Artificial Life in Virtual Environments

1
Artificial Life in Virtual Environments

A Virtual Reality Framework for Artificial Life
Simulations
Rogério Perino de Oliveira Neves
Marcio Lobo Netto
SVR'04
2
Objectives

• Build a customizable experiment development
  framework in Multi-Agent Context
• Apply Virtual Reality technologies to improve the
  visualization of Artificial Life experiments
• Implement experiments within the developed
  framework, research on Artificial Life field

3
About Artificial Life

• Combines biology and computer science to create
  synthetic models of living systems evolution
• A tentative to elucidate the logical structure
  (in a most general form) of biological evolution
• Provides techniques to simulate the natural and
  biological apparatus to solve humanistic
  engineering problems
• Originally dominated by computer scientists,
  nowadays studied by researches of almost all areas

4
Artificial Life

• The expression was first introduced by
  Christopher Langton in 1987, when was used as a
  conference name held inLos Alamos, New México,
  about The Synthesis and Simulation of Living
  Systems.

Artificial
life The proceedings of an Interdisciplinary
Workshop on the Synthesis and Simulation of
Living Systems September, 1987, Los Alamos, New
Mexico, Addison-Wesley Pub.
further information...
5
Key concepts

• The initial definition considered two types
• Weak AL simulation of biological phenomena
• Strong AL re-creation of life in the computer
• Possible attacks
• Bottom-up
• Emergent behavior rather than pre-defined
• Simple Rules rather than complex
• Top-down
• Project / Knowledge base / Composition

6
Attacks

• Bottom-Up
• Observed in the nature
• No planning
• Comes from emergence/evolution
• Generally associated to strong AL
• Top-Down
• Humanistic procedures
• Comprehends planning/foresight
• Generally associated with weak AL

7
Multi Agent Systems Theory
Autonomous agents
Biological agents
Robotic agents
Computational Agents
Artificial Life agents
Electronic agents
...
Artificial biological agents
Search agents
Entertainment agents
Viruses
Intelligent agents
...
8
Examples of Artificial Life Programs
9
State of the Art in Artificial Life
Neves,
Rogério Karl Sims videos, http//www.lsi.usp.br/
rponeves/research/sims, access 18/09/2003
further information...
10
Distinct types of Artificial Life research

• Origins of Life, self-organization and
  self-replication
• Development and replication
• Evolutionary dynamics and adaptation
• Autonomous agents and robots
• Communication, cooperation and social behavior
• Simulation, synthesis tools and methodologies

11
Motivation

• Weak visualization and Interaction
• most programs provides a poor representation of
  data
• the programs allow only the change of some
  parameters
• Hard code access
• When available, sources are in low-level, non OOP
  (ASM, C)
• Support to parallel architectures
• Multi-Agent ?? Multi-thread
• performance improvement in concurrent execution
• Full Real Environment simulation
• support 3D vector mathematics to operate objects
  in the scene
• Apply the State of the Art in visualization
  technologies, computer graphics through
  accelerator boards and VR to the exibition of the
  Virtual Environment

12
The CAVE VR System
Hight End
13
Other VR Systems
Lower Cost
14
Examples of Artificial Life Programs
15
Visual improvements
Cellular dynamics
16
Project features

• OOP Paradigm
• easy object/agent description/operation
• Cross-platform execution capability
• Open source philosophy
• Simulation of a true 3D space with vector
  dynamics
• easy manipulation of objects into 3D space
• Visualization in VR, immersive environments
• Multiple Visual VR device support
• Concurrent execution of programs
• Internet execution

17
Development resources

• Java / Java3D API (from SUN)
• Personal Computers
• Graphical Workstations (Silicon Graphics)
• Multi-processed systems (SPADE project)
• Cluster of PCs
• CAVE
• Visualization devices (from monitors to CAVES)
• GB Ethernet Network

18
Dependencies
Techniques
Other tools

• Object Oriented Paradigm
• Multi Agent Systems
• Vector Mathematics
• Discrete Time-Dynamics
• Concurrent Programming
• Computer Graphics
• Networking

• State Machines
• Non-linear Systems
• Chaotic Dynamics
• Ordinary differential equations
• Fuzzy Logic
• Artificial Neural Networks
• Evolutionary Search
• Genetic Algorithms

