TGD3351 Game Algorithms TGP2281 Games Programming III. in my own words, better known as Game AI

Transcription

1 TGD3351 Game Algorithms TGP2281 Games Programming III in my own words, better known as Game AI

2 An Introduction to Video Game AI

3 A round of introduction

4 In a nutshell B.CS (GD Specialization) Game Design Fundamentals Game Physics Game Algorithms 3D Game Programming Game Production B.MM SEGD Games Programming I (Game Fundamentals) Games Programming II (3D Game Dev) Games Programming III (Game AI)

5 Course matters Classes: (L) Tuesdays, (T) Thursdays, Consultation hours: Tuesdays, AR4004 (Lab) All other times by appointment Refer to GameAlgo course website for more details MMLS For announcements To upload only selected materials

6 Textbook and References AI for Game Developers, Bourg & Seeman Artificial Intelligence for Games, Millington Algorithms and Networking for Computer Games, Smed & Hakonen Programming Game AI by Example, Buckland Learning XNA 4.0, Reed

7 Disclaimer This course DOES NOT teach you programming or tell you what technologies are out there It is about algorithms and AI in games

8 C# for Games If you already are familiar with Java, this isn t that far away Why not Unity3D? UDK? The good Allows flexibility of designing game classes.net framework (great support if you ask me) The bad XNA Game Studio is dead (not really, but there s no dev anymore)

9 MonoGame Managed code (C# and other.net languages) Cross-platform Very-very-very similar to XNA (hence tutorials reusable) Open-source Growing community

10 Coursework Distribution A 100% Coursework subject B.CS Coursework Game Project 60% Test 20% Quiz/Homework 20% SEGD Coursework Game Project 70% Test 15% Quiz/Homework 15%

11 Project All students MUST hand in the project Project must be done in pairs (3-in-a-group or solo* not allowed) Game with distinctive AI features Details will be ready when the guidelines are out 4 checkpoints: Game Proposal, Progress Milestones #1, #2, Final Submission Demos for Milestones (W6, W10) Presentation for Game Proposal (W3) & Final Submission (W14)

12 Project You are to propose an original game that will best showcase your AI & technical accomplishments Contribution statement At the end of project, submit a contribution statement through Distribute 100 points between you and your partner Highlight your contributions to the project

13 Other Coursework Test Written test on your understanding of AI theory, concepts, application (no coding!) Quiz/Homework Class quizzes OR Short homework pieces in the form of theory or concept questions To be handed up on the same day or after a few days

14 Course policies 1. Collaboration Policy (Honour Code) Acknowledge the people you ascertain help from (other students, external parties) Acknowledge material taken from elsewhere Acknowledge source of code used No acknowledgement or referencing is same as saying I did everything myself when you did not.

15 Course policies 2. Late Policy Every student given a total of 2 late days to use on any coursework submission (except for Test) If a project submission is late, late days consumed for both project members No need informing to take late days. Your late days will be automatically deducted (if available) when you are late. Once the late days run out, any late submissions will be penalized 2% (of total coursework) per day late. No submissions allowed after the hard deadline.

16 About coding 1. This is not a C# or XNA or MonoGame course 2. You are expected to be able to pick up quickly Self-learning time is important Materials/guidance will be provided 3. What you can do for project depends on how much you are able to learn up

17 Toilet break Any further questions about the course?

18 AI Topics Game AI Fundamentals Movements Basic Movements Complex Movements & Flocking Pathfinding Basic Pathfinding A* Pathfinding and Extensions Decision Making Finite State Machines Decision Trees and Rule-based AI Fuzzy Logic Probability and Randomness

19 Topics Advanced Topics Tactical AI Turn-based AI: Game Trees & Tournaments* Neural Networks and Beyond* * Tentative

20 Let s look back in time

21 Early AI in Games Atari s Pong one of the first video games ever created and released to the public Tracking AI: How does it work?

22 Board games Video games Backgammon on Atari 2600 Pathfinding AI

23 Space Invaders Pattern AI. Moving and reacting based on patterns

24 Obsolete? Are these early ideas in Game AI still used in today s games? What do you think and WHY?

25 First-Person Shooters Quake 3 How do AI bots find cover under fire? Way points? Line of sight computations

26 First-Person Shooters Halo 3 Acting together: Team or group-based AI Coordination: Fixed / Adaptive

27 Platform Games Johnny Platform 100 levels of different difficulty, puzzle based challenges Arbitrarily sliding platforms Enemies that are well timed and moving strategically Platform game are more player-centric rather than AIcentric

28 Warcraft 2 Real-Time Strategy (RTS) What are strategies consider in an RTS game? Managing resources / economy Scripting

29 Europa Universalis IV Strategy Games Player competes with the computer AI to expand power through military, diplomacy and colonial wealth Makes strategic decisions and preferences, decides when to go war, manages economy and tech development Real time

30 Simulation Games The Sims Artificial life Programmable / adaptive behaviours, emotions Open-ended concept

31 Black and White Simulation Games Artificial life, evolution God-role: A creature can be taught and raised by the player Metacritic (90%), GameSpot (9.3/10)

32 Action RPG Games Elder Scrolls Oblivion & Skyrim Radiant AI : NPCs can make choices and engage in unpredictable behaviors with the world around them, e.g. Eat at this city at 2PM Dynamic creation of quests placed into the world Organic feel

33 Board Game / Turn-based AI Othello/Reversi Use information in game to determine best moves (optimal solutions) in the near future Minimax and other searching techniques Heuristics to speed up calculations

34 Game-Playing AI Genetic Algorithms & Neural Networks NN with 14 inputs, 20 hidden neurons, 5 outputs trained by a genetic algorithm

35 Game-Playing AI Genetic Algorithms & Neural Networks Code:

36 Reinforcement Learning Training a neural network (NN) to learn better Atari Breakout game Reward-based learning

37 Reinforcement Learning (DQN) Paper:

38 Coding: Get started Get Visual Studio Community (2013 or 2015). Sign up for Microsoft account to make it free (just need renewal from time to time) Get MonoGame (install or fork it) Start on some simple tutorials Series.aspx game-development-using-c-xna-and-monogame-3-0-building-a-shootergame-walkthrough-part-1-overview-installation-monogame-3-0-projectcreation/

39 Any Questions?

40 We shall play some games in this week s lab

41 Play & Tell Pong Pacman Galcon Class Discussion: Can you identify (in your own words) the game AI that is in these games?