All About Thresholds: An Overview of Haptic Perception of ...ece511/lecture...Haptics-e, 4(1). ENACTIVE ’07 Conference, Grenoble, France, November 20, 2007 Hong Z Tan • 12 Stiffness/Compliance

Hong Z Tan • 1ENACTIVE ’07 Conference, Grenoble, France, November 20, 2007

Hong Z. TanAssociate Professor

Electrical and Computer EngineeringMechanical Engineering (courtesy)Psychological Sciences (courtesy)

Faculty FellowEnvision Center for Data Perceptualization

Purdue [email protected]

All About Thresholds:An Overview of Haptic Perception of Mechanical Properties


Enactive Knowledge“Enactive knowledge is neither symbolic nor iconic. It is natural and intuitive, based on

experience and the perceptual consequences of motor acts. Enactive knowledge is information gained through perception-action interactions with the environment.”


Motivation

My own research focuses on haptic perceptionAny effective enactive interface should capitalize on our exquisite sensitivity to object properties through physical interactions with the environmentUnderstanding our sensitivity to mechanical properties can lead to enactive interfaces that match the human sensory-motor capabilitiesIn this context, haptic perception is an integral component of enactive knowledge


Outline

Overview of human discrimination thresholds for mechanical properties of objectsAn example of using thresholds for interface engineeringPredicting perception of complex signals from the perceptual thresholds for the components


When you use a force display…

F0 (force)magnitudedirection

K (stiffness / compliance)x (displacement / length)

linearangular (α, joint-angle position)

B (viscosity)

L&&& ++++= xMxBKxFF 0

Discrimination threshold The smallest difference between two stimuli that can be reliably detected;

Just Noticeable Difference(JND).


Force Magnitude Perception

Pang, Tan, & Durlach (1991). Manual discrimination of force using active finger motion. Perception & Psychophysics, 49(6), 531-540.



Force JND with Fixed Displacement

Problem:Confounding

Mechanical workcues! Average JND: 8%


Force JND with Roving Displacement

Randomizedmechanical work cues

Average JND:7% (fixed displacement)

14% (roving displacement)


Related Work by Lynette Jones

Photo courtesy of Lynette Jones

Contra-lateral limb matching of force; JND = 7%(Jones, 1989)

Jones (1989). Matching forces: Constant errors and differential thresholds. Perception, 18, 681-687.


Force Direction Perception

Barbagli, Salisbury, Ho, Spence, & Tan (2006). Haptic discrimination of force direction and the influence of visual information. ACM Transactions on Applied Perception, 3(2), 125-135.

Force-direction JND:H only: 25.6°

H with congruent V: 18.4°H with incongruent V: 31.9°



Force-direction JND is Independent of Reference F Direction

Tan, Barbagli, Salisbury, Ho, & Spence (2006). Force-direction discrimination is not influenced by reference force direction (Short Paper). Haptics-e, 4(1).


Stiffness/Compliance Perception

Tan, Durlach, Beauregard, & Srinivasan (1995). Manual discrimination of compliance using active pinch grasp: The roles of force and work cues. Perception & Psychophysics, 57(4), 495–510.

Jones & Hunter (1990). A perceptual analysis of stiffness. Experimental Brain Research, 79, 150-156.

Jones & Hunter (1990)

Contra-lateral limb matching of stiffnessStiffness JND = 23%

L&&& ++++= xMxBxKFF 0


Average Compliance JNDs(1,536 trials per point)

Preloaded spring;fixed displacement;

confounding force cues;no more mechanical work cues

Compliance JND:15% − 99%


Terminal Force Hypothesis

⎟⎟⎠

⎞⎜⎜⎝

⎛Δ+

Δ=Δ

ΔΔ

00

00

00

0

0

/)(1/)(

2)(

JND force equivalent )(JNDcompliance )(

CCCC

CDF

FC

Compliance JND:15% − 99%

Terminal-Force JND:5.15 % (s.d.=0.76%)


Length Perception

Durlach, Delhorne, Wong, Ko, Rabinowitz, & Hollerbach (1989). Manual discrimination and identification of length by the finger-span method. Perception & Psychophysics, 46(1), 29–38.



