Hierarchical and Heterogenous Reactive Agents for RealTime Application

1
Hierarchical and Heterogenous Reactive Agents for
Real-Time Application
Eurographics 2003

• C. Niederberger
• M. Gross
• ETH Zurich

2
Outline

• Introduction
• Related Work
• System Overview
• Core Components
• Agents
• Common Knowledge Base
• Agent Generation
• Definition of Agent Instances
• Group Behavior
• Results
• Conclusions and Future Work

3
Introduction

• Autonomous agents
• secondary characters in computer games
• various individuals
• complex behaviors
• group organization

4
Introduction

• Issues
• a framework for autonomous agents
• agent definition / agent group definition
• time complexity

5
Related Work

• Behavior modeling
• Flocks, herds and schools A distributed
  behavioral model Reynolds 1987
• Artificial fishes Physics, locomotion,
  perception, behavior Tu and Terzopoulos 1994
• SimCity, The Sims Electronic Arts Inc
• designed for a specific individual
• Behavior modeling with hierarchical inheritance
• Modeling behavior with personalities Blando et
  al 99
• Hierarchical agent control a framework for
  defining agent behavior Atkin et al 2001
• Group behavior
• A model of human crowd behavior Musse and
  Thalmann 97

6
System Overview

• Objective
• autonomous agents with reactive behavior based on
  situation recognition
• the creation of large numbers of secondary
  characters with individual and group behavior
• guarantee a minimal and constant frame-rate

7
System Overview
8
Core Components

• Agent Engine
• the main controller for all agents
• maintains a list of all agents
• keeps maximum time-per-run
• decides which agents should be activated
• round-robin scheduling
• Agents
• Common Knowledge Base

9
Agents

• Sensors
• Situations
• conditions
• actions

10
Common Knowledge Base

• the central part of system
• stores the knowledge of all agents in one large
  container.
• KBAgents, sensors, situations, actions,
  conditions, and attributes

11
Common Knowledge Base KBAgent

• KBAgent container of the whole knowledge of the
  instance

12
Common Knowledge Base Situation
13
Common Knowledge Base Action
14
Agent definition

• Multiple Inheritance

15
Group definition

• Abstract group

16
Group definition

• Modulo rules
• specify regular patterns

17
Group definition

• Recursive definition

18
Group Behavior

• Leader-follower situation

19
Results
2. generic family
1. basic behavior
3. recursively defined group
4. complex behavior
20
Conclusion

• A framework for autonomous agents with reactive
  behavior
• behavior definition for different behaviors
  through weighted multiple inheritance
• mechanisms for group definition and group
  behavior
• Supporting minimal frame rates independent of the
  number of simulated agents
• Useful for interactive simulations involving
  multiple agents, such as games