CS 680: GAME AI INTRODUCTION TO GAME AI. 1/9/2012 Santiago Ontañón

Transcription

1 CS 680: GAME AI INTRODUCTION TO GAME AI 1/9/2012 Santiago Ontañón

2 CS 680 Focus: advanced artificial intelligence techniques for computer games. Goal: cover state-of-the-art AI techniques that are 1) Recently used in commercial games, 2) Open problems in the game AI research community.

3 Outline Structure of the course Intro to Game AI Examples of Game AI Basic Game AI techniques Course Scope Paper presentations Projects

4 Outline Structure of the course Intro to Game AI Examples of Game AI Basic Game AI techniques Course Scope Paper presentations Projects

5 Structure of the Course 3 blocks: Real-time Strategy Games, Drama Management, Procedural Content Generation Flexible content Students need to: Complete 3 group projects (one per block) One of them will be your focus, and you will write a paper about it. Read 1 paper and present it in class on-line students will prepare a powerpoint/keynote/pdf presentation with audio Every 3 hour class will be divided in: 2 student paper presentations (20 minutes each) Lecture Project support session

6 Grading Projects: 70% Focus project: 40% Secondary projects: 15% each Paper presentation: 20% Attendance and class participation: 10%

7 Prerequisites Basic knowledge of artificial intelligence: Heuristic Search (e.g. A*) Game tree Search (e.g. minimax) Readings: Artificial Intelligence: A Modern Approach (Russell & Norvig) (Chapters 3,5,18)

8 Outline Structure of the course Intro to Game AI Examples of Game AI Basic Game AI techniques Course Scope Paper presentations Projects

9 What is Game AI? Artificial Intelligence for Computer Games Different from traditional AI Traditional AI: Optimality, efficiency Game AI: Fun, artificial stupidity

10 What is Game AI? Intersection of games and AI: Games AI Two (three) main communities working on it: Academics: Artificial Intelligence community: how can games help us have better AI (AI centric) Computer Game scholars: how can AI help us have better/more interesting/new forms of games?(games centric) Game industry: Their goal is to make games that sell more units (games centric)

11 Outline Structure of the course Intro to Game AI Examples of Game AI Basic Game AI techniques Course Scope Paper presentations Projects

12 Examples of Game AI Pac-Man (1980) First ever game to feature AI AI: finite state machine

13 Examples of Game AI Pac-Man (1980) First ever game to feature AI AI: finite state machine

14 Examples of Game AI Double Dragon AI: Finite State Machines

15 Examples of Game AI Chess AI needs to provide a collection of difficulty levels. Only one: hardest (to be played only against grand-masters), falls into the realm of traditional AI.

16 Examples of Game AI The Secret of Monkey Island And Then There Were None Dialogues, storytelling

17 Examples of Game AI Left 4 Dead 2 AI Director adjusts game pace to ensure desired dramatic effects

18 Examples of Game AI Black & White Uses machine learning to simulate a learning creature

19 Examples of Game AI Starcraft II Strategy, planning, pathfinding, economics, etc.

20 Types of Game AI Inside the game: Character control Director (drama management) During game development: Help in behavior/content design After game deployment: Analysis of game data

21 Types of Game AI Inside the game: Character control Director (drama management) Covered in this course During game development: Help in behavior/content design After game deployment: Analysis of game data

22 Outline Structure of the course Intro to Game AI Examples of Game AI Basic Game AI techniques Course Scope Paper presentations Projects

23 Basic Game AI Techniques Scripting: Finite State Machines, Behavior Trees Path-finding: A* Game Tree Search: Minimax Driving games: Steering behaviors [Physics: inverse kinematics] Plus a lot of fine-tuning

24 Scripting Most used game AI technique by far (99% of the games) Predefine behaviors of characters in the game Examples: Classic games (Pac-man, space invaders.), platformers (mario, turrican, etc.), strategy games (Dune II, Starcraft, etc.) Etc. Standard techniques: Finite-state machines Behavior trees Rule-sets High-level languages (e.g. lua)

25 Path-finding For navigating in complex maps Examples: Strategy games First-person shooters (Doom, Half-Life, etc.) RPGs (Skyrim) Standard techniques: A* Waypoint maps

26 Game Tree Search Adversarial games Examples: Board games (chess, checkers, backgamon, etc.) Standard techniques: Alpha-beta search (minimax)