]598.0log250.0)[3.4/(L)( 000 +−=Δ LLJND

JND=1mm@ 10mm ref

(10%)

JND=2.4mm@ 80mm ref

(3%)


Finger Joint-Angle Perception (α)

Tan, Srinivasan, Reed, & Durlach (2007). Discrimination and identification of finger joint-angle positions using active motion. ACM Transactions on Applied Perception, 4(2), Article 10, 14 pp.

PIP joint JND: 2.5°, regardless of PIP or MCP joint positionMCP joint JND: 2.0°−2.7°, better when PIP/MCP is straight

Derived JND at fingertip: ≈ 2.2 mm



Perception of Viscosity

Contralateral limb-matching procedure


Jones & Hunter (1993). A perceptual analysis of viscosity. Experimental Brain Research, 94, 343-351.

ViscosityJND = 34%


Summary of JNDs

VariableForce

Magnitude ………..Direction ………....

Stiffness ………………Displacement

Linear ………….....Angular …………...

Fingertip …......Viscosity ……………....

JND

5−10 %25−35°22 %

3−10 %2.0−2.7°2.2 mm34%



Outline

Overview of human discrimination thresholds for mechanical properties of objectsAn example of using thresholds for interface engineeringA study on perception of complex waveforms, using threshold measurements


Haptic Data Compression using Force JND

Hinterseer & Steinbach (2006). A psychophysically motivated compression approach for 3D haptic data. Proceedings of Haptics Symposium ‘06, 35-41. Slide courtesy of Peter Hinterseer

t

Do nothing Transmit new value

1-D

2-D

Fprevious

Fnew

deadzone deadzone

Fprevious

Fnew


(cont.)

3-D spherical deadzoneUser testing showed that packet rate can be reduced by up to 90% for the transmission of force magnitude

Further reduction in packet rate may be achieved by incorporating the discrimination thresholds for force direction


Outline

Overview of human discrimination thresholds for mechanical properties of objectsAn example of using thresholds for interface engineeringA study on perception of complex waveforms, using threshold measurements


Perception of Texture Gratings

Mini-stick3-Degrees of FreedomNominal 1μm resolutionForce commands updated at 2kHz

Question:Can the perception of complex texture gratings be predicted from that of simple texture gratings?

Cholewiak & Tan (submitted). A Frequency Analysis of the Detectability and Discriminability of Virtual Texture Gratings. Submitted to ACM TAP.


Two Types of Texture Gratings

Sinusoidal grating Square-wave grating

)2sin()(sin λπxAxh =⎩⎨⎧

≤−>

=0)2sin(,0)2sin(,

)(λπλπ

xifAxifA

xhsquare

x

y

z x

y

z


Components of a Square-Wave

Fundamental Frequency Component A*4/π* sin(2πx/λ)

3rd Harmonic Component A*4/(3π)* sin(3*2πx/λ)

5th Harmonic Component A*4/(5π)* sin(5*2πx/λ)

A





A*4/π

A

> A


Sinusoidal Threshold Results


Square-Wave Threshold Results


Ratio Results


But …

Our results means that when the amplitudes are very small (i.e., near detection threshold), a square-wave grating should feel like a sinusoidal grating!Indeed, we were able to show that at low amplitudes, a square-wave grating was indistinguishable from a sinusoidal grating untilthe amplitude of its 3rd harmonic component exceeded detection threshold


Concluding Remarks

Detection/discrimination thresholds go a long way towards our understanding of human perception of mechanical properties of real or virtual objectsThresholds for simple signals can be used to predict those for complex signalsThey are useful for determining the resolution of human-computer interfaces, for designing data compression algorithms, etc.Thresholds can be useful for the design of enactive interfaces!


Acknowledgment

CollaboratorsFederico BarbagliG. Lee BeauregardSteven A. CholewiakNat DurlachCristy HoXiao-Dong PangCharlotte ReedKen SalisburyCharles SpenceMandayam Srinivasan

Funding AgenciesOffice of Naval ResearchNational Institutes of HealthFairchild FoundationMIT G.Y. Chu FellowshipNaval Air Warfare Center Training Systems DivisionOxford McDonnell-Pew Centre for Cognitive NeuroscienceFord Motor CompanyNational Science Foundation

Documents

All About Thresholds: An Overview of Haptic Perception of ...ece511/lecture...Haptics-e, 4(1). ENACTIVE ’07 Conference, Grenoble, France, November 20, 2007 Hong Z Tan • 12 Stiffness/Compliance