John S Gero Agents – Agent Simulations AGENT-BASED SIMULATIONS.

Slides:



Advertisements
Similar presentations
A predictive Collision Avoidance Model for Pedestrian Simulation Author: Ioannis Karamouzas et al. Presented by: Jessica Siewert.
Advertisements

Controlling Individual Agents in High Density Crowd Simulation N. Pelechano, J.M. Allbeck and N.I. Badler (2007)
Flocking and more.  NPC groups can move in cohesive groups not just independently ◦ Meadow of sheep grazing? ◦ Hunting flock of birds? ◦ Ants? Bees?
Behavioral animation CSE 3541 Matt Boggus. Material recap and trajectory Geometric – Artist specifies translation and rotation over time Physically based.
Better Group Behaviors in Complex Environments using Global Roadmaps O. Burchan Bayazit, Jyh-Ming Lien and Nancy M. Amato Presented by Mohammad Irfan Rafiq.
Flocks, Herds, and Schools: A Distributed Behavioral Model By: Craig Reynolds Presented by: Stephanie Grosvenor.
1 CO Games Development 2 Week 22 Flocking Gareth Bellaby.
G. Folino, A. Forestiero, G. Spezzano Swarming Agents for Discovering Clusters in Spatial Data Second International.
OBJECT-ORIENTED THINKING CHAPTER Topics  The Object-Oriented Metaphor  Object-Oriented Flocks of Birds –Boids by Craig W. Reynolds  Modularity.
Optimizing Flocking Controllers using Gradient Descent
Crowd simulation Taku Komura. Animating Crowds We have been going through methods to simulate individual characters We have been going through methods.
Kristen Gardner. Outline Swarm Intelligence Flocking Basic Steering Swarms Applications Animals Vehicles People.
University of Texas at Austin CS 378 – Game Technology Don Fussell CS 378: Computer Game Technology Dynamic Path Planning, Flocking Spring 2012.
The Vector Field Histogram Erick Tryzelaar November 14, 2001 Robotic Motion Planning A Method Developed by J. Borenstein and Y. Koren.
1 Reactive Pedestrian Path Following from Examples Ronald A. Metoyer Jessica K. Hodgins Presented by Stephen Allen.
Social Force Model for Pedestrian Dynamics 1998 Sai-Keung Wong.
SITUATED AGENTS. John S Gero Agents – Situated Agents Basic Ideas Interaction not just encoding Construction not just recall Cognitive Science Dewey (1896):
1Notes  Assignment 2 is out  Flocking references  Reynolds, “Flocks, Herds, and Schools…”, SIGGRAPH’87  Tu and Terzopoulos, “Artificial Fishes…”, SIGGRAPH’94.
Presenter: Robin van Olst. Avneesh SudRussell Gayle Erik Andersen Stephen GuyMing Lin Dinesh Manocha.
Presenter: Robin van Olst. Prof. Dr. Dirk Helbing Heads two divisions of the German Physical Society of the ETH Zurich Ph.D. Péter Molnár Associate Professor.
Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin.
Evaluating and re-evaluating agent modeling: simulation and design January 11 th, 2007 Arch 484: Design Computing Seminar ? Daniel Belcher.
Agent Based Modeling (ABM)
Real-time Crowd Movement On Large Scale Terrains Speaker: Alvin Date:4/26/2004From:TPCG03.
Steering Behaviors For Autonomous Characters
A Crowd Simulation Using Individual- Knowledge-Merge based Path Construction and Smoothed Particle Hydrodynamics Weerawat Tantisiriwat, Arisara Sumleeon.
CS Reinforcement Learning1 Reinforcement Learning Variation on Supervised Learning Exact target outputs are not given Some variation of reward is.
Crowd Simulations Guest Instructor - Stephen J. Guy.
Artificial Intelligence in Game Design Camera Control.
Navigating and Browsing 3D Models in 3DLIB Hesham Anan, Kurt Maly, Mohammad Zubair Computer Science Dept. Old Dominion University, Norfolk, VA, (anan,
