Automatic generation of tower defense levels using PCG

Tower defense is a popular subgenre of real-time strategy game requiring detailed level design and difficulty balancing to create an enjoyable player experience. Because level production and testing are both time-consuming and labor-intensive, we propose and implement a framework to automate the process. We first analyze the three main components, or "building blocks", of the popular tower defense game Kingdom Rush: Frontiers (KRF), i.e. road maps, tower locations and monster sequences. We then automatically create new building blocks in the style of the original game, utilizing techniques from Procedural Content Generation (PCG), and assemble them to create new levels. We also add a fourth block: automated testing via a Monte Carlo search algorithm, to ensure the generated content is playable. We focus on KRF because it is a popular video game in the tower defense genre, and highlights some of the challenges of designing appropriate PCG and playtesting algorithms for a commercial video game.

[1]  Simon M. Lucas,et al.  A Survey of Monte Carlo Tree Search Methods , 2012, IEEE Transactions on Computational Intelligence and AI in Games.

[2]  Matthew Guzdial,et al.  Explainable PCGML via Game Design Patterns , 2018, AIIDE Workshops.

[3]  Simon M. Lucas,et al.  Evolving mario levels in the latent space of a deep convolutional generative adversarial network , 2018, GECCO.

[4]  Julian Togelius,et al.  Artificial Intelligence and Games , 2018, Springer International Publishing.

[5]  Julian Togelius,et al.  Sentient Sketchbook: Computer-aided game level authoring , 2013, FDG.

[6]  John E. R. Staddon,et al.  The dynamics of behavior: Review of Sutton and Barto: Reinforcement Learning : An Introduction (2 nd ed.) , 2020 .

[7]  Rhio Sutoyo,et al.  Dynamic Difficulty Adjustment in Tower Defence , 2015 .

[8]  Michael Mateas,et al.  Tanagra: Reactive Planning and Constraint Solving for Mixed-Initiative Level Design , 2011, IEEE Transactions on Computational Intelligence and AI in Games.

[9]  Adam M. Smith,et al.  Addressing the fundamental tension of PCGML with discriminative learning , 2018, FDG.

[10]  Stefan Freyr Gudmundsson,et al.  Human-Like Playtesting with Deep Learning , 2018, 2018 IEEE Conference on Computational Intelligence and Games (CIG).

[11]  Julian Togelius,et al.  Evolving maps and decks for ticket to ride , 2018, FDG.

[12]  Julian Togelius,et al.  Evolving Playable Content for Cut the Rope through a Simulation-Based Approach , 2013, AIIDE.

[13]  Rayner Alfred,et al.  Automated Evaluation for AI Controllers in Tower Defense Game Using Genetic Algorithm , 2013 .

[14]  John Kolen,et al.  Exploring Gameplay With AI Agents , 2018, AIIDE.

[15]  Julian Togelius,et al.  Procedural Content Generation via Machine Learning (PCGML) , 2017, IEEE Transactions on Games.

[16]  Daniele Loiacono,et al.  TrackGen: An interactive track generator for TORCS and Speed-Dreams , 2015, Appl. Soft Comput..