Goal-Oriented Game AI Purpose and Utilizations Research Organized by: Evan Schipellite.

Slides:



Advertisements
Similar presentations
Modelling with expert systems. Expert systems Modelling with expert systems Coaching modelling with expert systems Advantages and limitations of modelling.
Advertisements

Heuristic Search techniques
Curriculum Development and Course Design
8.3. A GENT AND D ECISION M AKING AI Agent driven AI and associated decision making techniques.
ARCHITECTURES FOR ARTIFICIAL INTELLIGENCE SYSTEMS
R005 Creating an interactive product Type of product and Software Choice.
Relating Error Diagnosis and Performance Characteristics for Affect Perception and Empathy in an Educational Software Application Maria Virvou, George.
Artificial Intelligence in Game Design Intelligent Decision Making and Decision Trees.
Intelligent Agents: an Overview. 2 Definitions Rational behavior: to achieve a goal minimizing the cost and maximizing the satisfaction. Rational agent:
Leroy Garcia 1.  Artificial Intelligence is the branch of computer science that is concerned with the automation of intelligent behavior (Luger, 2008).
Artificial Intelligence in Game Design Problems and Goals.
Chapter 14: Artificial Intelligence Invitation to Computer Science, C++ Version, Third Edition.
Constraining Autonomous Character Behavior with Human Concepts Presentation by Jonathan Martin Based on Chapter 3.1 of AI Game Programming Wisdom 2, by.
Artificial Intelligence Chapter 25 Agent Architectures Biointelligence Lab School of Computer Sci. & Eng. Seoul National University.
High Level Architecture Overview and Rules Thanks to: Dr. Judith Dahmann, and others from: Defense Modeling and Simulation Office phone: (703)
Artificial Intelligence in Game Design
Artificial Intelligence in Game Design Complex Steering Behaviors and Combining Behaviors.
Artificial Intelligence in Game Design
Game Theory, Social Interactions and Artificial Intelligence Supervisor: Philip Sterne Supervisee: John Richter.
Artificial Intelligence in Game Design Goal-Oriented Behavior.
GAME PLAYING 1. There were two reasons that games appeared to be a good domain in which to explore machine intelligence: 1.They provide a structured task.
Computational theory techniques in interactive video games.
Artificial Intelligence in Game Design Lecture 17: Goal Oriented Action Planning.
3 States and a Plan: The AI of F.E.A.R. Jeff Orkin Monolith Productions/ MIT Media Lab.
Artificial Intelligence in Game Design Lecture 8: Complex Steering Behaviors and Combining Behaviors.
Finite State Machines using Alice Stephen Cano CIS 4914 Senior Project Wednesday December 5th.
Control of Dynamic Discrete-Event Systems Lenko Grigorov Master’s Thesis, QU supervisor: Dr. Karen Rudie.
Combining Neural Networks and Context-Driven Search for On- Line, Printed Handwriting Recognition in the Newton Larry S. Yaeger, Brandn J. Web, and Richard.
Introduction to Machine Learning, its potential usage in network area,
Organizing Students to Practice and Deepen Knowledge
Management Contemporary Gareth R. Jones Jennifer M. George
Practical.
Decision Making: Decision Tree & State Machines Session 07
Othello Artificial Intelligence With Machine Learning
SNS College of Engineering Department of Computer Science and Engineering AI Planning Presented By S.Yamuna AP/CSE 5/23/2018 AI.
N-Tier Architecture.
Artificial Intelligence in Game Design
What You Will Do Describe the specificity principle and how it applies to a fitness program. Design goals for your fitness program.
Enemy and Friendly AIs Richard Gesick.
Artificial Intelligence Chapter 25 Agent Architectures
Artificial Intelligence Lecture No. 5
Manufacturing system design (MSD)
Personal Management Skills
Othello Artificial Intelligence With Machine Learning
Chapter 10 Verification and Validation of Simulation Models
© James D. Skrentny from notes by C. Dyer, et. al.
Teaching with Instructional Software
Artificial Intelligence
Course Instructor: knza ch
Algorithm An algorithm is a finite set of steps required to solve a problem. An algorithm must have following properties: Input: An algorithm must have.
Identifying Confusion from Eye-Tracking Data
Material Requirements Planning (MRP)
Making Healthy Decisions
CIS 488/588 Bruce R. Maxim UM-Dearborn
The use of Neural Networks to schedule flow-shop with dynamic job arrival ‘A Multi-Neural Network Learning for lot Sizing and Sequencing on a Flow-Shop’
Chapter Three Goal Setting for Teenagers
Artificial Intelligence Lecture No. 28
Knowing When to Stop: An Examination of Methods to Minimize the False Negative Risk of Automated Abort Triggers RAM XI Training Summit October 2018 Patrick.
Artificial Intelligence Chapter 25. Agent Architectures
Presented By: Darlene Banta
What You Will Do Describe the specificity principle and how it applies to a fitness program. Design goals for your fitness program.
Computer Graphics Lecture 15.
EE 492 ENGINEERING PROJECT
Artificial Intelligence Chapter 25 Agent Architectures
Personal and social development
Lesson 3 Consumer Science
Self-Managed Systems: an Architectural Challenge
Chapter 12 Analyzing Semistructured Decision Support Systems
Goals and Objectives.
M & E Plans and Frameworks
Presentation transcript:

Goal-Oriented Game AI Purpose and Utilizations Research Organized by: Evan Schipellite

What is GOB?  Goal-Oriented Behavior (GOB)  Make use of Motives / Desires to guide AI Characters  Examine the most ‘pressing’ goals and choose an action  The action reflects two key aspects of the algorithm  The action serves to satisfy the current goal of importance  The action results in the least ‘negative aftermath’