Ioannis Karamouzas, Roland Geraerts, Mark Overmars Indicative Routes for Path Planning and Crowd Simulation.
DARPA Mobile Autonomous Robot SoftwareLeslie Pack Kaelbling; March Adaptive Intelligent Mobile Robotics Leslie Pack Kaelbling Artificial Intelligence.
Robot Crowd Navigation using Predictive Position Fields in the Potential Function Framework Ninad Pradhan, Timothy Burg, and Stan Birchfield Electrical.
Flow Fields Hao Li and Howard Hamilton. Motivation for Flow Fields Multiple AI algorithms in a computer game can produce conflicting results. The AI must.
L – Modelling and Simulating Social Systems with MATLAB Lesson 5 – Introduction to agent-based simulations A. Johansson & W. Yu ©
PSY105 Neural Networks 1/5 1. “Patterns emerge”. π.
Liang, Introduction to Java Programming, Eighth Edition, (c) 2011 Pearson Education, Inc. All rights reserved Event Driven Programming, The.
EXIT = Way Out Julian Dymacek April 29. Escape Panic Paper Dr. Dirk Helbing, Illes J. Farkas, Dr. Tamas Vicsek Point mass simulation Uses psychological.
SI 2008: Study of Wave Motion July 19, 2008 Martin Bobb Joseph Marmerstein Feibi Yuan Caden Ohlwiler.
Computer Architecture Lecture 26 Fasih ur Rehman.
Adrian Treuille, Seth Cooper, Zoran Popović 2006 Walter Kerrebijn
Artificial Intelligence in Game Design Complex Steering Behaviors and Combining Behaviors.
Curiosity-Driven Exploration with Planning Trajectories Tyler Streeter PhD Student, Human Computer Interaction Iowa State University
Computer Animation Rick Parent Computer Animation Algorithms and Techniques Behavioral Animation: Knowing the environment Flocking.
Controlling Individual Agents in High-Density Crowd Simulation
Crowds (and research in animation and games) CSE 3541 Matt Boggus.
REFERENCES: FLOCKING.
Chapter 9 – Additional Scenarios. Marbles Collision Detection The Marbles scenario does not use any of the built-in Greenfoot collision detection.
Computer Animation Rick Parent Computer Animation Algorithms and Techniques Behavioral Animation: Crowds.
Crowd Self-Organization, Streaming and Short Path Smoothing 學號: 姓名:邱欣怡 日期: 2007/1/2 Stylianou Soteris & Chrysanthou Yiorgos.
Computer Animation Rick Parent Computer Animation Algorithms and Techniques Behavioral Animation: Crowds.
Autonomous Robots Robot Path Planning (3) © Manfred Huber 2008.
Artificial Intelligence in Game Design Lecture 8: Complex Steering Behaviors and Combining Behaviors.
Reinforcement Learning AI – Week 22 Sub-symbolic AI Two: An Introduction to Reinforcement Learning Lee McCluskey, room 3/10
Simulating Crowds Simulating Dynamical Features of Escape Panic & Self-Organization Phenomena in Pedestrian Crowds Papers by Helbing.
CSCI 4310 Lecture 5: Steering Behaviors in Raven.
ROB SAUNDERS, JOHN S. GERO
11/13/03CS679 - Fall Copyright Univ. of Wisconsin Last Time A* Improvements Hierarchical planning Pre-Planning.
Students: Yossi Turgeman Avi Deri Self-Stabilizing and Efficient Robust Uncertainty Management Instructor: Prof Michel Segal.
Crowds (and research in computer animation and games)
Computer Animation Algorithms and Techniques
Crowd Modelling & Simulation
Flocking Geometric objects Many objects
Swarm simulation using anti-Newtonian forces
CIS 488/588 Bruce R. Maxim UM-Dearborn
Crowds (and research in computer animation and games)
CIS 488/588 Bruce R. Maxim UM-Dearborn
Application to Animating a Digital Actor on Flat Terrain
FLOSCAN: An Artificial Life Based Data Mining Algorithm
Hiroki Sayama NECSI Summer School 2008 Week 2: Complex Systems Modeling and Networks Agent-Based Models Hiroki Sayama
Presentation transcript:

