Solving Large-Scale Pursuit-Evasion Games Using Pre-trained Strategies

Authors: Shuxin Li, Xinrun Wang, Youzhi Zhang, Wanqi Xue, Jakub Černý, Bo An

AAAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The empirical evaluation shows that our approach significantly outperforms the baselines in terms of speed and scalability, and can solve even games on street maps of megalopolises with tens of thousands of crossroads a scale beyond the effective reach of previous methods. and 5 Empirical Evaluation To evaluate the effectiveness of our two-stage algorithm, we compare it against the vanilla PSRO and NSGZero (Xue, An, and Yeo 2022), which is the state-of-the-art algorithm for solving pursuit-evasion games, in games with different sizes of graphs3. Furthermore, we perform several ablation studies to identify the effects of each component of our algorithm we introduced.
Researcher Affiliation Academia 1 School of Computer Science and Engineering, Nanyang Technological University, Singapore 2 Centre for Artificial Intelligence and Robotics, Hong Kong Institute of Science & Innovation, Chinese Academy of Sciences
Pseudocode Yes Algorithm 1: Pre-train Policy Model and Algorithm 2: PSRO with Pre-trained Policy Model
Open Source Code No The paper does not provide any explicit statements or links indicating that the source code for the methodology is openly available.
Open Datasets Yes For the synthetic graphs, we generate different sizes of grid worlds, and for the real-world graph, we extract the graph from the Singapore map via OSMnx (Boeing 2017).
Dataset Splits No The paper does not explicitly provide specific dataset split information (e.g., percentages, sample counts for training, validation, or testing).
Hardware Specification Yes Experiments are performed on a workstation with a ten-core 3.3GHz Intel i9-9820X CPU and NVIDIA RTX 2080 Ti GPU.
Software Dependencies No The paper mentions algorithms like PPO and MAPPO, and tools like OSMnx, but does not specify version numbers for any software dependencies.
Experiment Setup Yes The values of all parameters can be found in the Appendix.