Guiding and Interacting with virtual crowds in realtime

1
Guiding and Interacting with virtual crowds in
real-time

• Soraia Raupp Musse, Fabien Garat and Daniel
  Thalman
• In Proc. of Eurographics workshop on
• Computer animation and Simulation 1999
• Jaeho Kim, VR Lab. KAIST

2
Contents

• ViCrowd
• Interaction method
• Multi-client architecture
• Conclusion
• Discussion

3
ViCrowd

• Crowd
• Virtual human agent (agent)
• Sensors, memory, perception, motor
• Action, reaction
• Group
• A group of agents
• Crowd
• A set of groups

4
ViCrowd

• Control
• Guided interactive control
• Programmed pre-defined behaviors using script
  language
• Autonomous rule-based behaviors

5
ViCrowd

• A priority of behaviors
• Avoiding collision with obstacles
• Walking
• Reaction from panic event, e.g. looking for the
  exit doors
• Go to some position

6
ViCrowd

• A priority of behaviors
• Avoiding collision with obstacles
• Walking
• Reaction from panic event, e.g. looking for the
  exit doors
• Go to some position
• Guided gt Autonomous gt Programmed gt Inherent

7
Interaction
8
Interaction

• Motion
• Action
• Events
• Reactions
• Status
• Environment information

9
Interaction Entities

• Motion
• Go to specific position
• Action
• Apply a specific posture
• Events
• Can be activated by the external control
• Reaction
• Some parameters of reaction can be sent via
  external controller

10
Interaction Entities

• Internal status of groups
• E.g. the group emotion
• Environment information
• Declare obstacles that have to be avoided
• Declare regions where the crowd can walk

11
Interaction Paradigm

• Information exchanged for interacting with a
  group or an agent
• Selection
• Motion
• Action
• State
• Events/reactions
• Request
• Knowledge

12
Interaction Paradigm

• Selection paradigm
• To select a/an group/agent depending on some
  conditions (status, location, goal)
• Motion paradigm
• To define a specific motion to be applied by the
  selected entity
• Action paradigm
• Actions to be applied by the crowd
• A body posture , interaction with objects

13
Interaction Paradigm

• State paradigm
• Set the parameters to change the internal state
  of agents/groups
• Behavior data, Quantitative data, Internal status
  data
• Density paradigm
• Events/Reactions paradigm
• Activate/Deactivate events/reactions

14
Interaction Paradigm

• Request paradigm
• E.g. Position, orientation, emotion of group,
  goal of group, the current group behaviors
• Environment paradigm

15
Multi-Client Architecture

• Server-client architecture
• Server
• Coordinator
• Client
• Behavior clients
• ViCrowd client
• Rule-based behavior system (RBBS) client
• Database client
• Environment client (ENVIR)
• Interface clients
• Texture user interface (TUI)
• Dataglove client, speech recognition client,
  sound client

16
(No Transcript)
17
Communication Examples

• Example 1
• Client Name Textual User Interface
• Group Name Group1
• Stimulus BE gt Emotion
• Response ViCrowd APPLIES STATE BE lt Emotion
• EndResponse

18
Communication Examples

• Example 2
• Stimulus GOTO SPECIFIC_LOCATION
• Response ViCrowd APPLIES MOTION GOTO lt NAME AT
  63150 -5000 -12540 1 0 0
• Response ViCrowd WAIT_FOR (REQUEST GOALREACHED
  FOR lt NAME) TRUE
• EndResponse

19
Client Synchronization

• How can we synchronize stimulus when different
  clients send information to same guided group at
  same time ?
• Three levels of priority
• High-level non-interruptible stimuli
• Middle-level continuous and interruptible
  stimuli
• Low-level non-continuous and interruptible
  stimuli

20
(No Transcript)
21
A Case Study
22
Conclusion

• ViCrowds
• Specially focus on guiding crowds
• Interaction paradigm
• Multi-client architecture
• Future work
• More autonomy for the server
• Automatic decision of priorities of each stimulus

23
Discussion

• Guided, Programmed, Autonomous ?
• Motion, Action ?
• Interaction paradigm ?

24
Data type of communication

• Five kinds of data to send to others clients
• Confirmation data (ACK)
• E.g. Goal reached
• High-level behavior (HLB)
• E.g. Group5 goes to the restaurant
• Low-level behavior (LLB)
• E.g. Group5 GOTO position (X, Y, Z)
• Asking data (ASK)
• E.g. ASKPOS (asking for position) Restaurant
• Occurrence of crowd events (EVE)
• E.g. Event PANIC occurred

25
(No Transcript)