Dr Maurice Hendrix Researcher Link workshop, Brazil, 21 Mar 2017 Personalisation in educational games and virtual environments: Difficulty balancing, procedural content generation and VR Dr Maurice Hendrix Researcher Link workshop, Brazil, 21 Mar 2017
Outline Educational games Personalisation VR Open questions
42% of Americans play regularly Average 13 years experience Computer Games Electronic game involving interaction with a computer Large proportion of population plays Large industry for entertainment games 26% 27% 44% 56% 17% 30% AGE of Game Players 26% under 18 years 30% 18-35 years 17% 36-49 years 27% 50+ years GENDER of Game Players 56% male 44% female 42% of Americans play regularly Average gamer is 35 years old Average 13 years experience
Types of Serious Games
Educational Games Games often used as they are seen as engaging Issues: Competing with commercial games Different players (styles, knowledge, gaming abilities) Depending on game design: representation on a screen has limitations for immersion Often scenario based and very limited replayability
VR Adaptation
Personalisation What to adapt in a game? Depends on game design Content Choose from alternative Procedural generation
Personalisation What to adapt to? Learning styles? Controversial Background knowledge / skills? Many games stand-alone or poorly integrated (don’t know background) Csikszentmihalyi's Flow? Difficulty balancing is a ‘low hanging fruit’
Case study: obesity prevention (pegaso eu project)
6 steps for difficulty balancing Identify what in the game indicates difficulty by examining the gameplay and victory conditions. Identify what of these elements we can influence? If an implementation exists, locate variables in code. Consider whether the game features multiple mechanics and if so, if they are dependant or independent. Decide how variables will be used to calculate difficulty. Decide upon sensible starting values.
6 steps for difficulty balancing Indicators of difficulty zombie mechanic: number killed, score, nights survived Food matching: amount of food, XP, nights survived We can influence: Amount, and speed of zombie enemies faced. Amount of possible matches on boards and size of boards Variables located in code (scripts in Unity 3D engine) Multiple mechanics. Semi-independant Decide how variables will be used to calculate difficulty. Weighted average Decide upon sensible starting values. Empirical with a few volunteers
Evaluation? Also conducted with small platform game Case studies show how steps can be used by developers Participants like adapted versions better But small scale and brief so no meaningful data on learning (yet)
Case study: Kayaking VR rehab Kayak vr rehab learnig
Kayaking VR rehab Need to influence the environment (as with platformer) Conducted using procedural generation Motion sickness can be a problem
Open questions & further directions Game engine plugin (Unity 3D) Is engine independent framework possible? In VR – how does generation of content effect cyber sickness? Can we adapt to a persons susceptibility? Practical way to get background knowledge / skills or dos it need VLE integration?