Automatic level Generation for Tower Defense Games
Yu Du, Jian Li, Xiao Hou, Hengtong Lu, Simon Cheng Liu, Guo Xianghao, Kehan Yang, Tang Qinting · 2019 IEEE 3rd Information Technology, Networking, Electronic and Automation Control Conference (ITNEC) · 2019
We created a new method to automatically generate levels for a commercial-grade tower defense gameKingdom Rush:Frontiers (KRF) by means of Procedural Content Generation(PCG). Our research focuses on path generation and monster sequence generation. Firstly, we designed a mathematical representation of the game's path using geometric rules, and then we used search algorithms to develop an algorithm, which can generate new paths that are similar to the paths in original games. We implemented monster sequence generation with genetic algorithm in which we designed a representation of genes that reused some human-designed elements. In addition, we also designed a fitness function to make sure the generated level has moderate difficulty and playability. Finally, we automatically generated new levels for KRF with the combine of path generation and monster sequence generation. We used a turing test to show that our PCG levels are hard to be distinguished from the original levels.