🤖 AI Summary
Existing benchmark platforms for job shop scheduling (JSP), flow shop scheduling (FSP), flexible job shop scheduling (FJSP), and their variants—such as assembly constraints, sequence-dependent setup times, and online job arrivals—lack standardization, hindering fair and reproducible algorithm evaluation.
Method: This paper introduces the first open-source machine scheduling benchmark library, systematically unifying six major scheduling problem classes. It provides a standardized instance suite (>10,000 instances), reproducible experimental environments, and Gym-compatible interfaces supporting offline/online and deterministic/stochastic settings. Implemented in PyTorch/NumPy, it integrates OR-Tools, instance generators, and performance evaluation toolchains to accommodate both learning-based and classical optimization methods.
Contribution/Results: The library enables rigorous, reproducible comparison of over 20 baseline algorithms, significantly improving evaluation fairness and experimental transparency. It has already been adopted by multiple top-tier conference studies.
📝 Abstract
We introduce an open-source GitHub repository containing comprehensive benchmarks for a wide range of machine scheduling problems, including Job Shop Scheduling (JSP), Flow Shop Scheduling (FSP), Flexible Job Shop Scheduling (FJSP), FJSP with Assembly constraints (FAJSP), FJSP with Sequence-Dependent Setup Times (FJSP-SDST), and the online FJSP (with online job arrivals). Our primary goal is to provide a centralized hub for researchers, practitioners, and enthusiasts interested in tackling machine scheduling challenges.