19
Java Java3D

• Java
• Cross-platform capability
• Internet compliant
• Built over the OOP paradigm
• Concurrent Threads programming support
• Extensible Reliable
• Java3D
• New standard in VR development
• Hi-level interface to OpenGL/DirectX
• Scene description through scene-graphs
• Extends Java features

20
Java3D scene-graph example
21
Visualization Interactivity

• Directed, but not limited to
• Ordinary 3D boards
• Professional Render Boards
• Ordinary or Stereo Displays
• Displays with Shutter Glasses
• Head Mounted Displays (HMD)
• CAVES
• Other VR devices

• Mouse
• Keyboard
• Gloves
• Wands
• Other tracking devices

Needs Java3D version support for output and
picking behavior classes
22
Experiments in ALiVE
Environment
Virtual scene
Agent
Actor
Agent
Actor
Agent
Actor
Visualization device
UI / Interaction
23
Levels of operability
User Interface
Custom User Interface
Runtime Interface/Interaction
User Classes
Hi-Level/Pseudo Code
Project Scope
A.L.I.V.E. Framework
Super classes
Java/Java3D
Mid-Level/Language Code
Byte code
Java VM/Machine Code
24
Framework architecture
RenderClient Subset
25
UML diagram
26
UML diagram
27
Render Client parallel operation
Server
Env
Scene
Multicast Packages
RenderClient
RenderClient
RenderClient
28
Environment configuration
29
UI
30
Agent diagram example
31
Demo code
DEMO CODE
32
Developed Experiments

• Program test
• ALGAE Evolution / Adaptation
• Predator-Prey system
• Fish Schooling
• Flocking
• Biological demos
• Cellular dynamics
• Fungus growth
• Lymphocytes virus
• Mitosis

33
Demo Experiments
Fungus
Virus infection
Bacteria Simulation
Human Blood Cells
34
Predator-Prey System
Predator Prey System
35
Predator sight
FILTER
W1
Neural Net
R
W2

ACT
RADIATION
G
B
W3
Predator Prey System
RGB Filter
36
Predator-Prey population graphs
37
Predator-Prey population graphs
38
Fish Schooling
Evolutive Neural Networks
39
Flocking
Emergence of Complexity
40
Performance -1
41
Conclusions

• The project applies the representational power of
  Virtual Reality to the visualization of
  Artificial Life environments
• The developed framework provides a quick
  prototype development tool
• The experiments included demonstrates the
  framework capabilities, serving as template to
  users new experiments

42
Conclusions

• Making the project available in Sourceforje.net,
  users can contribute for improving the framework
• The experiments within the framework can be
  published thought the internet for faster
  interaction between researchers
• Ordinary people outside the scientific community
  can experiment the experimental virtual lab, a
  scientific divulgation tool
• The experiments evolved can take advantage of
  emerging visualization technologies without code
  modification

43
Possible employments

• Artificial Life experiment development
• Demonstrations of biologic processes
• Problem solving in sciences/engineering
• System training in robotics
• Simulation of genetic / evolutionary systems
• User-Assisted, Agent-Oriented pattern search in
  multi-dimensional space data ensembles
• Upcoming, future technologies (such
  nanotechnologies)

Neves,
Rogério P. O. and Netto, Marcio L. Evolutionary
Search for Optimization of Fuzzy Logic
Controllers 1st International Conference on
Fuzzy Systems and Knowledge Discovery, Volume I,
on Hybrid Systems and Applications I
further information...
44
Proposal to future works

• Interaction through sensitive devices
• Apply knowledge to solve problems in engineering
• Automated generation of Intelligent Control
  Systems, autonomous training method for robots or
  device controlling
• Ensemble (pre) processing through Multi-Agent
  data mining
• Image processing with evolvable Neuro-Fuzzy
  apparatus

45
Related Documents

• Rogério Neves, ALIVE Project Site thesis
• http//www.lsi.usp.br/rponeves/
• Official ALIVE Project site
• http//www.lsi.usp.br/alive/
• ARTLIFE Site, Artificial Life group
• http//www.lsi.usp.br/artlife/
• Questions doubts
• rponeves_at_lsi.usp.br

46
LOGO