Checking the difficulty of evolutionary-generated maps in a N-Body inspired mobile game (OLA2020)
0 likes57 views
This document discusses using an evolutionary algorithm to procedurally generate maps of varying difficulty for a mobile game called Gravityvolve, which simulates the n-body problem in physics. An experiment was conducted with 30 players voting on the difficulty of 21 generated maps both with and without a help line. The results showed that maps were often voted as too easy, and the help line made maps seem easier than intended. The map generation algorithm needs improvement to better match the difficulties felt by players.
Checking the difficulty of evolutionary-generated maps in a N-Body inspired mobile game (OLA2020)
- 1. Checking the difficulty of evolutionary-generated maps in a N-Body inspired mobile game C. López-Rodríguez, A.J. Fernández-Leiva, R. Lara-Cabrera, A.M. Mora, P. García-Sánchez
- 2. The N-Body Problem • In physics, it is the problem of predicting the individual motions of a group of celestial objects interacting with each other gravitationally. 2 plus.maths.org Youtube: phuebu1 Wikipedia
- 3. Gravityvolve Game • Physics-based Android game. • A kind of ‘space golf’ In order to complete a map, the player will have to push the particle from its initial position until it reaches the circumference's position. 3 Search for “GravityVolve” in Google Play
- 4. 4 Gravityvolve Game Search for “GravityVolve” in Google Play
- 5. 5 Gravityvolve Game Search for “GravityVolve” in Google Play
- 6. 6 Evolutionary Procedural Content Generation • A tool for generating maps automatically. • Depending of the chosen difficulty. • Based on an Evolutionary Algorithm.
- 7. 7 Evolutionary Procedural Content Generation • Algorithms that autonomously and dynamically create content (maps, items, characters, stories). • A creative and time-saving tool. • A way to do adaptive (to the player) games. • Potentially infinite games. • The generated content always fulfils certain restrictions.
- 8. 8 Evolutionary Procedural Content Generation • An Evolutionary Algorithm is a metaheuristic inspired in natural evolution. • It a population-based method. • Optimization algorithm.
- 9. 9 Evolutionary Procedural Content Generation Wikipedia i population initialization f(x) evaluation of individuals Se selection mechanism Cr crossover operator Mu mutation operator Re replacement policy
- 10. 10 Evolutionary Procedural Content Generation • Individual • Steady-state EA. • Binary Tournament selection. • Specific Mutation operator (random increase/decrease of coordinates, mass or angle). • Fitness: feasible distribution, solvable map.
- 11. 11 Evolutionary Procedural Content Generation • Crossover operator Parents Offspring
- 12. 12 Difficulty Evaluation • We aim to evaluate the difficulty of the generated maps. • The player will vote (EASY, MEDIUM, HARD) after completing a map. • With and without the help line. • Checking the correlation validate the PCG algorithm.
- 14. 14 Difficulty Evaluation Experiment • 30 players / 21 maps per player. • With and without the help line (HL). (after one week) • 448 votes with help line. • 210 votes without help line.
- 15. 15 Difficulty Evaluation Experiment Voted difficulties are mainly EASY Voted difficulties become harder
- 16. 16 Difficulty Evaluation Experiment Not well adjusted in some maps (voted EASY) Voted difficulties become harder, but still not good
- 17. 17 Difficulty Evaluation Experiment Not well adjusted in most of maps Voted difficulties become harder, but still not good
- 18. 18 Difficulty Evaluation Experiment HL makes the maps easier than expected HARD difficulty has not been well adjusted
- 19. 19 Difficulty Evaluation Experiment Maps with HARD and MEDIUM difficulty show higher error. HL has a big impact on the difficulty perception. Blue EASY Red MEDIUM Green HARD EASY = 1, MEDIUM = 2, HARD = 3
- 20. 20 Conclusions • Players tend to vote mainly EASY. Most of times using the Help Line. • Players normally vote easier difficulty that they have felt shame (compromise their own ability). • Difficulty asignment must be ‘softered’ for MEDIUM and HARD maps. FUTURE WORK • The difficulty assignment method must be improved to better fit the players’ feeling.
- 21. Thanks! (astronomic) Questions? Contact info: ╺ carlillo_lopez@hotmail.com ╺ afdez@lcc.uma.es ╺ raul.lara@upm.es ╺ amorag@ugr.es ╺ pablo.garciasanchez@uca.es 21 Template from SlidesCarnival GravityVolve Game for Android https://play.google.com/store/apps/details?id=com.carlos.gravityvolve&hl=es