Sensors for Prosthetic Grasping

1
Sensors for Prosthetic Grasping

• Seth A. Frear
• December 7, 2006

2
What is so difficult about grasping something
with a prosthetic hand?
3
The main difficulty comes from determining
whether or not something is slipping.
4

• The quality of a prosthetic arm is determined
  not only by the mechanical complexity, but also
  by the presence of a sensory system that will
  allow for grip optimization and the ability to
  quickly and efficiently complete tasks.

5
Examples of Robotic Hands
6
Different approaches to detecting and
compensating for slip

• Vibration sensors
• Force sensing resistors
• Force vector control
• Optical sensors for motion detection

7
Three papers

• Experimental development of a sensory control
  system for an upper limb myoelectric prosthesis
  with cosmetic covering (Andrea Tura Claudio
  Lamberti Angelo Davalli Rinaldo Sacchetti)
• - force sensing resistors
• - optical sensors
• Thick-film force, slip and temperature sensors
  for a prosthetic hand (A Cranny D P J Cotton P
  H Chappell S P Beeby N M White)
• - vibration sensors
• The tactile slip sensor Integration of a
  miniaturized sensory device on a myoelectric hand
  (G Puchhammer)
• -force vector control

8
What is myoelectric?

• A myoelectric signal is an electrical impulse
  that produces muscle contraction in the body
• Typically these signals have frequencies ranging
  from a few hertz to about 300 hertz and voltages
  ranging from around 10 microvolts to 1 millivolt

9
Paper one

• The first paper discusses the fitting of a 2
  degree of freedom Otto Bock hand with a sensorial
  system

10

• Without a sensory system visual feedback is the
  only thing the wearer can use to control the
  prosthesis. This can be tiring and slow
• To improve upon this the authors of this paper
  wanted to develop a system with involuntary
  feedback so the wearer would not have to
  explicitly control every aspect of the gripping
  motion

11
Visual feedback vs. involuntary feedback
12

• The sensory control system the authors used
  incorporates force-sensing resistors.
• -Force sensing resistors consist of two thin
  films of polymers, one is covered by a
  network of interlaced electrodes and the other
  by a semiconductor material. When no force is
  applied the resistance is high when force is
  applied, the semiconductor comes is contact with
  the electrodes creating a drop in resistance.

13
Force-sensing resistor setup
14
Operation of the sensory hand

• Myoelectric impulse is generated by flexing
  muscle
• On receiving the impulse the hand begins to close
• The closing action stops once a pre-determined
  contact threshold is reached
• Once contact has been reached automatically , the
  user can further increase grip strength by
  flexing muscle again.

15
What about slipping?
16
Optical sensors were incorporated

• If motion is detected the grip strength of the
  hand is automatically increased
• The optical sensors allow the contact threshold
  to be lowered so delicate objects can be gripped
  without damage

17
Examples
18
Problems with this approach

• Robustness and durability of the FSR
• Varying of the film offset of the FSR resulting
  in inaccuracies in the signals
• Optical sensors only function well in optimal
  light setting
• Optical system was not actually installed on the
  prosthesis (it was too big)

19
Paper two

• The second paper discusses the use of thick-film
  piezoresistive static strain sensors for grip
  strength and piezoelectric dynamic sensors
  (vibration sensors) to detect slipping

20

• The sensors are located on stainless steel
  cantilever structures fastened to the finger tips
  of the prosthetic hand
• Temperature sensors are also included to allow
  for temperature compensation of the force sensors
  as well as prevent thermal damage to the
  prosthesis

21
Sensor setup
22
Static force sensor average normalized output
23
Dynamic and thermal sensors
24
Strain distribution
25
Problems with this approach

• Delicate sensor setup
• Inadequate sensor sensitivity
• Sensor drift
• Very smooth surfaces may not provide enough
  vibration for the sensor to detect

26
Paper three

• The third paper discusses the use of force vector
  control in the gripping of an object
• Force vector control has actually been
  implemented in prosthetic hands currently in use

27
Force vector control

• In most cases the coefficient of friction of an
  obstacle is not known

28
A common range of coefficients of friction are
established creating a force cone as illustrated
below
29
As more weight is added the force vector changes
but is still within the friction limits
30
The weight is about to exceed the friction limits
31
To keep the object from slipping additional
gripping force is exerted
32
Problems

• Does not work well with objects that are very
  smooth or rough (out of the coefficient of
  friction range)

33
Conclusion

• All the examples discussed today use automatic
  grip adjustment no sensory feedback to the
  actual person controlling the hand
• A method of transmitting feeling to the person
  controlling the prosthesis would be optimal