Integrating Realistic Human Group Behaviors into a Networked 3D Virtual Environment

1
Integrating Realistic Human Group Behaviors
into a Networked 3D Virtual Environment
MAJ Thomas E. Miller US Army September 2000
2
FLATLAND

• To
  
  The Inhabitants of SPACE IN GENERAL
  
  This Work is Dedicated
  
  By a Humble Native of Flatland
  
  In the Hope that
  
  Even as he was Initiated into the
  Mysteries
  Of THREE Dimensions
  
  Having been previously conversant
  With
  ONLY TWO
  So
  the Citizens of that Celestial Region
  May
  aspire yet higher and higher
  
  Thereby contributing
  
  To The Enlargement of THE IMAGINATION .
  Dedication from
  Flatland A Romance of Many Dimensions
  by A. Square (Edwin A. Abbott, 1884)

3
Outline

• Motivation Goals
• Problem Statement
• Background
• Development Issues
• Results
• Recommendations for Future Work
• Questions

4
Motivation

• 3D Visualization - Virtual sand tables,
  Situational awareness visualization, Virtual
  rehearsal systems and Networked After Action
  Review Tool
• Realistic small unit behavior modeling can
  provide insights into Tactics, Techniques
  Procedures (TTPs)
• Provide the ability to model dangerous, complex
  small unit actions
• Professionally relevant

5
Goals

• Virtual human avatars with articulated joint
  structure and a motion library to model realistic
  human movement
• Set of rule-based physical and logical behaviors
  implemented to execute limited tactical
  formations and activities
• Human entities aggregate into a group or mount
  other non-human entities and then separate back
  to individual entity control
• Source code platform independent, open source and
  worldwide deployable
• Visually compelling for acceptance

6
Problem

• Define and implement a means to include
  articulated virtual humans with an extensible
  motion library in an existing large-scale virtual
  environment (LSVE)
• Define and implement a means to aggregate and/or
  mount and disaggregate human entities
• Define and implement a set of realistic
  behaviors to model tactical formations and
  activities

7
Problem continued

• Make it networked
• Define a means to rapidly generate content
• Allow new entities to share geometry and
  behaviors interchangeably
• Make it platform independent, open source and
  visually compelling

8
Background

• The Legacy of NPSNET
• DIS-Java-VRML Capture the Flag
• Model Nancy.wrl An H-Anim 1.1 exemplar
  created by Cindy Ballreich
• Extensible 3D Specification and the Extensible
  Markup Language (XML) family of technologies

9
The Legacy
FOG-M, VEH and Moving Platform Simulator 1986 -
1990
 
NPSNET I III, NPSNETSTEALTH and NPSNET IV (DIS -
C - Performer) 1990 - 1997
NPSNET HUMAN PROJECT
Human Animation 1.1 Specification
Web3D Consortium
DIS Java - VRML 1996 - Present
Extensible 3D Specification
?
NPSNET V (? Java - ?) Present
DIS Java X3D Present
GeoVRML
10
DIS-Java-VRML

• A way to implement networked, shared virtual
  worlds using widely available technology and
  tools
• Web-based delivery of content
• PC-based platforms
• Standard networking technologies
• Standard 3D engines
• The objective is all Web browsers everywhere,
  not high-end workstations

11
DIS-Java-VRML
12
DIS-Java-VRML

• Entity State Protocol Data Unit (ESPDU) and
  Articulated parameters are adequate for basic
  humanoid use
• Group
• children
• DEF Nancy_EspduTransform EspduTransform
• marking "Nancy 40"
• readInterval 0.25
• address "224.2.181.145" multicast
• port 62040
• siteID 0
• applicationID 1
• entityID 40
• translation -4950 -200 2125
  traceOffset 0.0 3 0.0
• traceSize 1 1 1
• traceColor 0 0 1 Blue Team
• traceJava FALSE

13
Capture The Flag

• Networked 3D virtual environment using
  DIS-Java-VRML
• Geolocated at Fort Irwin, CA
• Consists of a Red and Blue team, each trying to
  capture the opponents flag and return safely
  with it to their base
• Significant student contribution to code base

