Rotational Skin Stretch for Wearable Haptics: A new approach to tactile display

1
Rotational Skin Stretch for Wearable Haptics A
new approach to tactile display Biomimetics and
Dexterous Manipulation Laboratory Karlin Bark
2
Motivation
I

• Goal Develop a tactile display that can be
  mounted anywhere on a users body to provide a
  rich communication channel between the user and
  the environment.

ntroduction

• Position or Motion Feedback for Motion Training
  Exercises

• Haptic Feedback for amputees

3
Portable and Wearable Haptic Devices
I

• Vibrotactile Devices
• Easy to implement
• Good at event cues
• Annoying
• Desensitization

ntroduction
4
Current Skin Stretch Devices
I

• Focus on fingertip skin stretch
• None for large scale deformation
• Unexplored area of haptic feedback

ntroduction
Tiny Actuators deflect to induce skin stretch
2
5
Skin stretch haptic feedback
I

• Skin stretch is a natural part of our sense of
  touch
• It is a known contributor to our sense of
  proprioception and kinesthesia Edin1995,
  Collins2005

ntroduction
skin stretches around joints
6
Skin sensors
I
ntroduction
7
Questions
I

• Research has focused on developing a way to use
  skin stretch for a wearable interface
• How can we produce skin stretch? Design
• What happens to our skin when it is stretched?
  Mechanics
• How well can we perceive skin stretch?
  Perception
• Is there an application where skin stretch is
  appropriate? Application

ntroduction
8
How can we produce skin stretch?
9
Type of Motion
D
Linear
esign
People qualitatively preferred rotational stretch
due to the higher perceived magnitudes
10
Type of Contact
D
How many contact points? TWO
esign
Testing patterns of contact
11
Type of Contact
D
How to attach to skin?
esign
Avoiding slip on skin was a challenge. To reduce
the likelihood of producing slip contact, a skin
safe, double sided adhesive was used.
Contact pads attached with tape
12
Type of Contact
D
General size and shape of contact pads

• Requirements
• Small enough to fit on the body in a compact form
• Large enough to reduce slip
• Large enough to avoid painful sensations
• Round to prevent contact with sharp corners

esign
13
Location of Stretch Application
D
Where on the body? arms? legs? shoulder?
The Forearm A relatively large concentration
of skin stretch receptors
esign
Vallbo1995
14
How can we produce skin stretch?
D

• 2 Point, rotational stretch
• Rotates about a central axis
• Average range of motion /- 45 deg
• Torques applied range from 0 - 200 mNm
• Adheres to skin using skin safe adhesive
  (Red-e-Tape)

esign
15
End Effector Design
D

• Two different end effector designs were used for
  testing Fixed and Free Rotating contact pads
• By allowing the contact pads of the end effector
  to rotate freely, the strains and stretch induced
  differs
• More comfortable, less strain, decreased quality
  of overall perception

esign
16
Benchtop Skin Stretch Device
D

• Developed to apply controlled, rotational skin
  stretch

esign
indicate degrees of freedom
17
Wearable Skin Stretch Device
D
esign
Torque - 0.6 Nm
15-150 deg/s
115 g
1 deg resolution
18
What happens when skin stretch is applied?
19
Goal Mechanics of Skin Stretch
M

• What happens to our skin when skin stretch is
  applied?

• Measured the amount of skin stretch using two end
  effectors Fixed and Free
• What differences in the method of inducing skin
  stretch affect our perception?

echanics
Motion capture system in Prof. Delps Motion
Laboratory at the Clark Center
20
Methods Experiment Setup
M

