KnowledgeDriven,Interactive Animation of Human Running

1
Knowledge-Driven,Interactive Animation of Human
Running

• Armin Bruderlin
• ATR Meida Integration Communications Research
  Labs
• Tom Calvert
• Simon Fraser University
• Graphics Interface'96, Proceedings, Toronto,
  Canada, May 1996, pp. 213-221
• ReporterPo-Yen Shen

2
Outline

• Introduction
• Bipedal Running Concepts
• Running Algorithm
• Results
• Conclusions

3
Introduction
High Level
Low level
4
Introduction

• Some higher,above joint-manipulation level,motion
  control to animate human locomotion.
• 1.Inverse kinematics approach
• 2.Physically-based approach.
• 3.Hybrid Kinematic-dynamic approach.
• Drawbacksnot perform in real time or
• plain and lacking expression

5
Introduction

• This paper introduces an algorithm incorporating
  several levels of knowledge to overcome the
  disadvantages mentioned above.

6
Bipedal Running Concepts

• A bipedal running stride consists of 2 steps.
• A running step is made up of a single support
  state and a flight state.
• We will assume the motion of the legs to be
  symmetric so that we can restrict the analysis to
  one step.

7
Bipedal Running Concepts
8
Bipedal Running Concepts

• A running stride can be characterized by four
  high-level parameters
• 1. velocity (v)
• 2. step length (sl)
• 3. step frequency (sf)
• 4. flight height (H) where v sl
  sf. (1)

9
Bipedal Running Concepts

• There are relationships between these parameters.
• sl 0.1394 0.00465v
• where v lt 400 m/min.
• sl 0.1394 (0.00465level)v,
• where
• -0.001(poor) lt level lt0.001(skilled)

10
Bipedal Running Concepts

• Leg length also influence the relationship
  between v and sl.
• Froude numberv²/(2gl),lleg length
• l is proportional to body height
• ?Froude equality
• v²/body_heightv²/1.8

11
Bipedal Running Concepts

• y1the vertical position of the pelvis at
  toe-off
• y2the vertical position of the pelvis at
  heel-strike

12
Bipedal Running Concepts

• 0-180 steps/min,equation 5 is used.
• 180-230 steps/min,equation 4 is applied.
• Above 230 steps/min, flight time is kept constant

13
Bipedal Running Concepts
14
Running Algorithm

• The motion control algorithm has been
• developed with four design goals to make it a
  useful tool for an animator
• ease of motion specification
• interactive
• real time feedback
• ease of customizing and personalizing motion

15
Running Algorithm

• To meet these goals,three levels of knowledge are
  incorporated.
• Empirical knowledge
• Determines the kind of control parameters and
• how they are interrelated.
• Physical knowledge
• Determines the center of the body moves during a
  running stride.
• Limb-coordination knowledge
• To set up state-constraints and phase-constraints

16
Running Algorithm
17
Running Algorithm

• Parameters and attributes
• the interface to the user and control the current
  running stride.
• Parameters define the basic running stride
• Attributes change the expression or personality
  of the stride.
• A change in one of the parameters causes a change
  in the other parameters to maintain a natural
  stride,as well as in some attributes
• A change in any of the attributes is independent
  of the parameters.

18
Running Algorithm
19
Running Algorithm
20
(No Transcript)
21
Toeoff leg length
Xtoe
x1x2-vTflight
Step length
22
Running Algorithm

• x1 x2 vTflight
• y1 v(rad² - (x1 Xtoe)² )
• With key frames at HRS, TOR and HSL,the
  translation of the pelvis can be determined.
• During flight,the pelvis follows a parabolic
  trajectory,with 0 t Tflight x x1
  vt y y1 v(2gH) t 1/2gt² ( ??????)

23
Running Algorithm

• During support,the pelvis moves along an
  interpolating cubic spline segment,whose four
  control points are at HRS,mid-support,TOR and
  mid-flight.
• The kinetic and potential energy are both lowest
  about the mid-support.
• The calculation of the orientation of the
  pelvisrotation and listduring the current step
  completes the state constraints.
• This is done by linear interpolation between
• The four keyframes in Figure 7.

24
Running Algorithm
25
Running Algorithm
26
Results

• Snapshots of a few sample runs are shown below.

27
Results
28
Results
29
Conclusions

• Pulls together different knowledges and
  techniques to provide a high-level motion
  control.
• Conveniently create a large variety of human
  running styles
• In real time
• The creative control over the motion remains with
  the animator.
• Extend to running alone inclines and arbitrary
  paths.

30
Q A

• Thank you!