14
Capture The Flag
15
Nancy.wrl

• Nancy.wrl is a canonical Human Animation 1.1
  specification exemplar
• Created by Cindy Ballreich
• Approved for non-commercial usage with
  appropriate credit and maintenance of name and
  logo

16
Nancy
17
Nancy Walking in Fort Irwin
18
X3D Specification

• From the Web3D Consortium Press Release
• The X3D standard will define interoperable,
  light-weight components for 3D Web and broadcast
  applications. It combines a run-time delivery
  engine and VRML97-inspired file format with XML
  bindings, with strong industry support for a
  proven standardization process

19
X3D Specification

• Generates VRML on the fly
• Behaviors react to user input
• Viewpoint animation and navigation assistance
• Supports DIS and behavior protocols

20
Development Issues

• Define and Implement Mounting
• Define Behaviors
• Interfaces for Single and Aggregated Entities
• Animation / Motion Library
• Rapid Content Generation
• Demonstration Scenarios

21
Mounting

• Basic Mounting Algorithm
• Move the entity as if it were independent
• Determine new vector from aggregation element to
  entity
• Move the aggregation element
• Add the entitys relative vector to the
  aggregation elements new position to determine
  new entity position

22
Mounting

• Use DIS to send the mounted entities entity
  state updates
• Maintain true position in a non-networked ESPDU
  data structure to ease transition between
  independent and aggregated control
• Uses sophisticated, run-time VRML scene graph
  switching to control the display of aggregated
  entities in order to dispense with possibly
  complicated network handshaking
• Mounting algorithm works either way.

23
Define Behaviors

• Movement Formations
• Wedge
• Line
• Column / File
• Primitives for Air Assault Operations
• Load
• Unload Helicopter

24
Behaviors

• Behaviors are defined through a series of rules
  and goal positions
• For example, the goal position for the first
  entity in a line is 90 degrees from the current
  heading of the aggregation element and faced to
  the same heading
• Each entity has a specific rule to determine the
  correct goal position in each behavior
• Once the entity determines its goal position, the
  entity will incrementally move to the goal
  position and then correctly maintain that
  position by matching the velocity vector of the
  aggregation entity

25
Single Entity Interface
26
Aggregated Entities Interface
UNDER ACTIVE DEVELOPMENT!
27
Animations/Motions

• A KNEEL motion was added to the motion library
  as a proof of concept for behavior interchange

28
Rapid Content Creation

• Using X3D Edit tool, capable of redefining VRML
  Humanoid using Protos (works in all browsers)
• Native tag set to support H-anim 1.1
  specification created (currently does not work in
  CosmoPlayer 2.1)
• Both examples were made DIS compliant through
  the creation of an DISJavaVRML tag set
  (essentially DIS-Java-X3D)
• With the creation of new XSL translators, can
  now create DIS-Java-Java3D or other file formats

29
(No Transcript)
30
(No Transcript)
31
Demonstration Scenarios

• Tactical Movement Formations
• Transitions between movement formations
• Near Term Future Work Air Assault Operations
• Far Term Future Work - Battle Drills

32
Wedge
33
Results

• Human entities can be aggregated and
  disaggregated in an existing LSVE
• Content can be rapidly generated through X3D
• Behaviors, motions and geometries can be
  developed, included and interchanged

34
Recommendations for Future Work

• Design adaptive agent humans for analytics and
  emergent behavior TTP discovery
• Possible use of RELATE architecture discussed by
  LCDR Roddy, USN, and LT Dickson, USN
• Increase behavior and motion libraries
• Scale aggregation through higher tactical levels
• Adapt mounting aggregation for deployment load
  planning for ships and airplanes
• 3D load planning tool accounting for weight and
  cube support by weight and balance logic for
  loaded vehicle
• Provide 3D manifest visualization with
  loading/unloading plan

35
Thesis Organization

• Chapter l Introduction
• Chapter ll Related Work
• Chapter III Problem Statement
• Chapter IV Initial Development Efforts
• Chapter V Human Animation 1.1 Specification in
  X3D
• Chapter VI Single User Interface (Team Control
  Panel)
• Chapter VII Mounting of Human Entities
• Chapter VIII Conclusions and Recommendations
• Thesis Advisor Don Brutzman
• Second Reader LT Patrick Mack, USN

36
(No Transcript)
37

• Questions
• ?