GOB in Contemporary Games  Goal-Oriented Behavior allows for simple AI construction  It can be useful, making simple AI appear realistic  Not for all games, as it pertains more to ‘undirected’ AI  Generally, GOB is useful when a Game:  Needs numerous side-characters that can appear life-like  But, these characters are not directly controllable by gameplay  Needs to recreate a simulation (People, Animals, Plants)  Needs to provide AI with sub-actions within a state-machine  RTS units may begin working on the most pressing tasks if not specifically tasked by the player

The Sims 3 (The Sims Studio)

Ghost Master (Sick Puppies)

Prison Architect (Introversion Software)

So, What’s Going on in These Games?  Player interacts with units that are not fully controllable.  However, these units can be persuaded  Sims 3  Player works to satisfies the needs / desires of the residents  Ghost Master  Player positions independent ghosts in order to construct the best scaring environment  Prison Architect  Player monitors the psychology of the prisoners to improve the facility and plan their schedule  The Purpose is to create simple AI to choose and act in realistic ways

Goals  “goals describe some desired state of the world… we want to reach“ (Jeff Orkin)  Goals can be consider insistences (Ian Millington).  As a Goal’s rate increases, its insistences declares it a priority  Goals represent traits or states that we want to complete, or reduce the insistence of

Actions  “actions are defined in terms of preconditions and effects” (Jeff Orkin)  "present the character with a suite of possible actions and have it choose the one that best meets its immediate needs” (Ian Millington)  An Action serves as a means to complete a Goal.

 ‘Sleep’ is the highest priority. 3 actions are available.  The best route involves ‘Use Bed’  It will reduce the insistence of ‘Sleep’ the most  But, what if the numbers are not so obvious?

 ‘Hunger’ is highest priority. There are 4 possible actions.  However, two actions involve satisfying ‘Hunger’  The difference between ‘Hunger’ and ‘Bath’ is not much  In this case, overall insistence is taken into account  The AI does not simply choose the highest priority  It analyzes the overall discontentment (Ian Milliginton)

 All Goal values (Hunger, Sleep, Bath) are analyzed  The AI notes their resulting values from each action  Therefore,  ‘Order Pizza’ would result in (0, 4, 6)  ‘Eat Snack’ would result in (0, 2, 3)  ‘Eat Snack’ is chosen to minimalize discontentment

How does the GOB Algorithm Work?  Ultimately, the Answer is: In Many Different Ways  Essentially, the algorithm is constructed by taking into account Goals and Actions.  The AI evaluates the discontentment that would result from any given action, thus providing the best action

How does the GOB Algorithm Work?  To generate the best action  Utilize the current status of Goals and possible Actions  Then loop through the Actions, searching for the least discontentment result  Discontentment can be calculated by adding  All of the current Goal values  All of the resulting Goal changes in the Action

The Result of GOB  In its simplest form, the highest rated goal may lead to an action  Additional complexities can be developed by the programmer  Add in time durations to Actions  Additional changes are addeds to discontentment based on the time it may take to complete a task  Resolve incidents where Actions become unavailable  Have a unit continue another action while waiting  Have units take into account distance / obstacles  Actions may carry more weight depending on position, health, status, inventory, etc.  Depending on the game, these features can be further developed to make the AI appear independent

Example Adjustments  AI can ‘claim’ facilities, thus limiting their availability  AI can ‘claim’ the closest available facilities

Example Adjustments  If desired Actions are unavailable  The AI can revaluate their actions and choose the next best  If no Actions are available, the AI can wander instead

Basic Goal-Oriented Behavior in Practice

Additional GOB Information  "goal instances can be dynamically generated and destroyed“ (Denis Papp)  AI should be able to react to changes in the environment and their possible needs  "needs to look at the situation in which it wants to be, and then work backwards from it to calculate the best way of achieving the desired result“ (Ben Hardwidge)  Essentially, the AI should take what it wants to do, and then figure out which path will net that result in the best possible way.  Be careful with how much knowledge is provided.  Ex: If the Ai is trying to get to work, should it know about the delayed train before it arrives at the station?

Additional GOB Information  Planning System  “benefit is the ability to decouple goals and actions, to allow different types of characters to satisfy goals in different ways. The second benefit of a planning system is facilitation of layering simple behaviors to produce complex observable behavior“ (Jeff Orkin)  Design the GOB algorithm to be reusable by different units  Complex behaviors can result from simple combinations  "A goal is active if it is currently reasonable to execute it“ (Denis Papp)  Should the AI even complete a goal if it is not high enough priority?  Ex: Should the AI sleep just because there are not other pressing tasks?

Sources  Millington, Ian, and John David Funge. Artificial Intelligence for Games. Burlington, MA: Morgan Kaufmann/Elsevier, Print.  Orkin, Jeff. “Three States and a Plan: The A.I. of F.E.A.R.” Monolith Productions / M.I.T. Media Lab, Cognitive Machines Group. Web..  Hardwidge, Ben. “How AI in Games Works: The Planning System.” bitGamer. Web. 5 March,  Papp, Denis. “A Goal-Based Architecture for Opposing Player AI.” The Association for the Advancement of Artificial Intelligence. Web..

Image Sources  Sims 3  3-gameplay-the-sims jpg 3-gameplay-the-sims jpg  Ghost Master  hostmaster_Screen_4.jpg? hostmaster_Screen_4.jpg?  Prison Architect  preview-2.jpg preview-2.jpg