Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Showcase of My Research on Games & AI "till the end of Oct. 2014"

1,128 views

Published on

A presentation showcasing my research on Games and Artificial Intelligence (till the end of Oct. 2014) at IT University of Copenhagen, Copenhagen, Denmark.

Published in: Technology
  • Be the first to comment

Showcase of My Research on Games & AI "till the end of Oct. 2014"

  1. 1. I will talk about •Games, only games. –STYX Engine: Adaptive Personalized Content Generation for FPS Games –Ropossum: Evolutionary-based Authoring Tool for Physics-based Games –NEXT: 2D Puzzle, Progressively Generated –iNversion: 2D Puzzle, Progressively Generated –4 more new games between now and the beginning of 2015. –Supervised Projects with Noor Shaker •InfiSpelunky: Infinite Procedurally Generated Spelunky •Kinect and Immersive 3D Environment •Weebee on a Mission: A Serious Game
  2. 2. Adaptive Personalized Content Generation for FPS GamesFourth Year Project, 2012-2013
  3. 3. Marry someoneand break your addiction Employedor NOT Employed people are 2Xmore likely to feel anxiety and 2.25Xlikely to experience anger Singlevs. Married You are X2likely to be addicted on angry birds if you are single than if you are married
  4. 4. 6% Very anxious 17% Somewhat anxious 22% Neither 32% Somewhat relaxed 23% Very relaxed Player Emotion After Playing Angry Birds
  5. 5. Simple Easy Just right Tough 3% 6% 60% 31% Player Reported Difficulty for Angry Birds
  6. 6. Publications •A Quantitative Approach for Modeling and Personalizing Player Experience in First-Person Shooter Games,in the Extended Proceedings of the 2013 Conference on User Modeling, Adaptation and Personalization(UMAP 2013), 2013. •Personalizing Content Generation in First Person Shooter Games through Player Modeling.Submitted on Oct. 2014 to The Scientific World Journal, Special issue in "Recent Advances in Intelligent Techniques for Games", 2015.
  7. 7. Demo
  8. 8. The Big Picture of Player Modeling Game Player Player Experience Model Adaptation Model
  9. 9. Enforced Controllable features Gameplay features Prediction of player’s emotion Exhaustive search Towards Adaptation
  10. 10. So, how to do it in FPS game?
  11. 11. Levels Design Preference Learning Model Adaptive Content Generation Model level1 level2 Adapt level20 Adapt Adapt level21 levelN Adapt
  12. 12. Levels Generation Black and White
  13. 13. Items Placement (SOM)
  14. 14. Data Collection A B •60 players at F.I.T.E of Damascus, Syria
  15. 15. Building the Models Enforced Controllable features Gameplay features Prediction of player’s emotion Exhaustive search
  16. 16. Two Players, Engagement
  17. 17. Facial Expression
  18. 18. Facial Expression
  19. 19. Facial Expression
  20. 20. Facial Expression
  21. 21. Facial Expression
  22. 22. Facial Expression
  23. 23. Facial Expression
  24. 24. Facial Expression
  25. 25. Facial Expression
  26. 26. Facial Expression
  27. 27. Facial Expression
  28. 28. Nonverbal Signatures of Engagement in Super Mario Bros2013
  29. 29. Expressions during Gameplay
  30. 30. Behavioral Features •Visual Reaction (VR) –Bias of head on the x-axis compared to first frame(Avg. + STD) –Bias of head on the y-axis compared to first frame(Avg. + STD) –Left eye closed (Avg. + STD) –Right eye closed (Avg. + STD) –Mouth open (Avg. + STD) •Facial Expression (FE) –Angry % (Avg. + STD) –Happy % (Avg. + STD) –Sad % (Avg. + STD) –Surprised % (Avg. + STD)
  31. 31. Behavioral features Players’ reports of affect NeuroevolutionaryPreference Learning
  32. 32. Results
  33. 33. Publication Noor Shaker and Mohammad Shaker. Towards Understanding the Nonverbal Signatures of Engagement in Super Mario Bros, in Proceedings of the 2014 Conference on User Modeling, Adaptation and Personalization (UMAP 2014),2014.
  34. 34. Ropossum: Evolutionary-based Authoring Tool for Physics-based GamesGraduation Thesis, 2013