• A grid of seventy 1.5 mm markers were placed on
  the forearm to measure the displacement of the
  skin as rotational stretch was applied

echanics
21
Methods Stimulus
M

• A ramp rotation of 10, 20, and 30 degrees was
  applied in ascending order at a speed of 80 deg/s
  using both the fixed and free rotating contact
  pads

echanics
22
Data Analysis
M

• Three parameters were analyzed
• Absolute displacement of markers
• Torque applied Measured from force sensor
• Calculated strains and strain energy
• Relation of torque
• to displacement

echanics
23
Results Absolute Displacement
M

• Fixed end effector results in larger
  displacements, particularly at points farther
  from the contact points

echanics
24
Results Torque
M
echanics
arm moved slightly

• The torques applied using the fixed pads are
  higher
• Corresponds to higher magnitude perception

24
25
Results Strain Energy
M

• As expected, the overall strain energy correlates
  with theincrease in torque and skin displacement
• Fixed contacts also show local regions of higher
  strain energy near the contacts

echanics
E 320 kPa Agache80
26
What happens when skin stretch is applied?
M

• A variety of mechanical factors are changing and
  contribute to our perception
• Absolute displacement of skin
• Possibly activating a greater quantity and region
  of skin mechanoreceptors Olausson00
• Magnitudes of torque
• Results in higher perceived magnitude of stretch
• Amounts of strain energy
• Possibly increased mechanoreceptor activity
  Dandekar98
• We can utilize this knowledge, combined with our
  perception studies using both end effectors to
  develop a better stimulus

echanics
27
How well do we percieve skin stretch?
28
Basic Psychophysical Metrics
P

• Absolute Detection Threshold
• The minimum stimulus level necessary for
• a stimulus to be detected
• Humans known to accurately detect linear
• directions of skin stretch at movements as
• low as 0.3 mm Olausson98
• Discrimination Threshold
• The minimum difference required for a user to
  distinguish between two stimuli
• Can you distinguish the stretch associated with
  10 degrees of rotation from 12 degrees?
• with movement to distant location in between

erception
29
Methods Discrimination Threshold
P

• The Three-Interval One-Up Three-Down Method
  Levitt71
• There are three stimulus presentations per trial
• Two of the intervals contain the reference
  stimulus
• One randomly-selected interval contains the test
  stimulus
• Subjects task is to indicate which interval (1,
  2 or 3) contains the signal that is larger
• Tested two different reference stimuli at two
  different angular velocities
• 10, 30 degrees
• 20 deg/s, 80 deg/s

erception
30
Methods 3 Up 1 Down
P
erception
31
Results Discrimination Threshold
P
erception

• Discrimination threshold is significantly
  different (plt4x10-5) at a reference of 10 degrees
  compared to 30 degrees
• Changing velocity had little effect on thresholds

32
Implications
P

• Discrimination thresholds change over the range
  of rotation
• Changing the velocity has little effect on static
  perception of stimulus
• How important is velocity to the stimulus
  overall?
• Provided a sense of stretch resolution, but
• Discrimination thresholds vary depending on
  experiment setup and concentration level
• Not ideal parameters for practical use
• Highly controlled experiment
• Tested our ability to compare two distinct stimuli

erception
33
Skin Stretch Feedback Mapping
P

• Goal Determine how well users can linearly map
  skin stretch feedback to rotational position of
  virtual arm
• Why? If users can map skin stretch feedback
  linearly to position, it can be used for feedback
  of motion or proprioceptive information
• Develop a task where users are asked to correlate
  skin stretch feedback with the rotational motion
  of a virtual object

erception
34
Methods
P

• Two different tasks were tested
• A task where the subject actively positioned the
  virtual object using the feedback
• A where the subject sat passively as feedback was
  given and reported the perceived position

erception
35
Methods Active Positioning Task
P
erception

Afferent/Efferent Loop
35
36
Methods Passive Perception Task
P
erception
No Loop!
37
Methods
P

• Experiment Details
• Subjects were given 1-2 minutes of training to
  learn the mapping (vision and haptic feedback are
  given) and control
• Subjects set their desired range of rotation
• Test stimuli spanned positive and negative
  degrees of rotation in increments of 2 degrees
• Ex. Range of -40 to 30 deg Stimulus levels of
  -40, -38, -36 ... 26, 28, 30

erception
Active
Passive
38
Results Active Positioning, one subject
P
erception
39
Results Active Positioning
P