John S Gero Agents – Agent Simulations AGENT-BASED SIMULATIONS

John S Gero Agents – Agent Simulations Simulations in Design l Visual Simulations l e.g. renderings and models l Mathematical Simulations l i.e. systems of equations l Physical Simulations l e.g. wind tunnels l Computational Simulations l e.g. finite element analysis l Agent-based Simulations l e.g. crowd simulations

John S Gero Agents – Agent Simulations Simulating Crowds l Craig Reynolds’ Flocking Algorithm l A subclass of Reynolds’ Steering Behaviours l Extended flocking algorithms for games l Additional behaviours for goal-oriented path-following l The Social Force Model l Simulated crowd behaviour based on empirical results

John S Gero Agents – Agent Simulations Flocks, Herds and Schools 1. Separation. Steer to avoid flockmates. 2. Cohesion. Steer to move toward the average position. 3. Alignment. Steer toward the average heading. 4. Avoidance. Steer to avoid running into obstacles. (a) Separation (b) Cohesion(c) Alignment (d) Avoidance Steering behaviours used in Reynold’s model of flocking.

John S Gero Agents – Agent Simulations The Social Force Model 1. Pedestrians are motivated to move as efficiently as possible to a destination. 2. Pedestrians wish to maintain a comfortable distance from other pedestrians. 3. Pedestrians wish to maintain a comfortable distance from obstacles. 4. Pedestrians may be attracted to other pedestrians or objects (e.g. posters).

John S Gero Agents – Agent Simulations SITUATED ANALYSIS Pedestrians may be attracted to other pedestrians or objects. Pedestrians try to maintain a comfortable distance from obstacles like walls. Pedestrians try to maintain a comfortable distance from other pedestrians. Pedestrians try to move as efficiently as possible to a destination. Description of situated social force pedestrian obstacle destination attraction repulsion Designing doors

John S Gero Agents – Agent Simulations Narrow door

John S Gero Agents – Agent Simulations Wide door

John S Gero Agents – Agent Simulations Two doors

John S Gero Agents – Agent Simulations INTEREST IN EMERGENT BEHAVIOUR

John S Gero Agents – Agent Simulations Agent-Centric Design Evaluations l Efficiency l Inefficiency is measured with respect to the deviation of an agent’s actual walking speed from it’s desired walking speed. l Comfort l Discomfort is measured with respect to the number of changes in direction that have to be made by an agent to navigate a space. l Non-homogenous crowd simulations l e.g. simulating crowds of pedestrians with different desired walking speeds.

John S Gero Agents – Agent Simulations A Curious Agent The architecture of a curious agent.

John S Gero Agents – Agent Simulations Detecting Novelty 1. How often similar patterns have been experienced. 2. How similar these patterns have been. 3. How recently these patterns have been experienced.

John S Gero Agents – Agent Simulations Calculating Interestingness hedonic value = reward + punish

John S Gero Agents – Agent Simulations The Curious Social Force Model l Extends the Helbing and Molnar’s Social Force Model l Adds an additional rule to the Social Force Model: Pedestrians are motivated to move towards potentially interesting areas l Models curious exploratory behaviour as a social force l Uses the same simple model of locomotion as flocking and the social force model to move agents l Incorporates learning and curiosity into the agent model l Situates agents in past experience to detect novelty in new experiences

John S Gero Agents – Agent Simulations Situated Design Evaluations l Agent-centric evaluation of designs l Evaluations of interestingness depends upon the hedonic function which may vary from one agent to another. l Evaluations situated in experiences of agents l Interestingness depends upon novelty detection which in turn depends upon the long term memory of the agent. l Situatedness and changing evaluations l The experience of a design changes how an agent will evaluated it in the future.

MIT Class Spring 2002 An Example Design Problem: Curating a Gallery

John S Gero Agents – Agent Simulations Implementation l Sensing l Simple vision through raycasting l Perceiving l Perception of colours as hues l Learning & Novelty Detection l Self-organising maps l Planning & Moving l Generating and combining social forces

John S Gero Agents – Agent Simulations Sensing Simple vision implemented using raycasting.

John S Gero Agents – Agent Simulations Perceiving l Simple perception of hues l The artworks in the gallery are modelled as blocks of colour, the agents are only interested in the hues of these artworks allowing the sampled environment to be represented as a vector of single values (angles on the colour wheel). A colour wheel.

John S Gero Agents – Agent Simulations Learning & Novelty Detection l Learning l 1D Self-organising map l Novelty detection l Approximates detection of novelty based on similarity of previous experiences, the past frequency of similar experience and time since the last similar experience.

John S Gero Agents – Agent Simulations Planning & Moving l The Curious Social Force Model l Motivational forces are generated for all perceived objects in the direction of the object with a magnitude proportional to the object’s interestingness, the forces are then combined into a single curious social force by averaging their magnitudes and directions.

John S Gero Agents – Agent Simulations Emergent Design Problems

John S Gero Agents – Agent Simulations Emergent Design Problems l Overcrowding to avoid uninteresting (radical) artworks l The agents in the first room become overcrowded because the artworks that are visible in the second room are too different from those in the first and generate a curious social force blocking entry to the second room. l Neglect of artworks because of improper sequencing l The agents pass quickly through the last room because the artworks in this room are too different from those in the previous room, encouraging a rapid exit.

John S Gero Agents – Agent Simulations One Possible Solution to Emergent Design Problems

John S Gero Agents – Agent Simulations One Possible Solution to Emergent Design Problems l Improving the progression of artworks l Artworks in second and third room are swapped so that the difference between artworks in successive rooms is minimised. l Improved flow of agents between rooms l Agents are drawn into each new room as a result of the greater interest the agents have in experiencing similar-yet- different artworks to those that they have already seen. l Increased interest, efficiency and comfort l Better design improves situated and agent-centric evaluations.