  35. 35. -Popular physics based game -Can generate endless levels -All levels should be playable -Opens the imagination of all players to design, test, modifytheir own levels and helpthem achieve that.
  36. 36. Automatic Generation of Content forPhysics-based GamesPart of Graduation Thesis, 2013
  37. 37. Placing Components on Canvas
  38. 38. If the game is 5 sec long, then we have 5 * 60 = 300 update Given we have 5 actions, then we need to evaluate 243 e+10 states
  39. 39. Simulation-based Agent for Generating Playable ContentPart of Graduation Thesis, 2013
  40. 40. 205 explored nodes Solution Tree
  41. 41. Result 470.1 ±525.4 secfor generating a playable level
  42. 42. Samples
  43. 43. Graduation Thesis Publications •Evolving Playable Content for Cut the Rope through a Simulation-Based Approach, in Proceedings of Artificial Intelligence and Interactive Digital Entertainment (AIIDE 13), 2013. •Ropossum: An Authoring Tool for Designing, Optimizing and Solving Cut the Rope Levels, in Proceedings of Artificial Intelligence and Interactive Digital Entertainment (AIIDE 13), 2013. •Automatic Generation and Analysis of Physics-Based Puzzle Games, in Proceedings of the 2013 IEEE Conference on Computational Intelligence and Games (CIG 2013), 2013. Nominated for best paper award.
  44. 44. Demo
  45. 45. Projection-based Agent for Generating Playable Content for Physics-based Games2014
  46. 46. Projection-based Agent
  47. 47. Projection-based Agent
  48. 48. Projection-based Agent
  49. 49. Projection-based Agent
  50. 50. Projection-based Agent
  51. 51. Projection-based Agent
  52. 52. Projection-based Agent
  53. 53. Projection-based Agent
  54. 54. Demo
  55. 55. Supervised Projects2013 and 2014
  56. 56. InfiSpelunkyA Procedural Method for Automatic Generation of SpelunkyLevelsWalaaBaghdadi, Fawzya Shams Eddin, RawanAl-Omari, ZienaAlhalawani, Mohammad Shaker and Noor Shaker2013
  57. 57. Spelunky Level Structure
  58. 58. Evolutionary-based Generation
  59. 59. Controlling Difficulty 10%90%
  60. 60. Results Analysis, Solution Path Length 10%50%90%
  61. 61. Weebee on a MissionA Serious Game for Better Understanding the Behavior Differences Between ChildrenRawanAl-Omari, WalaaBaghdadi, ZienaAlhalawani, Mohammad Shaker and Noor Shaker2014
  62. 62. On a Mission
  63. 63. A Survey Survey Intelligent robot (a bot) Game 6% 11% 83% In case you want a product to inspect and change your child behavior, what would it be?
  64. 64. A Survey Discuss it Indirect Influence Ignore it 41% 52% 6% If there is a problem in your child behavior, what would do you do?
  65. 65. Questionnaire to Game Scenario Question Answer
  66. 66. Questionnaire to Game Scenario Question Game Scenario
  67. 67. Questionnaire to Game Scenario “Do you help other kids in need?”
  68. 68. Game Environments Park School Kitchen
  69. 69. Data Collection 8-12 years old children 100 players [50 males, 50 females]
  70. 70. Utilizing Kinect Control for More Immersive Interaction with 3D EnvironmentsSaeed Hajali, Kinda Tarboush, Marah Halaweh, Mohammad Shaker and Noor Shaker2014
  71. 71. “Why using a stylus when God gave us five fingers!” Steve Jobs when introducing the first iPhone, 2010
  72. 72. Hand Gesture Voice Command Interaction with the Environment
  73. 73. The Immersion Process Player Sketch Generating basic Level Canvas Allow Edit and Generation of Content
  74. 74. Select Rotate Delete Move The gestures that we use
  75. 75. The Application
  76. 76. One more thing..
  77. 77. A Progressive, Evolutionary-based Approach for Generating Playable Content for Games 2014
  78. 78. Time(MS) Level Timeline
  79. 79. Level End LevelStart Time(MS) Level Timeline
  80. 80. 0 300 500 1000 1700 2800 Action02 Level End LevelStart Time(MS) Action04 Action03 Action05 Action01 Level Timeline
