Generating varied, stable and solvable levels for angry birds style physics games

This paper presents a procedural level generation algorithm for physics-based puzzle games similar to Angry Birds. The proposed algorithm is capable of creating varied, stable and solvable levels consisting of multiple self-contained structures placed throughout a 2D area. The work presented in this paper builds and improves upon a previous level generation algorithm, enhancing it in several ways. The structures created are evaluated based on a updated fitness function which considers several key structural aspects, including both robustness and variety. The results of this analysis in turn affects the generation of future structures. Additional improvements such as determining bird types, increased structure diversity, terrain variation, difficulty estimation using agent performance, stability and solvability verification, and intelligent material selection, advance the previous level generator significantly. Experiments were conducted on the levels generated by our updated algorithm in order to evaluate both its optimisation potential and expressivity. The results show that the proposed method can generate a wide range of 2D levels that are both stable and solvable.