip) using a maximum
600 mAmps/DC current at 48 VDC.
17. Thumb and Controllers
• We are also busily laying out the
thumb. That’s an added rotational axis.
• A multi-axis controller based on very
small slave controller/amp - which is a
SMAC commercial product already and
a Master Controller are in early test.
The MC will eventually handle up to 50
axis - there are 16 basic axis in our
proposed SMAC hand.
• So- by the fall a finger/thumb
combination controlled by our MC/CBC
will be operating and doing work inside
SMAC. Our target is to produce a
"hand" with controllers - at a price of
around $5000. Time will tell if this is
achievable - although a good indicator
is that a current single axis SMAC
actuator with built in controller can be
purchased for around $600 vs
$3000 15 years ago.
18. The Future
• Fast programming technology is
needed. Copy movement programming
is commonplace. We are investigating
other methods - such as tracking via
glove and non contact
tracking. Advances are being made.
• The possibility of a user-friendly
prosthetic hand is real. This can be
programmed to accomplish many
common day actions. IP is out on this.
• And - the technology opens up the
possibility for very small SCARA like
robots - much better suited for small
component assembly - which takes us
back to the Foxconn reference in the
beginning.
19. Conclusion
• A key hurdle holding back the application of
robots in the factory has been the lack of
technically capable robotic fingers and
hands.
• It is now quite possible that this hurdle can be
overcome. If this proves true the expansion of
robots into work areas hitherto barred by the
lack of robot dexterity will take place. That is
a big step.