  81. 81. 0 300 500 1000 1700 2800 Rope cut Level End LevelStart Time(MS) Rocket press OmNom feed Rope cut Level Timeline in Cut the Rope
  82. 82. 0 300 500 1000 1700 2800 Dock Down Level End LevelStart Time(MS) Collect Stomp Win Jump Level Timeline in Super Mario Bros
  83. 83. Why is it Progressive/Aggressive? Game Timeline Generator
  84. 84. Why is it Progressive/Aggressive? Game Timeline Generator Grammatical Evolution Individuals 0 300 500 1000 1700 2800 Rope cut Level End LevelStart Time(MS) Rocket press Aircush press OmNom feed Rope cut 0 300 500 1000 1700 3200 Rope cut Level End LevelStart Time(MS) Rocket press Bubble Press OmNom feed Rope cut 0 300 500 1000 1700 2800 Rope cut Level End LevelStart Time(MS) Rocket press Aircush press OmNom feed Rope cut
  85. 85. Why is it Progressive/Aggressive? Game Timeline Generator Game Simulator Grammatical Evolution Individuals 0 300 500 1000 1700 2800 Rope cut Level End LevelStart Time(MS) Rocket press Aircush press OmNom feed Rope cut 0 300 500 1000 1700 3200 Rope cut Level End LevelStart Time(MS) Rocket press Bubble Press OmNom feed Rope cut 0 300 500 1000 1700 2800 Rope cut Level End LevelStart Time(MS) Rocket press Aircush press OmNom feed Rope cut
  86. 86. Why is it Progressive/Aggressive? Game Timeline Generator Game Simulator Grammatical Evolution Individuals 0 300 500 1000 1700 2800 Rope cut Level End LevelStart Time(MS) Rocket press Aircush press OmNom feed Rope cut 0 300 500 1000 1700 3200 Rope cut Level End LevelStart Time(MS) Rocket press Bubble Press OmNom feed Rope cut 0 300 500 1000 1700 2800 Rope cut Level End LevelStart Time(MS) Rocket press Aircush press OmNom feed Rope cut Mapper Aircush press From TLE to Level Structure
  87. 87. Why is it Progressive/Aggressive? Game Timeline Generator Game Simulator Grammatical Evolution Individuals 0 300 500 1000 1700 2800 Rope cut Level End LevelStart Time(MS) Rocket press Aircush press OmNom feed Rope cut 0 300 500 1000 1700 3200 Rope cut Level End LevelStart Time(MS) Rocket press Bubble Press OmNom feed Rope cut 0 300 500 1000 1700 2800 Rope cut Level End LevelStart Time(MS) Rocket press Aircush press OmNom feed Rope cut Mapper Simulation, Mechanics, Game Rules Simulator Aircush press From TLE to Level Structure
  88. 88. Why is it Progressive/Aggressive? Game Timeline Generator Game Simulator Grammatical Evolution Individuals 0 300 500 1000 1700 2800 Rope cut Level End LevelStart Time(MS) Rocket press Aircush press OmNom feed Rope cut 0 300 500 1000 1700 3200 Rope cut Level End LevelStart Time(MS) Rocket press Bubble Press OmNom feed Rope cut 0 300 500 1000 1700 2800 Rope cut Level End LevelStart Time(MS) Rocket press Aircush press OmNom feed Rope cut Mapper Simulation, Mechanics, Game Rules Simulator Aircush press From TLE to Level Structure Assign Fitness Score
  89. 89. The Mapping Process Start with ropes Rope_cut Rope_cut Aircushion_press rope_cut(200) rope_cut(500) aircuhion_press(700) rocket_press(1100) OmNom_feed(0)
  90. 90. The Mapping Process Rocket_press Rocket_press Generated Level Structure rope_cut(200) rope_cut(500) aircuhion_press(700) rocket_press(1100) OmNom_feed(0)
  91. 91. Ok, this is coolBut is it just Cut the Rope that canbe progressivelygenerated?
  92. 92. Nope.
  93. 93. Did you apply it somewhere else?
  94. 94. Yeeeeah.
  95. 95. NEXT and iNversion visit: www.mohammadshaker.com
  96. 96. NEXTwww.mohammadshaker.com/next.html
  97. 97. NEXTwww.mohammadshaker.com/next.html
  98. 98. iNversionwww.mohammadshaker.com/inversion.html
  99. 99. Keep tuned cause more interesting stuff are coming!
  100. 100. Keep updated at: www.mohammadshaker.com Fournew games at the beginning of 2015[ they are.. ]
  101. 101. Keep updated at: www.mohammadshaker.com Fournew games at the beginning of 2015[ they are.. ]
  102. 102. Visit my website, play NEXT, iNversion. www.mohammadshaker.com@ZGTRShakerLet’s connect!
  103. 103. Thank you
  104. 104. Thank you Noona

×