• Subjects are able to linearly map perceived skin
  stretch to rotation of a virtual object
• Average residuals 5.8 deg

erception
40
Results Passive Perception
P

• Subject performance degrades when not in control
  of system
• There is a region of stretch near 0 degrees where
  subjects have difficulty in distinguishing the
  stimulus from zero.

erception
Fixed Contact
Free Contact
41
Results Passive, Experienced subjects
P

• However, results also indicate that with
  experience, subjects can learn to use the
  feedback in a passive setting as well.

erception
42
Skin stretch mapping findings
P

• How well do users linearly map skin stretch
  feedback to rotational position of virtual arm
• Active within 6 deg, (R2 values of 0.92)
• Passive not as good (too many direction errors)
• With several hours experience, 5-8 deg
• Clear difference between fixed and free end
  effectors more ambiguity with free at zero deg
• Correlates with skin stretch analysis
• Qualitatively, users preferred free rotating

erception
43
How well can people perceive skin stretch?
feedback?
P
highest level
erception
perception level

• Skin stretch can be especially effective in
  applications where users can use it as feedback
  in response to voluntary motions

44
For what applications is skin stretch feedback
appropriate?
45
Skin Stretch for Practical Use
A

• Can we use haptic feedback to provide the
  position of a robotic arm?
• Proprioceptive and kinesthetic feedback for
  amputees

pplication
46
Experiment Methods
A

• Goal Provide position information using haptic
  feedback
• Vibration and skin stretch
• Task Move virtual hand a specific distance with
  only haptic stimulation to indicate position
• Move a cursor by using a single axis force sensor
  as an input device
• Haptic feedback provided (vibration or skin
  stretch)
• Cursor is attached to virtual object

pplication
47
Method Cursor Dynamics
A

• Virtual object dynamics analogous to a 2-link
  robot arm constrained to single axis motion

pplication

• Object has configuration dependent dynamics
  making it difficult to use a purely open loop
  strategy

48
Method 1 Vibration Feedback
A

• Small C2 vibrotactor was strapped to subjects
  forearm, near the elbow joint

• Operating characteristics
• 250 Hz
• Logarithmically varying amplitude of cursor
  position
• consistent with previous workMurray2003

pplication
49
Method 2 Skin Stretch Feedback
A

• Benchtop skin stretch device
• Attached to forearm, near elbow joint

pplication
50
Experiment Procedure
A

• Subject presses a push button when desired cursor
  position is reached and stable

• Cursor starting position appears on screen

• Subject is asked to move cursor a specified
  distance

• Using the force sensor as an input, the subject
  can control the position of the cursor
• Haptic feedback correlating to cursor position is
  fed back to the subject

• Actual and desired cursor position are then shown
  to the subject and the next trial begins

Move 3 Units LEFT
pplication
actual end
starting point
desired end
51
Experiment Procedure
A

• Tested 4 feedback conditions
• Training Procedure before each condition
• Training with visual and haptic feedback for one
  minute
• 10 practice trials with haptic but no visual
  feedback were given
• 36 trials per condition per subject

pplication
52
Results Data Analysis
A
error bars standard error
pplication

• Addition of haptic feedback results in
  significantly lower absolute errors (10 subjects,
  ANOVA plt0.005 over 340 trials)
• Skin stretch feedback results in significantly
  smaller errors overall

53
Results Velocity Analysis
A
error bars standard error
pplication

• Subjects were only able to have a sense of
  velocity when skin stretch feedback was provided
  (plt1x10-9 over 340 trials)

