CSE 348 - AI Game Programming

Spring 2008

Announcements

** Our schedule for the whole semester is now complete. Each of you have several tasks (class presentation; projects as slave or master). Note that some of you have assignments as early as this Tuesday and that class presentations start in early February. Make sure that you are prepared with your tasks at the due date. These cannot be postponed. Please make sure that when you schedule outside events (e.g., interviews) these do not collide with the dates when you have an assigned task.

** Please read carefully the updated guidelines below for presentations and for projects (slave and master). Students must ensure that they fulfill these tasks at the scheduled dates

** Hall of Fame

• Project # 1. Robocode.
  Tournament winner: Aperture Science (Hinkey, Christopher M.; Karneeb, Justin T.; Gillespie, Kellen J). Aperture Science team consisted of different kinds of bots executing concrete strategies that in combination resulted in a solid team strategy without the need of implementing an explicit coordination between the bots. Aperture Science destroyed the competition in this 2-bracket tournament.
• Project # 2. Unreal Tournament.
  • Tournament winner: Aperture Science (Hinkey, Christopher M.; Karneeb, Justin T.; Gillespie, Kellen J). The reigning champions, the Aperture Science team, keep their throne for another tournament (is there anyone who can beat these guys?). They developed a chatter class to allow communication between bots. Bots follow initially a search strategy, followed by a dominations strategy when all nodes are found. They encoded 11 different tactics (they called this "cases"), and on-the-fly decide which tactic to follow.
  • Innovation winner: Reretaliate (Brander, Greg; Moukhine, Nick; Wojciechowski, Chris). Team Reretaliate carefully studied the retaliate algorithm that I had presented in class. Observed a flaw in the way it operates (e.g., doesn't take into acount the current's bot location). And went on to implement their own version of retalaite that addresses this flaw!. Not only this but the actual implementation is much cleaner than the original Retaliate.
• Project # 3. Call To Power 2.
  • Tournament winner: Act-Fail (Morrison, J., Ziegler, J., Auslander, B.). The new champions! Act-Fail had the best score (barbarian kills + summation of city sizes) in the non-customized map. Act-Fail builds a civilization by constructing a series of cities spawning in a balanced fashion from the starting city.
  • Innovation winner: Aperture Science (Hinkey, Christopher M.; Karneeb, Justin T.; Gillespie, Kellen J). Aperture Science builds an elegant solution for the game model and encapsulated low level messages. This encapsulates low level details from the user and allows the user to concentrate on high-level strategies. We will be using their code as an starting point of future iterations of this class.
• Project # 4. Wargus.
  • Tournament winner: Gryphon Society (Moukhine, N., Brander, G., and Wojciechowski, C.). Team Gryphon Society realized that mobility is a key for the game and no other unit provides betetr mobility than the gryphon. A mass of gryphons destroyed the competition.

Description

Over the past 20 years computer games have gone through a remarkable evolution in areas such as graphics and multi-player gaming. Another area where some evolution has been made is in the quality of the computer opponent. Despite this evolution, there is a big mismatch between what people in the game industry refer to as the "AI" (roughly, the program controlling the computer opponent) and what AI actually is ("...technology can be controlled especially if it is saturated with intelligence to watch over how it goes, to keep accounts, to prevent errors, and to provide wisdom to each decision"- Allen Newell, 1992). In this course, we will look at contemporary computer games, explore techniques for implementing the AI (from the perspective of the game industry) and study opportunities to use AI (from the research perspective) to enhance the gaming experience. Here are some of the topics that will be covered in the course:

* Pathfinding and Navigation Systems 
* Group Movement, Tactics, and Planning
* Applying Action Planning into Games 
* Adaptive Games
* FPS, Team-based RTS, and Turn-based games 
* Machine Scripting Language for Game Designers
* Player Modeling

Prerequisites: CSE 327 or CSE 109.

Trivia about Game AI

Reading Material

Book: (recommended)

       Title: AI Game Programming Wisdom series

       Author: Steve Rabin (Editor)

       Publisher: Charles River Media

Communication

All announcements, handouts, etc. will be posted in this web site:

www.cse.lehigh.edu/~munoz/CSE348

Topics

• Introduction
• Tue. Jan 22.: Presentation Robocode project (Emily Cohen, John Mercer, and David Heefner) AI Game Research at Lehigh and Automated Planning
• Thur. Jan. 24: Automated Planning paradigm: HTN Planning
• Tue. Jan 29.: Machine learning and a case study of using reinforcement learning in games
• Thur. Jan. 31: Robocode project due;
• Tue. Feb. 5: Finite State Machines in Games (Stanislav Tsanev); Presentation Unreal Tournament project (Bryan Auslander Jonathan M. Ziegler, and James Morrison)
• Thur. Feb. 7: Scripting languages (Karneeb, Justin T) and expert systems
• Tue. Feb. 12: First Person Shooters (Xavier Otero Keil)
• Thur. Feb. 14: Squad tactics (Bryan Auslander)
• Tue. Feb. 19: Pathfinding with the A* algorithm (Joseph Siefers)
• Thur. Feb. 21: General Pathfinding: tables and navigation (Jeremy David Christman)
• Tue. Feb. 26: General Pathfinding: waypoints (Sarunac, Alexandar D. )
• Thur. Feb. 28: Unreal Tournament project due; Presentation Call To Power 2 project (Responsible: Matt Dilts and Jeremy D. Christman)
• Tue. Mar. 4: Spring break
• Thur. Mar. 6: Spring break
• Tue. Mar. 11: Camera control (Christopher M. Hinkey)
• Thur. Mar. 13: Player Modeling (Jim Morrison)
• Tue. Mar. 18: Role Playing Games (Nick Moukhine)
• Thur. Mar. 20: Individual NPC behavior (Jonathan Michael Ziegler )
• Tue. Mar. 25: Real-Time Strategy games (Dave Heefner)
• Thur. Mar. 28: Achieving gaming goals (Kellen J. Gillespie)
• Tue. April 1: Call To Power 2 project due; Presentation Wargus Project (Xavier Otero Keil, Joe Siefers, and Kristen Gardner)
• Thur. April 3: Map, World, Wall generation (Mercer, Jonathan C.)
• Tue. April 8: Story line, drama (Brander, Gregory M.)
• Thur. April 10: Crowd/swarm simulations (Kristen Gardner)
• Tue. April 15: Racing games (Emily Cohen )
• Thur. April 17: Sports Games (Chris Wojciechowski )
• Tue. April 22: Learning (Matt Dilts )
• Thur. April 24: Wargus project due; RECAP
• Last Day of Classes: April 25, Friday

Student's Work

This is a work-heavy course. The following activities will be performed by the students taking this course:

• Oral presentations. Each student will prepare and present in class sections from the book series  AI Game Programming Wisdom. Guidelines:
  • The students are required to meet with the instructor 1 week before the day of the presentation. For this meeting, the student must have completed his/her presentation using MS Powerpoint. 
  • The presentation should go beyond the assigned readings. It should include other sources. Consult the references that are cited in the chapter(s). Find your own sources.
  • If possible, it will include demo software (the CDs that come with the book are a good starting point)
  • Prepare for a 1 hour presentation. Presentations not well prepared are easy to spot. Start going through the material sooner rather than later.

• Written report. Each student will prepare a written report (around 15 pages) about the topics covered in his/her class presentation. This report will be distributed among classmates for reference material for the final exam together with the powerpoint presentations (including the instructor). Guidelines:
  • Use own words. Do not copy-paste from sources
  • due: the second class after your presentation
  • be as clear as possible. This will be the main source for preparing the exam

• Class participation. Attendance to class is required.

• Final exam. There will be a written exam about the topics covered in the course. The exam will not be a problem solving exam (e.g., write an algorithm to accomplish a certain task). Rather, the exam will test the overall knowledge from the topics covered in the course (e.g., open questions about the material). The main source for this are your classmates written reports, their class presentations, and the instructor's class presentations. Here is the first group and here is the second group of reports. Here is the third group.

• Programming projects. We will implement various programing projects in groups of 2-3 depending on enrollment, as follows:
  1. Project 1: Robocode (teams) (due: January 31st; Responsible: Emily Cohen, John Mercer, and David Heefner)
  2. Project 2: Unreal Tournament - Tournament (due: February 28; Responsible: Bryan Auslander Jonathan M. Ziegler, and James Morrison) Updated
  3. Project 3: Call To Power 2 (due: April 1st; Responsible: Matt Dilts and Jeremy D. Christman)
  4. Project 4: Wargus (due: April 24th; Responsible: Xavier Otero Keil, Joe Siefers, and Kristen Gardner)
  Guidelines:
  • Students will will form groups of 2-3 persons. We will try to maintain the groups throughout the semester. For each project there will be one and only one master group and the rest of the groups will be slave groups.
  • Master groups will be responsible for all activities related to managing the project, as follows:
    • Setting up the software making sure that it runs
    • Designing the competition (objective, scenario)
    • Creating a web page describing the competition
    • Presenting the project in class. This presentation should include a live demo of the system
    • Providing support to slave groups
    • Administrating the gaming competition on the due date
    • Master groups should not participate in the competition they are responsible for.
    • No group should play the role of master group in more than one project
    • Since not all will be able to be "master" groups, extra credit in the form of additional points to the final exam will be given to those who assume this role, if project was successfully lead by the master group. Some may required more effort than others and therefore will get more points.
    • In the remote event of an unsuccessful project because of omission of master group to accomplish the tasks indicated above, this will affect negatively the score of the final exam for all members of that master group (note: this have NEVER happened before)
  • Slave groups will be responsible for working on the project and participating in the competition.
    • Projects should be e-mailed to the master group on the due date (11:59PM EST), the day before the competition is held in class.
    • Presentation: In class I will select randomly one member of each slave group to make the presentation of the project. Bring powerpoint slides and if possible install them before the class starts. Do not explain what is the project about. Explain your solution.
    • A printout of the code, together with a 5-page report must be turned in for the class competition.

  Percentage distribution: presentation (13%), Written report (10%), 4 programming projects (13% each. Total: 52%), Final Exam (25%)

  ---

  Last update: Fri. April. 25 11:00:33 EST 2008