27 Basic Game AI Techniques Not the goal of this course These basic Game AI techniques are covered in the intro to Game AI course (next quarter)

28 Outline Structure of the course Intro to Game AI Examples of Game AI Basic Game AI techniques Course Scope Paper presentations Projects

29 Course Scope Advanced AI techniques: recent theoretical developments with new applications to commercial games. Real-Time Strategy Games Drama Management Procedural Content Generation

30 Course Scope Advanced AI techniques: recent theoretical developments with new applications to commercial games. Real-Time Strategy Games Flexible smart architectures for RTS game AI Real-time path finding for large maps, with large number of units Complex intelligent decision making Drama Management Procedural Content Generation

31 Course Scope Advanced AI techniques: recent theoretical developments with new applications to commercial games. Real-Time Strategy Games Drama Management Self-adapting games Player modeling Automatic interestingness evaluation Procedural Content Generation

32 Course Scope Advanced AI techniques: recent theoretical developments with new applications to commercial games. Real-Time Strategy Games Drama Management Procedural Content Generation Automatic map and level generation Automatic story and plot generation

33 Outline Structure of the course Intro to Game AI Examples of Game AI Basic Game AI techniques Course Scope Paper presentations Projects

34 Paper Presentations Each student will have to read a paper and give a minutes presentation in class about it Presented papers will concern related topics to what was covered in class the previous week For copyright reasons, I ll give you the private URL for downloading the papers in class

35 Outline Structure of the course Intro to Game AI Examples of Game AI Basic Game AI techniques Course Scope Paper presentations Projects

36 Projects Group projects (2 people per group) 3 projects Each group will pick one as their focus project Focus project: The group will write a 6-page (AAAI proceedings style format) about their project Students are encouraged to consider submitting their paper to a Game AI conference if they consider the obtained results interesting. Non-focus projects: Each group will have up to 20 minutes of time at the end of the course to present them in class

37 Projects Goal The goal of the projects is not to implement some techniques and see that they work The goal is to understand their strengths and limitations Before starting the implementation of a project think what is that you want to understand, and don t lose time with aspects of the implementation that are irrelevant to your experiments. You will be required to experiment with your implementation and present experimental results showing strengths and limitations of the implemented techniques

38 Projects Experimentation The goal of your experiments is not to show that the algorithm you implemented is the best The goal is to understand the algorithm: When does it work, and when does it not? Why? What could be done to improve it? Does it achieve the desired effect? Results showing an algorithm doesn t work do not mean your project is bad.

39 Project 1: Real-Time Strategy Games You will have to implement an AI bot for a real RTS game. I suggest using one of these: Starcraft S3

40 Project 1: Real-Time Strategy Games Your AI bot needs to: Play the complete game: Doesn t have to be very good. Use, for example, a simple finite-state machine technique to design the overall strategy of your bot. Implement one of the advanced techniques we will study in class concerning: real-time path-finding, or decision making. Experiment with your bot: Make it play against the built-in AI, or against other existing Ais Compare the performance of your bot before and after implementing the advanced technique: Does it play better or worse? Is it more or less fun to play against?

41 Project 2: Drama Management Implement a Drama Manager on top of a game engine. I will provide you with one, but feel free to use another. The game engine comes with a specific game defined. Feel free to modify it or extend it if you need additional elements in your game to test your DM

42 Project 2: Drama Management Identify what do you want your Drama Manager to accomplish (control difficulty, lead the player to more enjoyable plots, etc.) That will determine what actions do you want the DM to be able to perform in the game Decide what technique you will use (search-based, CBR, etc.), and if it is search-based you will need to define the function to optimize Experiment with your DM: Ask some friends to play the game with or without it, and see their reactions. Does the DM achieve its goal? Is the game more fun?

43 Project 3: Procedural Content Generation Add procedural content generation techniques to a game engine. Use the same game engine you used for Project 2 You can choose whether you want the system to generate maps/rooms or story

44 Project 3: Procedural Content Generation Determine exactly the extent of what you want to be procedurally generated: Given a story, generate maps? Given a story and a high-level map, generate rooms? Given a map and rooms, generate a story? Etc. Pick a PCG technique (search-based, CBR, etc.) Experiment with the result: Is the generated content better than the original one? Was it cheaper to generate? Are the games with content generated procedurally as fun as the original one?