Friction & Workspaces CPSC 599.86 / 601.86 Sonny Chan University - - PowerPoint PPT Presentation

friction workspaces
SMART_READER_LITE
LIVE PREVIEW

Friction & Workspaces CPSC 599.86 / 601.86 Sonny Chan University - - PowerPoint PPT Presentation

Friction & Workspaces CPSC 599.86 / 601.86 Sonny Chan University of Calgary Todays Agenda Rendering surfaces with friction Exploring large virtual environments using devices with limited workspace [From C.-H. Ho et al., Presence


slide-1
SLIDE 1

Friction & Workspaces

CPSC 599.86 / 601.86 Sonny Chan University of Calgary

slide-2
SLIDE 2

Today’s Agenda

  • Rendering surfaces with friction
  • Exploring large virtual

environments using devices with limited workspace

[From C.-H. Ho et al., Presence 8(5), 1999.]

slide-3
SLIDE 3

Coulomb Friction

What is it, and how do we render it?

slide-4
SLIDE 4

Coulomb Friction

  • Friction force proportional to normal force
  • Static (sticking) friction:
  • Kinetic (sliding) friction:

Ff = µFN Fs ≤ µsFN Fk = µkFN

slide-5
SLIDE 5

Rendering Friction

  • Basic case:

avatar surface

µ = µs = µk

FN = −Fc

Fc

slide-6
SLIDE 6

Rendering Friction

  • Do we move the avatar?

surface

α FN = −Fc cos α Ff = −Fc sin α

???

sin α > µ cos α tan α > µ

Move avatar if

slide-7
SLIDE 7

Rendering Friction

  • How far do we move the avatar?

α α = tan−1 µ

surface how far?

slide-8
SLIDE 8

Friction Cone

α α = tan−1 µ

slide-9
SLIDE 9

Friction Cone in 3D

device position avatar

slide-10
SLIDE 10

Static & Kinetic Friction

  • Case where static and kinetic

friction are distinct:

  • Two different coefficients of

friction gives two friction cones:

  • static friction cone
  • dynamic friction cone

α = tan−1 µ β = tan−1 µk β α

µs 6= µk

slide-11
SLIDE 11

Static & Kinetic Friction

  • Do we move the avatar?

β α

slide-12
SLIDE 12

Static & Kinetic Friction

  • Do we move the avatar?

β α

how far?

slide-13
SLIDE 13

Static & Kinetic Friction

  • Do we move the avatar?

β α

slide-14
SLIDE 14

Static & Kinetic Friction

  • Do we move the avatar?

β α

slide-15
SLIDE 15

Static & Kinetic Friction

  • Do we move the avatar?

β α

slide-16
SLIDE 16

Static & Kinetic Friction

  • Do we move the avatar?

β α

slide-17
SLIDE 17

Coulomb Friction: Summary

  • Friction force is proportional to normal (contact) force
  • Construct friction cone(s) from coefficients of friction
  • With a proxy algorithm, we can render forces of static and kinetic friction
  • In general, μs > μk
  • When do we switch between static and kinetic friction cones?
  • This is the crux of Assignment #2, Part II
slide-18
SLIDE 18

Friction Demo

slide-19
SLIDE 19

Workspace Management

slide-20
SLIDE 20

Your Haptic Device

How much workspace does the Novint Falcon haptic device have?

< 10 cm

slide-21
SLIDE 21

How can we deal with this limitation?

slide-22
SLIDE 22

Strategy #1: Absolute Position Scale

rA = s rD/W + rW

  • natural, direct mapping
  • navigate large virtual

environments

  • or small ones
  • loss of spatial resolution
  • serious adverse effects on haptic

performance and stability

  • fixed workspace

PROS 🙃 CONS 🙂

slide-23
SLIDE 23

How does your computer mouse interface handle this?

< 25.6”

slide-24
SLIDE 24

( ) ( ) ⋅ trajectory of the avatar trajectory of the end-effector

vd

va ⎝ ⎠ ⎛ ⎞ ⋅

trajectory of the device trajectory of the avatar

[From F. Conti & O. Khatib, Proc. IEEE World Haptics Conference, 2005.]

slide-25
SLIDE 25

Strategy #2: Ballistic Control

slide-26
SLIDE 26

∆rA = s(vD) ∆rD/W

Strategy #2: Ballistic Control

  • adaptive spatial resolution
  • span large workspaces when

moving fast

  • precise control when slow
  • workspace gets shifted or offset
  • essentially requires “indexing” to

correct offset

  • can use a button or switch on device

PROS 🙃 CONS 🙂

slide-27
SLIDE 27

How about mimicking these devices?

slide-28
SLIDE 28

Strategy #3: Rate Control

  • infinite workspace!
  • no indexing needed
  • viable for Assignment #3
  • cannot perform fast motions in

different directions

  • scratching a rough surface
  • tapping a hard object
  • force feedback is awkward

PROS 🙃 CONS 🙂 vA = s rD/W

slide-29
SLIDE 29

Workspace Drift Control

François Conti & Oussama Khatib, 2005 rd ra fa

workspace drift workspace drift trajectory of the avatar

slide-30
SLIDE 30

Main Ideas

  • Take advantage of our imprecise

proprioception

  • Move the centre of the virtual

workspace to coincide with the centre of the device workspace

  • Use the device operator’s own

motion to apply workspace drift

slide-31
SLIDE 31

Initial condition

⎝ ⎠ ⎛ ⎞ ⋅ ⋅ ⋅

physical workspace boundaries of the device. virtual workspace avatar virtual object 2-dof haptic device graphical display

  • f the avatar

[From F. Conti & O. Khatib, Proc. IEEE World Haptics Conference, 2005.]

slide-32
SLIDE 32

Motion toward object

⎝ ⎠ ⎛ ⎞ ⋅ ⋅ ⋅

[From F. Conti & O. Khatib, Proc. IEEE World Haptics Conference, 2005.]

slide-33
SLIDE 33

⎝ ⎠ ⎛ ⎞ ⋅ ⋅ ⋅

velocity and direction of local interaction avatar-object

vwa

the workspace drift.

Interacting with object

Want to move workspace centre to location of interaction

[From F. Conti & O. Khatib, Proc. IEEE World Haptics Conference, 2005.]

slide-34
SLIDE 34

⎝ ⎠ ⎛ ⎞ ⋅ ⋅ ⋅

direction of hand drift.

  • vwd

vwa

Drifting the workspace

[From F. Conti & O. Khatib, Proc. IEEE World Haptics Conference, 2005.]

slide-35
SLIDE 35

Workspace Drift Control

  • Drift only occurs when the operator’s

hand is in motion

  • Velocity of workspace drift is proportional

to the speed of hand motion

  • Can be combined with ballistic control or

rate control if needed

vW = s kvDk rD/W

Figure 7 - Moving an avatar along a straight line during a workspace drift requires compensating for the shift by slightly moving the end-effector in the

  • pposite direction of the workspace drift.

motion of the avatar motion of the hand avatar workspace drift

slide-36
SLIDE 36

Workspace Management: Summary

  • Many feasible strategies exist for managing a device’s limited workspace
  • Absolute position scale is easy and direct, but loss of resolution is bad
  • Ballistic control works well for mice, but requires “indexing” or clutching
  • Rate control makes haptic interaction difficult, but can be effective when used

carefully in specific workspace regions

  • Workspace drift control tries to combine the best qualities of these strategies