54
For what applications can skin stretch be used?
A

• Skin stretch is effective in providing motion
  feedback
• Conveys a sense of both velocity and position
  within a single stimulus
• More intuitive mapping than vibration for motion
• Can convey directions
• Preferred qualitatively
• Combined with our other results, skin stretch
  shows promise in feeding back motions and
  velocities of a single object

pplication
55
Summary of Contributions
C

• Developed a new method of providing haptic
  feedback, rotational skin stretch
• Constructed a benchtop and wearable skin stretch
  device
• Provided framework towards understanding what
  contributes to our perception of skin stretch
• Measured mechanical properties as skin is
  stretched
• Quantified perception characteristics of skin
  stretch
• Difference thresholds and effect of velocity on
  static perception
• Active vs. Passive perception performance
• Shown that skin stretch can be effective for
  motion (proprioceptive) feedback
• Provides a sense of velocity and static position

onclusions
56
Future Avenues to Explore
C

• Continue to study the benefits of using skin
  stretch for myoelectric prosthesis and other
  applications
• Motion training applications are currently being
  studied
• Re-design the wearable
• device to be even more
• compact and lightweight
• Explore effectiveness of
• skin stretch on other parts
• of the body
• Test different methods of
• skin stretch (motions,
• patterns)
• Test skin stretch in a
• more dynamic environment

onclusions
57
Future Work
C

• Re-design a better end effector to combine the
  best aspects of the fixed contact pads and free
  contact pads
• Improved perception at low rotations
• Less energy/torque/displacement at higher
  rotations for increased comfort

onclusions
courtesy of Gayle Lee
58
Acknowledgements

• Professor Mark Cutkosky
• Defense committee Drs. Scott Klemmer, Larry
  Leifer, Joan Savall and Sheri Sheppard
• My family and friends
• The BDML
• Jason Wheeler, Li Jiang, everyone who was a
  participant!!
• National Science Foundation, Tekes, Stanford
  Mechanical Eng. Department
• Scott Delp, Sam Hamner, Dustin Hatfield

59
References

• 1 Jones, L., Nakamura, M., and Lockyer, B.,
  Development of a Tactile Vest, Proceedings of the
  12th International Symposium on Haptic Interfaces
  for Virtual Environment and Teleoperator Systems
  (HAPTICS04)
• 2 Pasquero, J., Luk, J., Levesque, V., Wang,
  Q., Hayward, V., and MacLean, K. E. (2007)
  Haptically Enabled Handheld Information Display
  with Distributed Tactile Transducer. IEEE
  Transactions on Multimedia, 9(4), pp 746-753
• 3 B. Edin and N. Johansson. Skin strain
  patterns provide kinaesthetic information to the
  human central nervous system. Journal of
  Physiology, (487)243251, 1995
• 4 D Collins, K Refshauge, G Todd, and S
  Gandevia. Cutaneous receptors contribute to
  kinesthesia at the index ?nger, elbow, and knee.
  Journal of Neurophysiology, 9416991706, May
  2005.
• 5 Agache, P, Monneur, C, Lévéque, JL, de Rigal,
  J Mechanical properties and Young's modulus of
  human skin in vivo. Arch Dermatol Res 1980 269
  221232
• 6 H Olausson. Tactile directional sensibility
  peripheral neural mechanisms in man. Brain Res,
  866(1-2)178187, Jun 2000
• 7 Dandekar,K., Srinivasan, M. A., Role of
  mechanics in tactile sensing of shape. Touch Lab
  Report 2. rle tr-604. mit. cambridge. Touch Lab
  Report 2, RLE TR-604, Jan 1997.
• 8 Levitt, H. (1971). Transformed Up-Down
  Methods in Psychoacoustics. The Journal of the
  Acoustical Society of America, 49(2), 467-477.
• 9 K Kuchenbecker, N Gurari, and A Okamura.
  E?ects of visual and proprioceptive motion
  feedback on human control of targeted movement.
  Rehabilitation Robotics, Jan 2007.
• 10 A. Murray, R. Klatzky, and P. Khosla.
  Psychophysical characterization and testbed
  validation of a wearable vibrotactile glove for
  telemanipulation. Presence Teleoperators and
  Virtual Environments, 12(2)156 182, April
  2003.

60
the end